United States
Environmental Protection
Agency
Administration And
Resources Management
(PM-273)
21M-3004
June 1991
Indoor Air Quality And
Work Environment Study
Multivariate Statistical Analysis
Of Health, Comfort And
Odor Perceptions As Related To
Personal And Workplace
Characteristics
Volume 4
Printed on Recycled Paper
-------
Indoor Air Quality and Work Environment Study:
EPA Headquarters Buildings
Volume IV
MULTIVARIATE STATISTICAL ANALYSIS
OF HEALTH, COMFORT, AND ODOR PERCEPTIONS
AS RELATED TO
PERSONAL AND WORKPLACE CHARACTERISTICS
Atmospheric Research and Exposure Assessment Laboratory
US Environmental Protection Agency
Research Triangle Park, North Carolina 27711
JUNE 1991
-------
CONTENTS
CONTENTS 11
TABLES v
FIGURES v11
VOLUME IV REPORTING AND ANALYSIS TEAM v1i1
TECHNICAL TEAMS FOR VOLS. I-III 1x
ACKNOWLEDGEMENTS x
EXECUTIVE SUMMARY ES-1
Objectives ES-1
Statistical Methods ES-2
Results ES-5
Discussion ES-9
Conclusions ES-14
Recommendations ES-14
1. INTRODUCTION 1
1.1 Objectives 1
1.2 Description of Buildings 3
1.3 Questionnaire Design 4
2. DATA DESCRIPTION 5
3. DATA PREPARATION 7
3.1 Health, Comfort, and Odor Variables 7
3.2 Personal, Workstation, and Location Variables 8
3.3 Ventilation, Spatial and New Carpet Variables 9
it
-------
4. METHODS OF ANALYSIS 14
4.1 Principal Components Analysis 14
4.2 Linear Regressions 14
4.3 Logistic Regressions 16
5. RESULTS 17
5.1 Principal Components Analysis 17
5.2 Linear Regression 18
5.3 Logistic Regressions 18
5.4 Comparison of Linear and Logistic Results 20
6. DISCUSSION 26
6.1 Limitations of This Study 26
6.2 Relationships Among the Health Symptoms 28
6.3 Agreement Among the Regression Analyses 29
6.4 Variables Associated with Multiple Health Factors 29
6.5 Variables Associated with Two or Three Health Factors .... 36
6.6 Variables Associated with a Single Health Factor 38
6.7 Possible Gender-Related Variables 38
6.8 Variables Associated with Multiple Comfort/Odor Factors ... 40
6.9 Variables Associated with One or Two Comfort/Odor Factors . . 41
6.10 Effect of Location 41
6.11 Effect of New Carpet 44
6.12 Variables with Little Effect on Health, Comfort or Odor ... 45
6.13 Comparison with Other Studies 47
iii
-------
7. CONCLUSIONS 54
7.1 Methodology 54
7.2 Health Symptom Frequencies 55
7.3 Health and Comfort Effects 55
8. RECOMMENDATIONS 59
8.1 Improving Working Conditions 59
8.2 Implications for Future Research 61
9. REFERENCES 64
APPENDICES
A. EPA STUDY QUESTIONNAIRE A-l
B. IDENTIFICATION OF AIR HANDLING UNITS FOR EPA BUILDINGS B-l
C. METHODS AND RESULTS OF PRINCIPAL COMPONENTS ANALYSES OF
HEALTH, COMFORT, ODOR, AND PSYCHOSOCIAL QUESTIONS C-l
D. NUMBER OF MEN AND WOMEN AT EACH LEVEL OF THE HEALTH, COMFORT,
AND ODOR INDICES AND ASSOCIATED BINARY VARIABLE EMPLOYED
IN THE LINEAR AND LOGISTIC REGRESSIONS D-l
E. RESULTS OF LOGISTIC REGRESSIONS E-l
F. RESULTS OF LOGISTIC REGRESSIONS USING ALTERNATIVE DEFINITION
OF POSITIVE RESPONSE F-l
iv
-------
TABLES
ES-1. DEPENDENT VARIABLES IDENTIFIED BY PRINCIPAL COMPONENTS ANALYSIS
AND USED IN ALL REGRESSIONS ES-19
ES-2. VARIABLES SHOWING SIGNIFICANT ASSOCIATIONS WITH AT LEAST FOUR
OF THE 12 HEALTH FACTORS ES-20
ES-3. VARIABLES SHOWING SIGNIFICANT ASSOCIATIONS WITH AT LEAST THREE
OF THE 10 COMFORT AND ODOR FACTORS ES-21
1. FREQUENCIES BY GENDER FOR HEALTH SYMPTOMS EXPERIENCED LAST YEAR ... 69
2. PERSONAL AND WORKSPACE CHARACTERISTICS, BY GENDER 72
3. INDEPENDENT VARIABLE DEFINITIONS AND SOURCE OF INFORMATION 74
4. QUESTIONS THAT WERE NOT INCLUDED IN THE ANALYSIS 78
5. AIR HANDLING UNITS AT WATERSIDE MALL, AREAS SERVED,
AND ESTIMATED FLOW RATES IN CUBIC FEET PER MINUTE (CFM) 81
6. ESTIMATED SUPPLY AIR FLOW RATES AND NUMBER OF OCCUPANTS SERVED
BY AIR HANDLING UNITS ON EACH FLOOR OF THE TOWERS, FAIRCHILD,
AND CRYSTAL MALL 82
7. PRINCIPAL COMPONENTS ANALYSIS OF HEALTH SYMPTOMS 83
8. PRINCIPAL COMPONENTS ANALYSIS OF COMFORT CONCERNS 84
9. PRINCIPAL COMPONENTS ANALYSIS OF ODORS 85
10. PRINCIPAL COMPONENTS ANALYSIS OF PSYCHOSOCIAL VARIABLES 86
11. NUMBER OF RESPONDENTS ASSIGNED TO EACH CELL IN THE BINARY
VARIABLES CREATED FOR THE LOGISTIC ANALYSIS 87
12-33. RESULTS OF LINEAR AND LOGISTIC REGRESSIONS FOR 12 HEALTH
SYMPTOMS AND 10 COMFORT AND ODOR VARIABLES 88
12. HEADACHE AND NAUSEA 90
13. STUFFY NOSE/SINUS CONGESTION, RUNNY NOSE, SNEEZING, COUGH 91
14. WHEEZING, SHORTNESS OF BREATH, CHEST TIGHTNESS 92
15. DRY, ITCHING OR TEARING EYES, SORE BURNING EYES, BLURRY VISION ... 93
16. SORE THROAT, HOARSENESS, DRY THROAT 94
17- SLEEPINESS, UNUSUAL FATIGUE 95
-------
18. CHILLS AND FEVER 96
19. ACHING MUSCLES OR JOINTS, PAIN IN BACK, SHOULDER, NECK, WRISTS ... 97
20. DIFFICULTY CONCENTRATING, REMEMBERING; DEPRESSION, TENSION 98
21. DIZZINESS 99
22. DRY OR ITCHY SKIN 100
23. PROBLEMS WITH CONTACT LENSES 101
24. HOT STUFFY AIR 102
25. DRY AIR 103
26. COLD DRAFTY AIR 104
27. HUMID AIR 105
28. ODOR OF PAINT, CARPET CLEANING, OTHER CHEMICALS 106
29. ODOR OF COSMETICS, FOOD, BODY ODOR 107
30. ODOR FROM PHOTOCOPIES, PRINTING 108
31. ODOR FROM NEW CARPET OR DRAPES 109
32. MUSTY/DAMP ODOR, DIESEL EXHAUST 110
33. TOBACCO SMOKE ODOR Ill
34. VARIABLES APPEARING AT THE P < 0.01 SIGNIFICANCE LEVEL FOR
EACH OF THE 12 HEALTH FACTORS 112
35. VARIABLES ASSOCIATED WITH AT LEAST FOUR OF 12 HEALTH FACTORS ... 113
36. VARIABLES ASSOCIATED WITH ONE TO THREE HEALTH FACTORS 114
37. VARIABLES APPEARING AT THE P < 0.01 SIGNIFICANCE LEVEL FOR
EACH OF THE 10 COMFORT AND ODOR FACTORS 115
38. VARIABLES ASSOCIATED WITH AT LEAST THREE COMFORT/ODOR FACTORS ... 116
39. VARIABLES ASSOCIATED WITH ONE OR TWO COMFORT/ODOR FACTORS 117
40. VARIABLES ASSOCIATED WITH NO HEALTH, COMFORT, OR ODOR FACTORS ... 118
41. SYMPTOM PREVALENCE IN MANHATTAN FAMILIES
COMPARED TO EPA EMPLOYEES 119
vi
-------
42. COMPARISON OF BRITISH, DANISH, NETHERLANDS,
AND US BUILDING STUDIES 120
43. FINDINGS OF THREE EUROPEAN BUILDING STUDIES 121
FIGURES
1.ESTIMATED PER CAPITA SUPPLY AIR FLOW RATES (CFM/PERSON)
OF AIR HANDLING UNITS AT EPA HEADQUARTERS BUILDINGS 124
2.RELATIONSHIP BETWEEN THREE SETS OF LINEAR REGRESSIONS
AND TWO SETS OF LOGISTIC REGRESSIONS 125
vii
-------
VOLUME IV REPORTING AND ANALYSIS TEAM
US Environmental Protection Agency
Lance Wallace
C.J. Nelson
Mel Kollander
Pierce Foundation
Brian Leaderer
University of Maryland College of Medicine
Rebecca Bascom
Research Triangle Institute
George Dunteman
VOLUME IV PEER REVIEWERS
Steven Colome, Ph.D.
Integrated Environmental Services
Irvine, CA
Claudia Miller, M.D., M.S.
Univ. of Texas Health Sciences Center
San Antonio, TX
Peder Skov, M.D.
Clinic of Occupational Medicine
Copenhagen, Denmark
viii
-------
TECHNICAL TEAM FOR VOLS. I-III
This study of indoor air quality and work environment was conducted by
three technical teams representing multiple organizations. It was jointly
developed and carried out at EPA headquarters and the Library of Congress
Madison Building under the auspices of these teams working independently of
both management and unions at both EPA and the Library of Congress.
Overall project coordination was provided by two technical team leaders:
Kevin Teichman at EPA and Lawrence Fine at NIOSH.
EPA
NIOSH
REPORTING AND ANALYSIS TEAM
C. J. Nelson, Statistician
Mel Kollander, Survey Stat
Lance Wallace, Envir. Sci.
Brian Leaderer, Envir. Sci.a
Rebecca Bascom, M.D.e
Andy Clayton, Statistician*
George Dunteman, Statistician
Anne Fidler, Epidemiologist
Joseph Hurrel, Psychologist
Brian Leaderer, Envir. Scientist8
John Selfridge, Envir. Scientist3
MONITORING TEAM
Ross Highsmith, Chemist
Lance Wallace, Envir. Sci.
Tom Lumpkin, Chemist
Steve Hern, Biologist
Vinson Thompson, Chemist
Ken McLauchlan, Prof. Engineer
Linda Stetzenbach, Microbiologist0
Richard Gorman, Industrial Hygienist
Michael Crandall, Industrial Hygien.
Rebecca Stanevich, Ind. Hygienist
Brian Leaderer, Envir. Scientist8
John Selfridge, Envir. Scientist8
Mel Kollander, Survey Stat.
Lance Wallace, Envir. Sci.
Cecil Brenner, Stat.
Robert P. Clickner, Stat.d
Stephen K. Dietz, Stat.
SURVEY DESIGN TEAM
Anne Fidler, Epidemiologist
Thomas Wilcox, Physician
Joseph Hurrel1, Psychologist
Richard Hornung, Statistician
Brian Leaderer, Envir. Scientist8
John Selfridge, Envir. Scientist8
b
c
d
e
f
John B. Pierce Foundation, Yale University.
Kansas State University.
Consultant.
University of Nevada, Las Vegas.
Westat.
University of Maryland, Baltimore.
Research Triangle Institute.
Dr. Selfridge is now at
ix
-------
ACKNOWLEDGEMENTS
The work reported in this volume could not have been done without the
contributions of the large number of persons and institutions listed in the
first three volumes of this series.
Additional contributions to Volume IV were made by Michael Shamus of
EPA, who devoted many hours to tracing the maze of air handling units in the
Waterside Mall building; Robert Clickner of Westat, who provided data files on
locations of respondents, number of persons in rooms, and other essential
data; and EPA unions and management, who provided data on newly carpeted areas
in the Headquarters buildings. Andrew Clayton of RTI influenced the choice of
methods for the spatial analysis.
We are appreciative of the many thoughtful comments of the peer
reviewers. Jeff Davidson of the EPA Health and Safety Division provided not
only comments but also relevant journal articles. Helpful comments were also
received from Karen Hogan of EPA and from William Hirzy of the National
Federation of Federal Employees. Reviews were also provided by scientists and
engineers at EPA's Atmospheric Research and Exposure Assessment Laboratory:
Ross Highsmith, William C. Nelson, Andrew Lindstrom, and Matthew Somerville.
-------
EXECUTIVE SUMMARY
This volume presents the results of a multivariate statistical analysis
of a questionnaire administered in February-March of 1989 to 3955 EPA
employees working in three buildings in the Washington, DC area. Three
earlier volumes in this series have been released. Volume I describes the
development of the questionnaire and presents the descriptive results, but
without interpretation or statistical analysis. Volumes II and III deal with
a component of the study involving indoor air quality measurements and a
supplementary questionnaire administered to a subset of employees working near
the monitoring locations. This report presents a detailed statistical
analysis of the employee survey questionnaire described in Volume I, together
with additional data collected on ventilation and carpet installation; it does
not deal with the supplementary questionnaire or the environmental monitoring
results.
OBJECTIVES
The main goal of this analysis was to identify the personal or workplace
characteristics associated with employees' health and comfort. If such
characteristics can be identified, steps can be taken to provide better
working conditions and possibly improve health and productivity.
Personal characteristics such as age, sex, job category, and pay grade
may affect health. Also, "psychosocial" factors such as stress can impact
health. Therefore, a set of 25 personal characteristics, including seven
psychosocial variables, was included in the analysis.
A large number of workplace characteristics could affect health and
comfort, including ventilation, temperature, humidity, dust, noise, crowding,
maintenance, etc. A total of 29 workplace characteristics were included.
Finally, there was concern that some areas within buildings might have
higher symptom rates than other areas. Therefore, several additional analyses
were conducted to determine whether any spatial variation not explained by
other factors was occurring. This analysis included an additional set of 66
variables describing the locations served by different air handling units, and
five variables describing areas where the new carpet had been installed.
ES-1
-------
STATISTICAL METHODS
Multiple regression analysis was used to investigate the effects of
personal and workplace factors on reported health, comfort, and odor concerns
occurring during the year prior to administering the questionnaire. Multiple
regression can be used to analyze the effects of a large number of factors
simultaneously, and determine which of these factors have the strongest
association with the health effect or other variable of interest.
Most of the data used in the regressions were derived from the
questionnaire, which contained 68 questions and required about 30 minutes to
complete. The complete questionnaire is reproduced in Appendix A. Some
additional data were collected from EPA payroll and telephone directory files,
engineering records on air handling units, and management records on the
installation of new carpet during the period 1987-88.
To reduce the large number of questions on health symptoms, comfort
concerns, and odors to a manageable number of factors, a principal components
analysis (PCA) was first carried out. This type of analysis can be used to
combine separate variables into a single factor without losing too much
information. The PCA reduced the 32 health symptoms to 12 factors.
In most cases, separate symptoms dealing with a single body organ or system
formed a single factor. For example, four questions on eye irritation (dry
eyes, sore eyes, burning eyes, and blurry vision) formed one factor. In a
similar fashion, 10 questions on indoor air quality were reduced to four
factors, and 15 questions on odors were reduced to six factors.
These 22 health, comfort, and odor factors, defined in Table ES-1,
became the dependent variables used in all regressions. To avoid repeating
all the symptoms contained in a factor, we refer to the factor by a short
phrase such as "chest problems," which stands for the three health symptoms
"wheezing," "shortness of breath," and "chest tightness."
The independent variables included the 25 personal and 29 workplace
factors that were identified from the first questionnaire. Additional
independent variables included new carpet and ventilation. Because of the
reactions of many employees to the installation of new carpet in 1987-88,
offices and corridors with new carpet were identified and included as a
possible explanatory variable. Two measures of spatial variation were
ES-2
-------
employed: a "coarse-grained" location variable consisting of the eight
building "sectors" defined in Volume I (the East and West Towers, the second
and third floors of Waterside Mall, its Northeast and Southeast extensions,
Crystal Mall and the Fairchild building); and a "fine-grained" variable
consisting of 66 distinct areas within the three buildings. These areas were
characterized by the unique air handling unit (AHU) serving the area, or if an
AHU served areas on different floors, by the combination of the floor and AHU.
Three sets of linear regressions were then run on each of the 22 health,
comfort, and odor factors. The first set was designed to test for differences
between the three buildings, or between the eight sectors. The second set of
linear regressions tested for differences among the 66 areas. The third set
of regressions tested for an effect of the estimated supply airflow associated
with each AHU.
All three sets of regressions also included all of the other personal
and workplace characteristics. Some characteristics were significantly
associated with health symptoms in all three of the regressions, indicating
that no matter what measure of spatial variation or building ventilation was
used, these characteristics remained important.
Following the three sets of linear regressions, a final set of logistic
regressions was carried out. Logistic regressions are often used to determine
the increased risk associated with each factor of significance.
All the regressions were run separately for men and women to minimize
the gender-related interactions among certain variables that could otherwise
complicate the interpretation of the regressions. For example, women
employees had less formal education, received less pay, had fewer private
offices, and reported more health symptoms than men. Analyzing the genders
separately eliminates these gender-related interactions and makes
interpretation of the results clearer. However, one drawback of the separate
analyses is that the relative risks for men and women cannot be calculated.
All regressions on health symptoms were run both with and without the
comfort and odor factors as independent variables. Health could be affected
by indoor air quality (e.g., headaches arising from hot stuffy air), but both
could be affected by other variables (e.g., both the headache and the hot
stuffy air might be caused by an inoperative AHU). In many cases, the same
ES-3
-------
personal or workplace characteristics remained significant whether or not the
comfort and odor variables were included.
These decisions (separate regressions by sex; including and excluding
the comfort and odor variables on the 12 health factors) led to a total of 68
separate regressions within each of the three sets of linear regressions and
one set of logistic regressions. Since each regression included between 20
and 136 personal and workplace characteristics, some thousands of associations
were investigated. When such a large number of tests are made, some
associations will appear significant even though they are due to chance alone.
To reduce the number of these "false positives," we chose to consider
associations significant only if there was no more than one chance in 100 (p <
0.01) that the association was due to chance. However, we have published in
Appendix E of this report all the associations tested in the final logistic
regressions. Thus readers may determine for themselves what associations
appeared at any level of significance.
Since the three sets of linear regressions ("coarse-grained," "fine-
grained," and flow rate-based) gave similar results, only the set including
the "fine-grained" spatial variables was selected for comparison with the
logistic regressions. These comparisons may be found in Tables 12-33 in the
text; they are summarized in Tables 34 and 37 in the text.
ES-4
-------
RESULTS
Health
Of more than 100 variables tested, only 11 were associated with multiple
(more than three) health factors (Table ES-2). These 11 variables included
two workplace characteristics (dust and glare); three personal characteristics
(sensitivity to chemical fumes, allergy to molds, lack of college education);
two psychosocial factors (workload and conflicting job demands); two comfort
factors (hot stuffy air and dry air); and two odor factors (the odor of fresh
paint, carpet cleaning, and other chemicals; and the odor of cosmetics, body
odor, and food).
Workplace Characteristics
The workplace variable that appeared most often in relation to
significantly increased frequency of health symptoms was a dusty workspace.
Dust was often associated with increased frequency of nasal congestion, throat
symptoms, dizziness, and dry skin. Other associated symptoms included
headache, chest problems, eye irritation, and fatigue.
Glare was associated with headache, eye irritation, and (particularly
strongly--p<0.0001) with muscle aches and upper body pain.
Personal Characteristics
Persons reporting sensitivity to chemical fumes were more likely to
report headache, eye irritation, throat symptoms, fatigue, difficulty
concentrating, and dizziness. Other symptoms sometimes associated with
sensitivity to chemical fumes were nasal congestion, chest symptoms, and
bodily aches and pains. In total, sensitivity to chemical fumes was
significantly related to health factors more often than any other variable: 24
of 48 regressions for women and 22 of 48 for men.
Persons reporting mold allergies had more frequent eye irritation and
bodily aches and pains. Nose and throat symptoms were sometimes increased, as
were headaches and fatigue (men only) and dry skin (women only).
Persons (especially women) without a college education were more likely
to report chest problems, chills and fever, headaches, dizziness, and fatigue.
ES-5
-------
Psvchosocial Factors
Persons with a heavy workload reported increased frequency of headaches,
eye irritation, and difficulty concentrating.
Persons reporting conflicting work duties or demands were more likely to
experience symptoms such as difficulty concentrating. Other symptoms included
nasal congestion, chest symptoms, dizziness, chills and fever, and muscular
pain.
Comfort Factors
Persons reporting hot stuffy air reported more symptoms of headache,
stuffy nose, chest tightness, eye irritation, difficulty concentrating,
fatigue, and dizziness.
Persons reporting dry air reported more eye, nose, and throat problems,
and dry skin.
Odor Factors
Persons reporting the odor of fresh paint, carpet cleaning, and other
chemicals were more likely to report headache, throat problems, chills and
fever, and dizziness. Women reporting this odor were more likely to report
nasal congestion, chest problems, fatigue, and difficulty concentrating.
Women reporting the odor of cosmetics, body odor, and food smells were
more likely to report chills and fever, bodily aches and pains, and difficulty
concentrating. Men reporting this odor were more likely to report eye
irritation.
Comfort and Odor
Four regressions (linear/logistic; male/female) were run on each of the
10 comfort and odor factors. By far the most frequently appearing variable
was dust, appearing at the 0.01 level of significance in 20 of 20 regressions
for men and 15 of 20 for women. Dust was associated with all four of the
comfort conditions (hot, cold, dry, or humid air) and all six of the odor
factors (Table ES-3). Four other workplace variables--noise, glare, nearby
water leaks, and the use of a fan--were associated with a smaller number of
comfort and odor factors.
Two personal characteristics--sensitivity to chemical fumes and
conflicting job demands--were also associated with multiple comfort and odor
factors. Sensitivity to chemical fumes was associated particularly strongly
ES-6
-------
with odors of paint and other chemicals, printing and photocopying, and
tobacco smoke. It was also associated with the odor of new carpet, but among
women only. Persons reporting conflicting Job demands reported being bothered
by printing and photocopying odors, cosmetics and body odor, and damp, musty
odors.
Ventilation
Several efforts were made to relate ventilation characteristics to
health and comfort. Supply airflow rates were measured for about 30 of 50
AHUs. (Supply air is the total air supplied to building occupants by the
ventilation system, consisting of the sum of the fresh outdoor air and
recirculated air.) Concurrent temperature measurements of the supply air,
outdoor air, and recirculated air allowed the outdoor airflow rates to be
estimated for these 30 AHUs. However, the missing data on outdoor airflow
rates at the other 20 AHUs made it impossible to test for the effect of fresh
airflow on health and comfort symptoms.
As a backup calculation, the measured supply airflow rates for most of
the AHUs were supplemented with the estimated supply airflow rates for the
remaining AHUS. These values were calculated either from previously measured
supply airflow rates or from the manufacturer's rated capacity for the supply
air fans. The estimated values were then divided by the number of people
served by each AHU to provide an estimated per capita supply airflow rate.
This variable was then included in the third set of 68 linear regressions. No
effect was found on any health, comfort, or odor factor using these estimated
flows.
Spatial Variation
No effect on health was noted for any of the "coarse-grained" location
variables; that is, neither the building nor any of the large "sectors" within
a building (e.g., East or West Towers, Southeast Mall) had any consistent
association with either increased or decreased frequency of any of the 12
health factors investigated. Among the 66 "fine-grained" locations, only two
appeared to be related to increased frequency of some symptoms: the Library
area of the mall (AHU 21), associated (among men only) with flu-like symptoms
(chills, fever, muscle/joint aches and pains); and the part of the 2800
ES-7
-------
corridor of Waterside Mall served by AHU 16, associated (again among men only)
with complaints of eye irritation.
However, certain comfort concerns and odors were affected by particular
locations. The Southwest Mall reported significantly higher frequency of
odors of paint and other chemicals, engine exhaust and musty/damp odors.
Tobacco smoke odors were reported on the second and third floors of the West
Tower. Reports of cold drafty air were elevated among people served by AHU 14
on the second floor of Waterside Mall, and reports of dry air were elevated
among people served by AHU 16 on that floor. Reports of humid air were
elevated in the Library area (served by AHU 21) of Waterside Mall. Workers in
the garage and lower garage areas (AHU G&LG) reported being bothered by odors
of photocopying and printing processes; this area includes the EPA print shop.
Carpet
Persons in areas where the new carpet had been installed about a year
before the survey reported increased frequency of throat symptoms (sore
throat, dry throat, hoarseness). An effect on dizziness was also noted, but
only for men in some regressions.
ES-8
-------
DISCUSSION
Methodology
This is the first large-scale building study to employ an objective
measure (PCA) to assess the way in which health symptoms cluster together.
Previous studies have subjectively grouped symptoms into rather large
clusters, such as a "mucous membrane" category involving eye, nose, and throat
symptoms. The more objective PCA employed in this study indicated that each
of the 12 health factors identified in this study had a somewhat different set
of associated personal and workplace characteristics (see Tables 12-23 in the
text). If any of the 12 health factors had been further combined, some of the
associations may not have been observed, although associations common to the
combined factors would still be observable.
Thus an important feature of the study has been the objective
identification of health symptom clusters. These findings should be tested in
future studies to determine whether these clusters of symptoms have broad
validity.
Another important feature of this study was the fact that it attempted a
census (rather than a sample) of all EPA employees at the three buildings.
Even though the census was not complete, the 80% coverage allowed an
investigation of spatial variation within the buildings, including the effects
of ventilation and carpet installation. Although some preceding studies, such
as the Danish Town Hall Study (Skov et al.. 1989) attempted a census, this is
the first study to attempt an extensive spatial analysis within buildings.
Symptom Prevalence
We have been unable to locate baseline statistics on the prevalence of
acute symptoms in a working population in non-problem buildings. The closest
substitute we have found is several studies of acute symptoms in healthy urban
populations. Frequencies of nine symptoms among the EPA employees appeared
elevated by factors of 3-12 over the rates reported in these studies.
Although we recognize that these comparisons are not exact (the earlier study
populations included children and other persons not in the work force; our
results were for a single week in winter, when rates of respiratory disease
are typically twice as high as in spring and summer), we conclude, as did the
ES-9
-------
authors of the earlier studies, that the burden of minor acute symptoms is
high. For example, about a quarter of a million headaches per year could be
calculated among the 5000 EPA employees. Headaches were identified by the
respondents as the main cause of time lost from work. Allowing a few minutes
of lost productivity per headache results in extensive productivity losses due
to this single symptom.
Workplace Characteristics
Dust was identified as the characteristic contributing most powerfully
to a wide variety of health, comfort, and odor concerns. Dust has previously
been identified as a proximate cause of health problems in offices or schools.
The primary cause, presently unknown, could be physical irritation, allergens
or endotoxins. The Danish Town Hall Study found that measured dust and
surrogates for dust (fabric-covered surfaces and shelf space) were among the
most highly correlated variables with reported health symptoms (Skov et al..
1990; Gravesen et al.. 1990). Another study found that intensive cleaning of
carpets and wet dusting reduced health symptoms in an office building for at
least the following two months (Norback and Torgen, 1989).
The problem of dust may be more extreme at Waterside Mall than at the
other two EPA buildings. A substantially higher percentage of people reported
dusty conditions at the Waterside Mall building (24% compared to 16% and 12%
at the other two buildings). Also, a much higher percentage of employees
at Waterside Mall than at the other two buildings brought up problems of
maintenance (including lack of dusting or vacuuming, presence of vermin, etc.)
as the first item they mentioned on the essay question (EPA, 1989, p. 5-43).
Dust was also the single strongest variable affecting comfort and odor
factors. Since odors may be carried by particles, or by gases associated with
particle emissions, as in tobacco smoke, printing processes, painting, etc.,
the observed direct relationships of odors with dust may have a physical
explanation.
Glare was the other workplace variable affecting a number of health
factors. The influence of glare on headache and eye irritation is
understandable. However, the strongest effect of glare was on muscle pain,
including neck and shoulder pain. This may be explained if people working
ES-10
-------
with computers and word processors adopt and maintain uncomfortable positions
to minimize the effects of glare.
Noise was associated with several comfort and odor factors, including
particularly cold drafty air.
Water leaks near the respondents' office area were associated with
increased reports of damp and musty odors. Water leaks were also associated
with reports of unusual fatigue and difficulty concentrating (men only).
These findings support the possibility that molds are implicated in some
health symptoms, consistent with the finding that allergies to molds were
broadly associated with a number of symptoms.
Although new carpet showed few relationships with health symptoms, it is
of interest due to the widespread reports of health problems following
installation in 1987-88. People in areas where the new carpet was installed
reported higher frequencies of throat symptoms (sore throat, dry throat,
hoarseness). Since the period of recall was one year, which extended back to
the time of installation, the question of whether the respondents were still
experiencing these symptoms, or whether they had simply experienced them
earlier and were now recovered, could not be resolved. (Analysis of the "last
week" question could resolve this point.) Nonetheless, this finding provides
some support for the hypothesis that carpet installation can result in these
symptoms.
The finding that spatial location had no discernible effect on health
does not mean that there were no differences in frequency of health symptoms
among buildings or locations within buildings, but only that such differences
could be largely explained by other, possibly more directly related,
variables. For example, since dusty offices were more prevalent in the
Waterside Mall building than in the other buildings, the frequency of those
health symptoms significantly related to dust would be expected to be higher
at Waterside.
The finding that ventilation capacity and AHUs also had no discernible
effect on health symptoms should also be interpreted with caution. It is
likely that other variables, more directly related to health problems, have
captured whatever variability was associated with the AHUs. For example, hot
stuffy air was associated with a number of health symptoms. Hot stuffy air is
ES-11
-------
likely to have been associated with certain AHUs as well, although this
analysis has not been carried out.
Personal Characteristics
The variable with the most widespread influence on health was
sensitivity to chemical fumes. The questions used to define this factor
included a list of possible sources of irritation (e.g., tobacco smoke, fumes
from new carpets, paint, etc.) and a question on whether the respondent
considered herself or himself "especially sensitive" to any of these chemical
mixtures. About a third of the respondents (29-32%) at each building answered
"Yes." Women were more likely to report sensitivity to chemical fumes than
men (37% vs. 25%).
The hypothesis that Waterside Mall employees would be more likely to
report sensitivity to chemical fumes, either because they had been physically
sensitized by the carpet installation or because they were influenced by the
extreme publicity and/or by fellow workers' complaints, was not confirmed
(Waterside Mall 32%, Crystal Mall 32%, and Fairchild 29%).
Sensitivity to chemical fumes was also significantly related to reports
of odors, particularly odors of paint and other chemicals, tobacco smoke, and
photocopier and printing processes. This is consistent with anecdotal reports
of severe reactions to paint, cigarette smoke, and other odorous products.
Several psychosocial variables were associated with health symptoms, and
also with comfort and odor concerns. The most commonly occurring psychosocial
variable was a measure of conflicting job duties or demands on employees by
their supervisors or peers. This variable was associated with difficulty
concentrating, bodily aches and pains, chills, and nasal and chest symptoms,
as well as many odors. Other important psychosocial variables were workload,
associated with eye irritation and difficulty concentrating, and job
dissatisfaction, associated with fatigue and difficulty concentrating. Since
all of the three most important psychosocial factors affected only about half
of the health symptoms, it appears that the bulk of the reported health
problems at the EPA buildings cannot be attributed to psychological factors
alone.
The importance of mold allergies for a number of health symptoms
indicates that molds may play a part in the reported health symptoms at the
ES-12
-------
EPA buildings. Molds and dustiness may both be results of inadequate
maintenance.
Comfort and odor factors
Reports of dry air and hot, stuffy air were often associated with health
symptoms. The reports of dry air were supported by the inspections of the
AHUs, which found no AHU with an operable humidifier. The measured indoor
relative humidities averaged about 23%, which is below recommended standards.
However, these measurements were made during the winter and may not reflect
conditions at other times.
With respect to the reports of stuffy air, however, little evidence was
found of inadequate ventilation. The design capacity of the supply air fans
was calculated both per capita and per unit area, and was found to be adequate
for most areas in Waterside Mall. The main exception was the print shop at
Waterside Mall--but this area has had a new AHU installed since the monitoring
period. However, it is recognized that these calculations are incapable of
determining the actual delivery of fresh outdoor air to individual employees
at all locations.
The odor factor having the most significant associations with health
symptoms was the factor comprised of the odors of paint, carpet cleaning,
pesticides, and other chemicals such as glues and cleansers. Some of these
odors are associated with maintenance activities or with renovation.
ES-13
-------
CONCLUSIONS
Analysis of the 32 health symptoms indicated that they clustered into 12
factors, largely by single body systems such as eye, nose, throat, chest, and
central nervous system. This finding, if confirmed, would be useful in
designing future questionnaires.
Symptom levels appeared to be elevated at all buildings compared to
findings from previous studies of acute symptoms in healthy urban populations.
However, statistics on symptom frequencies in adult working populations more
comparable to the EPA respondents could not be located.
Major variables affecting more than three health factors included two
workplace characteristics (dust and glare) and several personal
characteristics (sensitivity to chemical fumes, allergy to molds, workload,
and conflicting job demands). Comfort and odor factors associated with
multiple health symptoms included hot stuffy air, dry air, and the odor of
fresh paint and other chemicals.
RECOMMENDATIONS
The importance of dust, mold allergies, the odor of paints and other
chemicals, and nearby water leaks in affecting multiple health symptoms and
comfort/odor concerns points to building maintenance or renovation as a
crucial factor in the problems at the EPA buildings, particularly the
Waterside Mall building. One recent study indicates that intensive carpet
cleaning and wet dusting can have a beneficial effect on employees' health.
Another suggests that removal of carpeting improved the health of students in
two Swedish schools. Therefore it is recommended that attention be given to
ways of improving building maintenance, particularly to reduce dust, clutter,
and conditions conducive to growth of mold. The design of new buildings
should stress adequate building maintenance and reduced likelihood of dust
buildup. This should include consideration of the relative advantages of
carpeting and fabric-covered partitions vs. other more easily cleaned
surfaces, the use of gratings or walk-off mats to reduce track-in of dust and
dirt, and techniques to allow for easy replacement or cleaning and drying of
water-soaked carpet. The complexity of the issue of building design and
ES-14
-------
maintenance is recognized. For example, carpets may be cleaned with chemicals
whose residue has been shown to cause health problems, or they may be soaked
during steam cleaning to the extent of encouraging mold growth. On the other
hand, noncarpeted surfaces such as tile or linoleum may allow unacceptable
noise and can be slippery following cleaning.
Some employees stated in their questionnaire responses that they had to
keep unused files in their offices as a historical record. This suggests that
an adequate archiving policy and provision of external storage space could
reduce the level of dust and clutter. We are aware that EPA management has
taken steps recently (Operation Cleanup) to reduce clutter, and we recommend
continued vigilance.
Since many sources of dust are associated with essential office
processes such as filing, photocopying, and the like, providing supplemental
air cleaning and filtration may also be considered. Since this would also
require supplemental ventilation, or at least recirculation, the same system
could reduce employees' complaints of hot stuffy air. The system should
include both charcoal filters for volatile organic compounds (VOCs) and high-
efficiency filters to reduce fine particle accumulation. However, if persons
work in proximity to common indoor sources (e.g., computers, printers) careful
attention to the design of the airflow patterns would be required. Whether
these systems work as advertised is not clear; adequate testing of systems
should be carried out. In the case of major indoor sources such as copiers,
separate ventilation systems are recommended.
A second factor of broad importance in affecting health was glare. With
the increased use of computers, it appears that reducing glare has the
potential of reducing complaints of headache, eye irritation, and neck and
shoulder pain. Providing glare screens for all computers may be helpful, as
would providing information to employees on the importance of reducing glare.
An important finding of this study has been that headache is the single most
common cause of absenteeism and lost work time (EPA, 1989). Thus if reducing
glare resulted in reducing headache prevalence, employee absenteeism and lost
work time may also decrease. Inspections of workstations to improve ergonomic
design could reduce the complaints noted in this study regarding uncomfortable
workstations. Such improvement of employee working conditions could also be a
cost-effective means of increasing productivity.
ES-15
-------
Neither the environmental measurements nor the results of these analyses
support a recommendation to replace the present air handling systems with more
powerful systems. (Possible exceptions are the units in the garage and lower
garage areas, the Library area, and the Southwest Mall, which were all
associated with comfort or odor problems. We are aware that the garage and
Library units have been replaced.) However, the findings of widespread
health problems associated with complaints of hot stuffy air do support a
recommendation that supplemental ventilation be supplied, particularly to
locations where people are reporting problems. This additional ventilation
could include additional outdoor air, or it could consist of supplemental
recirculated air, preferably including filtration capable of removing both
particles and VOCs. Supplying additional fresh air alone might not deal with
the problem of dust buildup due to internal sources.
This study suggests that carpet installation was associated with
increased symptoms of throat problems. It is not known, however, which aspect
of the installation process may be implicated. One possibility is emissions
of organic chemicals, either from the carpet, the carpet adhesive, or, if
molding is involved, the molding itself or the molding adhesive. Another
possibility is the raising of dust, molds, allergens, or endotoxins while
ripping up old carpet or molding. This suggests that carpets be installed
without employees present and that some sort of minimum waiting period ensue
before they reenter their offices to allow dust to settle and organic chemical
concentrations to decline. However, this study provides no guidance on the
optimum length of the minimum waiting period. If the carpets can be aired out
in the warehouses for a week or two before installation, the initial high rate
of organic emissions should be greatly reduced. (We understand that EPA
management has in fact adopted this procedure in recent carpet installations.)
Sensitivity to paint, carpet, adhesives, tobacco smoke, automobile
exhaust, and other chemical mixtures was associated with more health symptoms
than any other variable. It was also one of the strongest factors associated
with headache, the major cause of time lost from work identified in the
questionnaire responses. Therefore, steps to reduce the likelihood of these
chemical fumes or vapors from reaching employees may improve productivity.
(This is the case regardless of whether such chemical exposures have a
physical or psychological effect.) Such steps include a no-smoking policy
ES-16
-------
that is adequately enforced (this was a common complaint in some essays,
particularly in the Fairchild building); steps to seal off the underground
garage from the work areas, through enforcement of the fire door policy and
perhaps through some work on the elevator shafts and stairways; and reduction
of activities such as painting, cleaning, and other maintenance or renovation
work during working hours. Pollution prevention through the use of low-
emitting adhesives, alternative cleansers, and other nontoxic products would
reduce the indoor sources of VOCs. Since sensitivity to chemical fumes is so
little understood, we recognize that further research is required. However,
prudent steps to reduce chemical exposures, and to communicate with employees
so they can recognize that effective actions are being taken, seem justified.
Among the psychosocial variables, the most important seemed to be
conflicting job demands. This suggests that particular attention be paid to
establishing clear areas of responsibility and lines of authority to reduce
the likelihood for multiple, possibly contradictory, demands to be placed on
employees. Other important psychosocial variables included workload and job
dissatisfaction. However, job satisfaction was generally high for the
employees as a whole, and many employees stated that although their workload
was high, they were committed to their jobs.
In terms of future research, the questionnaire and data base collected
in this study will form a major resource. The identification of clusters of
health symptoms should be validated on different populations. The
questionnaire should be revised according to the experience in this study and
retested on small populations. Research on methods of collecting settled dust
should be undertaken to allow this component to be added to future monitoring
studies. Additional analysis of the most important variables (dust, hot
stuffy air) could be undertaken to pinpoint those areas of the buildings with
problems. An analysis of existing data to identify organizational units with
psychosocial problems could be fruitful, provided problems with
confidentiality can be overcome.
In summary, the findings that dust, water leaks, mold allergies, and the
odor of fresh paint and chemicals are associated with health and comfort
factors suggest that improved building operation and maintenance procedures
are needed. The finding that glare is associated with increased incidence of
headache, the leading cause of employee absenteeism and loss of work time, as
ES-17
-------
well as with bodily aches and pains, suggests that reduction of glare could
improve employee productivity. The finding that hot, stuffy air and dry air
are also associated with many health problems suggests that increased effort
in providing stable temperature regimes, improved airflow, and relative
humidities in the approved range would also be desirable. The finding that
sensitivity to a variety of chemical mixtures is associated with many health
symptoms indicates that reduction of chemical exposures may be indicated
through enforcement of smoking policies; controlled use of paints, glues, and
cleansers during non-working hours; airing out carpets before installation and
providing a minimum waiting period before reentry; and sealing off pathways
for auto exhaust to reach the workspace. For maximum effectiveness, any such
actions should be clearly communicated to employees. The finding that several
psychosocial variables (particularly conflicting job demands) contribute to
acute symptoms suggests that increased attention should be paid to defining
job duties and responsibilities to reduce the stresses due to conflicting job
demands.
ES-18
-------
Table ES-1. DEPENDENT VARIABLES IDENTIFIED BY PRINCIPAL COMPONENTS ANALYSIS
AND USED IN ALL REGRESSIONS
HEALTH SYMPTOMS
1. headache; nausea
2. stuffy nose/sinus congestion; runny nose; sneezing; cough
3. wheezing; shortness of breath; chest tightness
4. dry, itching, tearing eyes; sore eyes; burning eyes; blurry vision
5. sore throat; hoarseness; dry throat
6. sleepiness; unusual fatigue
7. chills; fever
8. aching muscles; back pain; shoulder/neck pain; hand/wrist pain
9. difficulty concentrating; difficulty remembering; tension; depression
10. dizziness
11. dry itchy skin
12. problems with contact lenses
COMFORT CONCERNS
1. air too hot; air too stuffy; too little air movement
2. air too dry
3. air too cold; too much air movement
4. air too humid
ODORS
1. fresh paint; carpet cleaning; pesticides; chemicals
2. cosmetics; food; body odor
3. photocopier; printing process
4. new carpet; new drapes
5. musty/damp odor; fishy smells; engine exhaust
6. tobacco smoke
ES-19
-------
Table ES-2. VARIABLES ASSOCIATED8 WITH AT LEAST FOUR OF THE 12 HEALTH FACTORS
Variable N_ Health Factors Significantly Associated
Workplace Characteristics
Dust 11 headache; nasal, chest, eye, throat symptoms;
fatigue; chills; difficulty concentrating1*;
dizziness; dry skin; contact lens problems
Glare 5 headache; eye symptoms; fatigue; difficulty
concentrating; paine
Personal Characteristics
Sensitivity to 9 headache; nasal, chest, eye, throat symptoms;
Chemicals fatigue; pain; difficulty concentrating; dizziness
Mold Allergies 7 headache; nasal, eye, throat symptoms; fatigue; pain;
dry skin
No College 5 headache, chest symptoms, fatigue, chills and fever,
dizziness
Psychosocial Characteristics
Conflicting 6 nasal, chest symptoms; chills and fever, pain,
Demands difficulty concentrating, dizziness
Workload 5 headache, eye symptoms, pain, difficulty
concentrating, dizziness
Comfort and Odor Characteristics'*
Odor of Paint, 8 headache; nasal, chest, throat symptoms; fatigue;
Chemicals chills; difficulty concentrating; dizziness
Hot Stuffy Air 7 headache; nasal, eye, chest symptoms; fatigue;
difficulty concentrating; dizziness
Dry Air 4 nasal, eye, throat symptoms; dry skin
Odor of Cosmetics 4 eye symptoms, chills, pain, difficulty concentrating
8 Significant at p < 0.01 in at least two (of eight) regressions
b Includes difficulty concentrating/remembering, depression, tension
c Includes aching muscles, back pain, shoulder/neck pain, hand/wrist pain
d Significant at p < 0.01 in at least two (of four) regressions
ES-20
-------
Table ES-3. VARIABLES SHOWING SIGNIFICANT* ASSOCIATIONS WITH AT LEAST THREE OF
THE 10 COMFORT AND ODOR FACTORS
Variable N Odor and Comfort Factors Significantly Associated
Workplace Characteristics
Dust 10 all four comfort concerns and all six odor factors
Glare 5 hot air, dry air, odors of paint, cosmetics, dampness
Noise 5 hot air, dry air, cold air, odors of paint, cosmetics
Use Fan 4 hot air, humid air, cold air [-], odor of cosmetics
Water Leaks 3 odors of dampness, cosmetics, tobacco smoke
Personal Characteristics
Conflicting 6 dry air, cold air, odors of cosmetics, photocopying,
Demands dampness, tobacco smoke
Chemical 5 humid air, odors of paint, photocopying, new carpet,
Sensitivity tobacco smoke
8 Significant at p < 0.01 in at least two (of four) regressions
ES-21
-------
1. INTRODUCTION
In March of 1989, a study of indoor air quality was carried out in three
buildings occupied by the US Environmental Protection Agency at Washington,
DC. A questionnaire was distributed to all 4,900 employees. The
questionnaire contained 68 questions on health and comfort concerns, workplace
conditions, and personal characteristics of the respondents. Nearly 4000
employees answered the questionnaire, for an 80% response rate. A few weeks
later, environmental measurements were taken in the buildings, and a
supplemental questionnaire was administered to a subset of about 500 employees
who worked in the areas monitored.
Three EPA reports have been published on this study. Volume I (EPA,
1989) describes the survey results. Volume II (EPA, 1990a) summarizes the
results of the environmental monitoring. Volume III (EPA, 1990b) analyzes the
relationships between the environmental monitoring results and the responses
to the supplemental questionnaire.
This report is a statistical analysis of the responses to the first
questionnaire. It covers most of the questionnaire material, including most
of the health symptoms, comfort concerns, and personal and workplace
characteristics. It includes none of the environmental measurements nor any
of the responses to the supplementary questionnaire.
1.1 Objectives
The fundamental goal of the analysis was to determine relationships
between employee-reported health symptoms and comfort concerns and personal
and workplace characteristics. If such relationships can be established, it
may be possible to improve the relevant working conditions and thereby reduce
employee concerns, sickness and absenteeism. Improved working conditions may
also result in increased productivity.
Several previous large-scale studies of office buildings (Skov and
Valbjorn, 1987; Skov et ai., 1989, 1990; Wilson and Hedge, 1987; Zweers et
al., 1990) have tentatively identified some personal and workplace
characteristics that may be associated with health symptoms. For example, in
all of these studies, women reported higher symptom rates than men, although
1
-------
the root cause remains unknown. However, no other variable out of the many
variables tested in these studies was consistently identified in all studies
as being related to health or comfort. Also, none of the previous studies
examined the spatial variation in health symptoms within a single building;
such variation might be caused by differences in ventilation or areas with new
carpet, water damage, or other possible localized sources of problems.
Therefore, the present study was designed to test the widest possible
number of health symptoms, comfort concerns, odors, workplace characteristics,
personal factors, and spatial variables that might affect employees. A
detailed questionnaire was developed that took approximately 30-45 minutes to
complete. Additional data were collected on building ventilation
characteristics, employee density, and locations where new carpet had been
installed.
The study was designed to test a large number of hypotheses. Some of
the major objectives were as follows:
1) Determine which health symptoms are most closely correlated with other
health symptoms. Identifying symptoms that "cluster" together, perhaps
because of the same or similar causes, would allow a smaller number of
health symptoms to be included on future questionnaires.
2) Determine which workplace characteristics affect health and comfort.
This would indicate which improvements in working conditions should be
targeted as having the greatest likelihood for improving employee well-
being. Some of these workplace characteristics were of great interest
to employees and management, including ventilation and new carpets:
2a) Determine the effect of ventilation on health and comfort.
Complaints of hot, stuffy, and dry air have been common, but the
effect of these ventilation conditions on health is unknown.
2b) Determine which symptoms may be related to the installation of new
carpet. The carpet installation of 1987-88 was associated with an
outbreak of symptoms and employee concerns. The statistical
analysis would help determine which symptoms, if any. were
elevated among employees in the newly carpeted areas.
3) Determine which personal and psychological characteristics affect health
and comfort. This objective could identify areas such as job
satisfaction or workload that could be marked for improvement.
-------
Most of the information used was obtained from the questionnaires. Thus
the personal and workplace characteristics, like the health symptoms and
comfort concerns, are subjective; no independent confirmation of these
characteristics has been made, with the exception of the newly carpeted areas,
for which management records were used.
To avoid awkwardness, we use terms like "dust" or "glare" as shorthand
for the more precise "office was reported dusty during last year" or "reported
being bothered by glare." We also use the terms "sensitivity to chemical
fumes" or "sensitivity to chemicals" to refer to a specific set of questions
on our questionnaire; these terms should not be taken to refer to the syndrome
known as "Multiple Chemical Sensitivity,* nor should their use be taken to
indicate acceptance of the cause-effect relationship implied. Readers with
questions regarding the precise meaning of a variable may look up the exact
wording of the question in the original questionnaire (Appendix A) .
1.2 Description of Buildings
The main EPA Headquarters building, with about 4000 employees, is
located in Southwest Washington, DC and is known as Waterside Mall. The
remaining two buildings, each with about 500 employees, are known as Crystal
Mall and the Fairchild Building. Crystal Mall is located in Crystal City,
VA., just across the Potomac River from Washington. Fairchild is located a
few blocks from Waterside Mall, on South Capitol St.
Waterside Mall was originally constructed in 1970; EPA took occupancy in
1971. Two 12-story buildings (the Bast and West Towers) are connected by a
central three-story mall. The ground floor of the mall is commercial space
housing restaurants and shops. Although the second floor of the mall was also
designed as commercial space, with 10 to 12-foot ceilings finished to 8-feet,
it was transformed into office space soon after the building was built. The
third floor of the mall and the towers were designed for standard offices.
Over the years, additions have been made to the building. The Southeast Mall
was built in the late 1970s, and the west side of the Mall, consisting of
three floors and a basement, was built in 1971. The Northeast Mall, five
floors and two basement floors, was occupied by EPA in 1984. Thus the
Waterside Mall complex comprises at least three separate buildings (the two
-------
Towers and the Mall), with several additional architecturally distinct areas.
Crystal Mall is a 12-story office building constructed about 1970 in
Crystal City, VA. BPA leases five floors housing about 500 workers. The
Fairchild Building is an eight-story structure that was built in the late
1970s. EPA has leased five floors since 1980, again for about 500 employees.
For the last few years, including the period of recall on the
questionnaire, all EPA Headquarters buildings have had a smoking policy, with
smoking restricted to outdoors and designated rest rooms.
1.3 Questionnaire Design
A new questionnaire was designed for this study by research scientists
from EPA, from the National Institute of Occupational Safety and Health
(NIOSH). and from Yale University/Pierce Foundation. The group included
environmental scientists, statisticians, survey design specialists,
physicians, epidemiologists, industrial hygienists, and specialists in
measuring environmental chemicals, bioaerosols, and building-related
parameters such as air exchange rates. Existing questionnaires were reviewed,
including questionnaires for medical conditions, psychosocial variables, and
previous indoor air quality investigations. When possible, standardized
questions from these questionnaires were used. In some cases, new questions
were added. The questionnaire went through a number of drafts over a period
of several months, and was pretested twice before the final version was
completed. The resulting questionnaire (Appendix A) includes 68 questions
(some with multiple parts) addressing health, comfort concerns, odors, and
personal and workplace characteristics. For a detailed discussion of the
questionnaire design process, see Volume I (EPA 1989).
Other reports on the questionnaire design and results from both the BPA
buildings and the associated study in the Madison Building, Library of
Congress (LOG), are contained in Fidler fit fil. (1990), Leaderer fit al. (1990),
Nelson fit ai.. (1990) , Hall fit al. (1990) , and Hurrell fit al. (1990) . Reports
on monitoring results in the EPA and LOG buildings are contained in Crandall
fit Al- (1990), Persily and Dols (1990), and Selfridge fit ai. (1990).
-------
2. DATA DESCRIPTION
An early decision was made to have separate statistical models for men
and women, based on consistent findings from previous studies that women
report symptoms at a higher rate than men. Accordingly, the symptom
frequencies are presented separately for men and women for the 32 health
symptoms (Table 1).
About 3900 responses were available, divided almost evenly among men
(1900) and women (2000). The symptoms that occurred "often" or "always" in
the previous year at the highest rates were sinus congestion (females 36%,
males 26%); headache (28%, 13%); fatigue (27%, 13%); dry skin (27%, 16%); and
eye irritation (26%, 15%).
Personal and workplace characteristics showed differences among male and
female respondents (Table 2). Women were younger and less well educated than
men: 24% of women were under 30 compared to only 8% of men; more than 90% of
the men had college degrees, but only half of the women did. Pay grades
followed suit, with three-quarters of the men in the highest pay grade (GS 13
and above), but only a third of the women. More women than men smoked
tobacco--18% compared to 12%.
Medical conditions showed similar rates among men and women--rates for
asthma, eczema, and allergies to dust, molds, animals, and pollen were within
a few percent for both sexes. The reported asthma rates of 9% for men and
women more than doubled the nationwide rates of 3% for men and 4% for women in
1988 (NCHS, 1989). About 37% of women considered themselves unusually
sensitive to certain chemical fumes or vapors, as compared to 25% of men.
Seven measures of stress were evaluated. Men and women scored similarly
on most measures, including workload, job satisfaction, and conflicting job
demands. However, women were about twice as likely as men to report external
(not job-related) stress factors, such as caring for a family member. Nearly
twice as many women as men lived alone (25% vs. 15%) while more than twice as
many men as women (26% vs. 12%) described themselves as living with others and
being the sole provider.
About four of five men had an enclosed office, but only half of the
women did. About 40% of women shared their workspace with two or more
-------
persons, compared to 14% of men. More men reported a window in or near their
workspace (49% vs. 38% of women).
Women worked more hours at computers than men: 27% of the women spent
more than four hours a day on the computer but only 15% of the men. Women
also reported being bothered by glare more than men: 53% vs. 42%. Women also
worked more often with chemical-containing office products such as adhesives
and whiteout: 24% worked one or more hours per day with such chemical
mixtures compared to only 11% of men. Three times as many women as men (15%
vs. 5%) reported working more than one hour at a copy machine.
Men and women had equal levels of complaint about noise, with more than
half of each finding their work environment too noisy. Women complained of
dusty working conditions more frequently than men (56% compared to 40%) .
-------
3. DATA PREPARATION
The variables used in the analysis are listed in Table 3, together with
the identification of the questions from which they were constructed. Some of
the questions on the questionnaire were not used in the analysis. These
questions, and the reasons for dropping them from the analysis, are shown in
Table 4.
Other questions were either accepted as is, or were combined or
otherwise receded in some way. Coding of many binary questions followed
standard conventions ("No" - 0, "Yes" - 1). Decisions on coding the remaining
questions are explained below.
3.1 Health, Comfort, and Odor Variables
The questions on health symptoms had two parts, one dealing with how
often the symptom occurred during the previous year ("never," "rarely,"
"sometimes," "often," or "always") and the other with how many days it
occurred last week. It was decided to analyze the "last year" question but
not the "last week" question, since the latter might not be representative of
typical conditions. Another reason was that many of the other questions on
the questionnaire covered lengthy periods. For example, people were asked if
new carpet, new furniture, etc., had been installed within the past 6 months
or one year. These questions might have to be omitted in analyzing the "last
week" question. Finally, analyzing the "last week" question would have
doubled the number of computer analyses, which was not feasible economically.
Nonetheless, we recognize that there are certain advantages to analyzing the
"last week" question (responses are apt to be more accurate; the "number of
days" response lends itself better to quantitative analysis). Also, running
the "last week" analyses could serve as a check on the stability of the
conclusions reached from analyzing the "last year" questions.
A final question was "Does the symptom usually change when not at work?"
with three possible responses: "gets worse; stays same; gets better."
Responses of "gets better" would normally be considered to be work-related.
However, some building-related symptoms may get worse after leaving work
(e.g., muscle and back pains), others may stay the same for long periods of
-------
time (e.g., dry skin), and some non-building-related symptoms may improve
after leaving work because of natural causes (e.g., a headache that begins in
the morning before coming to work and gets better after leaving work). All of
these situations would lead to misclassification. Also, since this question
was placed immediately after the "last week" question, the possibility exists
that some respondents interpreted the question as referring to last week
rather than last year, as intended. For these reasons, this question was not
used in this analysis.
Some previous studies have analyzed only symptoms that were thought to
be work-related, using questions similar to those discussed above to separate
work-related from non-work-related symptoms. Other studies, like this one,
have analyzed all symptoms without regard for whether they are work-related.
An interesting contribution to future studies of this sort would be to repeat
the analysis on only the work-related symptoms. If the two analyses are
fundamentally similar, identifying work-related symptoms may be unnecessary.
Comfort questions were also in two parts, "last year" and "last week",
each with identical 5-point scales ranging from "never" to "always." Odor
questions dealt only with odors noticed at the workstation during the last
year, using the same five-point scale.
3.2 Personal and Workstation Characteristics
Initially, two age variables were used: age and (age)2, to allow for
nonlinear effects. After determining that these variables were collinear,
they were replaced in all later analyses by a three-way categorization: <30,
30-39, 40+. These categories have been used by other investigators (Wilson
and Hedge 1987, Skov and Valbjorn 1987), who found some effects related to
age.
Three questions on the comfort of the chair or desk at the workstation
and the adjustability of the chair were combined into one discomfort index.
Three pay grade categories were created: < GS9, GS9-GS12, > GS12.
The original 3-level question on smoking (none, light, heavy) was
receded into a binary variable (smokers - 0, nonsmokers - 1) because of the
small size of the smoking group (12% of men, 18% of women).
-------
Three types of workspace were considered, based on answers to Question
I.la: enclosed offices, cubicles with half-height partitions, and open areas.
Three types of space sharing were also considered, based on responses to
Question I.lb: single occupant, sharing with one other person, and sharing
with two or more persons. These variables were kept separate for the initial
multiple least-squares regressions, but for the final logistic regressions,
they were combined. This was done because it was felt that an occupant of a
private office might have different characteristics from an occupant of a
cubicle, although both would respond that they did not share their workstation
with anyone else. Thus the three types of space were crossed with the three
types of space sharing to form nine "cells." These nine cells were then
combined into three categories. The first category consisted of persons in a
private office. The second category was made up of two cells: persons who
shared an enclosed office with just one other person, and persons in private
cubicles with half-height partitions. The last category consisted of the
remaining six cells: persons sharing an office with two or more workers (one
cell), persons sharing cubicles with one or more others (two cells), and
persons in open areas (three cells). Two binary variables were created from
the three categories.
Question 1.7 on bookcases, desks, and other items in the workstation was
recoded into several variables. One was intended to approximate a "shelf
space" variable, of the type developed in the Danish Town Hall study (Skov and
Valbjom, 1987), and hypothesized to be a measure of the total dust-catching
area in the workplace. This variable was the sum of the positive responses to
questions on the presence of a desk, bookcases, file cabinets, and other
furniture. Another variable, intended to test for possible effects of
chemicals emitted by wood products (e.g. formaldehyde or terpenes such as a-
pinene), was the sum of positive responses to questions on wood desks,
bookcases, and furniture.
3.3 Ventilation, Spatial, and New Carpet Variables
Data on air handling units (AHUs), the locations they serve, and the
areas with new carpet of the type associated with EPA employee health and
comfort concerns were acquired and added to the data base for analysis. This
-------
allowed us to test for the possibility that the ventilation system or
installation of new carpet could affect health, comfort, or odor perceptions.
3.3.1 Ventilation
The most desirable information on ventilation would be actual long-term
continuous measurement of outdoor airflow rates. This information could then
be combined with employee densities to calculate a per capita outdoor airflow
rate (in cubic feet per minute per person, or cfmp), which could be compared
directly to the recommended flow rates contained in ventilation standards such
as Standard 62-89 of the American Society of Heating, Refrigeration, and Air
Conditioning Engineers (ASHRAE).
However, this ideal situation was not possible in this study. A year-
long monitoring effort, preceded by an extended study design, with appropriate
financial resources, was mounted to provide this information for the
associated Library of Congress study, but there was no time to develop the
same program for the EPA buildings. Nonetheless, some useful information was
collected on the ventilation system at the EPA buildings.
Approximately 50 large AHUs bringing in outside air were identified in
the Waterside Mall building (Table 5). Additional units were identified in
the Crystal Mall and Fairchild buildings, but because information was lacking
about the location of the respondents' workplaces in the specific areas served
by each unit, the AHUs in these two buildings were treated as a single unit in
each building. An additional 50 or so small air conditioners were identified
in the Waterside Mall building; however, since the completeness of this data
base was questionable, and since these units did not affect the air exchange
rate, they were not considered further. Appendix B provides a detailed
discussion of the identification process and sources of information for
ventilation data.
For the following discussion, several terms need to be defined. "Supply
air" is the air supplied to building inhabitants through the ventilation
ducts. It includes both outdoor air and recirculated air. "Design flow rate"
is the flow rate, in cubic feet per minute (cfm), that the supply air fans are
theoretically capable of maintaining.
10
-------
The supply airflow rate was estimated for each AHU. For about 30 of the
AHUs, the actual supply airflow rate was measured during the monitoring period
in March of 1989. The outdoor airflow rate for each of these AHUs was
calculated by measuring temperature differences between supply air, outdoor
air, and recirculated air. These measured outdoor airflow rates could have
been used to calculate an average per capita outdoor airflow rate for these
locations only, but the results would refer only to the short period when the
measurement occurred. Also, these values would have applied to only about
half the EPA employees. For these reasons, the outdoor airflow rates were not
used in the statistical analysis.
For two additional AHUs in the East Tower, a 1986 "air balancing" report
provided a measured supply airflow rate. For the remaining 18 or so AHUs,
only the design flow rates were available. Good agreement (generally within
15%) was noted between the design and measured supply airflow rates of most of
the AHUs, both in the 1986 and 1989 measurements. Therefore it is expected
that the use of the design flow rates rather than measured flow rates for
these remaining AHUs did not introduce major errors.
By summing up either the measured or design flow rates within a given
building or sector of a building, and dividing by the area, a supply airflow
rate per unit area (cfm/ft2) was estimated (Appendix B). The resulting values
could then be compared to a typical design value of one cfm/ft2.
To obtain an estimated per capita supply airflow rate (cfm per person) ,
the computerized EPA directory for the period of the survey (February-March
1989) was employed to determine the number of persons in each office served by
a particular AHU. This supply airflow rate is not to be confused with the
outdoor airflow rate referred to in ASHRAE Standard 62-89.
The AHUs in the East and West Towers of Waterside, and in the Crystal
and Fairchild buildings, serve portions of all floors in each building. Thus,
the per capita supply airflow rate of these AHUs could be estimated by floor
(Table 6). The frequency distribution of estimated per capita supply airflow
rates for the EPA employees is shown in Figure 1.
The above information was used to determine locations with relatively
high or low per capita supply airflow rates. It was also used to determine if
any AHUs were associated with health symptoms or comfort concerns, and if such
associations were a function of the supply airflow rate. (For example, if
11
-------
some air handling systems were actually sources of dust, mold, or other
allergens, symptoms might increase with supply airflow rate.)
3.3.2 Spatial Variables
A question of interest is not only whether some entire buildings have
health or comfort problems that are significantly different from other
buildings, but also whether areas within buildings might also show significant
differences. Volumes I-III employed an a priori categorization into eight
"sectors" based on architecture and time of construction. These "sectors"
consisted of the Crystal Mall building, the Fairchild Building, the East and
West Towers of Waterside Mall, and four portions of the central mall: the
second floor and third floor, and the Northeast Mall and Southeast Mall. (At
times, the small Southwest Mall was considered to be a separate sector; at
other times, this was considered part of the second floor of the Mall.) In
this analysis, we added a ninth "sector": the garage and lower garage areas
of Waterside Mall. These nine sectors, most having several hundred employees,
formed the "coarse-grained" location variables that were tested for
differences in employee health, comfort, and odor concerns.
However, we also wished to test for the possibility that smaller spatial
units might display significantly different symptom prevalence rates. Since
the AHUs tile the work areas in the central mall, and since most of them serve
20 or more respondents of each sex, they were selected as the "fine-grained"
spatial variables. (One exception is the eight small AHUs [A-H] on the second
floor of the mall; these were combined into two groups.) However, in the
Crystal and Fairchild buildings, and in the West and East Towers, the AHUs
serve vertical slices of each building. Employing the AHUs as a spatial unit
in these areas could make it impossible to detect problems occurring on a
single floor. Therefore the AHUs in these four areas were subdivided by
floor. Because the number of respondents in the West and East Towers was too
small to allow individual floors to be included, adjoining floors were grouped
into two- or three-floor units (four-floor units for the "peripheral" AHUs).
The result was a set of 66 variables that served both as a measure of
spatial variation and an indicator of the AHU associated with each space.
These variables could then be used to test the hypothesis that AHUs or other
12
-------
spatially varying factors could be associated with variation in health,
comfort, or odor responses.
3.3.3 New Carpet
In late 1987 and early 1988, about 27,000 yards of carpet (covering
roughly a third of the available floor space) were laid in two of the three
EPA buildings; only Crystal Mall received no new carpet. Health symptoms were
reported by about 100 employees shortly after installation. One hypothesis
was that offgassing from the carpet, the adhesive, or other operations
associated with the installation process could be involved. Although the
questionnaire was administered about a year after the installation, it was
thought desirable to test whether persons in the newly carpeted areas showed
any higher symptom prevalence or comfort concerns than other respondents.
Two types of carpet were employed, one for offices and one for
corridors. About 17,000 yards were glued down, and the remaining 10,000 were
tacked down. Reasoning that carpet in one's office would be more likely to
affect one than carpet in a nearby corridor, and that carpet adhesive could
have something to do with the outbreak, we created a total of five variables:
no new carpet; glued down carpet in corridors on the same floor as the
respondent; glued down carpet in the respondent's office; tacked down carpet
in same-floor corridors; and tacked down carpet in the office. The total
number of persons located in offices or on floors with the new carpet was
2034, (about half of all respondents). This is somewhat more than the
fraction of area taken up by the new carpet (about a third) but because of
including all persons on floors where the carpet was installed only in
corridors, a somewhat larger fraction would be expected.
13
-------
4. METHODS OF ANALYSIS
The basic hypothesis of this study is that health, comfort, and odors
may depend on parameters associated with the person, the building, or the
individual workstation. Thus a multivariate method of analysis was employed
to test all these possibilities simultaneously.
4.1 Principal Components Analysis
The first step of the multivariate analysis was to examine the
relationships among the many health, comfort, odor, and psychosocial questions
to reduce them, if possible, to a smaller number of factors that would account
for most of the observed variance. Principal components analyses (PCA) were
performed separately on these four groups of questions.
4.2 Linear Regressions
The health, comfort, and odor factors identified by PCA became the
dependent variables in all three sets of linear regressions. Independent
variables were selected from the personal and workplace characteristics,
location, airflow, and new carpet variables. Once the independent variables
had been selected for each set of linear regressions, regression diagnostics
were performed using the SAS routine COLLIN (SAS users' manual, 1989) to test
for collinearity and matrix singularities.
A core set of independent variables in all linear regressions was the 54
personal and workplace characteristics (except for some minor changes based on
results from the COLLIN analyses of each set of regressions). Other
independent variables on location, airflow, and new carpet areas were used in
one or two of the three sets of regressions according to the hypothesis being
tested.
Since our results and those of previous studies have shown considerable
differences among men and women in symptom prevalence, and since many
variables (e.g., pay grade, type of work, workstation environment) are related
to gender, separate regressions were run for each sex. This has the effect of
accounting for gender-related interactions without specifically including
interaction variables.
14
-------
Although perceptions of comfort and odor may be considered as dependent
variables, affected by personal or workplace characteristics, they may also
act as independent variables, having an effect on health. (It is also
possible that the relationship is reversed--that the state of one's health
affects one's perception of comfort.) Therefore, in order to test this
potential association of comfort and odor with health, an additional set of
regressions was run on the health factors only, again separately by sex, with
the comfort and odor variables augmenting the other independent variables.
4.2.1 First Set of Linear Regressions
The initial set of linear least-squares multiple regressions was aimed
at testing whether there were differences between buildings, or between
building sectors. Each regression included eight building sector variables
and 55 personal and workplace variables.
4.2.2 Second Set of Linear Regressions
The second set of linear regressions was designed primarily to test the
hypotheses that AHUs or recently carpeted areas might be associated with
increased prevalence of health, comfort, or odor complaints. Also, since the
AHUs corresponded, for the most part, to distinct contiguous areas, these
regressions were able to test whether individual locations (on a finer-grained
scale than the building sectors used in the preliminary set of regressions)
were associated with increased reports of health, comfort, or odor problems.
Thus the 66 AHU/floor variables replaced the eight building sector
variables, and four variables representing the carpeted areas were added. All
other variables remained in the regressions. Thus the second set of
regressions, besides testing these new hypotheses, provided a test of the
stability of the previous findings. If most of the previously identified
variables continued to be significant, it would be evidence that the models
are stable.
4.2.3 Third Set of Linear Regressions
To test whether the estimated supply airflow rates of the AHUs had any
effect on health, comfort, or odors, a third set of linear regressions was
carried out. These regressions replaced the 66 AHU/floor variables with a
15
-------
single continuous variable reflecting the estimated per capita supply airflow
rate in cubic feet per minute per person. (To test for nonlinear effects, the
square of the flow rate was also included as a variable.)
4.3 Logistic Regressions
A set of logistic regressions was carried out to estimate the magnitude
of the effects on health and comfort of personal and workplace
characteristics. For these logistic regressions, all of the health, comfort,
and odor factors were reduced to binary variables. That is, people were
classed into two groups: those who reported the factor sufficiently often
(according to some criterion) and those who did not. Then the relative risk
of falling into the "positive" group can be calculated for each personal or
workplace characteristic.
Since different criteria could be used to classify a symptom as
positive, a second set of logistic regressions was carried out to test the
effect of different classification criteria on the results.
Thus a total of five sets of regressions were carried out: three sets
of linear multiple regressions and two sets of logistic regressions. The
relationships between these regressions are shown in Figure 2.
16
-------
5. RESULTS
5.1 Principal Components Analysis
The first PGA analysis identified nine principal factors among the 31
health symptoms experienced over the past year (Table 7). Twenty-nine of the
31 symptoms could be assigned to one of these factors on the basis of a
loading greater than 0.5. The two remaining symptoms (dizziness and dry skin)
had low loadings on all nine factors. One symptom--problems with contact
lenses--applied only to a subset of respondents and therefore could not be
included in the PGA. Thus a total of 12 health factors were identified from
the original 32 health questions. The analysis was repeated for the "last
week" questions on health with essentially the same results.
In a similar manner, a total of four comfort factors were isolated from
the original 10 questions on comfort (Table 8), and six odor factors were
isolated from the original 15 questions on odors (Table 9).
The fourth PGA was performed on a set of 22 psychosocial questions (Part
IV of the questionnaire). These questions had been drawn from a larger set of
previously validated questions on job satisfaction, workload, conflicting job
demands, etc. (Hurrell et aJL.f 1990). Another set of questions was added on
such external factors as child care responsibilities, caring for an elderly or
disabled person, volunteer activities, etc., to provide a measure of external
stress. These questions arose from previous investigations by NIOSH but have
not been formally validated. The PCA analysis confirmed that the questions in
each of the seven psychosocial factors were closely related and represented
relatively independent (uncorrelated) factors (Table 10).
The methods and results of the principal components analyses are
described more fully in Appendix C.
An index was created for each of the 12 health, four comfort, and six
odor factors by summing the responses for each of the questions comprising the
factor. Responses of "never" or "rarely" were assigned a value of 0,
"sometimes" 1, "often" 2, and "always" 3. Thus each index extended from 0 to
values of 3, 6, 9, 12, or 15 depending on how many questions were combined to
form the factor. The number of males and females associated with each level
of the health, comfort, and odor indices is displayed in Appendix D.
17
-------
Females reported higher symptom frequencies on all 12 health factors.
Females also reported more discomfort on all four comfort conditions, finding
the air to be hotter, colder, drier, and more humid than men.
On the other hand, for four of the six odor factors, the complaint rates
were similar for men and women. Men were almost as likely as women to report
odors from copying machines, carpets, musty or damp smells, and tobacco smoke.
Women were more likely to report odors of new paint or other chemicals, and
cosmetics.
5.2 Linear Regressions
These 22 health, comfort, and odor indices were used as the dependent
variables for all three sets of linear regressions. An additional 12
regressions in each set were run on the health indices using the comfort and
odor indices as independent variables, as described above. Since all
regressions were run separately by gender, a total of (22 + 12) X 2 - 68
regressions were run in each of the three sets of linear regressions.
In the first set of linear regressions, a number of personal and
workplace characteristics were significantly associated with health, comfort,
or odor concerns. Most of these associations continued to be significant in
the second and third set of linear regressions, and also in the logistic
regressions. Therefore we have chosen to present the results from only one of
the three sets of linear regressions (the set employing the AHU/floor and new
carpet variables) together with the results from the first set of logistic
regressions.
5.3 Logistic Regressions
The 22 health, comfort, and odor factors were again employed as
dependent variables in the logistic regressions. Since these factors were
represented as multilevel indices in the linear regressions, they had to be
transformed to binary variables for the logistic regressions. Thus a choice
was necessary regarding what level of each index to consider the dividing line
between those who exhibited the symptom sufficiently often and those who did
not. For certain of the rarer symptoms, too stringent a definition would lead
to too small a number of persons with the symptom to allow effective
statistical analysis. Therefore, it was decided to "cut" each index at the
18
-------
point that would allow at least 4-5% of the male or female population to be
classified as exhibiting the given set of symptoms. Since men reported lower
symptom prevalence than women, the possibility of separate "outpoints" for men
and women was considered; however, it was decided to keep the same cutpoints
for both men and women. The number of men and women assigned to each level of
the 22 binary dependent variables is shown in Table 11.
Since the cuts are made on an index that includes four possible levels
for each answer, it is not possible to unambiguously interpret a given index
level as representing, say, a "sometimes" or "often" response. For example,
on a four-question index, a score of "4" could be made by answering "always"
to one question, "sometimes" to another, and "rarely or never" to the
remaining two. However, the same score could be achieved by answering
"sometimes" to all four questions, or "often" to just two questions. Since
previous studies have used a different system to judge the intensity of the
response, (i.e., requiring at least an "often" or "always" response to at
least one symptom among the multiple-symptom complex) , we reran a set of
logistic regressions for the "eye irritation" factor using this selection
criterion.
The complete results of the logistic regressions, including the odds
ratios and their 99% confidence limits, are provided in Appendices E and F.
(The 68 tables of Appendix E refer to the 68 logistic regressions using the
"index cutpoint" criterion described above; the four tables in Appendix F
correspond to the four regressions on the eye irritation factor using the
"often or always" criterion.)
In general, the magnitudes of the odds ratios for variables
significantly associated with health outcomes were less than 2. That is, few
variables were strong enough to cause a doubling of the risk of any health
outcome. Exceptions included doctor-diagnosed asthma, associated with a 3- to
4-fold increase in the risk of having chest-related problems; eczema, which
increased dry skin reports by factors of 2.5-5; and wearing contact lenses,
which increased the risk of reporting eye irritation by factors of 2.2-2.3.
19
-------
5.4 Comparison of Linear and Logistic Results
The second set of linear regressions included 44 regressions with 126
variables, and 24 regressions with 136 variables. Thus 8,808 separate tests
of significance were performed. The logistic regressions added more than
1,000 additional tests. Whenever so many tests are performed, some
associations will appear by chance to be significant ("false positives").
To reduce the number of "false positives," we considered variables significant
only if the likelihood of an association appearing by chance was less than 1%
(p < 0.01). (The use of the more common value of p < 0.05 could lead to an
unacceptably large number of "false positive" findings. For example, if no
true associations existed, and all variables in the second set of regressions
were unrelated, the expected number of "false positive" findings at the 0.05
level would be 440.) Even at the 0.01 level, as many as 88 false positives
would be expected.
The results of the second set of linear regressions and the set of
logistic regressions are compared in Tables 12-33. All variables that
appeared at least once at the p < 0.01 level of significance are listed in
these tables. Those with the largest number of significant associations are
briefly discussed below.
5.4.1 Health Factors
5.4.1.1 Headache and Nausea
Sensitivity to chemical fumes appeared at a significant level in all
eight possible regressions (Table 12). Workload was significant in six of
eight regressions, but not for females in the logistic regressions. Glare was
always significant for females but rarely for males. Dust and noise were
often significant in the models without comfort and odor, but did not retain
their significance in the models including these variables as explanatory
variables. Allergies to molds were significant for men only. Among the
comfort variables, hot stuffy air appeared at the p < 0.001 level in all four
possible regressions. The odor of paint and carpet cleaning was also
significant in three of four regressions.
20
-------
5.4.1.2 Stuffy Nose, Runny Nose, Sneezing, Cough
Dust and allergies to dust were often significant, with mold allergies
appearing slightly less often (Table 13). Allergies to pollen were
significant for females only. Hot stuffy air was also significant. Women
younger than 30 reported significantly fewer nasal symptoms than those older
than 30.
5.4.1.3 Wheezing, Shortness of Breath, Chest Tightness
Pre-existing asthma had by far the most predictive power for these
symptoms (Table 14). Dust was significant in several regressions, but became
less so in the regressions including the comfort and odor variables. Females
who were sole providers, reported conflicting job demands, and had no college
degree were also likely to have a higher prevalence of chest symptoms. Among
comfort and odor variables, only hot stuffy air was significant for both
sexes, although the odor of fresh paint and carpet cleaning was highly
significant for females.
5.4.1.4 Dry, Itching or Tearing Eyes, Sore Eyes, Burning Eyes, Blurry Vision
A heavy workload and wearing contact lenses to work were significant in
all regressions (Table 15). Sensitivity to chemical fumes, glare, and mold
allergies were also significant in most regressions, although the latter two
were not significant for males in the logistic regressions. Among the comfort
complaints, hot stuffy air and dry air were often associated with these eye
irritation symptoms. Odors from photocopying processes were also
significantly associated with eye symptoms.
5.4.1.5 Sore Throat, Hoarseness, Dry Throat
Sensitivity to chemical fumes, the presence of tacked down new carpet in
offices, dusty offices, and mold allergies were significantly associated with
throat symptoms in many of the regressions that excluded the comfort variables
(Table 16); however, all these variables lost explanatory power when the
comfort variables were added to the regressions. Dry air was very strongly
associated (p < 0.001) with throat symptoms. The smell of fresh paint was
also related to these complaints.
21
-------
5.4.1.6 Sleepiness, Unusual Fatigue
Dissatisfaction with the job and sensitivity to chemical fumes were
sometimes associated with fatigue and sleepiness (Table 17). Hot stuffy air
was strongly associated with these symptoms.
5.4.1.7 Chills and Fever
Women without college degrees were more likely to report these flu-like
symptoms (Table 18). Persons with these complaints were more likely to use
portable heaters. Cold air was significant in all regressions. The odor of
paint was significant in three of four regressions.
5.4.1.8 Aching Muscles or Joints, Pain in Back, Shoulder, Neck, Wrists
Glare entered at the 0.001 level in all eight regressions (Table 19).
Mold allergies also appeared often. An uncomfortable workstation was
significant in many cases. Females under 30 and those in the highest pay
bracket were significantly less likely to report these symptoms. The odor of
cosmetics/body odor was associated with increased frequency of symptoms.
5.4.1.9 Difficulty Concentrating, Remembering; Depression, Tension
A heavy workload and dissatisfaction with the job were the main
variables associated with these psychological and neurological symptoms (Table
20). Other important psychosocial variables included measures of conflicting
job demands and role ambiguity. Sensitivity to chemical fumes was often
significantly related to these symptoms. Hot stuffy air was also often
significant. Females reporting odors of paint/carpet cleaning and
cosmetics/body odor had higher frequencies of symptoms.
5.4.1.10 Dizziness, Light-Headedness
Sensitivity to chemical fumes was the variable most commonly associated
with this symptom (Table 21). Among females, a dusty office and a heavy
workload were significant. Males working in areas with new carpet laid in the
last year (both self-reported and as appearing in the building records) showed
significantly increased prevalence rates in the linear regressions but not in
the logistic regressions. Hot stuffy air was a significant variable in the
22
-------
logistic regressions but not in the linear regressions. The odor of paint and
carpet cleaning was a significant factor for both men and women.
5.4.1.11 Dry, Itchy Skin
Pre-existing eczema was the most significant variable in all eight
regressions (Table 22). Dust was often significant, as was dry air. Females
using portable heaters reported dry skin more often.
5.4.1.12 Problems with Contact Lenses
Few variables were consistently associated with contact lens problems
(Table 23). This is partly due to omitting some variables by error in the
second set of logistic regressions. Dissatisfaction with the job sometimes
appeared significant. Among males, a long time at the workstation was
sometimes associated with increased problems. Among females, dust and dust
allergies were sometimes significant.
5.4.2 Comfort Factors
5.4.2.1 Hot Stuffy Air
Dust and use of a fan were strongly associated with complaints of hot
stuffy air (Table 24). Many other variables were associated with this
complaint, including lighting problems and a heavy workload. Lack of a window
was significant for males only.
5.4.2.2 Dry Air
Dust, noise, and an uncomfortable workstation were significantly
associated with reports of dry air (Table 25). AHU 16, on the second floor of
the mall, was also associated with dry air.
5.4.2.3 Cold Drafty Air
Noise and the use of a portable heater were reported more often in areas
with cold drafty air; use of a fan was significantly less likely in such areas
(Table 26). Respondents at the Crystal Mall and Fairchild buildings reported
cold drafty air consistently less often, whereas several areas in the
Waterside Mall building (e.g., the AHU 14 area on the second floor) reported
cold air significantly more often.
23
-------
5.4.2.4 Humid Air
Dust, the use of a fan, and the lowest pay grade (
-------
5.4.3.6 Tobacco Smoke
Sensitivity to chemical fumes and location on the lower floors of the
West Tower, the Northeast Mall, the eleventh floor of Crystal Mall, and the
garage locations of Waterside Mall were significantly related to tobacco smoke
odors (Table 33). These odors were also associated with dusty offices.
25
-------
6. DISCUSSION
6.1 Limitations of This Study
As with all attempts to complete a census, nonresponse may cause a bias.
Although the response rate was high, and was quite comparable to previous
large-scale building studies, 19% of EPA employees did not complete the
questionnaire. It may be argued that these people had fewer health problems
and therefore felt no need to complete the questionnaire. On the other hand,
nonresponse was highest (38%) among low-paid wage-grade employees (EPA, 1989).
If these employees have somewhat higher frequencies of health problems, as
indicated in nationwide statistics, the resulting bias would be in the
opposite direction.
Employees were asked to characterize the frequency of their health
symptoms over the past year by using somewhat vague terms such as "rarely" or
"sometimes." Relying on memory over a full year is risky, and the vagueness
of the descriptive terms invites misclassification. The most likely result of
these basic limitations is to make it more difficult to detect associations
between health effects and the workplace.
An attempt was also made to identify work-related symptoms, by asking if
the symptom "gets better when not at work." As previously discussed, some
work-related symptoms such as muscle pain may get worse on leaving work,
whereas some non-work-related symptoms, such as a morning headache, may get
better on leaving work; therefore this question was not used in the analysis.
To the extent that this decision jumbled together work-related and non-work-
related symptoms, it also makes it more difficult to detect associations with
the workplace.
All employee responses were subjective. None of the health symptoms or
descriptions of workplace characteristics were verified, with the exception of
the locations of the new carpet and AHUs. This means that all associations,
even those that are highly significant, must be considered tentative; they are
relations between perceptions of health and perceptions of the workplace.
However, the large number of responses required to create a significant
association means that the perceptions are at least widely shared, and may
therefore have a core of truth. Also, the consistency of relationships across
26
-------
different regressions, and the verification of expected relationships (e.g.
asthma was associated with wheezing; new carpet odors were reported most often
in areas with new carpet) provide additional support for the reliability of
the questionnaire responses.
However, consistency of response does not guarantee freedom from bias.
It is likely that persons suffering symptoms will search their environment for
a cause, and may therefore be more likely to observe and report workplace
problems. This "wish bias" (Wynder ££ aj.. 1990) could result in an apparent
but false relationship between symptoms and one or more workplace
characteristics. Since we did not verify reported workplace characteristics
such as dust and glare, we cannot rule out this possibility. On the other
hand, these findings may be testable: areas reported as dusty could be
investigated and compared to other areas, assuming no changes since the
questionnaire was administered. Also, some results, such as the importance of
workplace dust, have been established in previous studies (Skov et al., 1990)
on the basis of objective measurements.
A related methodological limitation is the difficulty of distinguishing
physical/chemical/biological causes from psychological/social ones, when the
investigator is forced to depend on self-reported symptoms and unverified
perceptions of workplace conditions. For example, more symptoms may be
reported under conditions of intense stress and publicity, such as surrounded
the carpet installation at Waterside Mall. This phenomenon has been referred
to by some as Mass Psychogenic Illness, or MPI (Hall and Johnson, 1989). We
attempted to deal with this phenomenon in several ways. First, we included a
set of psychosocial indicators to identify job stresses such as heavy
workloads and conflicting duties. These indicators were allowed to compete
with other personal and workplace characteristics in determining the potential
causes of health symptoms. Secondly, we included several buildings that were
not affected by the carpet incident (Fairchild received no new carpet, and the
Library of Congress Madison building was also unaffected by this incident).
Comparison of symptom frequencies at these buildings would allow the MPI
hypothesis to be tested. Thirdly, since the concept of chemical sensitivity
is thought by some to have a strong psychological component, we asked people
in all four buildings if they believed they were especially sensitive to
carpet fumes and other chemical mixtures. If the carpet installation process
27
-------
had caused either a chemical or a purely psychological sensitivity, the
frequency of persons reporting sensitivity should have been elevated in
Waterside Mall and perhaps Crystal Hall compared to the Fairchild and Library
of Congress buildings.
6.2 Relationships among the Health Symptoms
The PCA for the 32 health symptoms showed that they clustered into a
smaller number of factors that could generally be associated with a particular
bodily system such as eyes, nose, throat, central nervous system, muscular
system, etc. That is, individual symptoms that might appear to be quite
different (runny nose vs. sinus congestion; back pain vs. neck pain) tended to
vary in a similar manner if they referred to the same bodily system. If this
observation is correct, it implies that future questionnaires could be
simplified by asking just one question on each of these health factors,
instead of three or four. However, it should be noted that the questions were
ordered on the questionnaire by body systems (all eye symptoms followed by all
throat symptoms, etc.). This leaves open the possibility that respondents
answered what they perceived to be similar questions in a consistent way,
possibly creating a bias. A kind of inertia may set in as one goes down the
page, such that answering "sometimes" to a question on sore throat makes it
more likely that the same answer will be provided to the following questions
on hoarseness and dry throat. For example, this may be the reason that
headache and nausea, two symptoms not necessarily strongly related, formed a
factor on the PCA--they were listed in first and second position.
To our knowledge, this is the first time that PCA has been employed in
large-scale studies of Sick Building Syndrome. The results obtained here
indicate that previous subjective groupings of symptoms such as "mucous
membrane" or "indoor air quality" symptoms may include some symptoms from
different groups, which may have different causes. For example, "mucous
membrane" symptoms might include eye, nose, and throat symptoms; yet these
systems were quite separate in the PCA. Grouping such symptoms could make it
more difficult to identify potential causes affecting only one or two of the
symptoms in the group.
28
-------
6.3 Agreement among the Regression Analyses
All three sets of linear regressions displayed excellent agreement
regarding the variables that were significant for both men and women. Of 57
associations that achieved the p < 0.01 level of significance for both men and
women in the first set of linear regressions, 49 (>80%) achieved the same
level in the second set of linear regressions. This is evidence that the
replacement of the eight building sectors by 66 AHU-based spatial variables
and five carpet variables did not adversely affect the stability of the
relationships between the most powerful explanatory variables and the health,
comfort, and odor factors. When all 66 AHU-based location variables were
replaced by a single continuous variable--estimated supply airflow of the
AHUs--once again most of the significant variables in the first two
regressions continued to be significant.
The logistic regressions confirmed the general level of stability
displayed by the three preliminary linear regressions. Again most of the
associations (>60%) occurring at the p < 0.01 level for both men and women in
the linear regressions remained so in the logistic regressions.
As measures of goodness of fit, the linear regressions typically
accounted for about 15-25% of the total variation (R2 - 0.15-0.25). The
logistic regressions generally resulted in a 70-80% chance of ordering any
pair of respondents correctly with respect to their probability of
experiencing symptoms (concordance ratio - 0.7-0.8). Both of these goodness-
of-fit measures indicate moderate success in predicting which groups of
respondents are more likely to suffer health symptoms, given their personal
and workplace characteristics.
6.4 Variables Associated with Multiple Health Factors
The results of the second set of linear regressions and the logistic
regressions on the 12 health factors (Tables 12-23) are summarized in Table
34. A total of 96 regressions were performed, 48 for each sex. Thus any
personal, workplace, or location variable could appear a maximum of 96 times.
(Since the comfort and odor variables were included in half of the
regressions, they can appear only 48 times.) The total number of times a
29
-------
variable appears at the p < 0.01 level of significance is provided for each
sex separately in the last two columns. Sensitivity to chemical fumes was
significantly related to health factors more often than any other variable:
in 24 of 48 regressions for women, and 22 of 48 for men. A dusty office was
the second most frequently significant variable, with 24 of 48 regressions for
women and 16 of 48 for men. Among the comfort and odor variables, hot stuffy
air was significant in nearly half of all possible regressions: 10 of 24 for
women and 13 of 24 for men. The odor of paint and chemicals was significant
in 17 of 24 regressions for women but only 4 of 24 for men.
However, even at the p < 0.01 level, we might expect as many as 100
false positive findings, since we performed more than 10,000 tests of
significance in the 96 regressions. To reduce the chances for a false
positive, in the following discussion we shall require that a variable reach
the 0.01 level of significance for at least two of the eight possible
regressions run on any given health factor. It can be shown that this is
equivalent to accepting about three false positives out of every hundred
associations that meet the criterion. We can then rank the variables by how
many different health factors they are associated with.
Using this criterion, three classes of variables may be differentiated:
those that are significantly associated (in at least two of eight regressions)
with multiple (more than three) health factors, those associated with just two
or three health factors, and those associated with only one health factor.
These rankings are summarized in Tables 35 and 36.
Although up to 136 individual variables were considered in these
analyses, only about 11 turned out to be consistently associated with
increased prevalence of at least four of the 12 health factors (Table 35).
The 11 variables include two workplace characteristics (dust and glare); three
personal characteristics (sensitivity to chemical fumes, mold allergies, lack
of college education); two psychosocial factors (workload and conflicting job
demands); two comfort factors (hot stuffy air and dry air); and two odor
factors (the odor of fresh paint, carpet cleaning, and other chemicals; and
the odor of cosmetics, body odor, and foods).
30
-------
6.4.1 Workplace Characteristics.
6.4.1.1 Dust.
The workplace variable that appeared most often in relation to
significantly increased frequency of health symptoms was a dusty workspace.
Dust was significantly associated (at the p < 0.01 level for two or more
regressions) with 11 of the 12 health symptom complexes, missing only the
muscle and back pain syndrome (Table 35).
Since airborne particles were uniformly low throughout all buildings
(EPA, 1990a), it may appear contradictory that reports of dusty offices are so
strongly associated with health, comfort, and odor concerns. However, the
airborne particles measured were small particles of the type that can pass
through the nasal barriers and can exacerbate chronic respiratory problems
such as bronchitis and emphysema; visible dust on surfaces includes larger
particles and fibers that have been associated with acute upper respiratory
(eye, nose, throat) symptoms of the type reported. No measurements were made
of the amounts of carpet dust or settled dust on surfaces in the buildings.
Previous studies have also indicated the importance of dust in health
complaints. The Danish Town Hall Study (Skov and Valbjorn, 1987; Skov et al.,
1989) measured shelf space and fabric-covered surfaces, both measures of
available area for dust to collect, and found them to be the two variables
most highly correlated with health symptoms. These investigators also
measured settled dust in offices and found that it was significantly related
to health complaints (Skov et al.. 1990). Dust may contain pollen, mite
feces, or other allergens, leading to such health symptoms as headache, stuffy
or runny nose, irritated eyes, and dry throat (Gravesen et al.. 1990). In the
EPA study, all four of these symptoms showed a significant association with
reported dustiness of EPA workstations.
In this study, several questions were designed to test the possibility
that increased dust could lead to increased health problems. One question
concerned the number of desks, bookcases, and file cabinets in the
workstation; this was an attempt to estimate the "shelf factor," or the amount
of dust-gathering area in the workstation. Other questions concerned the
"fleece factor": the presence of carpet, drapes, and fabric-covered
partitions as possible dust-catchers. One question was simply the direct
question of how often the workplace seemed too dusty. The first question (on
31
-------
the "shelf factor") never appeared at the p < 0.01 level of significance. The
second group of questions (on the "fleece factor") also failed to be
significant for any health symptom. (One of these questions, on carpet, was
not employed in the regression because of too little variability: 96% of
respondents reported carpeting at their workstations.) Thus it appears that
in this study the direct question on how often the workstation is dusty is
superior to the indirect questions on the amount of shelf space or fabric-
covered area. Also, the measurements of the fabric-covered and shelf areas in
the Danish study are likely to be more accurate than the subjective estimates
of employees.
6.4.1.2 Glare.
Glare was significantly associated with headache, eye irritation,
fatigue, muscle pain, and difficulty concentrating (Table 35). By far the
strongest effect was on bodily aches and pains, including head, shoulder, and
neck pain. This may be due to maintaining a cramped or uncomfortable position
in order to avoid glare. This explanation would also account for the
increased reports of fatigue among workers bothered by glare.
The association of glare with headache, eye irritation and difficulty
concentrating seems logical. Since headache was the single most important
cause of lost work time (EPA, 1989), the relationship of glare to headache may
be an important economic factor; reducing glare could improve productivity by
reducing lost work time.
6.4.2 Personal Characteristics
6.4.2.1 Sensitivity to Chemical Fumes
Although sensitivity to chemicals is a little-understood and
controversial area, it was felt to be an interesting area of inquiry.
Therefore a question was included that asked "Do you consider yourself
especially sensitive to [tobacco smoke, fumes from copying machines, printing
processes, chemicals such as adhesives and cleansers, pesticides, new
carpeting, drapes, paint, and carpet cleaning]?" Little information exists on
the percentage of persons who consider themselves sensitive to chemicals; this
question would provide such data for the study population. Also, little is
known about the "epidemic" effect of publicity in possibly raising the
32
-------
prevalence of complaints among nearby workers. Since the Waterside Mall
building had experienced a recent highly publicized incident of worker
illnesses following carpet installation, the hypothesis was made that the
prevalence of sensitivity to chemical vapors at WSM might be higher than at,
say, the Fairchild building, where no carpet had been laid. This would be the
case whether there was a true chemical/biological effect or a purely
psychological effect. In fact, little difference was shown among the three
buildings in the prevalence of persons reporting unusual sensitivity to
chemical fumes: 32% at both Waterside and Crystal, 29% at Fairchild. As a
further test of this hypothesis, the same question was asked of the 2000-odd
workers at the Library of Congress Madison building; 33% reported chemical
sensitivity. Thus, the absolute prevalence appears to be rather high (nearly
a third of all respondents), and the number of persons considering themselves
unusually sensitive to chemicals appears to have been unaffected by either the
carpet installation itself or the ensuing publicity related to fellow workers'
complaints. (It is recognized that this analysis did not include the most
seriously affected workers--those working in the alternative workspaces.
Also, the question as asked could not distinguish the relative severity of
responses or the number of chemical mixtures associated with increased
sensitivity.)
The question on sensitivity to chemical fumes was the most commonly
appearing indicator of increased prevalence of health complaints. Nine of the
12 health factors were significantly associated with sensitivity to chemical
fumes in at least two of eight regressions: headache; eye, nose, throat, and
chest symptoms; fatigue; muscle pain; dizziness; and difficulty concentrating
and other central nervous system complaints (Table 35). All of these symptoms
have been linked to Sick Building Syndrome (Molhave 1987) , which some
investigators have hypothesized may be closely related to chemical sensitivity
(Ashford and Miller, 1990).
6.4.2.2 Mold allergies
Four types of allergies were recorded (mold, dust, animals, pollen), but
only the mold allergies were repeatedly associated with health symptoms. Mold
allergies were significantly associated (in at least two of eight regressions)
with eye, nose, and throat symptoms; muscle pain; headache; fatigue; and dry
33
-------
skin (Table 35). The number of significant associations involving mold
allergies could be an important clue with respect to health symptoms and
building conditions. Several recent studies have reported increased health
problems in homes, hospitals, and other buildings with mold or moisture
problems (Horey and Williams, 1990; Kalliokoski e_£ al., 1990). If dusty
working conditions provide a growth habitat for mold, the appearance of both
these variables as important explanatory factors might be expected. The very
low measured relative humidities may argue against this interpretation;
however, these measurements were made during the winter, when low indoor
humidities are expected. High humidities may be encountered during other
seasons. Also, a number of cases of water leaks, toilet overflow, water-
stained carpet and damp ceilings were reported in the essay section of the
questionnaire; therefore the conditions for mold growth may exist in some
areas.
6.4.2.3 Education
Persons--particularly women--with higher education reported fewer health
problems. Both men and women with college educations reported fewer chest
problems and flu-like (chills and fever) symptoms. Women with college degrees
reported fewer headaches, less fatigue, and less dizziness. This finding is
in general agreement with observed lifestyles of less educated and lower-paid
people (more smoking, lower nutrition, lower quality medical care) that may
lead to their higher prevalence of health problems (NCHS, 1989).
6.4.3 Psychosocial factors
6.4.3.1 Workload
Persons reporting a heavy workload were more likely to report headache,
eye irritation, and the group of symptoms comprised of difficulty remembering,
difficulty concentrating, depression, and tension. These are the symptoms
that one would expect to be related to working hard or in a concentrated
fashion under tight time schedules.
6.4.3.2 Conflicting Job Demands
Persons reporting conflicting demands were more likely to suffer from
nasal symptoms, chills and fever, and the psychological/neurological symptoms
34
-------
including tension, depression, difficulty concentrating, and difficulty
remembering.
6.4.4 Comfort Conditions
6.4.4.1 Hot Stuffy Air
Perceptions of hot or stuffy air were associated with more health
factors than any other comfort or odor concern. Many people complained of
this condition, particularly in the essay section of the questionnaire; it was
associated with increased incidence of headache, stuffy nose, eye irritation,
chest tightness, fatigue, difficulty concentrating, and dizziness (Table 35).
Although the room temperature and airflow measurements during the one-week
monitoring period in winter showed generally moderate temperatures and
adequate airflow, the log of building complaints shows that measured
temperatures have been high in some locations, and complaints of reduced
airflow have occurred in some areas.
Some locations in Waterside Mall had fewer complaints of hot stuffy air;
these included the Northeast Mall and the East and West Towers. These
locations had the highest estimated supply airflow rates per unit area.
6.4.4.2 Dry air
Perceptions of dry air were associated with increased reports of stuffy
nose, eye irritation, sore throat, and dry skin. Measured humidity levels
were indeed very low (about 25% relative humidity), and no AHUs were found in
Waterside Mall with operating humidifiers. However, the measurements were
made in the winter and it is not known whether humidity levels remain low all
year round.
6.4.4.3 Gender Differences in Perception of Comfort
As mentioned above, women reported more comfort problems than men,
finding the air to be hotter, colder, drier, and more humid. Since women may
wear lighter clothes than men, they might perceive the same air temperatures
as colder; but since they do more work with copy machines, printers, and other
local heat sources, they could also be exposed to higher localized
temperatures during part of the day. The influence of psychological pressures
associated with lower pay and worse working conditions could also influence
35
-------
perceptions of comfort. It may also be that a cultural factor inhibits men
from expressing discomfort.
6.4.5 Odor
6.4.5.1 Odor of Fresh Paint, Carpet Cleaning, and Other Chemicals
This odor factor was significantly associated with a number of health
symptoms: headache, stuffy nose, chest tightness, sore throat, fatigue,
chills and fever, difficulty remembering or concentrating, and dizziness
(Table 35). Most of these odors are associated with maintenance activities or
with renovation (painting, installing new carpet, etc.).
6.4.5.2 Odor of Cosmetics, Body Odor, and Non-Fish Food Odors
This odor factor was associated with eye irritation, chills and fever,
muscle pain, and the factor including symptoms of depression, tension,
difficulty concentrating, and difficulty remembering. These associations,
based on all EPA respondents, include some of the same symptoms mentioned in
the medical literature and in the essay question by persons reporting strong
reactions to perfumes, colognes, and other cosmetics.
6.5 Variables Associated with Two or Three Health Factors
In addition to the 11 variables affecting multiple health symptoms
described above, another dozen or so variables were associated with just two
or three health symptoms (Table 36). These include several personal/medical
characteristics (allergies to dust, under 30 years old, and wearing glasses);
two psychosocial factors (job satisfaction, sole provider for a family); seven
additional workplace variables (noise, use of a portable heater, window
visible from office area, walls recently painted, water leaks near workspace,
hours at work, and new carpet); one location variable (AHU 21 in the Library
area of Waterside Mall); and one comfort factor (cold drafty air).
6.5.1 Personal/Medical Characteristics
Dust allergies were strongly associated with stuffy nose and
occasionally associated with problems with contact lenses (Table 36). Wearing
glasses was occasionally associated with eye and nose irritation. Women under
36
-------
30 were less likely to report nasal symptoms and muscular aches and back pain.
This was the only observed effect of age, which indicates that age was not a
very important factor in most of the targeted health symptoms for this
population of office workers.
6.5.2 Psychological/social factors.
Dissatisfaction with the job was very strongly associated with the group
of symptoms comprised of difficulty concentrating, depression, and tension
(Table 36). It was also associated with fatigue, mainly among women. Women
who were sole providers for their families reported higher frequencies of
chest and throat symptoms and chills and fever.
6.5.3 Workplace Variables
Noise was associated with headache, muscle pain, and difficulty
concentrating (Table 36). Lack of a window was associated (among men only)
with eye and throat irritation and fatigue. Use of a portable heater was
associated with chills and fever and dry skin. Water leaks were associated
with tiredness and the factor including difficulty concentrating, depression,
and tension. Persons with recently-painted walls reported more chest symptoms
and eye irritation. Long hours at work were associated with contact lens
problems and flu-like symptoms. Tacked-down new carpet in hallways was
associated with dizziness.
6.5.4 Location Variable
The only location variable associated with more than one health symptom
was AHU 21 in the Library area of Waterside Mall. Hen in this area reported
higher frequency of flu-like symptoms (chills and fever, muscle/joint aches
and upper-body pains).
6.5.5 Comfort Factor
Cold, drafty air was associated with nasal congestion, throat symptoms,
and chills and fever.
37
-------
6.6 Variables Associated with a Single Health Factor
An additional 14 variables were associated (p < 0.01 in at least two
regressions) with a single health symptom (Table 36): pre-existing asthma
(chest symptoms); eczema (dry skin); allergy to pollen (nasal symptoms);
uncomfortable workstation, lights too dim, and living alone (aching
muscles/upper body pain); wearing contact lenses, hours at computer, and odor
of photocopying (eye irritation); odor of new carpets (headache); new carpet
tacked down in office (throat symptoms); role ambiguity (central nervous
system symptoms); and external stress (fatigue). One location variable (AHU
16 on the second floor of the mall) was associated with eye symptoms among men
only.
6.7 Possible Gender-Related Variables
A few variables appeared significant primarily for one gender.
6.7.1 Females
The odor of paint and other chemicals was much more likely to be
significantly associated with women's health symptoms than with men's (17
significant associations for women compared to four for men--Table 34).
Several factors may contribute to this result. More women than men reported
working with chemicals (24% vs. 11%). Also, more women had their workstations
in open areas where chemicals are more likely to be used (40% vs. 14% for
men), whereas more men were in private offices where the levels would be lower
than near the sources (46% vs. 21% for women).
Women with college degrees reported fewer headaches, fatigue, dizziness,
chills and fever, and chest symptoms than other women. Since only 6% of males
lacked college degrees, the number may have been too small to allow detection
of differences in symptom prevalence among men.
Young women (under 30) were less likely to report muscular aches and
upper-body pains than older women. Since 24% of females but only 8% of males
were under 30, again the numbers of males involved may have been too small to
detect symptom prevalence differences. Males of any age reported fewer
muscular aches and pains than females; the number of males reporting the five
38
-------
pain-related symptoms as occurring "often" or "always" in the previous year
was about half the number of females.
Women reporting that they were the sole providers for their family were
more likely to report throat and chest problems as well as chills and fever.
Many more men reported being sole providers (26% vs 12% of women); however,
since this is a more customary situation in society, and since men in general
earn more than women, there may be fewer stresses involved for males.
Women working in EFA's garage and lower-garage level offices were more
likely to report chest symptoms and muscle pain than women in other locations.
These areas include the print shop and the maintenance and supplies areas,
which may involve more physical labor than in other areas. They may also have
more chemicals in use or in storage, and more exposure to automobile exhaust.
Women reporting recent repainting of nearby walls were more likely than
men to report chest and eye symptoms.
6.7.2 Males
Lack of a window was a contributing factor to higher rates of eye and
throat symptoms and feelings of fatigue among men. Somewhat fewer women than
men had access to a window (38% to 49%). but this difference is too small to
explain why women were not affected by this variable.
Men working near the EPA Library (AHU 21 on the second floor of the
mall) reported increased rates of flu-like symptoms (muscle aches/upper-body
pains and chills and fever). This area was also associated with increased
reports of damp or musty odors. Roof leaks were noted in this area, and
biological sampling carried out by EPA management confirmed higher levels of
biological contamination (J. Davidson, personal communication). This AHU has
been replaced.
The odor of new carpets was associated with headache, chest and throat
symptoms, fatigue, dizziness, and dry skin among men but not among women.
Water leaks near the workstation were associated (among men but not
women) with feelings of fatigue and the central nervous system (CNS) symptoms
of depression, tension, and difficulty concentrating/remembering. About the
same number of men as women (13% vs. 12%) reported water leaks occurring in
the previous year; thus the reason for an association with men but not with
women is not due to sample size.
39
-------
6.8 Variables Associated with Multiple Comfort and Odor Factors
A summary of the variables significantly associated with comfort and
odor concerns (Tables 24-33) is provided in Table 37. The most frequently
appearing variable was dust, appearing at the 0.01 level of significance in 15
of 20 regressions for females, and 20 of 20 for males.
Again requiring that a variable be significant in at least two of the
possible regressions (to avoid false positives), we can group variables by
whether they affect multiple comfort/odor factors or only one or two.
Dust was by far the factor most broadly associated with comfort and odor
factors. By the criterion above, it was significantly associated with all 10
factors. In most of the cases involving odors, the reason for the association
appears evident: tobacco smoke, photocopying and printing processes, and
diesel fumes are themselves sources of dust particles, and fresh paint is
often associated with dust-raising renovation. Dusty offices were also
associated with conditions of hot stuffy air, dry air, and cold drafty air.
Other workplace characteristics often associated with comfort and odor
concerns were noise, glare, nearby water leaks, and the use of a fan (Table
38). Noise was strongly associated with cold drafty air, which may sometimes
be associated with renovation. Glare was associated with hot stuffy air.
Water leaks were associated with musty or damp odors and with odors of
cosmetics and tobacco smoke. (Since smoking is allowed only in some rest
rooms, and since cosmetics are often applied in rest rooms, the association of
these odors with water leaks may be expected.) Fans were employed more often
when the air was hot, stuffy and humid, and less often under cold, drafty
conditions.
Two personal characteristics were often associated with odor concerns,
but seldom with comfort concerns. Sensitivity to chemical fumes was related
to increased reports of odors of paint and other chemicals, photocopying and
printing process fumes, and tobacco smoke (Table 38). Persons reporting
conflicting job demands were more likely to report odors of cosmetics and body
odor, photocopier and printing process odors, and musty or damp odors.
Two contiguous spatial locations were associated with multiple comfort
or odor concerns. The Southwest Mall (AHU 15 and SW-2) showed significantly
increased reports of cold air, odors of paint and other chemicals, and the
40
-------
odor factor including damp or musty smells and engine exhaust. An adjacent
area on the second floor of the mall (AHU 16) was also associated with cold
air, dry air, and the damp/musty/engine exhaust odors.
6.9 Variables Associated with One or Two Comfort/Odor Factors
With the exception of the variables discussed above, most of the
remaining variables associated with comfort and odor were quite specific,
often appearing with only one comfort or odor condition (Table 39). For
example, in all four sets of regressions, heater use was associated with cold
drafty air. The presence of new drapes and carpets was associated with the
odor of new drapes and carpets, as was the presence in the respondents'
offices of tacked-down new carpet. The presence of a copy machine was
associated with the odor of photocopying processes. Similarly, recently
painted walls were associated with the odor of paint, and the use of chemicals
at the workspace was associated with the odor of paint and other chemicals.
Although all of the above associations may sound rather unsurprising, if
not excruciatingly obvious, they are an important indication that the
questionnaire is sound and that the answers to these questions may be relied
upon. That is, since these responses display the expected relationships with
comfort and odor prevalence, they indicate that other observed relationships
based on the questionnaire responses may be trustworthy.
Seven other variables were associated with just one or two comfort or
odor conditions. Low pay was associated with reports of humid air. An
uncomfortable workstation was associated with reports of dry air. Complaints
of dim lighting were associated with reports of hot stuffy air and dry air. A
private office, no access to a window, and a heavy workload were all
associated with increased reports of hot stuffy air. Mold allergies were
associated with reports of damp, musty odors.
6.10 Effect of Location
6.10.1 Effect of location on health
The main goal of the first set of linear regressions was to determine
whether buildings and/or building sectors differed in the prevalence of
41
-------
health, comfort and odor reports. The results showed few differences between
buildings or building sectors. None of the 12 health factors showed a
significant relationship with any building or building sector.
The second set of regressions, employing a finer-grained measure of
spatial variation (the areas served by the AHUs), confirmed the general lack
of a "spatial effect" on health symptoms and extended this finding to apply to
AHUs as well: few individual AHUs (or the spaces they serve) were associated
with increased reports of health symptoms at the 0.01 level of significance.
The small number of associations noted were significant for one sex but not
the other. In view of the large number of spatial variables (66) tested in 12
sets of four regressions (male/female; with/without comfort and odor
variables), most of these associations may be due to chance. To reduce the
possibility of reporting chance associations, only those that were significant
at p < 0.01 for at least two of the four regressions are discussed below.
Men in the Library area (AHU 21) had significantly elevated frequency of
flu-like symptoms, including chills and fever as well as aching muscles and
joints and upper body pain. (This AHU was replaced shortly after the survey.)
Men on a portion of the third floor of the Mall (AHU 9) also had higher
frequency of aching muscles/upper body pain.
Women in the garage areas (AHUs G.LG) showed higher rates of chest
symptoms. Women in the Southeast Mall (AHU SE 2-1, 2-2) reported higher
frequencies of difficulty concentrating and dizziness. Women on the lower
floors of the West Tower (AHU WT FL1-3) and the 8th and 9th floors of the East
Tower (AHU ET1 FL 8-9) reported higher rates of dry skin complaints.
Some locations were associated with significantly reduced rates of
health symptoms. Men on the lowest floor of the Northeast Mall (AHU NEG-1,2)
and women on the next higher floor (AHU NEG-4) reported significantly fewer
throat symptoms than in other areas. Women in Crystal Mall showed lower
frequency of chills and fever.
The logistic regressions confirmed the general lack of a building effect
on health variables, with the possible exception of AHU 16 on the second floor
of the mall, which was associated with increased levels of eye irritation.
The lack of a "building effect" on health symptoms does not necessarily
mean that there are no differences in health symptom prevalence rates among
buildings. For example, one of the most important variables associated with
42
-------
health complaints was dust. People who reported that their workstation was
too dusty were significantly more likely to have headaches, nasal problems,
cough, eye irritation, and sore or dry throat. These people were more heavily
concentrated in WSM than in the other two buildings; 24% of WSM respondents
reported that their workstation was often or always too dusty, compared to 16%
at Crystal and only 12% at Fairchild. Since dust was separately included as a
variable, the "building" variable did not reflect this difference in
dustiness; rather, the building variable includes only whatever differences
remain after all other specific variables (which may differ between buildings)
are accounted for. Therefore, the fact that little or no "building effect"
was observed means that no remaining difference between buildings (after
removing the differences due to the specific variables included in the
regression) was strong enough to account for variations in health symptom
prevalence.
Similarly, the lack of any consistent association of the AHUs with
health symptoms should not be interpreted to mean that ventilation had no
effect on health; in fact, some of the ventilation-related comfort variables
(hot stuffy air, dry air) had extensive associations with health symptoms.
To summarize, our finding of little or no residual effect of location
implies that most or all of the spatial variability was captured by the other
existing variables in our model. Thus we need not look for additional causes
of spatial variability. Instead, we can analyze the spatial variability
associated with certain important variables (e.g., dust, sensitivity to
chemical fumes, hot stuffy air) to identify locations in possible need of
corrective action.
6.10.2 Effect on Comfort and Odor
In the first regression, which included the "coarse-grained" building
and building sector variables, a clear "building effect" was observed for one
comfort and two odor factors. The Waterside Mall building was significantly
more likely to be associated with complaints of cold drafty air than the other
two buildings. Also, the second floor at Waterside was significantly more
likely to be associated with the odor complex including musty/damp, fishy, or
engine exhaust odors. (This floor had a much higher complaint rate for engine
exhaust odors than any other sector.) Finally, the building sector including
43
-------
the two office areas on the garage and lower garage levels of Waterside Mall
were more likely to be associated with reports of printing or photocopying
odors.
In the second regression, which employed the "fine-grained" AHD/floor
combinations, certain AHUs (or factors associated with the spaces they serve)
were strongly associated with comfort and odor factors. These included two
AHDs on the Southwest Mall (increased reports of paint and other chemical
odors, together with damp/musty and engine exhaust odors), and some serving
the second floor of the mall (complaints of dry air and/or musty/damp odors
from AMDs 2 and 16) . Cold air complaints were significantly increased in some
of the Waterside Mall areas (e.g. AHD 14 on the second floor of the mall), and
significantly decreased in the Crystal Mall building. This agrees with the
respondents' essays, which included many complaints of cold air at Waterside
Mall and hot stuffy air at Crystal Mall. Printing process odors were evident
in the print shop location (61) . Odors of tobacco smoke were associated with
the first floors of the West Tower (AHD WTl PL 1-3) and the basement level of
the Northeast Mall (NEC-1,2) . The Library area of Waterside Mall (AHD 21) was
associated with reports of humid air. Many of these building effects on
comfort and odor factors were also confirmed by the logistic regressions.
€.11 Effect of New Carpet
Although new carpet showed only a few strong associations with health
symptoms, it is discussed here because of the intense interest in the carpet
situation at Waterside Mall. The new carpet variable was the only one for
which objective data existed to check against the questionnaire responses.
Evidence that the areas with new carpet had been reasonably well identified
was provided by the very high correlation of the associated variables (new
carpet, tacked or glued, in offices or in nearby corridors) with reported
presence and odor of new carpet.
The two health factors that were most closely associated with new carpet
were throat symptoms and dizziness. Both men and women in offices that had
received the new carpet (tacked-down) reported higher prevalence of throat
problems in the past year, significant at the 0.01 level in three of eight
44
-------
possible regressions, and at the 0.05 level in an additional three
regressions. Complaints of dizziness or light-headedness were also elevated
(among men only) in areas with new tacked-down carpet in nearby hallways. In
addition, one other health factor (headache and nausea) was associated with
the odor of new carpets (males only).
Since the period of recall for the health symptoms was one year, and
since the carpet had been installed roughly one year before, it is not known
which of the respondents, if any, were still experiencing symptoms; however,
these results provide some support for relating some health symptoms--most
likely throat symptoms--to areas with the new carpet.
Another study (Norback and Torgen, 1989) also noted a significant effect
of carpet on dry throat and sore throat complaints, as well as eye irritation
and headache. Workers in two schools with wall-to-wall carpets had 2-3 times
the prevalence of these complaints as workers in four schools without carpets.
Since the carpets were 8-10 years old and were never cleaned with chemicals or
dry shampoo, emissions of VOCs were unlikely to be the cause of the
complaints. Also, since the area in question has low humidity for several
months each winter, mites and mite allergens were also ruled out. The author
concluded that dust accumulation on the carpets was the most likely cause of
the association (Norback, 1990).
6.12 Variables with Little Effect on Health, Comfort, or Odor
Many personal and workplace characteristics showed little or no effect
on the reported health symptoms and comfort/odor concerns. For example, as
discussed above, little spatial variability was noted for any health symptom.
Neither the coarse-grained spatial variables (buildings or building sectors)
nor the fine-grained variables (AHU by floor) showed consistent associations
with any health symptom, with the exceptions mentioned above. Smoking status,
type of office, time working at a copying machine, presence of new furniture,
airflow capacity, certain psychosocial indices, several comfort/odor factors,
and several other variables that have been suspected to be associated with
health symptoms seldom showed significant associations. Table 40 summarizes
the variables that showed few or no associations with health, comfort, or odor
factors. Some are discussed further below.
45
-------
Five questions concerned the presence of laser printers and copy
machines and the number of hours spent at the workstation or working with
computers/word processors and copy machines. None of these variables had much
effect on health symptoms. Time at the workstation was associated with chills
and fever and with contact lens problems, but only for men, while time in
front of a computer was associated with eye and nasal irritation, but only for
women. Fewer men than women worked 5 hours or longer near computers (298 men
compared to 551 women).
Several questions dealt with office surroundings, such as presence of
fabric-covered partitions, live plants, and new carpet, drapes, furniture or
equipment. Two additional variables, "shelf space" and wood surfaces, were
created from these questions. None of these variables except the new carpet
were associated with health symptoms.
Two questions concerned recently painted or recently rearranged walls.
Neither variable had much effect on health symptoms, although the recently
painted walls variable was sometimes (females only) associated strongly with
eye and chest symptoms.
Two of the seven psychosocial indices used in this study (conflicting
job demands and workload) showed multiple significant associations with health
symptoms. Several other variables (job satisfaction, external stress, and
role ambiguity) were associated with one or two symptoms. However, two other
variables (job control and utilization of skills) had little effect.
Although allergies to molds, dust, and pollen were important explanatory
variables for some symptoms, allergies to animals or to other allergens were
not associated with any health symptom.
Age was not often an important explanatory variable. When it did
appear, it was for women only: women under 30 had fewer nasal symptoms and
muscular aches and pains.
Pay grade was seldom important as an explanatory variable. Females in
the highest pay grade reported significantly fewer muscle and back pains;
females in the medium pay grade (GS9-12) reported dry skin more often.
In the first two sets of linear regressions, separate variables were
included for the type of office and the number of persons sharing the office.
For the final logistic regression, these variables were combined into one
46
-------
variable with three levels: private offices, private cubicles or offices with
exactly two people, and all other combinations. No matter which set of
variables was employed, the type of office and space sharing had little effect
on health symptoms. One exception was fatigue (among men only), which was
increased in open areas or offices with two or more workers.
Two variables were created to describe respondents' living conditions:
living alone and living with others as sole provider. Neither variable
provided much explanatory power for health or comfort symptoms. Living alone
was associated with increased reports of muscle aches and pains among men
only, while being the sole provider was associated with increased rates of
chest, throat, and flu-like symptoms among women only.
Smoking status was not associated with any of the health or comfort
symptoms.
6.13 Comparison with Other Studies.
The normal prevalence of the 32 target symptoms is not well known. For
example, the annual Health Interview Survey carried out by the National Center
for Health Statistics (NCHS) includes only a few of the symptoms (dry skin,
fever, "frequent" headaches) used in our questionnaire, and prevalence rates
are reported only for those symptoms that receive medical attention or
restrict activity (NCHS, 1988). Thus it is not possible to compare NCHS data
with data from this study.
6.13.1 Family-Based Studies.
One study (Lebowitz et al., 1972a) followed 418 families in Manhattan
for several years, collecting weekly reports on 14 symptoms experienced by all
members of the family over the previous week. Nine of the symptoms were
similar to those in this study. These nine symptoms were reported 3-12 times
more often by EPA workers than by the Manhattan subjects (Table 41).
The reasons for these large differences are unclear. In the Manhattan
study, information was supplied by one person (the mother in 85% of cases) for
all other family members; this could lead to underreporting of certain less
noticeable symptoms (such as headache, eye irritation, and aching muscles, the
three symptoms with the greatest differences) in other family members.
47
-------
However, other studies with different methodologies resulted in prevalence
estimates similar to those of Lebowitz (Buck, 1956; Dingle 1964). Also, these
family studies included children, who typically suffer higher rates of minor
upper respiratory symptoms, and persons not in the work force, who are also
generally considered to be at a lower level of health. These factors would
tend to increase the prevalence of most of these symptoms in families compared
to a working population. Since our study took place in a week in the winter,
when respiratory infections are about twice as common as in spring or summer
(NCHS, 1988), it is likely that the respiratory symptoms were more frequent
than in a typical week; however, this factor of 2 would not explain the much
larger factors of 3-12 observed, nor would this explain the differences in the
non-respiratory symptoms.
Lebowitz concluded that his study had found "a remarkably high burden of
illness in a healthy urban population--in fact, far more than normally
imagined" (Lebowitz e£ aj.., 1972b). Since the rates in the EPA employee
population appear to be several times those found by Lebowitz, it may be that
EPA employees also are carrying a heavy burden of minor illnesses. For
example, about half of EPA employees reported a headache the previous week
with an average duration of 2.0 days. This corresponds to about 5000
headaches per week among the 5000 employees, or a quarter of a million
headaches per year. If each headache results in a few minutes of lowered
productivity, the total time lost would be several man-years. About 15% of
EPA employees reported leaving work early or not coming in due to headache
(EPA, 1989). Some other symptoms, such as sinus congestion, were even more
common than headache. About 5% of employees reported losing work time due to
this symptom (EPA, 1989).
6.13.2 Large Building Studies
Three other studies have administered questionnaires to several thousand
office workers in an attempt to learn more about building-related health
complaints. These studies have taken place in Great Britain (Wilson and
Hedge, 1987), Denmark (Skov and Valbjom, 1987; Skov et al., 1989, 1990) and
the Netherlands (Zweers e_£ al., 1990; Preller, et al., 1990). All three of
those studies involved a large number of buildings (28 to 61), whereas the
present study included only a small number of buildings (3). Thus these
48
-------
investigators were able to compare, for example, buildings with natural
ventilation to those with mechanical ventilation, and buildings with spray
humidification to those with steam humidification, comparisons which our study
is unable to make. On the other hand, since our study involved a census
rather than a sample of employees, we were able to carry out a spatial
analysis of our buildings. (The Danish investigators also carried out a
census of 4369 employees in 28 buildings, but did not analyze spatial
variation within the buildings.) We were also able to test the effect of all
air handling systems employed in the buildings as well as the effect of new
carpet installed in two of the buildings in the previous year.
Most of the personal characteristics and work environment variables
measured in the previous studies were also measured in this study, allowing
multiple comparisons to be made among the four studies. The basic
characteristics of the four studies are listed in Table 42.
Many of the health symptoms included in the questionnaire were similar
from one study to the next; however, the wording of the questions was
different. For example, all four studies attempted to deal with the frequency
of symptoms and whether they were "work-related." However, all four had
different definitions both of frequency and of what constituted a work-related
symptom. Wilson and Hedge asked respondents if they had experienced any of a
set of 10 symptoms "on two or more occasions in the previous year," and if the
symptoms were better on days spent away from the building they worked in.
Skov and Valbjorn asked if a set of 12 symptoms occurred "several times a week
or more" and if the symptoms improved on weekends, holidays, and vacations.
Zweers et al. and Preller et al. asked if any of 18 symptoms occurred "daily
or weekly in the past 12 months and if improvement was noted when away from
work." The EPA study asked if any of a set of 32 symptoms occurred "never,
rarely, sometimes, often, or always" in the past year, and if they "usually
got better" on leaving work; a second EPA category asked if the symptom
occurred last week, and if so, on how many days. As can be seen, the minimum
criterion for a "positive" response varied from 2 days per year in the British
study to about 100 days per year in the Danish study. We attempted a direct
comparison of symptom frequencies in these studies but were unable to overcome
these fundamental differences in definition.
49
-------
6.13.2.1 Grouping of Symptoms
In all the studies, symptoms were grouped for analysis. In one paper
based on the British study (Raw e£ al. , 1990) , only the number of positive
symptoms was used as the dependent variable. In the Danish study, two
groupings were used: mucous membrane (eye, nose, throat) complaints and
"general" symptoms (headache, fatigue, etc.). The Netherlands investigators
grouped 17 symptoms into five factors.
Of the four studies, apparently only one (the present study) used an
objective method of grouping health symptoms. As described above, a PCA on
the original 32 health symptoms identified nine factors. Three additional
symptoms (dizziness, dry or itchy skin, problems with contact lenses) were
also analyzed. The nine factors and three individual symptoms are listed here
for comparison with those in the European studies:
1. Headache and nausea
2. Nasal (runny nose, stuffy nose/sinus congestion, sneezing, cough)
3. Chest (wheezing, shortness of breath, chest tightness)
4, Eyes (dry itching or tearing eyes, sore eyes, blurry vision, burning
eyes)
5. Throat (sore throat, hoarseness, dry throat)
6. Fatigue (unusual fatigue or tiredness, sleepiness or drowsiness)
7. Chills and fever
8. Muscle pain (aching muscles or joints, pain or stiffness in upper
back, pain or stiffness in lower back, pain or numbness in
shoulder/neck, pain or numbness in hands or wrists)
9. Neurological (difficulty remembering or concentrating, feeling
depressed, tension or nervousness)
10. Dizziness
11. Dry skin
12. Problems with contact lenses
For comparison, the five symptom groupings in the Netherlands study were
as follows:
1. Neurological (fatigue, headache, heavy feeling in the head,
listlessness, dizziness, problems with concentration,
forgetfulness, irritability)
50
-------
2. Skin (dry skin, skin rash, itchy or prickly skin)
3. Eyes (weary eyes, itchy, prickly, or watering eyes)
4. Nose/throat (blocked or runny nose, dry throat, sore throat)
5. Fever [sic; more appropriately, flu-like] (shivery, muscle or
joint pain)
Thus the Netherlands classifications mix some categories found to be
separate in the US analysis. For example, the Netherlands "neurological"
category includes symptoms from four different US categories--headache,
dizziness, fatigue, and problems with concentration and forgetfulness. Also
the Netherlands study combines nose and throat symptoms, which were found to
be separate factors in the US study.
The Danish study identified two symptom groupings:
1. Hucosal irritation (at least one work-related symptom of
irritation/dryness of the mucous membrane of the eyes, nose, or
throat);
2. General symptoms (at least one work-related symptom of headache,
abnormal fatigue, or malaise).
Again, the Danish classification, like that of the Netherlands study,
lumps categories found to behave as separate factors in this study. Because
of these differences in grouping symptoms, comparisons of significant factors
across studies may be flawed. Nevertheless, a list of variables found to be
significant (and those tested and found not significant) in the studies is
included in Table 43.
In the present study, we have chosen to require a significance level of
p < 0.01, because of the large number of tests being made. The three other
studies chose significance levels of 0.05. Since those studies also made
multiple comparisons, they may be expected to include a larger number of false
positive findings.
6.13.2.2 Comparison of Findings
In view of the different definitions of health symptoms, and the
different ways of grouping those symptoms for analysis, the findings of the
51
-------
studies may be expected to differ somewhat. However, if certain findings are
common to most or all studies, they may be viewed as having some
substantiation.
The only finding that was common to all three European studies was the
increased symptom rates reported by females. This increased frequency was
also noted in our study, but we made no formal tests to determine its
significance. This was because we believe that many factors interact with
gender (education, pay grade, type of workstation, number of persons sharing
workspace, etc.) and would have to be accounted for in a multivariate model.
By running separate models for men and women, we allowed for all such
interactions, at the cost of being able to calculate the odds ratio between
men and women.
A major explanatory variable isolated in both the Danish (Skov et al..
1990; Gravesen et al. . 1990) and our studies was dust. The likely reasons for
the importance of dust as a reservoir of allergens have been discussed above.
A few factors found significant in two of the three European studies
were also significant in the U.S. study. These included dissatisfaction with
the job and pre-existing allergies.
Some of the variables that we found to be highly significant were not
examined, or were examined in a less specific manner, in some of the other
studies. These include perceptions of dust and glare at the workstation,
allergies to molds, and sensitivity to chemical fumes. Other variables that
we found significant, such as the presence of new carpet, were not tested for
in the other studies.
Some factors that were found significant in some of the other studies
failed to reach significance in this study. For example, the type of
workstation and number of people sharing a room (an important factor in the
British study) seldom had any effect on health symptoms. Time spent at a
photocopier or in front of a video display terminal (important in the Danish
and Netherlands studies) also had little effect. The "shelf factor" found
important in the Danish study was not significant in this study. (However,
the Danish study actually measured the shelf factor, whereas our study
depended on subjective estimates; also, our inclusion of a direct question on
the presence of dust may have supplanted the more indirect measure represented
by the shelf factor.) Similarly, the "fleece factor" found important in the
52
-------
Danish study was not significant in this one, although again the Danish study
included objective measurements. Also, carpet is an important contributor to
the fleece factor, but 96% of the EPA floor area was carpeted, which reduces
the possible variation. The somewhat irregular age effects noted in other
studies, with younger and middle-aged workers sometimes reporting higher
symptom prevalence than older workers, were not replicated in this study; for
the few times age was important, younger workers showed lower prevalence.
Being a smoker had no effect on health symptoms in our study, although both
the Danish and Netherlands studies noted an effect. (The prevalence of
smoking in this study may have been lower than in the European studies; also
all buildings in this study had no-smoking policies in effect in the work
areas.)
53
-------
7. CONCLUSIONS
This study has allowed testing a number of hypotheses regarding the
association of personal and workplace characteristics with perceptions of
health, comfort, and odor. A new questionnaire included many variables that
have been tested in previous studies, but also included a number of new
variables tested here for the first time. The effects of location within the
building were tested as a possible factor of importance for the first time.
Information was also obtained on the locations where carpet had been installed
in the previous year, to test the hypothesis that carpet installation might
have health effects. The many AHUs in the buildings were mapped out and every
person was associated with an AHU. This made it possible to carry out a valid
multivariate statistical test of the hypothesis that AHUs, or the spaces
served by them, might be associated with health, comfort, or odor.
Conclusions regarding the methodology and the questionnaire will be discussed
first, followed by conclusions on health and comfort effects.
7.1 Methodology
The original 32 health symptoms were reduced to 12 factors using an
objective statistical method: PGA. These factors were often different from
those used in previous studies, which have apparently used subjective
groupings of symptoms. For example, some studies have grouped eye, nose, and
throat symptoms into a mucous membrane category; however, the PCA showed
clearly that these three types of symptoms were separate, at least for the EPA
employee population. Some studies have grouped headache, fatigue, and
difficulty concentrating or remembering into one syndrome; however, these
symptoms were also members of three different classes, at least for the EPA
employees. Our logistic regressions have also indicated that each of the
health and comfort factors identified by PCA had a different set of
potentially causative personal and workplace characteristics. This suggests
that regression analyses that lump symptoms from different classes have the
potential for missing significant associations due to misclassification.
Since the original PCA was carried out on the health symptoms occurring
"last year," the analysis was repeated on the questions referring to "last
54
-------
Materials Belong Tof
OPPT Library
401 M Street, SW (TS-793)
week" to determine whether the factors iWtt$Uff£^w&Cl<£Q46&in the same. The
same factors were identified. It remains possible that these factors were
influenced by the ordering of the health symptoms on the page of the
questionnaire. Although the symptoms were not grouped or otherwise labelled
on the questionnaire, they were ordered by body system; that is, most of the
eye symptoms were in proximity to each other, followed by the nasal symptoms,
etc. With some noteworthy exceptions, the PCA factors were very similar to
the a priori ordering. A future study in which questions are randomly ordered
on the questionnaire would be definitive in determining whether the factors
found by the PCA are robust.
PCA analyses were also carried out on 10 comfort questions, 15 odor
questions, and 24 "psychosocial" questions. In the case of the psychosocial
factors, the questions on the questionnaire were shortened versions of
extensively tested groups of questions on five or six psychosocial factors,
such as job satisfaction and conflicting job demands. The PCA analysis
confirmed that the shortened set of questions constituting each of the
psychosocial factors was self-consistent and that each factor was independent
of the others.
7.2 Health Symptom Frequencies
Attempts to compare symptom frequencies in this study to those observed
in other studies of workplace populations proved fruitless because of
different definitions of symptoms. However, a comparison to an earlier study
of a healthy urban population showed that EPA employees reported higher
frequencies of all nine of the symptoms common to both studies.
7.3 Health and Comfort Effects
The workplace variable associated with the largest number of health and
comfort factors was dust. Dust may provide a reservoir of allergens causing
mucous membrane irritation, as indicated by several recent studies. Related
variables were allergies to mold and to dust, strongly associated with some of
the same symptoms. Such allergies may be exacerbated by working in dusty
55
-------
conditions. About 24% of Waterside Mall workers reported dusty workspaces,
compared to 16% of Crystal Mall and 12% of Fairchild building employees.
Nearly a third of people at all three buildings reported that they were
unusually sensitive to a variety of chemical vapors. Such persons were
significantly more likely to report health symptoms and odors. Women were
more likely than men to report sensitivity to chemical fumes (37% vs. 25%),
but the number of associations of this variable with health symptoms and odor
problems was about the same in both sexes.
Glare at the workspace was strongly associated with several health
symptoms, including eye strain and upper body pain. (The latter may be
related to maintaining uncomfortable posture for long periods of time in order
to reduce or avoid glare.)
Among the comfort concerns, complaints of hot, stuffy air were most
likely to be associated with a large number of different health symptoms. Dry
air was also an important variable in many symptoms. Cold air and humid air
were seldom related to health symptoms.
Among the six classes of odors, the odor of paint, carpet cleaning, and
other chemicals was most likely to be associated with health symptoms. Odors
from cosmetics and food were also associated with a number of health symptoms.
The odor of new carpets and drapes was associated with headache. Musty or
damp odors were associated with flu-like symptoms (chills, fever, and muscle
and joint pain). The odor of photocopying processes was seldom associated
with health symptoms. The odor of tobacco smoke was also seldom associated
with any health symptom. Smoking is restricted in all three buildings, and
nicotine measurements showed little indication of cigarette smoke in the 50
locations measured.
Two psychosocial factors were often significantly related to health
symptoms, particularly the cluster of symptoms including difficulty
concentrating, tension, and depression. These factors were workload and
conflicting job demands. Three other psychosocial factors (job satisfaction,
role ambiguity, external stress) were related to one or two health symptoms.
Two remaining psychosocial factors (job control, utilization of abilities)
seldom appeared significant.
Persons in areas where new carpet had been installed in the last year
were more likely to report sore throat, dry throat, and hoarseness. Dizziness
56
-------
was also increased among men in these areas. Men reporting odors of new
carpet were also more likely to report these and other health symptoms.
Although none of these findings were individually incontrovertible, taken as a
whole they suggest that carpet or the carpet installation process (ripping up
old carpet and molding, which would increase dust levels; applying carpet and
molding adhesives, which would increase VOC levels; and installing new carpet
and molding) can be associated with certain health effects among a substantial
number of employees. (These effects are not limited to a small sensitive
group, since those workers who had reacted strongly and were working at
alternative workplaces were not included in this statistical analysis.)
Most of the variables described above were about equally important among
men and women. Some variables were stronger in one sex than in another.
Among women, conflicting job demands, lack of formal education, time at a VDT,
the odor of paint and other chemicals, and being the sole provider were
important indicators for some health symptoms. Young women reported
significantly fewer nasal symptoms and bodily aches and pains. Among men,
access to a window was a significant (protective) variable for three health
factors (fatigue, eye and throat symptoms). Time at the workstation, working
in the EPA Library area, and the odor of new carpets was also related to
several health symptoms among men.
After accounting for the effect of confounding variables, neither
buildings nor building sectors were consistently associated with increased
prevalence of any health symptom. Thus the slightly but consistently elevated
rates for certain buildings or building sectors reported in Volume I (EPA,
1989) are due to variations in the explanatory or confounding variables. For
example, the somewhat lower health symptom prevalence rates reported in the
Crystal building are due in part to the higher percentage of male workers at
Crystal (57% compared to 43% at Waterside): males reported consistently lower
rates than women for all health factors.
Similarly, after accounting for specific variables that vary spatially
(e.g. hot stuffy air), few of the 66 areas identified by their unique
AHU/floor combinations could be associated with health symptoms. The main
exceptions were the Library area of Waterside Mall (AHU 21) , associated with
increased reports among men of chills and fever and muscular aches/upper body
pain; and AHU 16 on the second floor of the mall, associated with eye
57
-------
symptoms, again among men only. Other areas occasionally associated with
health complaints were the garage and lower garage areas of Waterside Mall
(AHU G & LG) . and the Southwest Mall and one nearby location (AHU 14).
On the other hand, certain buildings, sectors, and locations were
strongly associated with some of the comfort and odor variables. For example,
the Waterside building was associated with more complaints of cold drafty air
than either of the other buildings; the print shop in Waterside was associated
with more reports of photocopying and printing process odors than other areas.
Although previous studies have found that the type of office and number
of persons in an office are associated with symptoms, this study found few
such associations.
Our attempt to include questions that would represent a "shelf factor"
and "fleece factor", found to be important in the Danish Town Hall Study (Skov
and Valbjom, 1987) was ineffective in producing relationships with health or
comfort symptoms. Since these factors may be surrogates for dustiness, it
appears that measuring settled dust or asking a direct question on dustiness
of the workplace may be more effective than indirect efforts to estimate the
amount of dust-catching surface.
In previous studies, age has often been found to be a factor related to
health symptoms. However, in this study it never appeared for men, and
infrequently for women. When it did, the direction of the effect was
different than in previous studies: women under 30 had significantly fewer
nasal symptoms and bodily aches and pains.
Unlike observations in some other studies, pay grade was rarely found to
be a significant health factor. Smoking status also did not appear to be an
important factor in this population, although smoking rates were very low.
58
-------
8. RECOMMENDATIONS
The following recommendations are offered for improving working
conditions and for planning future research.
8.1 Improving Working Conditions
The major variable affecting a wide variety of health, comfort, and odor
responses was dust. Dusty working conditions were significantly associated
with increased prevalence of nose and throat irritation, chest tightness,
fatigue, dizziness, and dry skin among both men and women. Dusty, dirty,
cluttered working conditions and lack of regular cleaning and vacuuming were
also identified as major problems in a large number of the essay responses
(Selfridge, 1991) for the Waterside Mall building.
Therefore, the main practical recommendation arising from this study is
to reduce dust in the workplace. One way to do this would be by designing
workplaces to be as dust-free as possible. Replacing open files and bookcases
with the type with retractable shelves might reduce opportunities for dust,
dirt and clutter to collect. Providing storage areas separate from
workstations would reduce dust and clutter near the workers. A public
information campaign informing workers that dusty conditions may be affecting
their health and providing helpful hints on reducing dust and clutter may be
useful. In terms of designing future buildings, a design inhibiting
possibilities for dust to collect and allowing easy maintenance of
workstations is indicated. (One of the many ways to reduce dust accumulation
is by using a "walk-off" grating or carpeted strip leading from the lobby
doorway to the stairs or elevators, that serves as a trap to remove dirt from
shoes. Studies in Germany (Deutsches Teppich-Forschungsinstitut, 1983) have
indicated that such walk-off zones can reduce the daily accumulation of dirt
in the workplace by a factor of 2-3, and can reduce the required frequency of
cleaning or shampooing in the workplace by the same factor.)
A second way to reduce dust is by improved maintenance. Some
respondents mentioned the fact that security regulations based on protecting
confidential business information have made it almost impossible to vacuum
certain areas of Waterside Mall; attention to overcoming these problems
59
-------
without compromising security may be required. However, conventional vacuum
cleaners are least efficient for the particle sizes (<2 Mm) of greatest
allergenic or asthmatic potential (Godish, 1989). Vacuuming may well raise
the airborne levels of these particles for hours following the operation.
Vacuuming after work hours would allow time for airborne particles to settle.
Vacuums with high efficiency (HEPA) filters may collect dust more efficiently.
An alternative to vacuuming is shampooing or steam cleaning of carpets.
Leinster et al. (1990) demonstrated reduced prevalence of dry eyes, stuffy
nose, dry throat, lethargy, and dry skin among employees for two months
following steam cleaning of carpets, wet cleaning of hard surfaces, and HEPA
vacuuming of carpets after steam cleaning. However, care must be taken not to
use too much shampoo or water; dry residues from too much shampoo have been
shown to cause mucous membrane irritation (Kreiss and Hodgson, 1984), and
soaked carpets would provide a mold-growth medium.
A third way to reduce dust is by filtering the air. Although outdoor
air is routinely filtered, major sources of dust also occur indoors, from all
the processes employed in a modern office. Systems supplying additional
ventilation and filtration of office air are available commercially. Some of
these systems employ HEPA filters to remove particles, charcoal filters to
remove VOCs, and powerful fans (>300 cfm) to assure that most of the air in
the breathing zone of workers has been cleaned (although dust generated in the
vicinity of the worker may still be inhaled.) A recent test of such a system
installed on two floors of an office building showed that over a two-month
period both airborne and settled dust was reduced, and various health symptoms
were also reduced among the employees on those floors compared to employees on
two other floors (Hedge, 1991). However, additional testing is necessary to
determine how well such systems perform over extended periods.
A second variable of major impact on certain health symptoms was glare.
Glare was the single most significant variable in the symptoms of aching
muscles and upper-body pain, and was also implicated in headache, fatigue, and
eye irritation symptoms. It seems likely that some of the problems with glare
have been exacerbated by the greatly increased use of computers and video
display terminals. Supplying all display terminals with glare screens could
be effective in reducing problems associated with glare. A "walk-through"
inspection of rooms to identify and eliminate sources of glare is indicated.
60
-------
The third variable widely implicated in health problems was allergy to
molds. Such allergies were very strongly associated with nose and throat
problems, and less strongly with eye irritation and headaches. Molds require
moisture to grow, and they have been found to be plentiful in homes with high
humidity, carpets that have had spills or seepage from water leaks, books and
magazines, cardboard cartons, etc. The essay question in this study elicited
several comments from workers regarding water leaks, overflowing restrooms,
carpets that had been soaked and not removed, etc. It seems possible that
such incidents may have provided the conditions for mold growth and subsequent
health problems of building occupants. Again, a "walk-through" inspection for
molds and replacement of carpets that have been previously soaked may be
called for.
This study provided limited support for the suspicion that the carpet
installation process can cause health problems. Throat symptoms (dry throat,
sore throat, hoarseness) were found to be significantly associated with areas
that had new carpet installed. Dizziness was also reported in areas with new
carpet, but only among men. These findings suggest that carpet be installed
with minimal use of adhesives and careful control of dust, at times when
employees are not present, and perhaps with a waiting period before employees
return.
8.2 Implications for Further Research
It appears that dust is a major factor in building-related health
symptoms. However, it is not known what component of the dust may be the
specific agent. Dust mites, molds, pollen, fibers, pesticides, volatile
organic chemicals, roach allergens, and allergenic macromolecules have all
come under suspicion. Further research to track down the specific agent seems
to be indicated. Methods are needed to measure settled dust in a reproducible
way and to determine the amount of settled dust that gets transferred to the
eyes and nose of the affected worker. Additional comparative studies of
different methods to reduce dust and their effectiveness in reducing health
and comfort complaints are also desirable.
Although this study provided only limited support for the idea that
carpet installation plays a role in employee health symptoms, it would appear
61
-------
that further research is justified. As with dust, the precise agent is not
known. It could be VOC emissions from carpet, from carpet adhesive, or from
molding and molding adhesive (which is usually replaced during installation of
the carpet.) VOCs at high concentrations have been found to cause throat
symptoms and dizziness. Since carpet installation is sometimes associated
with construction, dust or reentrainment of molds or other allergenic
particles could also be involved. Research to identify compounds emitted by
carpets and adhesives and the variation of their source strengths by brand and
type of carpet seems indicated. Research to understand the nature of dust
buildup in carpets and effectiveness of cleaning methods is also desirable.
This study was one of the first to carry out an objective analysis,
using PGA, of a large number of health, comfort, odor, and psychosocial
questions. The principal factors discovered in this way allowed similar
questions to be grouped and a more streamlined statistical analysis to take
place. However, the question of whether the factors found in this study have
universal validity needs to be studied. Since the factors matched the
ordering of the health questions, there is a possibility that some of the
findings are artifices of the questionnaire design. A survey of a different
population using the same questions in random order would clarify this
question. Ultimately, the knowledge of which symptoms group together would
allow a more streamlined questionnaire to be designed.
A related question concerns the design of future questionnaires for
building studies. A number of questions employed in this questionnaire were
highly useful; another group was not useful at all; and some questions were of
marginal use. A careful examination of the effectiveness of each question
needs to be carried out, with the ultimate product being a redesigned
questionnaire for use in future studies. This process may benefit from being
linked with ongoing efforts by the American Society for Testing and Materials
to create a standardized questionnaire (Hedge, 1990).
The data collected in this study will be a rich source of future
studies. Many analyses were suggested during this study that could not be
carried out. One of the most important future research opportunities would be
to repeat the analysis using only work-related symptoms. A comparison of the
two approaches could have considerable impact on future studies. For example,
if the basic findings remain unchanged, it could indicate that future studies
62
-------
need not attempt to determine work-related symptoms. Since most previous
studies have attempted to look at work-related symptoms, but have used
somewhat different definitions, this would provide a point of comparison
between all such studies, as well as those studies that did not use a "work-
related" criterion. Conversely, if large differences are found between the
regression results for work-related and non-work-related symptoms, it would
indicate that more attention needs to be paid to achieving useful and
comparable definitions of what constitutes a work-related symptom.
A second potentially fruitful area of additional statistical analysis
would be the "last week" questions for health and comfort conditions. Since
people's memory over seven days is far better than over a year, more accurate
indications of prevalence rates could be obtained. Also, with respect to the
specific question of whether persons in the newly carpeted areas have
experienced some long-lasting effect on health, a comparison of their most
recent health symptoms to those of persons in other areas would show whether
they are still experiencing elevated rates.
A third analysis could concentrate on the spatial variation of the
variables significantly associated with health: dust, hot stuffy air,
sensitivity to chemical fumes, and a few others. This could assist in
identifying locations where corrective action might have the greatest effect.
A fourth analysis on psychological and organizational factors may also
be useful. Confidentiality provisions required that all administrative
identifiers be removed before analysis; however, if some way were found to
allow organizational analysis without violating these provisions, it might be
possible to identify organizational components where an opportunity for
improving employee well-being and productivity exists.
Future studies to verify the relationships identified in this study
would be desirable. For example, the effect of dust on office workers could
be studied by making objective measurements of productivity before and after
initiating various dust-control measures (supplemental filtered ventilation,
improved cleaning procedures, etc.).
63
-------
9. REFERENCES
Ashford, N. and Miller, C., Chemical Exposures: Low Levels and High Stakes.
Van Nostrand Reinhold, New York, NY, 1991.
Buck, C., "Acute Upper Respiratory Infections in Families," Amer. J. Hygiene
63:1, 1956.
Crandall, M.S., Highsmith, R., Gorman, R., and Wallace, L., "Library of
Congress and U.S. EPA Indoor Air Quality and Work Environment Study:
Environmental Survey Results," in Walkinshaw, D. (ed.) Indoor Air '90:
Proceedings of the 5th International Conference on Indoor Air Quality and
Climate: Toronto. Canada. 29 July - 3 August 1990. Canada Mortgage and
Housing Association, Ottawa, Canada. Vol. 4, p. 597-602, 1990.
Deutsches Teppich-Forschungsinstitut e. V., "Concerning the Effectiveness of
Clean-off Zones", Test Report No. 83 037, Tufton B.V. Textielfabriek KS
Aachen, Germany, 7 July 1983.
Dingle, J.H., Badger, G.F., and Jordan, W.S., Illness in the Home. Western
Reserve Univ. Press, Cleveland, 1964.
EPA, Indoor Air Quality and Work Environment Survey; EPA Headquarters
Buildings. Volume I: Employee Survey. US Environmental Protection Agency,
Washington, DC, 1989.
EPA, Indoor Air Quality and Work Environment Survey: EPA Headquarters
Building. Volume II; Results of Indoor Air Environmental Monitoring Study.
US Environmental Protection Agency, Washington, DC, 1990a.
EPA, Indoor Air Quality and Work Environment Survey: EPA Headquarters
Buildings. Volume III: Relating Employee Responses to the Follow-Up
Questionnaire with Environmental Measurements of Indoor Air Quality. US
Environmental Protection Agency, Washington, DC, 1990b.
Fidler, A.T., Wilcox, T.G., and Hornung, R.W., "Library of Congress Indoor Air
Quality and Work Environment Study: Health Symptoms and Comfort Concerns," in
Walkinshaw, D. (ed.) Indoor Air '90: Proceedings of the 5th International
Conference on Indoor Air Quality and Climate: Toronto. Canada. 29 July - 3
August 1990. Canada Mortgage and Housing Association, Ottawa, Canada. Vol. 4,
p. 603-608, 1990.
Godish, T., Indoor Air Pollution Control. Lewis Publishers, Chelsea, MI, 1989.
Gravesen, S., Skov, P., Valbjorn, 0., and Lowenstein, H. , "The Role of
Potential Immunogenic Components of Dust (MOD) in the Sick Building Syndrome"
in Walkinshaw, D. (ed.) Indoor Air '90: Proceedings of the 5th International
Conference on Indoor Air Quality and Climate: Toronto. Canada. 29 July - 3
August 1990. Canada Mortgage and Housing Association, Ottawa, Canada. Vol. 1,
p. 9-13, 1990.
64
-------
Hall, E.M. and Johnson, J.V., "A Case Study of Stress and Mass Psychogenic
Illness In Industrial Workers," J. OCCUP. Med, 31:243-50, 1989.
Hall, H.I., Leaderer, B.P., Cain, W.S., Selfridge, O.J., Fidler, A.T., and
Wilcox, T., "Subjective and Physical Correlates of Irritation from the Library
of Congress/EPA Indoor Air and Work Environment Study," in Walkinshaw, D.
(ed.) Indoor Air '90: Proceedings of the 5th International Conference on
Indoor Air Quality and Climate: Toronto. Canada. 29 July - 3 August 1990.
Canada Mortgage and Housing Association, Ottawa, Canada. Vol. 4, p. 641-646,
1990.
Hedge, A., "Effect of filtering breathing-zone air in reducing Sick Building
Syndrome complaints," press announcement, Centercore, Inc., Plainfield, NJ,
1991.
Hedge, A., "Questionnaire Design Guidelines for Investigations of 'Sick'
Buildings" in Walkinshaw, D. (ed.) Indoor Air '90: Proceedings of the 5th
International Conference on Indoor Air Quality and Climate: Toronto. Canada.
29 Julv - 3 August 1990. Canada Mortgage and Housing Association, Ottawa,
Canada. Vol. 1, p. 605-610, 1990.
Hosmer, D. and Lemeshow, S., Applied Logistic Regression. Wiley, New York,
1989.
Hurrell, J.J., Sauter, S.L., Fidler, A.T., Wilcox, T.G., and Hornung, R.W. ,
"Job Stress Issues in the Library of Congress/EPA Headquarters Indoor Air
Quality and Work Environment Study," in Walkinshaw, D. (ed.) Indoor Air '90:
Proceedings of the 5th International Conference on Indoor Air Quality and
Climate: Toronto. Canada. 29 July - 3 August 1990. Canada Mortgage and
Housing Association, Ottawa, Canada. Vol. 4, p. 647-652, 1990.
Kalliokoski, P., Reponen, T., Nevalainen, A., and Katila, M.L., "Improper Use
of Ventilation May Cause Severe Health Hazards in Hospitals," in Walkinshaw,
D., (ed) Indoor Air '90: Proceedings of the 5th International Conference on
Indoor Air Quality and Climate: Toronto. Canada. 29 Julv - 3 August 1990.
Canada Mortgage and Housing Association, Ottawa, Canada. Vol. 4, p. 523-28,
1990.
Kreiss, K. and Hodgson, M., "Building-Associated Epidemics," Chapter 6 in
Walsh, P., Dudney, C., and Copenhaver, E. (eds.) Indoor Air Quality. CRC
Press, Boca Raton, FL, 1984.
Leaderer, B., Wilcox, T., Fidler, A. Selfridge, J., Hurrell, J. Kollander, M. ,
Clickner, R., Fine, L., and Teichman, K., "Protocol for a Comprehensive
Investigation of Building Related Complaints," in Walkinshaw, D., (ed.) Indoor
Air '90: Proceedings of the 5th International Conference on Indoor Air
Quality and Climate: Toronto. Canada. 29 July - 3 August 1990. Canada
Mortgage and Housing Association, Ottawa, Canada. Vol. 4, p. 609-614, 1990.
65
-------
Lebowitz, M.D., Cassell, E., and McCarroll, J., "Health and the Urban
Environment. XI. The Incidence and Burden of Minor Illness in a Healthy
Population: Methods, Symptoms, Incidence" Amer. Rev. Reso. Pis.. 105:824-34,
1972a.
Lebowitz, M.D., Cassell, E. and McCarroll, J., "Health and the Urban
Environment. XII. The Incidence and Burden of Minor Illness in a Healthy
Population: Duration, Severity, and Burden" Amer. Rev. Resp. Pis. 105:835-41,
1972b.
Leinster, P. Raw, G. , Thomson, N. , Leaman, A. and Whitehead, C. , "A Modular
Longitudinal Approach to the Investigation of Sick Building Syndrome," in
Walkinshaw, D. (ed.) Indoor Air '90: Proceedings of the 5th International
Conference on Indoor Air Quality and Climate: Toronto. Canada. 29 July - 3
Aueust 1990. Canada Mortgage and Housing Association, Ottawa, Canada. Vol. 1,
p. 287-292, 1990.
Melhave, L. (1987) "The Sick Buildings--A Sub-Population among the Problem
Buildings?" in Seifert, B., (ed.), Indoor JU.r'g7; Proceedings of the 4th
International Conference on Indoor Air Quality and Climate. Institute of Air,
Water, and Soil Hygiene, Berlin, Germany, Vol. 2, p. 469-474, 1987.
Morey, P. and Williams, C., "Porous Insulation in Buildings: A Potential
Source of Microorganisms," in Walkinshaw, D. (ed) Indoor Air '90: Proceedings
of the 5th International Conference on Indoor Air Quality and Climate:
Toronto, Canada. 29 July - 3 August 1990. Canada Mortgage and Housing
Association, Ottawa, Canada. Vol. 4, p. 529-534, 1990.
NCHS, Current Estimates from the National Health Interview Survey for1987.
National Center for Health Statistics, Hyattsville, MP, 1988.
NCHS, Current Estimates from the National Health Interview Survey for 1988.
National Center for Health Statistics, Hyattsville, MP, 1989.
Nelson, C.J., Leaderer, B. P., Teichman, K., Wallace, L., Kollander, M., and
Clickner, R.P., "Environmental Protection Agency Indoor Air Quality and Work
Environment Study: Health Symptoms and Comfort Concerns," in Walkinshaw, P.
(ed.) Indoor Air '90: Proceedings of the 5th International Conference on
Indoor Air Quality and Climate: Toronto. Canada. 29 July - 3 August 1990.
Canada Mortgage and Housing Association, Ottawa, Canada. Vol. 4, p. 615-620,
1990.
Norback, D. and Torgen, M., "A Longitudinal Study Relating Carpeting with Sick
Building Syndrome," Environ. Int. 15:129-135, 1989.
Norback, P., "Environmental Exposures and Personal Factors Related to Sick
Building Syndrome", Ph.P. dissertation, Uppsala University, Sweden, 1990.
66
-------
Persily, A. and Dols, W.S., "Ventilation and Air Quality Investigations of the
Madison Building: Preliminary Results," in Walkinshaw, D. (ed.) Indoor Air
IlOj Proceedings of the 5th International Conference on Indoor Air Quality
and Climate: Toronto. Canada. 29 July - 3 August 1990. Canada Mortgage and
Housing Association, Ottawa, Canada. Vol. 4 p. 621-626, 1990.
Preller, L. , Zweers, T., Brunekreef, B. and Boleij, J., "Sick Leave Due to
Work-Related Health Complaints Among Office Workers in the Netherlands," in
Walkinshaw, D. (ed) Indoor Air '90: Proceedings of the 5th International
Conference on Indoor Air Quality and Climate: Toronto. Canada. 29 July - 3
August 1990. Canada Mortgage and Housing Association, Ottawa, Canada. Vol. 1,
p. 227-230, 1990.
Raw, G. , Roys, M. , and Leaman, A., "Further Findings from the Office
Environment Survey: Productivity," in Walkinshaw, D. (ed) Indoor Air '90:
Proceedings of the 5th International Conference on Indoor Air Quality and
Climate: Toronto. Canada. 29 July - 3 August 1990. Canada Mortgage and
Housing Association, Ottawa, Canada. Vol. 1 p. 231-236, 1990.
SAS, Statistical Applications Manual. SAS Institute, Gary, NC, 1989.
Selfridge, O.J., Berglund, L.G., and Leaderer, B.P., "Thermal Comfort
Dissatisfaction Responses in the Library of Congress and Environmental
Protection Agency Indoor Air Quality and Work Environment Study," in
Walkinshaw, D. (ed.) Indoor Air '.90; Proceedings of the 5th International
Conference on Indoor Air Quality and Climate:Toronto. Canada. 29 July - 3
August 1990. Canada Mortgage and Housing Association, Ottawa, Canada. Vol. 4,
p. 665-670, 1990.
Selfridge, O.J., Snjimmyy of Responses to the Essay Question of the Employee
Survey Questionnaire. Supplement to Volume I of the Indoor Air Quality and
Work Environment Study: EPA Headquarters Buildings. EPA, Atmospheric Research
and Exposure Assessment Laboratory, Research Triangle Park, NC, 1991.
Skov, P. and Valbjorn, 0., "The 'Sick' Building Syndrome in the Office
Environment: The Danish Town Hall Study" in Seifert, B. (ed). Indoor Air '87:
Proceedings of the 4th International Conference on Indoor Air Quality and
Climatef Institute of Air, Water, and Soil Hygiene, Berlin, Germany, 1987.
Skov, P., Valbjorn, 0., Pedersen, B.V., and the Danish Indoor Climate Study
Group, "Influence of Personal Characteristics, Job-related Factors, and
Psychosocial Factors on the Sick Building Syndrome," Scand. J. Work Environ
Health 15:286-295, 1989.
Skov, P., Valbjorn, 0., Pedersen, B.V., and the Danish Indoor Climate Study
Group, "Influence of Indoor Climate on the Sick Building Syndrome in an Office
Environment," Scand. J. Work Environ. Health 16:363-71, 1990.
Wilson, S. and Hedge, A., The Office Environmental Survey: A Study of
Building Sickness. Building Use Studies, Ltd., London, England, 1987.
67
-------
Wynder, E.L., Hlggins, I.T., and Harris, R.E., "The Wish Bias," J. Clin.
Epidemiol. 43:619-21, 1990.
Zweers, T, Preller, L., Brunekreef, B., and Boleij, J., "Relationships Between
Health and Indoor Climate Complaints and Building Workplace, Job and Personal
Characteristics," in Walkinshaw, D. (ed) Indoor Air '90: Proceedings of the
5th International Conference on Indoor Air Quality and Climate: Toronto.
Canada. 29 July - 3 August 1990. Canada Mortgage and Housing Association,
Ottawa, Canada. Vol. 1, p. 495-500, 1990.
68
-------
Table 1.
FREQUENCIES BY GENDER FOR HEALTH SYMPTOMS EXPERIENCED LAST YEAR
Health Symptom
Headache
Nausea
Runny nose
Stuffy nose/sinus
congestion
Sneezing
Cough
Wheezing or whistling
in chest
Shortness of breath
Chest tightness
Dry, itchy, or tearing
eyes
Sore/strained eyes
Blurry, double vision
Burning eyes
Sore throat
Sex
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
Percent of Respondents Who
Experienced Symptom Last Year:
Jf Never Rarelv Sometimes Often Alwavs
1901
2031
1891
2015
1891
2015
1892
2023
1892
2020
1892
2024
1886
2020
1890
2021
1885
2020
1895
2021
1889
2021
1889
2019
1892
2020
1889
2024
13
7
60
48
15
15
12
12
14
14
20
18
68
68
69
61
71
67
32
25
26
25
67
56
46
37
26
27
37
21
30
32
34
27
25
19
41
31
47
37
23
20
20
21
19
18
24
18
26
18
19
19
24
21
44
35
38
44
8
17
37
37
37
33
37
41
28
34
7
9
8
13
8
11
29
31
32
33
10
17
21
26
25
30
12
25
1
3
12
16
21
27
7
12
5
9
2
2
2
4
2
3
13
21
13
20
3
7
8
12
5
8
1
3
0
0
1
5
5
9
1
2
1
2
0
1
0
1
0
0
2
5
2
5
1
2
1
4
0
1
(continued)
69
-------
Table I (continued)
Health Symptom
Hoarseness
Dry throat
Unusual fatigue or
tiredness
Sleepiness or drowsiness
Chills
Fever
Aching muscles or joints
Problems w/contact
lenses
Difficulty remembering
things
Dizziness/light
headedness
Feeling depressed
Tension or nervousness
Difficulty concentrating
Dry or itchy skin
Sex
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
Percent of Respondents Who
Experienced Symptom Last Year:
N JJever Rarelv Sometimes Often Alwavs
1889
2020
1885
2022
1895
2022
1892
2024
1893
2027
1891
2024
1893
2026
1856
1999
1893
2018
1892
2020
1895
2020
1891
2020
1895
2014
1889
2019
50
45
35
29
28
18
16
15
56
45
53
57
42
39
84
76
52
• 45
59
47
39
32
27
27
30
30
41
31
34
30
33
26
29
20
28
21
30
26
40
33
32
24
6
5
23
24
27
25
32
29
29
23
31
26
20
17
12
19
23
27
30
35
40
40
12
22
6
10
20
24
6
9
21
23
12
22
23
28
33
35
31
33
23
26
3
4
8
13
11
21
13
19
2
6
1
1
5
10
3
7
3
6
2
5
5
9
9
13
7
9
12
18
1
1
1
4
2
6
2
5
0
2
0
0
1
3
2
4
1
2
0
1
1
2
1
2
1
2
4
9
(continued)
70
-------
Table 1 (continued)
Health Symptom
Sex
N
Percent of Respondents Who
Experienced Symptom Last Year:
Never Rarely Sometimes Often Always
Pain or stiffness in
upper back
Pain or stiffness in
lower back
Pain or numbness in
hands or wrists
Pain or numbness in
shoulder/neck
M
F
M
F
M
F
M
F
1891
2021
1891
2018
1889
2019
1892
2021
52
43
41
39
72
63
58
48
25
19
25
20
17
16
20
18
17
24
24
27
8
14
16
21
6
11
8
10
3
6
6
11
1
3
2
4
1
2
1
3
71
-------
Table 2. PERSONAL AND WORKSPACE CHARACTERISTICS, BY GENDER (%)
Characteristics
Personal
Age:
Pay grade:
College degree
Smoke
Wear glasses
Wear contact lens
Allergies :
Asthma
Eczema
Live alone
< 30
30-40
>40
< GS 9
GS 9-12
> GS 12
Dust
Molds
Pollen
Animals
Live w/others--sole provider
Use regularly:
Worked at copier:
Fan
Heater
Desk lamp
None
<1 hour
>1 hour
Worked at computer : None
<1 hour
>1 hour
Worked with chemicals
Especially sensitive to chemical fumes
Female
24
40
36
35
30
35
52
18
56
23
46
30
39
17
9
8
25
12
44
22
43
14
71
15
13
17
70
24
37
Male
8
37
55
6
18
76
94
12
65
16
41
30
46
17
9
8
15
26
37
9
36
22
73
5
19
21
60
11
25
(continued)
72
-------
Table 2 (Continued)
Characteristics Female Male
Workspace
Private office 21 46
Private cubicle or share with one person 39 40
Open area or shared with two or more 40 14
Too dusty 56 40
Too much glare 53 42
Too noisy 55 55
Too dark 7 5
New carpet in office within past year
Respondents 17 17
Building records 17 16
New carpet in corridors (building records)
Glued down 31 30
Tacked down 8 8
Near:
Fabric-covered partitions 45 44
Laser printer 41 29
Photocopier 15 11
Live plants 53 39
Window visible 38 49
New furniture in past year 44 37
New equipment in past year 64 59
Walls recently painted 22 19
Walls recently rearranged 20 16
Water leaks last year 12 13
73
-------
Table 3.
INDEPENDENT VARIABLE DEFINITIONS AND SOURCE8 OF INFORMATION
Personal Characteristics
JOB SATISFACTION index of job satisfaction from IV. 1
CONFLICTING JOB DEMANDS index of conflicting demands from IV.4
JOB CONTROL index of control over work from IV.5
WORKLOAD index of workload from IV.6.a-d
UTILIZE SKILLS index of skill utilization from IV.6e-g
ROLE AMBIGUITY index of job definition from IV.6.h-k
(Above six "psychosocial" variables constructed according to
principal components analysis of Part IV of the main
questionnaire)
MEDIUM PAY
HIGH PAY
EXTERNAL STRESS
CONTACT LENS
GLASSES
ASTHMA
COLLEGE GRAD
ALLERGIC TO PLANTS
ALLERGIC TO ANIMALS
ALLERGIC TO DUST
ALLERGIC TO MOLDS
ALLERGIC: OTHER
CHEM SENSITIVITY
ECZEMA
LIVE ALONE
pay grades GS9-12 from EPA personnel records
pay grades GS13 and above from personnel records
index of family or home stress from IV.7
wears contact lenses at work from Il.l.b
wears glasses at work from II.2
doctor-diagnosed asthma from II.16.a
college graduate from V.3
allergic to pollen or plants from II.18.a
allergic to animals from Il.lS.b
allergic to dust from II.lS.c
allergic to molds from Il.lS.d
allergic to other allergens from Il.lS.e
sensitive to chemical fumes from 11.20
doctor-diagnosed eczema from 11.14
living alone from V.2
(continued)
74
-------
Table 3 (Continued)
SOLE PROVIDER living with others and sole provider from V.2
NOSMOKE current nonsmoker from II.3
LESS THAN 30 under 30 years of age from 11.21
OVER 40 over 40 years of age from 11.21
(Reference variable for above two groups are people between the
ages of 30 and 40 inclusive)
Workstation Characteristics
HOURS AT WORK hours spent per day at workstation from I.4.b
SHELF INDEX surface area index of office furnishings from 1.7: sum
of "yes" responses of items I.7.a-g (metal desk, wood
desk, metal bookshelves, wood bookshelves, file
cabinets, other metal furniture, other wood furniture)
WOOD INDEX wood present in office furnishings from 1.7: sum of
"yes" responses of items I.7.b, I.7.d, and I.7.g( wood
desk, wood bookshelves, and other wood furniture)
FABRIC PARTITIONS presence of fabric-covered partitions from I.7.h
LASER PRINTER presence of laser printer from I.7.J
PHOTOCOPIER NEAR presence of photocopy machine from I.7.k
PLANTS NEARBY presence of live plants from 1.7.1
HOURS AT VDT hours at work on computer or word processor from 1.9. a
1 HOUR AT COPIER one hour of work on copying machine from I.9.b
>1 HOUR AT COPIER over one hour of work on copying machine from 1.9.a.
(No work on copying machine was the reference group)
USED CHEMICALS used chemicals from I.9.e
USE FAN regular use of portable fan from I.10.a
USE HEATER regular use of portable heater from I.lO.c
USE LAMP regular use of desk lamp from I.lO.d
(continued)
75
-------
Table 3 (Continued)
NEW CARPET/DRAPES presence of new drapes or new carpets from I.ll.a-b
(sun of the two Items)
NEW FURNITURE
NEW EQUIPMENT
WALLS PAINTED
WALLS REARRANGED
WATER LEAKS NEAR
WINDOW NEARBY
LIGHTS TOO DIM
GLARE
NOISE
DUST
UNCOMNDX
OWNRMPVT
RMGTICUB
CAPACITY
presence of new furniture from I.ll.c
presence of new equipment from I.ll.d
walls recently painted from I.ll.e
walls recently rearranged from I.ll.f
water leaks In the last year from 1.12
presence of outside window from III.6
lights too dim from III.4
experience reflection or glare at workstation from
III.5.a. (1 if sometimes, often, or always, else 0)
work environment too noisy from III.l.k.1. (1 if
sometimes, often, or always, otherwise 0)
work environment too dusty from III.l.ra.1. (1 if
sometimes, often, or always, otherwise 0)
an index of workstation uncomfortableness ranging from
0 to 4 formed by combining responses from III.7.a and
III.T.b
a binary variable indicating private office from I.I.a
and I.l.b
a binary variable indicating either the sharing of an
enclosed office with Just one other person or working
in a private cubicle with half-height partitions from
I.I.a. and I.l.b. The remaining work space/space
sharing configurations served as the reference group.
a continuous variable associating a maximum per capita
airflow capacity (cfmp) with each person based on the
measured or rated airflow capacity of a given air
handling unit (AHU) and the number of persons served
by that AHU. Used only in third linear regression.
(continued)
76
-------
Table 3 (Continued)
flew Carpet Variables
The following four variables were constructed from information in a
memorandum from EPA management to EPA unions identifying the areas where new
carpet was installed in the winter of 1987-88.
GDOFFICE a binary (0,1) variable Indicating the presence of
recent glued down carpet in the respondent's office
GDCORRID a binary variable indicating the presence of recent
glued down carpet in respondent's corridor (GDOFFICE
took precedence over GDCORRID)
TSOFFICE a binary variable indicating the presence of recent
tacked down carpet in the respondent's office
TSCORRID a binary variable indicating the presence of recent
tacked down carpet in the respondent's corridor
(TSOFFICE took precedence over TSCORRID). No recent
carpet installation was the reference group.
Location Variables
L - a set of 8 variables representing two buildings and seven
"sectors" in the Waterside Hall building. (L1-L8: Crystal Hall
building, Fairchlld building, six sectors in Waterside Hall
including East and West Towers, Northeast and Southeast Hall, 2nd
and 3rd floors of the Hall). This "coarse-grained" spatial
variable used only in first linear regression.
AHU - a set of 66 indicator variables representing the particular air
handling unit serving a respondent. In the case of air handling
units serving several floors, an additional subdivision was
created Identifying the floor of the respondent. This set of
"fine-grained" spatial variables used in second linear regression
and in all logistic regressions.
* For most variables, the source of information was the questionnaire. The
exact wording of the question may be determined by consulting Appendix A.
Questions are identified by Roman numeral for Parts I-V of questionnaire;
Arabic numeral for question number; small letter(s) for subpart(s) of
question. Example: IV.l.a-d - Part IV, question 1, subparts a through d.
For the remaining variables, the main sources of information were EPA
personnel records (payroll and directory information) and management-union
records on locations of air handling units and new carpet installation.
77
-------
Table 4. QUESTIONS THAT WERE NOT INCLUDED IN THE ANALYSIS OF THE FIRST QUESTIONNAIRE
QUESTION
PART I
2. years of service with EPA
3b. hours per week in building
5. days in building last week
6. time of arrival at work
8. carpeting on floor of workstation
PART II
oo 4. smoke tobacco at workstation
5. smoke elsewhere at work
8. reduced ability to work
9. stayed home or left home early
10. seasons most bothered
lla. associate symptoms with work
lib. symptoms changed over past year
12. wheeze, cough, SOB, fever, chills
13. chest illnesses: missed work
15. wheezing episodes without fever
REASON FOR NOT INCLUDING
May not have been in same building
Little variation
Not related to last year symptoms
Little relation to health
97% had carpeting—no variation
No variation—smoking not allowed
Almost no variation
Could be dependent variable
Could be dependent variable
Not clear how to include in analysis
Not clear how to use in analysis
Cannot use in regression analysis
Unexpectedly high prevalence rate—unusable
Could be a dependent variable
Low prevalence rate
(continued)
-------
Table 4. (continued) QUESTIONS NOT INCLUDED IN THE ANALYSIS OF THE FIRST QUESTIONNAIRE
QUESTION
17. frequency of infections
19. sources of eye, nose, throat irritation
23-31. gynecological questions
PART III
1.1. too quiet
3. seasons you would like to adjust
physical conditions near workstation
9a. times per week went outdoors
9b. primarily to get fresh air
10. satisfaction with physical
conditions at workstation last week
11. satisfaction with overall conditions
at workstation last year
12. has environment improved in last year
13. does environment improve during the day
PART V
1. day of week survey completed
5. room number; phone number
6. essay question
REASON FOR NOT INCLUDING
Could be a dependent variable
Could be a dependent variable
Requires separate analysis
Few positive responses
Cannot use in regression analysis
May be a dependent variable
Modifies 9a.
Not useful in last year analysis
Could be dependent variable
Could be dependent variable
Not clear how to interpret
No relation to symptoms
Violates confidentiality
Not analyzable by regression
-------
Table 5. AIR HANDLING UNITS AT WATERSIDE MALL, AREAS SERVED, AND MEASURED OR ESTIMATED
SUPPLY AIR FLOW RATE CAPACITIES IN CUBIC FEET PER MINUTE (CFM)
00
o
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
ID
1
4
5
6
8
9
12
13
2
3
10
14
16
18
19
20
21
A
B
C
D
E
F
G
H
15
SW-2
Service
Sector
M3
M3
M3
M3
M3
M3
M3
M3
M2
M2
M2
M2
M2
M2
M2
M2
M2
M2
M2
M2
M2
M2
M2
M2
M2
SW,M2
SW
Area
Floor
3
3
3
3
3
3
3
3
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
Sample
Room
3402
3234
3201
3002
3702A
3702C
3600
3100
2107
2501
2701
2102
2817
2821
2603
2632
2902
2411
2123
2322
2517
2435
2522
2615
2612
2
256
Supply
(CFM)
15809
16105
9636
28440
11329
17635
11422
12036
10095
12566
7644
10282
10780
14400
7562
8921
15574
2100b
2255
2100b
2100b
3285
?
1650
1300
27868
6038
Air Capacity8
(Tons) Occupants
108
85
142
139
60
73
57
59
31
24
26
109
68
92
19
32
49
7
8C
14C
3
24
0
9c
12C
114
15 8
CFM
Occ.
146
189
68
205
189
242
200
204
326
524
294
94
159
157
398
279
318
300
282
150
700
137
— _
183
108
244
755
-------
Table 5.(cont.)
AIR HANDLING UNITS AT WATERSIDE MALL, AREAS SERVED, AND SUPPLY AIR
FLOW RATE CAPACITIES IN CUBIC FEET PER MINUTE (CFM)
No.
28
29
30
31
32
33
34
35
36
37
oo 38
*- 39
40
41
42
ID
SE2-1
SE2-2
SE3-1
SE3-2
NEG-1
NEG-2
NEG-3
NEG-4
NE1
NE2
NE3-1
NE3-2
LG-1
G-ld
ETB
Service
Sector
SE
SE
SE
SE
NE
NE
NE
NE
NE
NE
NE
NE
MLG
MG
ET
Area
Fl
2
2
3
3
B
B
G
G
1
2
3
3
LG
G
B
Sample
Room
214
256
300
312
1
15
4
13
102
202
300
308
LG100
G100
55
Supply
(CFM)
25325
3605
10665
12700
12800
12800
16000
8000
20800
20800
16000
16000
8000
9600
7219
Air Capacity
(Tons)
85
10
45
45
40
40
50
25
65
65
50
50
25
30
Occupants
197
30
85
83
11
46
38
49
160
192
58
84
66
133
38
CFM
Occ.
129
120
125
153
1160
278
421
163
130
108
276
190
121
72
190
a Supply air capacities in CFM for 31 AHUs provided from measurements in March, 1989;
design capacities for remainder (in tons) are based on information from M. Shamus, EPA.
Translation from tons to CFM for AHUs # 32-42 is calculated as one ton = 320 CFM.
b These capacities calculated by averaging the capacities of units B,D,G, and H.
c Number of people returning questionnaires, when this number is larger than the number
given by the EPA Directory.
d The sum of two air handling units, each of 15-ton capacity.
-------
Table 6. ESTIMATED SUPPLY AIR FLOW RATES AND NUMBER OF OCCUPANTS SERVED BY AIR
HANDLING UNITS ON EACH FLOOR OF THE TOWERS, FAIRCHILD, AND CRYSTAL MALI
£L
1
2
3
4
5
6
7
8
9
10
11
12.
Tot
Number of Occupants on Each Floor
East Tower West Tower
ET1 ET2 HI WT1 WT2
14 30 ---
30 36 --- 20 27
52
40
38
28
41
35
17
30
31
15
367
Supply Air
Tower
WT1
ET3
, East
6564
3002
37
30
24
19
29
42
30
18
35
12
19
32
30
40
17
49
32
27
19
342 265
Flow Rates
Tower -
d-12)
(4-11)
3,
Estimated Per Capita
FJ^
1
2
3
4
5
6
7
8
9
10
11
12
ET1
357
196
148
116
119
154
118
123
256
156
132
292
ET2
158
141
185
140
172
207
152
93
132
237
111
324
ET3
158
94
100
75
176
61
94
111
158
27
39
38
20
33
23
25
34
23
16
298
35
33
34
42
41
33
42
37
30
18
WT3
14
25
41
33
26
17
5
4
_o
370 165
by Floor (cfm) Estimated from
and Fairchild
WT2 5062
CM 17812
and Crystal Mall
d-12)
(2-11)
WT3
FC
Measured Supply Air
Capacities (cfm)
ET1 ET2
5000 4750
5890 5090
7670
4630
4540
4320
4840
4290
4360
4690
4100
4380
58710
Design
6855
4213*
4131*
3931*
4404*
3904*
3968*
4268*
3870
3890
53470
Capacities for
West
Air Handling Units:
4121
12645
(4-11)
d-8)
Supply Air Flow Rates (cfmp) :
WT1
328
274
168
173
328
199
285
263
193
285
410
WT2
187
145
153
149
120
123
153
121
137
169
281
WT3
294
165
100
125
159
242
824
999
—
Crystal (CM- 2)
Floor
2
7
8
10
11
No.
Occ.
115
128
138
127
94
Per Capita
Flow
Rate
155
139
129
140
189
Fairchild (FC)
2
3
5
6
7
8
106
116
106*
83
105
123
127
116
113
163
128
110
* Calculated values
82
-------
Table 7. PRINCIPAL COMPONENTS ANALYSIS OF HEALTH SYMPTOMS
FACTOR (PERCENT OF
VARIANCE
PAIN
CENTRAL
NERVOUS
SYSTEM
(CNS)
EYES
NOSE
CHEST
THROAT
FLU
FATIGUE
HEADACHE
EXPLAINED)
(9.9)
(9.8)
(8.9)
(8.2)
(7.7)
(7.0)
(5.6)
(4.8)
(4.3)
HEALTH SYMPTOM INCLUDED" IN FACTOR
aching muscles or Joints
upper back pain or stiffness
lower back pain or stiffness
shoulder/neck pain or numbness
hand/wrist pain or numbness
difficulty remembering
feeling depressed
tension or nervousness
difficulty concentrating
dry. itching, or tearing eyes
sore/strained eyes
blurry/double vision
burning eyes
runny nose
stuffy nose/sinus congestion
sneezing
cough
wheezing
shortness of breath
chest tightness
sore throat
hoarseness
dry throat
chills
fever
unusual fatigue or tiredness
sleepiness or drowsiness
headache
nausea
QUESTION
NUMBERb
u
cc
dd
ee
ff
w
y
z
aa
j
k
1
m
c
d
e
f
g
h
1
n
o
P
s
t
q
r
a
b
DIZZY (not selected)
DRY SKIN (not selected)
CONTACTS (not included)
dizziness
dry skin
problems with contact lenses
x
bb
8 Factor loading > 0.5
b Part II, Question 7, symptoms experienced "last year" in
building
83
-------
Table 8. PRINCIPAL COMPONENTS ANALYSIS OF COMFORT CONCERNS
FACTOR (PERCENT OF
VARIANCE EXPLAINED')
HOT STUFFY AIR (28)
DRY AIR
(20)
COLD DRAFTY AIR (16)
HUMID AIR
(10)
QUESTION
COMFORT CONCERN INCLUDED8 IN FACTOR NUMBERb
too little air movement b
want to adjust air movement c
too hot d
want to adjust temperature £
too stuffy j
too dry h
want to adjust humidity i
too much air movement a
too cold e
too humid g
8 Factor loading > 0.5
b From Part III, Question 1, conditions during last year
84
-------
Table 9. PRINCIPAL COMPONENTS ANALYSIS OF ODORS
FACTOR (PERCENT OF
VARIANCE EXPLAINED) ODOR INCLUDED8 IN FACTOR
PAINT (15)
COSMETICS (13)
COPIER (12)
CARPET (12)
MUSTY (10)
SMOKE (6.7)
chemicals (glue, cleansers...)
pesticides
cleaning of carpets , drapes . . .
paint
body odor
cosmetics (perfume, af ter- shave. .)
food smells (not including fish)
photocopying machine
printing press, binding
new carpet
new drapes or curtains
fishy smells
musty or damp basement smells
diesel or engine exhaust
tobacco smoke
QUESTION
NUMBER6
1
m
n
o
a
b
e
j
k
g
h
d
f
i
c
8 Factor loading > 0.5
b From Part III, Question 2, odors during last year at workstation
85
-------
Table 10. PRINCIPAL COMPONENTS ANALYSIS OF PSYCHOSOCIAL VARIABLES
FACTOR (PERCENT OF
VARIANCE EXPLAINED')
ROLE AMBIGUITY
(13)
WORKLOAD
(13)
JOB SATISFACTION
(ID
JOB CONTROL (9.3)
PSYCHOSOCIAL VARIABLE QUESTION
INCLUDED" IN FACTOR NUMBERb
clear on job responsibility 6h
can predict others' expectations 6i
work objectives well defined 6j
clear about others' expectations 6k
must work very fast 6a
must work very hard 6b
little time to get things done 6c
a great deal to be done 6d
satisfied with job la
would take same job if offered Ib
would take same job over any other Ic
would recommend job to friend Id
influence over amount of work 5a
influence over materials needed 5b
influence on policy, procedures 5c
influence on work environment 5d
CONFLICTING DEMANDS
(8.9)
UNDERUTILIZATION
OF SKILLS (8.5)
SALARY SATISFACTION0
(6.1)
bosses request conflicting tasks 4a
others request conflicting tasks 4b
requests conflict with work 4c
can use skills from school 6e
can do things one does best 6f
can use skills from experience 6g
satisfied with salary 2
satisfied with oppor. to advance 3
a Factor loading >0.5
b From Part IV, Questions 1-6
c This factor not used in analysis
86
-------
Table 11. NUMBER OF RESPONDENTS ASSIGNED TO EACH CELL IN THE BINARY VARIABLES
CREATED FOR THE LOGISTIC ANALYSIS
Dependent Variable Females Males
no ££s no yes
HEALTH SYMPTOMS
headache and nausea 1273 742 1568 322
stuffy nose/sinus congestion, runny nose 1447 558 1553 324
wheezing, shortness of breath 1704 308 1713 169
dry, itching, tearing eyes 1553 456 1648 232
sore throat, hoarseness, dry throat 1530 489 1604 276
sleepiness, unusual fatigue 1648 372 1705 183
chills and fever 1370 653 1580 308
aching muscles, back pain, neck pain 1609 399 1687 194
difficulty concentrating, depression, tension 1508 497 1544 337
dizziness 1449 571 1625 267
dry itchy skin 1487 532 1585 304
problems with contact lenses 301 176 225 80
COMFORT CONCERNS
hot stuffy air 1291 611 1396 443
dry air 1225 709 1384 479
cold drafty air 1048 896 1295 578
humid air 1360 597 1452 414
ODORS
fresh paint, carpet cleaning, other chemicals 1350 614 1471 381
cosmetics, food, body odor 1478 491 1626 248
photocopies, printing process 1758 220 1713 168
new carpet or drapes 1640 336 1564 312
musty/damp odor, engine exhaust, fishy smell 1498 480 1472 400
tobacco smoke 1687 303 1646 237
87
-------
Tables 12-33. RESULTS OF LINEAR AND LOGISTIC REGRESSIONS
The following tables present the results of the second set of linear
regressions side by side with the results of the logistic regressions. In
both sets of regressions, the dependent variables were 12 health factors, 4
comfort factors, and 6 odor factors. Thus there are 22 tables in all.
For the linear regressions, the independent variables included 56
questionnaire items dealing with personal and workspace characteristics, 66
variables representing air handling units or other spatial variables such as
floor within a building or tower, and four variables representing the carpet
installed the previous year--a total of 126 independent variables. The 10
comfort and odor variables were added to these 126 independent variables in a
separate set of regressions carried out on the 12 health factors. Both sets
were run for males and females separately.
For the logistic regressions, only those variables that had appeared at
the 0.05 level of significance in the linear regressions were included in the
model. Thus instead of 126 or 136 independent variables, most of the logistic
regressions included between 20 and 40 independent variables. The identical
procedure of running separate regressions by gender and including or excluding
the comfort and odor factors as independent variables for the 12 health
factors was followed.
Thus for each of the 12 health factors, 8 separate regressions were run
(linear/logistic, male/female, with/without the comfort and odor factors as
independent variables) . For each of the 10 comfort and odor factors
considered as dependent variables, 4 regressions (linear/logistic,
male/female) were run.
All independent variables that appeared at the 0.001 level of
significance for at least one regression are listed with a "+++" or " —"
sign, depending on the direction of the effect. If the level of significance
was between 0.001 and 0.01, the variables are marked with "++" or "--". If
any of these variables appeared in any other regression at a significance
level between 0.05 and 0.01, it was marked with a single "+" or "-".
For each health symptom, the personal and workplace variables have been
presented first, followed by the comfort and odor variables (the latter
appeared in only half of the regressions). For those interested in the
88
-------
effects of the spatial variables, the air handling unit (AHU) variables have
been separated from the other workplace variables.
To make it easier to identify the most important explanatory factors,
those with the largest number of significant associations are listed first.
89
-------
Table 12. HEADACHE AND NAUSEA
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE FEMALE MALE FEMALE MALE
LIN LOG LIN LOG LIN LOG LIN LOG
SENSITIVE TO CHEMICALS
WORKLOAD
GLARE
MOLD ALLERGIES ++ -H- +
DUST ++ + -H- ++
NOISE ++ +++ ++ +
EDUCATION -
EXTERNAL STRESS -H- + +
NONSMOKER -H- + +
ROLE AMBIGUITY ++ + +
AHU WT1FL67 --
AHU CM7 +
AHU ET1FL67 +
Comfort and Odor Variables
HOT, STUFFY AIR +++
DRY AIR + +
ODOR OF PAINTS, CARPET CLEANING +++ 4+ +
ODOR OF NEW CARPETS ++
LIN - Linear least-squares regression
LOG - Logistic regression
+++, - Significant at p < 0.001
++,— - Significant at 0.001 < p < 0.01
+,_ - Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
90
-------
Table 13. STUFFY NOSE/SINUS CONGESTION, RUNNY NOSE, SNEEZING, COUGH
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE FEMALE MALE FEMALE MALE
LIN LOG LIN LOG LIN LOG LIN LOG
ALLERGY TO DUST
DUST
ALLERGY TO MOLDS
ALLERGY TO POLLEN
LESS THAN 30
CONFLICTING DEMANDS +++ ++ -H- + NI + NI
GLASSES -H- NI NI -H- ++
HOURS AT COMPUTER ++ + + +
SENSITIVE TO CHEMICALS + ++ ++
AHU 4 -H- + + +
AHU 13 + ++
AHU 20 -H-
Comfort and Odor Variables
HOT STUFFY AIR
DRY AIR
COLD AIR
ODOR OF PAINT AND CARPET CLEANING
LIN - Linear least-squares regression
LOG - Logistic regression
+++, - Significant at p < 0.001
++,— - Significant at 0.001 < p < 0.01
+,- - Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
91
-------
Table 14. WHEEZING, SHORTNESS OF BREATH, CHEST TIGHTNESS
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE FEMALE MALE FEMALE MALE
LIN LOG LIN LOG LIN LOG LIN LOG
ASTHMA
CONFLICTING DEMANDS
DUST
NO COLLEGE EDUCATION
SENSITIVE TO CHEMICALS + +++ ++ +++
WALLS RECENTLY PAINTED ++ ++ ++
SOLE PROVIDER + ++ + ++
AGE < 30 -
NEW CARPET, TACKED DOWN, OFF. + ++
ALLERGY TO ANIMALS + +++
ALLERGY TO DUST
AHU G1LG1 ++ + NE ++ NE
AHU 16 ++ +
Comfort and Odor Variables
HOT STUFFY AIR + +++ +++ 4-H-
ODOR OF NEW CARPETS +++
ODOR OF PHOTOCOPIES AND PRINTING + ++
ODOR OF PAINT AND CARPET CLEANING
ODOR OF MUST/DAMPNESS +
LIN - Linear least-squares regression
LOG - Logistic regression
+++, - Significant at p < 0.001
++,— - Significant at 0.001 < p < 0.01
+,_ - Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
92
-------
Table 15. DRY, ITCHING OR TEARING EYES, SORE EYES, BURNING EYES, BLURRY VISION
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE FEMALE MALE FEMALE MALE
LIN LOG LIN LOG LIN LOG LIN LOG
CONTACTS WORN TO WORK
WORKLOAD
SENSITIVE TO CHEMICALS +++ +++ +++ +++ +++ ++ + +++
ALLERGIES TO MOLDS
GLARE +++ -H-+ +++ +++ +++ +
DUST +++ + +++ +++
WALLS RECENTLY PAINTED -H- +++ + ++
HOURS AT COMPUTER -H- + -H- +
GLASSES WORN TO WORK +++ +++ +
USE HEATER + -H- + + +
NO WINDOW -H- 4+
EXTERNAL STRESS ++
WATER LEAKS ++
AHU 16 + +++ -H
Comfort and Odor Variables
HOT STUFFY AIR ++
DRY AIR ++
ODOR OF PHOTOCOPYING ++
ODOR OF COSMETICS, BODY ODOR, FOOD
LIN - Linear least-squares regression
LOG - Logistic regression
-K+, - Significant at p < 0.001
-H-,— - Significant at 0.001 < p < 0.01
+,- - Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
93
-------
Table 16. SORE THROAT, HOARSENESS, DRY THROAT
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE FEMALE MALE FEMALE MALE
LIN LOG LIN LOG LIN LOG LIN LOG
SENSITIVE TO CHEMICALS ++ ++ +++ +4- ++ +
DUST +++ +++ +++ +++
ALLERGIES TO MOLDS + ++ +++ + + +++ +
NEW CARPET TACKED DOWN, OFF. ++ ++++++ +
NEW FURNITURE ++ ++ + + +
NO WINDOW ++ +++ + +
SOLE PROVIDER
GLARE
CONFLICTING DEMANDS
EXTERNAL STRESS ++
USE FAN
AHU NEG-1,2 - NE
AHU NEG-4
AHU ET3FL8-11
AHU SE32
Comfort and Odor Variables
DRY AIR
COLD AIR -H-
ODOR OF PAINT, CARPET CLEANING
ODOR OF NEW CARPET
ODOR OF COSMETICS, BODY ODOR
LIN - Linear least-squares regression
LOG - Logistic regression
+++, - Significant at p < 0.001
++,— - Significant at 0.001 < p < 0.01
+ ,- - Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
94
-------
Table 17. SLEEPINESS, UNUSUAL FATIGUE
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE
JOB DISSATISFACTION
SENSITIVE TO CHEMICALS
DUST
EXTERNAL STRESS
GLARE
ALLERGIES TO MOLDS
NO COLLEGE EDUCATION
NO WINDOW
WATER LEAKS
PRIVATE OFFICE
PRIVATE CUBICLE OR 1 MATE
LASER PRINTER NEAR
NOISE
TOO DIM
Comfort and Odor Variables
HOT STUFFY AIR
ODOR OF PAINT, CARPET CLEANING
ODOR OF COSMETICS, BODY ODOR
ODOR OF NEW CARPET
FEMALE MALE
LIN LOG LIN LOG
FEMALE MALE
LIN LOG LIN LOG
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
95
-------
Table 18. CHILLS AND FEVER
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE
USE HEATER
NO COLLEGE EDUCATION
CONFLICTING DEMANDS
HOURS AT WORK
DUST
SOLE PROVIDER
GLARE
SENSITIVE TO CHEMICALS
NEW CARPET
HIGH PAY
NOISE
AHU 21
AHU CM FL2
Comfort and Odor Variables
FEMALE
MALE
FEMALE MALE
LIN LOG LIN LOG LIN LOG LIN LOG
COLD AIR
DRY AIR
ODOR OF PAINT, CARPET CLEANING
ODOR OF DAMPNESS/MUSTY
ODOR OF COSMETICS, BODY ODOR
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
96
-------
Table 19. ACHING MUSCLES OR JOINTS, PAIN IN BACK, SHOULDER, NECK, WRISTS
VARIABLE
GLARE
ALLERGIES TO MOLDS
UNCOMFORTABLE WORKSTATION
LESS THAN 30
HIGH PAY
SENSITIVE TO CHEMICALS
WORKLOAD
NOISE
LIVE ALONE
DIM LIGHT
CONFLICTING DEMANDS
WATER LEAKS
DUST
SOLE PROVIDER
LASER PRINTER NEAR
AHU 21
AHU 9
AHU G1LG1
AHU FC FL6
Comfort and Odor Variables
COLD AIR
ODOR OF COSMETICS, BODY ODOR, OTHER FOOD
ODOR OF PAINT, CARPET CLEANING
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
FEMALE MALE FEMALE MALE
LIN LOG LIN LOG LIN LOG LIN LOG
+
+
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
97
-------
Table 20. DIFFICULTY CONCENTRATING, REMEMBERING; DEPRESSION, TENSION
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE FEMALE MALE FEMALE MALE
LIN LOG LIN LOG LIN LOG LIN LOG
JOB DISSATISFACTION +++ +++ +++ +++ +++ +++ +++ +++
WORKLOAD +++ +++ ++ +++ +++ +++ ++ +++
CONFLICTING DEMANDS ++ + +++ ++ -H- + +++ ++
SENSITIVE TO CHEMICALS ++ +++ + -H- + +++ + ++
ROLE AMBIGUITY + +++ -H- + -H- -H- ++
NOISE ++ + +++ -H- -H- NI NI
DUST +++ +++ -H- NI NI
WATER LEAKS + +++ + + ++
GLARE + + -H- + ++
LIVE ALONE ++ + +
USE FAN _____
EXTERNAL STRESS + ++ NI + NI
AHU SE 2-1 + + ++ +
AHU ET2 FL6-7 4- ++
Comfort and Odor Variables
HOT STUFFY AIR -H- + +-H- -H-f
ODOR OF COSMETICS, BODY ODOR ++ -H- +
ODOR OF PAINT, CARPET CLEANING -H- ++
ODOR OF TOBACCO SMOKE ++
LIN - Linear least-squares regression
LOG - Logistic regression
+++, - Significant at p < 0.001
-H-,— - Significant at 0.001 < p < 0.01
+,_ - Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
98
-------
Table 21. DIZZINESS
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE
SENSITIVE TO CHEMICALS
DUST
WORKLOAD
NEW CARPET, GLUED DOWN, CORRIDOR
TACKED DOWN, CORRIDOR
TACKED DOWN, OFFICE
FEMALE
LIN LOG
MALE
LIN LOG
FEMALE
LIN LOG
MALE
LIN LOG
NEW CARPET,
NEW CARPET,
NEW CARPET
CONFICTING DEMANDS
NO COLLEGE EDUCATION
GLASSES
ALLERGIES—OTHER
NO WINDOW
AHU FC 8
AHU ET2 FL 1-3
AHU ET2 FL 8-9
AHU ET3 FL 4-7
AHU WT1 FL 4-5
AHU WT1 FL 6-7
AHU WT1 FL 10-11
AHU WT2 FL 1-3
AHU WT2 FL 10-11
AHU WT3 FL 4-7
AHU SE 2-1
AHU SE 2-2
Comfort and Odor Variables
+ +
+ +
+ -H
NI
NI
NI
NI
HOT STUFFY AIR
ODOR OF PAINT, CARPET CLEANING
ODOR OF NEW CARPET
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
99
-------
Table 22. DRY OR ITCHY SKIN
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE
FEMALE
MALE
FEMALE MALE
LIN LOG LIN LOG LIN LOG LIN LOG
ECZEMA
DUST
USE HEATER
MEDIUM PAY
NEW CARPET, TACKED DOWN, COR
ALLERGY TO MOLDS
UNCOMFORTABLE WORKSTATION
GLARE
SENSITIVE TO CHEMICALS
AHU WT2 1-3
AHU ET1 FL 8-9
AHU ET B
AHU ET3 FL 4-7
AHU CM FL 2
Comfort and Odor Variables
DRY AIR
ODOR OF NEW CARPET
ODOR OF TOBACCO SMOKE
NI
NI
NI
NI
NI
NI
NE
NI
NI
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
100
-------
Table 23. PROBLEMS WITH CONTACT LENSES
COMFORT AND ODOR VARIABLES
EXCLUDED INCLUDED
VARIABLE
JOB DISSATISFACTION
HOURS AT WORKSTATION
DUST
DUST ALLERGIES
USE LAMP
JOB CONTROL
EXTERNAL STRESS
UNCOMFORTABLE WORKSTATION
WORKLOAD
Comfort and Odor Variables
ODOR OF COSMETICS, BODY ODOR
FEMALE
LIN LOG
MALE
LIN LOG
FEMALE
LIN LOG
NI
NI
MALE
LIN LOG
4- NI
++ NI
NI — NI
NI ++ NI
NI ++ NI
NI + NI
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
101
-------
Table 24. HOT STUFFY AIR
VARIABLE
FEMALE
MALE
LIN LOG LIN LOG
USE FAN
DUST
DIM LIGHT
GLARE
WORKLOAD
NOISE
NO WINDOW
ENCLOSED OFFICE/PRIVATE OFFICE
UNCOMFORTABLE WORKSTATION
ABILITIES UNDERUTILIZED
SENSITIVE TO CHEMICALS
MOLD ALLERGIES
NEW CARPET TACKED DOWN, OFFICE
NEW CARPET, GLUED DOWN, CORRIDOR
NEW CARPET TACKED DOWN, CORRIDOR
MID-HEIGHT CUBICLES
COPIER NEARBY
AGE > 40
AHU NEG-1, NEG-2
AHU NEG-4
AHU FC 8
AHU ET
AHU WT
+
+
LIN - Linear least-squares regression
LOG - Logistic regression
+++, - Significant at p < 0.001
++,— - Significant at 0.001 < p < 0.01
+ ,_ - Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
102
-------
Table 25. DRY AIR
VARIABLE
FEMALE
LIN LOG
MALE
LIN LOG
DUST
UNCOMFORTABLE WORKSTATION
NOISE
DIM LIGHT
GLARE
CONFLICTING DEMANDS
MOLD ALLERGIES
PRIVATE OFFICE
AHU 16
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
103
-------
Table 26. COLD DRAFTY AIR
VARIABLE
FEMALE
LIN LOG
MALE
LIN LOG
USE HEATER
NOISE
USE FAN
CONFLICTING DEMANDS
DUST
WINDOW
WORKLOAD
HOURS AT COMPUTER
HIGH PAY
AHU CM FL 2,7,8,10
AHU FC FL 7
AHU FC FL 6
AHU 14
AHU 16
AHU 15SW2
AHU 9
AHU NE 3-2
AHU WT1 FL 6,7,10,11
+
++
+
NE
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
104
-------
Table 27. HUMID AIR
VARIABLE
FEMALE
LIN LOG
MALE
LIN LOG
USE FAN
LOW PAY
DUST
SENSITIVE TO CHEMICALS
ALLERGIES—OTHER
CONFLICTING DEMANDS
NONSMOKER
AHU 21
AHU 10
AHU 2
AHU 20
AHU G1LG1
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
105
-------
Table 28. ODOR OF PAINT, CARPET CLEANING, OTHER CHEMICALS
VARIABLE FEMALE MALE
LIN LOG LIN LOG
WALLS RECENTLY PAINTED
SENSITIVE TO CHEMICALS
DUST
USED CHEMICALS AT WORKSTATION
NOISE
GLARE
WALLS RECENTLY REARRANGED
CONFLICTING DEMANDS
WINDOW
GLASSES
AHU 15 SW 2
AHU ET3 FL 8-11
AHU G1LG1
LIN = Linear least-squares regression
LOG - Logistic regression
+++, - Significant at p < 0.001
++.— - Significant at 0.001 < p < 0.01
+,- - Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
106
-------
Table 29. ODOR OF COSMETICS. BODY ODOR, FOOD
VARIABLE FEMALE MALE
LIN LOG LIN LOG
CONFLICTING DEMANDS
DUST
HIGH PAY
GLARE +-H- -H- +
NOISE + +++ +•
WATER LEAKS +
USE FAN
NO COLLEGE EDUCATION
WINDOW
USE LAMP -H- +
AHU CM 2,7 -H- +
AHU ETB - — NE
LIN = Linear least-squares regression
LOG = Logistic regression
+++, = Significant at p < 0.001
-H-,— - Significant at 0.001 < p < 0.01
+,- - Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
107
-------
Table 30. ODOR FROM PHOTOCOPIES, PRINTING
VARIABLE FEMALE MALE
LIN LOG LIN LOG
PHOTOCOPIER PRESENT
DUST
CONFLICTING DEMANDS
SENSITIVE TO CHEMICALS ++ +++ ++ +
PRIVATE OFFICE
WINDOW
ECZEMA ++ +
LIVE PLANTS ++ +
GLARE + ++
GLASSES
NONSMOKERS
WORKED > 1 HR ON PHOTOCOPIER ++
AHU G1LG1
AHU NEG3-4 — NE
LIN - Linear least-squares regression
LOG - Logistic regression
+-H-, - Significant at p < 0.001
++,— - Significant at 0.001 < p < 0.01
+ ,- - Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
108
-------
Table 31. ODOR FROM NEW CARPET OR DRAPES
VARIABLE
FEMALE MALE
LIN LOG LIN LOG
NEW CARPET PRESENT
NEW CARPET, TACKED DOWN,
NEW CARPET, GLUED DOWN,
DUST
NEW CARPET, TACKED DOWN,
NEW CARPET, GLUED DOWN,
WALLS RECENTLY PAINTED
SENSITIVE TO CHEMICALS
CONFLICTING DEMANDS
ROLE AMBIGUITY
AHU ET1 FL67
AHU NEG-1, NEG-2
AHU WT2 FL 6-7,10-11
AHU ET3 FL 8-11
OFFICE
OFFICE
CORRIDOR
CORRIDOR
++ +
NE
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
109
-------
Table 32. MUSTY OR DAMP ODOR, DIESEL EXHAUST
VARIABLE FEMALE MALE
LIN LOG LIN LOG
DUST
WATER LEAKS
CONFLICTING DEMANDS
USE LAMP ++ +
GLARE -H- -H-
ALLERGIES TO MOLDS -H-f ++
UNDERUTILIZATION OF ABILITIES +
NEW CARPET, TACKED DOWN, OFFICE -H-
AHU 15 SW 2 +++ -H-+ +-H-
AHU G1LG1 +++ + +
AHU 16 + -H-
AHU 2 -M-
AHU ET1 FL 10-11
AHU 19.G.H -H-
LIN - Linear least-squares regression
LOG - Logistic regression
+++, - Significant at p < 0.001
++,— = Significant at 0.001 < p < 0.01
+,- = Significant at 0.01 < p < 0.05
NE - Not estimated--mathematical singularity
NI - Not included
110
-------
Table 33. TOBACCO SMOKE
VARIABLE
SENSITIVE TO CHEMICALS
DUST
WATER LEAKS
CONFLICTING DEMANDS
ALLERGIES TO POLLEN
NONSMOKER
AHU WT1 FL1-3
AHU NEG-1,2
AHU CM FLU
AHU G1LG1
AHU 14
AHU CM NOFL
AHU 13
FEMALE MALE
LIN LOG LIN LOG
++ +
-H-
-H-+ +
LIN
LOG
NE
NI
Linear least-squares regression
Logistic regression
Significant at p < 0.001
Significant at 0.001 < p < 0.01
Significant at 0.01 < p < 0.05
Not estimated--mathematical singularity
Not included
111
-------
Table 34. VARIABLES APPEARING AT THE p < 0.01 SIGNIFICANCE LEVEL FOR EACH OF THE 12 HEALTH SYHPTOMS
HEAD
Independent Variable
F
M
NOSE
F
M
CHEST
F
M
EYES
F
M
THROAT
F
M
TIRED
F
M
CHILLS
F
M
ACHES
F
M
DEPRES
F
M
DIZZY
f
H
SKIN
F
M
LENS
F
M
TOTAL
F
M
Personal/Workplace (Max = 48 + 48)
Personal; physical /medical
sensitivity to chemicals
mold allergy
dust allergy
under 30
asthma
eczema
contact lens worn to work
glasses worn at work
pollen allergy
Personal; psycho logical /social
workload
conflicting demands
job dissatisfaction
no college education
sole provider
external stress
role ambiguity
hours at workstation
hours at computer
live alone
Workplace;
dusty office
glare
noise
use portable heater
no window
uncomfortable workstation
water leaks
walls recently painted
new carpet, tacked down, office
new carpet, tacked down, hallway
AHU 21 (Library area, WSM)
AHU 16 (2nd floor, USM)
lights too dim
Cohort/Odor (Max = 24 + 24)
hot stuffy air
dry air
cold air
odor of paint, chemicals
odor of cosmetics, body, food
odor of new carpets
odor of photocopying, printing
4
0
0
0
0
0
0
0
0
2
0
0
2
0
1
0
0
0
0
1
3
1
0
0
0
0
0
0
0
0
0
0
2
0
0
2
0
0
0
4
3
0
0
0
0
0
0
0
4
0
0
0
0
0
1
0
0
0
2
1
2
0
0
0
0
0
0
0
0
0
0
2
1
0
1
0
2
0
1
1
4
-4a
0
0
0
1
4
0
2
0
0
0
0
0
0
1
0
3
0
0
0
0
0
0
0
0
0
0
0
0
1
2
2
2
0
0
0
1
2
4
0
0
0
0
2
0
0
1
0
0
0
0
0
0
0
0
4
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
1
0
1
-1
4
0
0
0
0
0
4
0
2
2
0
0
0
0
0
1
0
0
0
0
0
0
3
1
0
0
1
0
1
0
0
2
0
0
0
2
0
0
0
4
0
0
0
0
0
0
0
1
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
1
1
4
4
0
0
0
0
4
2
0
4
0
0
0
0
0
0
0
2
0
1
4
0
1
0
0
0
3
0
0
0
0
0
2
1
0
0
0
0
2
3
2
0
0
0
0
4
0
0
4
0
0
0
0
1
0
0
0
0
2
1
0
0
2
0
1
0
0
0
0
3
0
2
2
0
0
2
0
1
2
1
0
0
0
0
0
0
0
0
1
0
0
2
0
0
0
0
0
2
1
0
0
0
0
0
0
2
0
0
0
0
0
2
2
2
0
0
0
3
2
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
2
0
0
0
2
0
0
0
1
0
0
0
0
0
2
0
1
1
1
0
4
0
0
0
0
0
0
0
0
0
0
4
2
0
1
0
0
0
0
2
1
0
0
0
0
0
0
0
0
0
0
1
2
0
0
2
1
0
0
1
2
0
0
0
0
0
0
0
0
0
1
0
0
1
0
0
0
0
1
1
1
0
2
0
2
0
0
0
0
0
0
2
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
2
0
3
2
0
0
0
0
0
2
0
0
1
0
0
0
0
0
0
0
0
0
0
0
2
2
2
0
0
1
0
0
0
0
0
0
0
0
0
1
0
1
0
0
0
2
0
0
0
1
1
3
0
0
0
0
0
0
2
0
0
0
1
2
1
0
0
0
1
2
0
-4
0
0
0
0
0
2
2
0
0
1
0
0
0
0
0
1
4
1
0
0
3
1
0
0
0
0
0
1
0
0
1
1
2
0
0
1
4
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
0
4
1
0
0
2
0
0
0
0
2
0
1
0
0
0
0
1
0
0
3
0
0
0
0
0
0
0
0
4
2
4
0
0
0
1
0
0
1
3
1
2
0
0
0
0
0
0
0
0
0
0
1
0
0
2
2
0
0
2
0
0
0
0
0
0
0
0
4
4
4
0
0
1
4
0
0
0
0
1
2
0
0
0
2
0
0
0
0
0
0
2
0
0
0
0
0
0
4
0
0
0
0
0
0
0
0
4
1
0
2
0
0
0
0
0
0
3
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
2
0
0
0
3
0
0
0
0
0
0
1
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
1
2
0
0
0
1
0
0
1
0
1
0
0
2
0
0
0
4
0
0
0
0
0
0
0
0
0
0
0
0
0
3
0
0
4
0
1
0
0
0
-3
0
0
0
0
2
0
0
0
0
0
1
0
0
0
0
4
0
0
0
0
0
0
0
0
0
0
0
0
0
2
1
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
1
0
0
0
2
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
1
0
1
0
0
1
0
2
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
24
10
7
9
4
4
4
3
4
16
14
9
11
7
2
1
0
3
1
24
14
4
6
0
4
1
6
3
3
0
1
2
10
7
7
17
8
0
2
22
15
4
0
4
4
4
3
0
13
7
6
2
0
5
5
4
0
2
16
10
7
3
7
3
5
0
2
2
4
3
1
13
8
2
4
4
7
2
Negative sign indicates significantly lower frequency of health symptoms
-------
Table 35. VARIABLES ASSOCIATED8 WITH AT LEAST FOUR OF THE 12 HEALTH FACTORS
Workplace Characteristics
Dust (headache; nasal, chest, eye, throat symptoms; fatigue;
chills; difficulty concentrating1*; dizziness; dry skin;
contact lens problems)
Glare (headache; eye symptoms; fatigue; difficulty concentrating;
painc)
Personal Characteristics
Chemical (headache; nasal, chest, eye, throat symptoms; fatigue;
Sensitivity pain; difficulty concentrating; dizziness)
Mold Allergies (headache; nasal, eye, throat symptoms; fatigue; pain; dry
skin)
No College (headache, chest symptoms, fatigue, chills and fever,
dizziness)
Psychosocial Characteristics
Workload (headache, eye symptoms, pain, difficulty concentrating,
dizziness)
Conflicting (nasal, chest symptoms; chills and fever, pain, difficulty
Demands concentrating, dizziness)
Comfort and Odor Characteristicsd
Hot Stuffy Air (headache; nasal, eye, chest symptoms; fatigue; difficulty
concentrating; dizziness)
Dry Air (nasal, eye, throat symptoms; dry skin)
Odor of Paint, (headache; nasal, chest, throat symptoms; fatigue;
Chemicals chills; difficulty concentrating; dizziness)
Odor of Cosmetics (eye symptoms, chills, pain, difficulty concentrating)
0 Significant at p < 0.01 in at least two (of eight) regressions
b Includes difficulty concentrating/remembering, depression, tension
c Includes aching muscles, back pain, shoulder/neck pain, hand/wrist pain
d Significant at p < 0.01 in at least two (of four) regressions
113
-------
Table 36. VARIABLES ASSOCIATED" WITH ONE TO THREE HEALTH FACTORS
Workplace Characteristics
Noise (headache, pain, difficulty concentrating)
No window (eye, throat symptoms; fatigue)
Walls painted (chest, eye symptoms)
Water leaks (fatigue, dificulty concentrating)
Use heater (chills and fever, dry skin)
AHU 21 (Library area) (chills and fever, pain)
New carpet, tacked
down, hallway (dizziness, reduced frequency of dry skin)
New carpet, tacked
down, office (throat symptoms)
Uncomfortable
workstation (pain)
Lights too dim (pain)
AHU 16 (2nd floor Mall) (eye symptoms)
Personal Characteristics
Allergy to dust (nasal symptoms, contact lens problems)
Under 30 (reduced frequency of nasal symptoms, pain)
Glasses (nasal, eye symptoms)
Contact lenses (eye irritation)
Asthma (chest symptoms)
Eczema (dry skin)
Pollen allergy (nasal congestion)
Psychosocial Characteristics
Job satisfaction (fatigue, difficulty concentrating, contact lens
problems)
Sole provider (chest, throat symptoms; chills and fever)
Hours at work (chills and fever, contact lens problems)
Hours at computer (eye symptoms)
External stress (fatigue)
Role ambiguity (difficulty concentrating)
Live alone (pain)
Comfort/Odor Conditions^
Cold air (nasal, throat symptoms; chills and fever)
Photocopy odor (eye irritation)
New carpet odor (headache)
8 Significant at p < 0.01 in at least two (of eight) regressions
b Includes difficulty concentrating/remembering, depression, tension
c Includes aching muscles, back pain, shoulder/neck pain, hand/wrist pain
d Significant at p < 0.01 in at least two (of four) regressions
114
-------
Table 37. VARIABLES APPEARING AT THE p < 0.01 SIGNIFICANCE LEVEL FOR EACH OF THE 10 COMFORT AND ODOR CONCERNS
Independent Variable _
HOT
f_M
Personal/Workplace (Max = 20 + 20)
Workplace:
dusty office
use fan
noise
glare
water leaks
lights too dim
no window
new carpet, tacked down, office
walls recently painted
uncomfortable workstation
photocopier present
use heater
new carpet .,
private office *
low pay
used chemicals at workspace
Location: AHU *
15SU2
16
61-LG1
NEG-1,2
14
WT1 FL 1-3
CM 2,7,8,10
2
21
Personal; physical/medical
sensitivity to chemicals
mold allergy
Personal; psychological/social
conflicting demands
workload
2
2
0
2
0
2
0
0
0
1
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
1
1
0
1
2
2
2
1
0
2
2
1
0
0
1
0
0
1
0
0
0
0
0
-1
0
0
0
0
0
0
0
0
2
Comfort Concerns
DRY COLD HUMID
F M F M F M
2
0
1
0
0
1
0
0
0
2
0
0
0
1
0
0
0
2
0
0
0
0
0
0
0
0
1
0
0
2
0
2
2
0
2
0
0
0
2
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
0
2
0
0
-2a
2
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
2
0
-2 -
0
0
0
0
2
0
2
-1
2
0
0
0
1
0
0
0
0
2
0
0
0
0
2
2
0
0
2
0
1
0
0
0
0
0
1
2
2
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
1
0
0
0
0
0
1
2
0
1
0
2
2
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
2
2
0
0
0
0
PAINT
L-M
2
0
1
0
0
0
0
0
2
0
0
0
0
0
0
2
2
0
1
0
0
0
0
0
0
2
0
1
0
2
0
1
2
0
0
1
0
2
0
0
0
0
0
0
2
1
0
0
0
0
0
0
0
0
2
0
0
0
BODY
F M
1
2
0
1
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
2
0
2
0
2
1
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
Odors
PRINT
F M
2
0
0
0
0
0
0
0
0
0
2
0
0
-1
0
0
0
0
1
0
0
0
0
0
0
2
0
2
0
2
0
0
1
0
0
1
0
0
0
2
0
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
2
0
Noticed
CARPET
F_ H
1
0
0
0
0
0
0
2
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
0
0
2
0
1
0
2
0
0
0
0
0
0
2
2
0
0
0
2
0
0
0
0
0
0
-2
0
0
0
0
0
0
0
0
0
DAMP
F__M
2
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
2
0
1
0
0
0
0
0
0
0
2
2
0
2
0
0
1
2
0
0
1
0
0
0
0
0
0
0
0
2
2
0
0
0
0
0
2
0
0
0
1
0
SMOKE
F__M
1
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
2
0
0
0
2
0
1
0
2
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
2
0
0
0
2
0
1
0
TOTAL
F M_
15
8
4
4
2
3
1
2
2
3
2
2
2
3
2
2
4
2
5
1
3
2
3
0
1
11
4
12
1
20
5
9
8
5
4
5
4
4
2
3
2
2
1
2
2
5
6
1
4
2
2
1
4
2
5
0
7
3
8 Negative sign indicates significantly lower frequency of comfort or odor concerns
-------
Table 38. VARIABLES ASSOCIATED" WITH AT LEAST THREE OF THE 10 COMFORT/ODOR
FACTORS
Workplace Characteristics
Dust (all four comfort concerns and all six odor factors)
Glare (hot air, dry air, odors of paint, cosmetics, dampness)
Noise (hot air, dry air, cold air, odors of paint and cosmetics)
Use Fan (hot air, humid air, cold air [negative], odor of cosmetics)
Water Leaks (odors of dampness, cosmetics, tobacco smoke)
Personal Characteristics
Chemical (humid air, odors of paint, photocopying, new carpet,
Sensitivity tobacco smoke)
Conflicting (dry air, cold air, odors of cosmetics, photocopying,
Demands dampness, tobacco smoke)
Locations
AHU 15, SW-2 (cold air, damp/musty odor, odor of paint & other chemicals)
AHU 16 (cold air, damp/musty odor, dry air)
8 Significant at p < 0,01 in at least two (of four) regressions
116
-------
Table 39. VARIABLES ASSOCIATED" WITH ONE OR TWO COMFORT/ODOR FACTORS
Workplace characteristics
Walls Recently Fainted
Used Chemicals at Workspace
Lights Too Dim
No Window
Frivate Office
Workload
Uncomfortable Workstation
Use of a Heater
Low Pay
Photocopier Present
New Carpet
New Carpet, Tacked Down, Office
Locations
AHU 2
AHU 14
AHU 21
AHU G1-LG1
AHU WT1 FL 1-3
AHU NEG-1,2
AHU CM 2,7,8,10
(odor of paint/chemicals; new carpet odor)
(odor of paint/other chemicals)
(hot air, dry air)
(hot air)
(hot air)
(hot air)
(dry air)
(cold air)
(humid air)
(odor of photocopying/printing)
(odor of new carpet)
(odor of new carpet)
(humid air, damp/musty odor)
(cold air)
(humid air)
(odor of photocopying/printing)
(odor of tobacco smoke)
(odors of tobacco smoke,
new carpet [negative])
(cold air [negative])
Personal Characteristics
Mold Allergy
(damp/musty odor)
8 Significant at p < 0.01 in at least two (of four) regressions
117
-------
Table 40. VARIABLES ASSOCIATED8 WITH NO HEALTH, COMFORT, OR ODOR FACTORS
Personal Characteristics
Allergic to Animals
Allergic: Other
Smoker
Psychosocial Variables
Job Control
Utilization of Skills
Workplace Characteristics
Building
Location in Building
Private Cubicle or Semiprivate Office
Open Area or Share Office with Two or More
Estimated Supply Air Flow Rate
Shelf Index
Wood Index
Fabric-Covered Partitions Nearby
Laser Printer Nearby
Photocopier Nearby
Plants Nearby
Hours at Photocopier
New Furniture
New Equipment
Walls Recently Rearranged
Location Variables
About 57 of 66 AHU/floor combinations
Comfort/Odor Factors
Humid Air
Musty/Damp Odor
Tobacco Smoke Odor
Significant at p < 0.01 in at least two regressions for each factor
118
-------
Table 41. SYMPTOM PREVALENCE IN MANHATTAN FAMILIES COMPARED TO EPA EMPLOYEES
Percent of days with symptom Ratio:
Symptom Manhattan Waterside Mall WSM/NYC
Eye irritation
Aching muscles
Headache
Nausea
Sore throat
Wheezing
Fever
Stuffy nose
Cough
1.8
1.2
2.4
0.6
1.8
0.77
0.70
8.0
5.2
21
13
21
4.4
11
4.0
3.0
30
16
12
11
8.7
8.0
6.1
5.2
4.3
3.7
3.1
Source: Manhattan data: Lebowitz, 1972a.
WSM data: calculated from Exhibit ES-5, p. ES-11, Volume I,
119
-------
Table 42. COMPARISON OF BRITISH, DANISH, NETHERLANDS AND US BUILDING STUDIES
Characteristic
Number of buildings
Type (Public/private)
Number of respondents
Response rate
Number of health symptoms
Number of symptom groups0
No. rooms measured for IAQ
UKa
47
Both
4373
92%
10
10
0
DKb
28
Public
3507
80%
12
2
14
Neth.c
61
Both
7043
74%
18
5
0
usd
3
Public
3955
81%
32
12
100
8 Wilson and Hedge, 1987.
b Skov et al.. 1989. 1990.
c Zweers e_t al. , 1990
d This study
e Number of symptom groupings for analytical purposes
120
-------
Table 43. FINDINGS OF THREE EUROPEAN BUILDING STUDIES
I. British study (Wilson & Hedge: Raw et al.)
(N - 4400 workers, 47 buildings, 90% response rate)
Self-reported productivity was decreased by:
1. Sharing office with five or more workers
2. Sharing office with one or more smokers
3. Lack of control over temperature, ventilation, and lighting
4. Lack of comfort
5. Three or more building-related symptoms (out of 10)
6. Age (but no regular pattern)
Self-reported productivity was unaffected by:
1. Gender
2. Job classification (an effect was noted but this was explained by the
number of people sharing the room; the authors concluded that "Managers
report higher productivity not because they are managers but because
they are less likely to share their offices.")
3. Length of time respondent had worked at the building
4. Time spent at a video display terminal
5. Hours per day (or days per week) at the office
6. When sunlight entered the office
7. Job stress
One or more Sick Building Syndrome- type symptoms (dry eyes, itching eyes,
stuffy nose, runny nose, dry throat, headache, lethargy) were increased by:
1. Working for a public rather than private organization
2. Air-conditioned buildings
3. Humidified buildings
4. Being female
(continued)
121
-------
Table 43. (Continued) FINDINGS OF THREE EUROPEAN BUILDING STUDIES
II. Danish study (Skov & Valbjom. 1987: Skov et al. . 1990)
(N - 3500 workers, 14 buildings, 80% response rate)
Mucous membrane irritation or "general symptoms" were increased by:
1. Being female
2. Hayfever
3. Migraine
4. Smoking
5. Not drinking coffee or drinking more than 6 cups per day
6. Indoor climate problems at home
7. Working > 40 hours per week
8. Being a clerk
9. Working 7 hours per day
10. Handling carbonless copy paper
11. Photocopying > 25 sheets per day
12. Spending more than 1 hour per day at a VDT
13. Not having varied work
14. Job dissatisfaction
15. Dissatisfaction with fellow workers
16. Workload
17. Lack of influence over job and high work speed
18. Floor dust
19. Organic fraction of floor dust
20. Carpets
21. Area of the office
22. Number of workstations in the office
23. Shelf factor
24. Fleece factor
25. High temperature
26. Age of building
27. Type of supply air ventilation system
Symptoms were unaffected by:
1. Office volume
2. Workplace area
3. Ventilation rate
4. Relative humidity
5. Temperature
6. Carbon dioxide
7. Formaldehyde
8. Static electricity
9. Volatile organic compounds
10, Man-made mineral fibers
122
-------
Table 43. (Continued) FINDINGS OF THREE EUROPEAN BUILDING STUDIES
III. Netherlands Study (Preller. et al.. 1990)
(N - 7000 workers, 61 office buildings, 75% response rate)
Self-reported absenteeism was increased by:
1. Humidification (either spray or steam)
2. Lack of control over temperature
3. Openable windows
4. Hours in front of a video display terminal
5. Being female
6. Being middle-aged (worst) or young (next worst)
7. Lack of education
8. Being a smoker
9. Having allergies or CNS symptoms
10. Lack of job satisfaction
Self-reported absenteeism was unaffected by:
1. Type of ventilation (natural vs. mechanical)
2. Reclrculation
3. Wearing of contact lenses
4. Passive smoking
5. Number of people in the workplace (> 10 vs. < 10)
6. "Fleece factor" (2 of the following present: wall-to-wall carpeting,
curtains, bookshelves)
123
-------
ESTIMATED SUPPLY AIR FLOW RATES
PROVIDED TO EPA EMPLOYEES
600
estimated per capita flow rate (cfmp)
600
cumulative frequency (%) of employees
Figure 1. Cumulative frequency distribution of estimated per capita
supply air flow rates in the three EPA Headquarters buildings.
Supply air flow rates were measured for about 30 air handling units
and estimated for the remaining 20 from design parameters. The
number of employees is about 5000.
124
-------
RELATIONSHIP BETWEEN THREE SETS OF LINEAR REGRESSIONS
AND TWO SETS OF LOGISTIC REGRESSIONS
1ST LINEAR REG. (N=68)
BUILDING/SECTOR
(8 LOCATION VARIABLES)
PERSONAL/WORKPLACE
(55 VARIABLES)
COMFORT/ODOR
(10 VARIABLES)
TOTAL: 73 VARIABLES
Ol
2ND LIN. REG. (68)
AHU/LOCATION
(66 VARIABLES)
NEW CARPET
(4 VARIABLES)
PERSONAL/WORK/
COMFORT/ODOR
(66 VARIABLES)
136 VARIABLES
1ST LOGISTIC (68)
INCLUDES ONLY
VARIABLES THAT
WERE SIGNIFICANT
AT P < 0.05 IN
SECOND LINEAR
REGRESSION ABOVE
17-59 VARIABLES
3RD LIN. REG. (68)
AIRFLOW CAPACITY
(2 VARIABLES)
PERSONAL/WORK/
COMFORT/ODOR
(66 VARIABLES)
68 VARIABLES
2ND LOGISTIC (4)
USES DIFFERENT
CRITERION TO
DEFINE SYMPTOMS
(i.e., AT LEAST
ONE SYMPTOM MUST
BE EXPERIENCED
OFTEN OR ALWAYS)
25-32 VARIABLES
Fig. 2. Relationships among the regressions. Each set of regressions except the last
involved 68 separate analyses. The number of observations in each regression ranged between
3800 and 3900 for most cases.
-------
APPENDIX A
EPA STUDY QUESTIONNAIRE
A-l
-------
PLEASE READ BEFORE
COMPLETING QUESTIONNAIRE
Many questions in the questionnaire concern either last week or last year. By
"LAST YEAR" we mean the 12-month period ending today. If you have worked
in the building for less than one year, answer the "LAST YEAR" questions only
for the part of the year that you worked in this building.
Please report your ACTUAL EXPERIENCES LAST WEEK even if last week was
unusual for you. By "LAST WEEK" we mean any or all days worked from last
Monday through Friday.
CONFIDENTIALITY
To protect your privacy, the identification for your questionnaire is the bar-code
label on the cover. The bar-code cannot be read by EPA computers or staff.
Additionally, the survey forms will be gathered by staff from Westat, Inc., an
independent survey research firm, and processed away from EPA. Your name
and other information necessary for the survey and analysis that might identify
you, such as your room and telephone number, will not be disclosed to
individuals, unions, or management of EPA. Reports of the survey will not give
your name, nor will data be presented in such a way that you, or anyone else,
could be identified.
STUDY SPONSORS AND ORGANIZATION
The study has been developed and is being conducted by the National Institute for Occupational
Safety and Health (NIOSH), the John B. Pierce Foundation Laboratory at Yale University, and
Westat, Inc. It is being managed by EPA and NIOSH, and is being supported by funds from EPA.
-------
PART I. DESCRIPTION OF YOUR WORKSTATION
This section asks you to describe your workstation.
Your answers to these questions will help us to
construct a picture of your work surroundings.
By WORKSTATION we mean your desk, office, cubicle,
or place that is your primary work area. This descrip-
tion is obvious for many people, but more difficult for
those whose jobs require them to move about the
building. If you do move about the building, your
workstation is the specific location where you spend
more time than any other single location. If your
workstation has been relocated, use the location
where you are now.
1. There are many different types of workstations.
Please check the categories that best describe
the space in which your current workstation is
located.
a. Type of space (Check one)
1. n Enclosed office with door
2. D Cubicle with floor to ceiling book-
cases or partitions and no door
3. H] Cubicle surrounded by mid-height
bookcases or partitions
4. [H Open office area
5. O Stacks (e.g., books or periodicals)
6. ED Loading dock, laboratory, copy
center, or print shops
7. n Work all around the building
8. D Other (specify)
b. Type of space sharing (Check one)
1. Q Single occupant
2. E] Shared with one other person
3. O Shared with two or more other
persons
4. CD Other (describe)
2.
How many years of service do you have with
EPA? (Enter number of months if less than one
year.)
3.
b.
4.
b.
years
months
How many years have you been working
in this building? (Enter number of months
if less than one year.)
years
months
During a typical week, how many hours do
you spend in this building?
hours per week
How many years have you worked at your
current workstation? (Enter number of months
if less than one year.)
years
months
During an average workday, how many hours
do you spend at your workstation?
hours per day
5. How many days did you work in this building last
week?
days last week
-------
6. What time do you usually:
a. Arrive at work :
b. Leave work :
c. Varies (describe)
AM PM
D D
D D
7. Which of the following Hems are pr
within 15 feet of your workstation?
'yes" for each Item.)
a. Metal desk
b. Wood or composition desk . .
c. Metal bookshelves or
bookcases
d. Wood or composition
bookshelves or bookcases . .
e. Re cabinet(s)
f. Other metal furniture
g. Other wood or composition
furniture
h. Fabric-covered partitions . . .
i. Portable humidifier
j. Laser printer ,
k. Photocopy machine
1. Uve plants
8. Is there carpeting on most or all of
your workstation?
1. D No
2. D Yes
>
esently located
(Check "no" or
No Yes
1 2
D D
D D
D D
D D
D D
D D
D D
D D
D D
D D
D D
D D
the floor at
9. During a typical day LAST WEEK, how much time
did you spend working with each of the following
Hems? (If you worked with an item at all, but less
than 1 hour, enter 1 hour per day.)
Hours
per day
a. Computer or word processor
with screen/kevboard
b Photocopy machine
c. Photographic developing
and processing
d. Printing processing (press,
binding materials etc ) ....
e. Other chemicals such as
glues, adhesives, cleansers,
white out, rubber cement,
pesticides etc
NOTE: If you have worked in this building for less
than a year, answer the following questions
for the part of the year that you worked in
this building.
10. Were any of the following Hems regularly used
at your workstation during the LAST YEAR:
(Check "no" or 'yes" for each item.)
No Yes
1 2
a Portable fan I I I I
b. Portable air filter, or cleaner,
or negative-ion generator — Q Q
c. Portable heater Q Q
d. Desk lamp fn [~]
-------
11. During the LAST YEAR (and since you've been in
your current workstation) have any of the following
changes taken place within 15 feet of your current
workstation? (Check "no" or yes" for each Hem.)
No
1
Yes
2
a.
b.
c.
d.
e.
f.
New carpeting ............ Q Q
New drapes or curtains — Q Q
Newfumlture ............. D D
New equipment, such
as a computer ............ D £]
Walls painted ..... : ....... []
Rearranged walls .......... Q
12. At any time during the LAST YEAR, have you
noticed evidence of new or continuing water
leaks from the ceiling, floors, walls, or pipes
near your workstation?
1.
2.
No
Yes
-------
PART II. INFORMATION ABOUT YOUR HEALTH AND WELL-BEING
This section asks questions about the status of
your health and well-being. Your answers to these
questions will help us construct a profile of the
health status of the employees in this building.
Please answer all the questions even if you don't
associate these health conditions with your work.
1.
2.
Do you wear contact lenses?
~~\ Never
1.
Go to 0.2
2. O Sometimes
3. D Often
4. CD Always
b. Do you wear contact lenses at work?
1. C] Never
2. dl Sometimes
3. D Often
4. D Always
c. If never worn at work, why?
During work, how often do you wear eyeglasses
(NOT including contacts) for close-up work?
1. n Never
2. O Sometimes
3. D Often
4. O Always
3.
Which of the following best describes your
history of smoking tobacco products such as
cigarettes, cigars or pipes?
1. Q
2. Q
3. Q
Never smoked -
Former smoker •
Current smoker
4. Do you smoke tobacco products at your
workstation?
1. D Never
2. [U Sometimes
3. Often
5. Do you smoke tobacco products elsewhere at
work?
1.
2.
3.
Sometimes
Often
6. In a typical 24 hour day, how many CIGARETTES
do you usually smoke?
1 Q None
2. [] 1 to 5
3. n 6 to 10
4. n 11to20
5. D 21 to 30
6. n 31 or more
-------
Please answer the three questions
to the right about each symptom
listed below, even if you believe
the symptom is not related to the
building.
(For each symptom, answer the first
question. If the response Is "never,"
go down to tf»e nexr symptom.)
b, nausea
ff fSs* -.»
c. runny nose
d. stuffy nose/sinus congestion
sneezing
•& ''"
m
C'
k. sore/strained eyes
g. wheezing or whistling in chest . .
h. shortness of breath
ilry, Itching, or tearing eyetf ;i\T»
* *'*"
L blurry/double vision
m. burning eyea
< f
f* f * f
n. sore throat
'' Sf ,
o. hoarseness
p. dry throat
q. unusual fatigue or tiredness ..
»jr* JS-*
r, sleeplnesss or drowsiness ...
Please indicate how often
during the LAST YEAR
you have experienced this
symptom while working
in this building.
Some-
N«ver Rarely times Often Always
1 2
D D
1 2
D D
345
D D D
345
D D D
Please indicate
how many days
LAST WEEK you
experienced this
symptom while
working in this
building.
(Fill In No. of days)
?f/iwW, ',
'' f ", '
Z'>
' ^"1^
Does the
symptom usually
change when
not at work?
Gets Stays Gets
Worse Same Better
123
123
D-D D
1 2 3
ODD
1 2 3
ODD
18 * 2 ', 3
D a., a
123
pan
1 2 3
ana
1 2 3
ana
1 2 3
:a'la * a
1 2 3
ODD
1 2 3
nan
1 2 3
nan
1 2 3
D a D
1 2 3
nan
1 2 3
D D a
1 2 3
nan
123
D D a
-------
7. (continued)
(For each symptom, answer the first
question. If the response is "never,"
go down to the next symptom.)
i~ chills
t fever,
u. aching muscles or joints
y. feeling depressed
z. tension or nervousness
aa. dHficurty concentrating .
bb. dry or Itchy skbi... ?f.'.
«; , , K»& X ' *&*'" i
cc. pain or stiffness in upper back .
dd. pain or stiffness in lower back .
ee. pain or numbness In
shoulder/neck
v. problems with contact lenses ...
w. difficulty remembering things
ff. pain or numbness fa
' <, hands or wrists ,;..
Please indicate how often
during the LAST YEAR
you have experienced this
symptom while working
in this building.
Some-
Never Rarely times Often Always
a
Please indicate
how many days
LAST WEEK you
experienced this
symptom while
working in this
building.
(Fill in No. of days)
Does the
symptom usually
change when
not at
Gets Stays Gets
Worse Same Better
2
D
t 2
D D
1 2
D D
1 2
D D
2
D
1 2
D D-
12
D D
1 2
D D
|P1' 2'<
D E3
3
D
3
-q
3
D
3
D
3
,,D
3
D
3
3^ P
-1 ' "2 3
ODD
""V 2 3
D D D
1 2 3
ODD
1 2 3
R,' D D
1^ 2 3
ODD
6
-------
NOTE: The next four questions (Questions 8-11) refer
to your symptoms described in Question 7.
If you reported that you never experienced
any of these symptoms, go to Question 12.
How often during the LAST YEAR have any of
your symptoms reduced your ability to work in
this building?
1. Q Never
2. Q] Rarely
3. Q Sometimes
4. D Often
5. D Always
9. a. Have any of your symptoms caused you to
stay home from work or leave work early
during the LAST YEAR?
1.
2.
3.
4.
Never
Sometimes
Often
b. Which symptoms?
10. In which season(s) are you bothered more by the
symptoms you reported in Question 7? (Check all
that apply.)
1. Q Winter
2. Q] Spring
3. Q] Summer
4. n Fall
5. n No relation to seasons
1 1. a.
Do you associate any of the symptoms you
reported in Question 7 with your work in this
building?
b. Have these symptoms:
1. £] improved over the last year
2. CH become worse over the last year
3. Q stayed the same
12. During the LAST YEAR, have you had an illness
in which you had repeated episodes of THREE
OR MORE of the following symptoms at the same
time: wheezing, cough, shortness of breath,
fever, chills, aching joints/muscles?
1.
2.
D NO
D Yes
13. During the LAST YEAR, have you had any chest
illnesses, such as bronchitis or pneumonia,
that have kept you off work, indoors at home,
or in bed?
1.
2.
D No
D Yes
14. Has a physician ever told you that you have, or
had, eczema?
1.
2.
D No
D Yes
15. During the LAST YEAR, have you had any
episodes of wheezing (whistling in the chest)
WITHOUT fever, or chills, or sore throat?
1.
2.
D No
D Yes
-------
16. a. Has • physician ever told you that you have,
or had, asthma?
1. D No
2. D Yes
b. In what year was it first diagnosed?
19
c. Have you had an asthma attack during the
LAST YEAR?
No
1.
2.
17. Comparing your health since working in this
building with your health before you began to
work in this building ...
a. ... do you have infections (e.g., colds, flu,
bronchitis, etc.)...
1. CD more frequently?
2. CH less frequently?
3. []] with the same frequency?
b. ... do your infections (e.g., colds, flu,
bronchitis, etc.) tend to...
1. O last longer?
2. O last a shorter amount of time?
3. last about the same amount of time?
18. Do you believe you are or may be allergic to
any of the following? (Check "no" or yes" for
each item.)
a. pollen or plants
b. animals
c. dust
d. molds
e. Other (specify) .
No
1
D
Yes
2
D
D
D
D
8
19. During the LAST YEAR, how often do you believe
you have experienced EYE, NOSE, THROAT, OR
RESPIRATORY IRRITATION at your workstation
from:
*' ALWAYS 'Si
OFTEN
,•#' .' SOMETIMES
RARELY
•\;i;'V NEVER «£
a. Tobacco smoke . . .
b. Fumes from a
photocopying
machine
c. Fumes from
printing processing
(press, binding
materials, etc.)
d. Fumes from other
chemicals such
as adhesives,
glues, cleansers, .
white out, rubber
cement, etc
e. Fumes from
pesticides
f. Fumes from
new carpeting
g. Fumes from
new drapes,
curtains, or
furniture
h. Fumes from
paint
i. Fumes from
cleaning of carpets,
drapes, or other
furnishings
j. Other (specify)
-,fl
a
^
t
p,
" * ••>
-
a
•'"
,
t
p.
t
a'
1
a
^
5^
f"
Q
i
ar
"
1
a
,
D
2
a
2
a
2
D
2
a
2
n
L__J
2
a
2
D
2
a
2
a
2
D
**• ^
I x ^
3
JP
H'
3
D
3
D
~w
3
a
3
n
i__j
' -3
'B
, *A
3
P
3
D
3
a
3
a:
4
D
4
D
4
D
4
D
4
h
L_J
4
D
4
D
4
D
4
D
4
«' J
5
D
jyp
- /
5
D
y<
&
s
q
%*
'*&m
" ^'
,*!|
'5j|
,q
5
S
a
^i
«j
8*
Q|
5
D
5
D
5
3
-------
20. Do you consider yourself especially sensitive to
any of the Hems in Question 19?
| 1. D No
2. D Yes
21. How old are you?
years
Go to Part III onpg. 11
Women working In office buildings have occasionally
reported patterns of gynecological or women's health
problems. The following questions have been included
to help sort out some of these issues in this building.
As with the rest of the questions in this survey, your
responses are entirely voluntary and will be kept
confidential.
23. During the LAST YEAR have you menstruated
(had a period)?
1
2.
D
D
Mn »
Yes
Go
to 0.29
24. How often during the LAST YEAR has your
period been regular? (By regular, we mean
your periods come about once a month, you
can usually predict when they will come plus
or minus 4 days, and each time they last about
the same number of days.)
1, D Never
2. D Rarely
3. D About half the time
4. D Often
5. D Always
25. a. How many days does your menstrual flow
(period) typically last?
days
b. During the last year, what was the LONGEST
period you had?
days
c. During the last year, what was the SHORTEST
period you had?
days
26. a. How many days does your cycle typically
last? (Count from the first day of one period
to the first day of the next.)
days
b. During the last year, what was the LONGEST
cycle you had?
days
c. During the last year, what was the SHORTEST
cycle you had?
days
27. How often during the LAST YEAR has there been
bleeding or spotting between your periods?
1. Q Never
2. EH 1 - 3 times
3. n 4-6times
4. D 7-9times
5. O 10 or more times
9
-------
28. •.
Some women experience menstrual
symptoms, such as headaches, weight
gain, irritability, cramping, breast
tenderness, or back pain. How often
have you experienced any of these
menstrual symptoms during the LAST
YEAR?
1.
2.
3.
4.
Never
Go to Q.29
1 ' 3 times
4 -6 times
7 -9 times
5. 0 10 or more times
b. When you experience these symptoms,
typically how severe are they?
1. Q Mild; could be ignored at times
2. O Moderate; pain, bloating, or mood
change noticeably present
3. O Severe; difficult to do most tasks
4. CH Extreme; incapacitating
29. During the LAST YEAR have you been . . .
(Check "no" or yes" for each Item.)
No
1
a.
b.
c.
d.
e.
Pregnant or nursing?
Taking birth control pills? .
Going through menopause
(change of We)?
Post-menopausal
(completed menopause)? ..
Taking estrogen replace-
ment therapy?
Yes
2
D
D
D
D
D
10
30. a. During the LAST YEAR have you been taking
hormones prescribed by a physician?
b. Specify what kind(s) and what they were
prescribed for.
31. a. Has a physician ever told you that you
had ... (Check "no" or yes" for each item.)
Year
No Yes First
1 2 Diagnosed
Fibroids? D D
cysts? ..... n n
Enlarged uterus? []] Q
If all are "no," go to Part III
b. Have there been noticeable changes during
the last year? (Check one box for each item.)
Other,
Decreased Increased No Specify
In Size In Size Change Below
1 234
Fibroids . . .
Cysts
Enlarged
uterus
Specify
D
D
D
D
D
D
D
D
D
D
D
D
-------
PART III. INFORMATION ABOUT YOUR PRESENT WORK ENVIRONMENT
M u. how you, worK
1. At your present workstation,
HOW OFTEN...
(Please check one box for
last year and one box for
last week.)
rt * /•
toolrttle
wement?
d. was the temperature
too hot?
I e. was the temperature
r too cold?
f. did you want
to adjust the
temperature?
j« "•" ^VX•
b- was ft too dry?
.. during the LAST YEAR
e"vronment « *«»
workstation. You
... during the LAST WEEK
1
^
i.
k.
i.
m.
WWPWA.W. - SA. 5
to adjust th«
/humidity?.... ;
-• 'vtfKA ^ ^ ^* •"** ">#•'&&•
-w. \ •. v. v?v % ^vis % -XX-.*
was the air too
stuffy?
was it too noisy?
was tt too quiet?
was the work
area too dusty?
n
1 *— '
1
1
D
!
D
1
* v
•" l_ j
rt*-A
2
2
D
2
D
2
"3
3
3
D
3
D
3
1 Never
•m
\ £-L
I ' ''§
1 1 v
• •• •. x**
I *
U1
Rarely
2
W*r .-
^ V '
/fiv ^4' / .,
§o
'''i
^
2
I £&f
f 3 '5
D ^
3
D
3
3
D
3
D
—
Often
f
4
D;
" -f
4
.P,
? ' ^
yj £•.;:
o '
•tJ
4
D
4
D
4
D
,A '^
* 1
-D
^
4
D
**
-^"4 ^
P
4
D
4
D
4
D
4
D
~
Always
/x
S
' D ,
t.
s '--A
+3t SfsA
,- -i •">'•? i
'
'
* f S '
"a '
5
D
5
D
5
W
a
' " Xrf3t>
D ?r
s
D
yV^.
^5
D
5
D
5
D
5
D
5
a
^— — ^.^
11
-------
2. During the LAST YEAR, how often, if at all, have
you noticed any of these types of ODORS at your
present workstation? (Check one box for each Item.)
• « ALWAYS
OFTEN
SOMETIMES ,
RARELY
V\ NEVER
§. Body odor
b. Cosmetics, such
as perfume or
after-shave
c. Tobacco smoke . . .
d. Fishy smells
e. Other food smells . .
f. Musty or damp
basement smells . .
g. Odors from
new carpet
n. Odors from new
drapes or curtains .
i. Odors from diesel
or other engine
exhaust
j. Odors from a
photocopying
machine
k. Odors from
printing processing
(press, binding
materials, etc.) ....
1
n
.- 3
s;>i^
^*
1
^
|3J
IS
1
°;
1
D
1
a
1
0
yf'
a-
s#
;W
Q
.* * »>
D
ft
2
a
2
1=1
2
D
2
a
2
D
2
D
2
a
2
D
2
a
2
D
2
D
3
D
>^ ••£
j*®
3
S
^
n
'V.
:ft'
3
D
D
3
a
-3
i:
*?•«
3t"
a
,,
3
a
S
4
D
4
a
4
a
4
a
4
a
4
D
4
a
4
a
4
a
4
a
4
a
1
s
O
•\
' "i!"
V
Jg^
S
' ^
It
'ir
S
s
a
s
s
O
s
Q
$
%
s
D
•*•
'&
12
M$< *'JP&?*&* f&ats ' • ALWAYS n
OFTEN
v* W SOMETIMES :
RARELY
NEVER
2. (continued)
1. Odors from other
chemicals such as
adhesives, glues,
cleansers, white
but, rubber cement,
pesticides, etc
m. Odors from
pesticides
n. Odors from clean-
ing of carpets,
drapes, or other
furnishings
o. Odors from
paint
p. Other unpleasant
odors (describe) . . .
*
r.
a.
Ii7
r^i«
t
%^A''
' l#>
Q;
1
D
V 1 V
*x
2
n
2
n
L_J
2
a
2
n
2
a
/'
> '
-,•.•",
\ y,*
>i «•
f
3
a
&
'[ f
fe
" *•-:
3
m-
'#&
n
3
^J;
4
D
4
1 1
4
D
4
n
4
D
&
•?•
\
,
'»
,*3
5
D
^1
-Pi'
?!
s
a
«'
a'
.
a
3. In which seasons would you most like to adjust
the physical conditions around your workstation?
(Check all that apply)
a. Air movement..
b. Temperature ...
c. Humidity
d. Odors
None Winter Spring Summer Fall
1 2
1 2
3
3
3
D
3
D
4
D
4
4
D
4
5
D
5
D
5
D
5
D
-------
4. Please rate the lighting at your workstation.
1. n Much too dim
2. n A little too dim
3. D Just right
4. D A little too bright
5. D Much too bright
5. a. Do you experience a reflection or "glare"
in your field of vision when at your
workstation?
1. d Never
2. d Sometimes
3. d Often
4. d Always
b. Where does the reflection or glare come
from? (Check all that appty)
1. Q Window, sunlight, outside reflection
2. d] Overhead fluorescent lights
3- D Video display screen and/or
reflections when looking at screen
4. D Desk lamp
5. CH Other (specify)
6. Can you see out an outside window from your
workstation?
1.
2.
N°
Yes
7. a. How comfortable is the chair at your
workstation?
1. d Reasonably comfortable
2. d Somewhat uncomfortable
3. d VerV uncomfortable
4. d Don't have one specific
chair » | Go to Q.8 \
b. Is your chair easily adjustable?
1. d No
3. d Not adjustable
8. How comfortable is the current set-up of your
desk or work table (that is, height and general
arrangement of the table, chair, and equipment
you work with)?
1.
2.
3.
4.
9. a.
Reasonably comfortable
Somewhat uncomfortable
Very uncomfortable
Don't have one specific desk or
work table
During the LAST YEAR, how many times
per week did you go outdoors, weather
permitting, during work hours (for lunch,
break, or other reasons)?
time(s) per week •
If zero, goto Q. 10
b. How many of these times did you go
outdoors primarily to get some fresh air?
time(s) per week for fresh air
13
-------
NOTE: The next four questions concern the overall
physical environment at your workstation,
that is, the air quality, temperature, light,
noise, odor, etc.
10. During the LAST WEEK, how satisfied were you
with the physical environment at your workstation?
1. Q Very satisfied
2. Q Somewhat satisfied
3- D Not too satisfied
4. Q Not at all satisfied
11. During the LAST YEAR, how satisfied were you
with the overall physical environment at your
workstation?
1. D Very satisfied
2. Q Somewhat satisfied
3. D Not too satisfied
4. D Not at all satisfied
14
12. During the LAST YEAR, has the overall physical
environment in the vicinity of your workstation:
1. Q improved
2. Q become worse
3. Q stayed the same
13. During a typical work day, does the overall
physical environment in the vicinity of your
workstation:
1. Q improve during the day
2. O become worse during the day
3. n staVthe same
-------
r
PART IV. CHARACTERISTICS OF YOUR JOB
This section asks you to describe your job in terms
of specific qualities. In order to gain a better under-
standing of your work environment, we would like to
know how you feel about your job situation. As stated
before, your responses will be kept confidential.
1. We would like you to think about the TYPE OF
WORK YOU DO IN YOUR JOB. (Check one box
for each statement)
a. All in all, how satisfied are you with your
job?
1. G Very satisfied
2. G Somewhat satisfied
3. G Not too satisfied
4. G Not at all satisfied
b. Knowing what you know now, if you had
to decide again whether to take the job
you now have, what would you decide?
Would you...
1. G Decide without hesitation to take the
same job
2. G Have some second thoughts
3. G Decide definitely not to take the same
job
c. If you were free right now to go into any type
of job you wanted, what would your choice
be? Would you...
1. G Take the same job
2. G Take a different job
3. G Not want to work
d. If a friend of yours told you he/she was
interested in working in a job like yours,
what would you tell him/her? Would you ...
1. Q Strongly recommend it
2. Q Have doubts about recommending it
3. Q Advise against It
2. How satisfied are you with your salary?
1. G Very satisfied
2. G Somewhat satisfied
3. G Not too satisfied
4. Not at all satisfied
3. How satisfied are you with your opportunity
for advancement at EPA?
1. G Very satisfied
2- Q Somewhat satisfied
3. G Not too satisfied
4. G Not at all satisfied
-------
4. Conflicts can occur in any Job. For example,
•omeone may ask you to do work in a way which
is different from what you think is best, or you
may find that ft is diffcutt to satisfy everyone.
HOW OFTEN do you face problems in your work
like the ones listed below? (Check one box for
each statement)
< VERY OFTEN, ,
FAIRLY OFTEN
SOMETIMES <*
RARELY OR NEVER
a. Persons equal in
rank and authority
over you ask you
to do things which
conflict.
b. People in a good
position to see if
you do what they
ask give you things
to do which conflict
with one another
c. People whose
requests should
be met give you
things which
conflict with
other work you
have to do
1
n
i__i
1
D
i
D
fj»,
If
'
2
n
Uw*'
#&*
'M^
Jty?
'K
'2-
a
-
f
*,
'
2
D
3
n
* *
3
a
3
D
/•
'
4
n
U*+4
v
- , ]
4
%]
'"*:
j. '•
ff
4
D
16
5. The next series of questions asks HOW MUCH
influence you now have in each of several areas
at work. By influence we mean the degree to
which you control what is done by others and
have freedom to determine what you do yourself.
(Check one box for each question)
VERY MUCH
MUCH
A MODERATE AMOUNT i
LITTLE
'.., ,<><* , •* VERY^ITTLE
a. How much
influence do
you have over
the amount of
work you do?
b. How much
influence do
you have over
the availability
of materials
you need to
do your work?
c. How much do
you influence
the policies
and procedures
in your work
group?
d. How much
influence do
you have over
the arrangement
of furniture and
other work equip-
ment at your
workstation?
1
n
£ #
* *~
•*"*,*,
*»f <
.. •*
1 ';
D!
,
t..
rV
:•
e
v.'SK
f-y
1
a
2
a
2
a
2
n
1 1 i
2
a
*i
'f
31
n
m
'''
'"*&
T*
V.
3|
D]
'
3
n
.!••»• i
3
D
4
D
4
a
4
n
L — *
4
a
f
' '
5
:S
I.:',
' *
-*"'h
'^
*',
a
T/
^ '?
5
D
U~*J
5
D
-------
6. The next series of questions asks HOW OFTEN
certain things happen at your job. (Check one
box for each question)
VEfTY OFTEN
FAIRLY OFTEN
OCCASIONALLY
a. How often does
your job require
you to work
very fast?
b.
c.
f.
How often does
your job require
you to work
very hard? ....
How often does
your job leave
you with little
time to get
things done? ..
d. How often is
there a great
deal to be
done?
e. How often does
your job let you
use the skills
and knowledge
you learned in
school?
How often are
you given a
chance to do
the things you
do best?
E3.
2
a
2
D
2
D
a
n
4
D
4
n
4
D
D
.-6
5
a
D
">"'Z',> «* VERY; OFTEN
FAIRLY OFTEN
',- , /', SOMETIMES
OCCASIONALLY
>?""-'?•$ '• ' „ RARELY ;
6. (Continued)
g. How often can
you use the
skills from
your previous
experience and
training?
h. How often are
you clear on
what your job
responsibilities
are?
i. How often can
you predict
what others
will expect
of you on the
lob?
j. How much of
the time are
your work
objectives well
defined?
k. How often are
you clear about
what others
expect of you
on the job?
jsff'f $&
^m
m
* y^sffi
•%"!
% '
''&?%
^%4>
'F''!*
- -,
-••,,
'' ^
* »
-,'*
•'.
vV
,,
\ »
u
*
1
a
2
n
t__i
2
a
2
a
2
a
2
D
-'*
-
^
'3 %
t i
s/ •"
•til
ip
,
3
IR:
'^ff-f-
f
< 3
a
•
^
3
D
^
3
D
4
n
1 *
4
a
4
a
4
a
4
D
r'
',*
M
n
•L— *
f
5
D
xi
5
D
A/
<**
5
a
5
D
17
-------
7. In order to better understand your responsibilities
outside your normal working day, the next series
of questions deals with other significant aspects
of your life. (Check "no" or "yes" for each question)
No Yes
1 2
Do you have children
at home?
Do you have major
responsibility for
childcare duties?
c. Do you have major
responsibility for
housecleaning duties?
d. Do you have major
responsibility for the
care of an elderly or
disabled person on a
regular basis?
Are you taking courses
for credit toward a
degree or a diploma?
f. Do you have a regular
commitment of five
hours or more per week,
paid or unpaid, outside
of this job? (Include
volunteer work, charitable
work, second job, etc.)
18
-------
PARTY. CONCLUDING QUESTIONS
This section concludes this survey. Your answers
to these questions, like your answers to the previous
questions, will be kept confidential. This information
is needed for statistical purposes.
1.
What day of the week did you complete this
survey?
1.
2.
3.
4.
Monday
Tuesday
Wednesday
Thursday
5. D Friday
2. Which of the following best describes your current
living and financial arrangements?
1 . Q Live alone, sole provider of rent/mortgage,
utilities, food, and other living expenses.
2- Q '-'ve alone- Dut receive assistance from
one or more others in paying rent/mortgage,
utilities, food, and other living expenses.
3. Q Live with one or more other persons, but
sole provider of rent/mortgage, utilities,
food, and other living expenses.
one or more otner persons who
help to pay rent/mortgage, utilities, food,
and other living expenses.
3. What is the highest grade you completed in
school?
1. '.Qj 8th grade or less
2. n 9m. 10th, or 11th grade
3. D High school graduate
4. D 2 years of college or Associate Degree
5. Q Bachelor's or technical degree
6. Q Some graduate work
7. n Graduate or professional degree
4. a. What is your pay plan and grade (e.g.,
GS-5, GM-14, SES-2, WG-2, etc.)?
b.
Which of the following best describes your
Job duties and responsibilities? (If more than
one applies, check the ONE box for the job
duties on which you spend the most time.)
1. n Managerial (such as administrator,
manager, etc.)
2. n Professional (such as engineer,
scientist, lawyer, etc.)
3. [^ Technical (such as technician,
programmer, etc.)
4. Q Administrative Support (such as
clerical, computer operator, etc.)
5. Q Service (such as health services,
food preparation, janitorial, etc.)
6. [U Craftsman (such as mechanic,
repairer, etc.)
7. [^ Operator or laborer
8. Q Other (specify)
The following information is needed so that your
workstation can be located within this building. This
is necessary so that we can relate your responses to
the air measurements that will be taken in a few weeks.
As with the rest of the questions in this survey, this
information will be kept confidential. Please tell us:
5. a. Your room number
b. Your workstation telephone number (your
direct or private number.)
19.
-------
6. Is there anything else you would like to tell us about environmental or health matters in this building?
If so, please use this space provided for that purpose.
Please put your completed questionnaire in the return envelope provided. Seal it and take it to one of the
return boxes located near the elevators and building exits.
PLEASE READ THE NEXT PAGE
20
-------
In a few weeks we plan to conduct air measurements in this building.
At that time people whose workstations are close to the air
measurement locations will be asked a few additional questions. You
may be recontacted at that time.
Thank you very much for your time and patience in filling out this
questionnaire.
2.1
-------
APPENDIX B
IDENTIFICATION OF AIR HANDLING UNITS FOR EPA BUILDINGS
B-l
-------
IDENTIFICATION OF AIR HANDLING UNITS FOR EPA BUILDINGS
Poor indoor air quality may be the result of the ventilation system.
The system may not supply enough fresh air, or it may be contaminated. In the
former case, symptoms might decrease as the air flow increases; in the latter
case, they may get worse with increasing air flow. On the other hand, if the
air flow is short circuited (a portion does not reach the occupant), the
amount of air flow might be irrelevant, and symptoms would show no dependence
on the flow rate, although they might show a relationship with individual air
handling units. In order to handle these possibilities, two variables were
created and added to the data set. One variable was created by identifying
every room with an air handling unit (AHU). The second variable consists of
a total air flow capacity associated with each AHU. The following section
provides a description of the information sources used to create these
variables.
All the air handling units (AHUs) that were bringing in outside air to
the Waterside Mall, Fairchild, and Crystal Mall buildings at the time of the
questionnaire census and monitoring survey (February-March 1989) were
identified. A total of 42 such units have been identified at Waterside.
(Another 41 air conditioning units were identified, but most of these have
been determined to provide recirculation only, without bringing in any outside
air.) At Fairchild, two units supplied the entire building. At Crystal, a
total of five units were identified.
Sources of Information
Several sources of information have been used to determine the airflow
capacities and rooms served by each AHU:
(1) Blueprints prepared in 1980-81 by Town Center Management
(2) A balancing report prepared in 1986
(3) Measurements performed during March 1989 by K. Maclachlan
Associates
(4) A report by Engineering Design Associates (Jan. 1990)
(5) Conversations with Mike Shamus of EPA's OARM
(6) A telephone conversation with John Mayoral, building manager at
Fairchild
(7) A telephone conversation with the office of Gary Wadkins,
chief engineer for Charles E. Smith (owner of Crystal Mall building)
B-2
-------
In some cases connected with the Waterside Mall building, these sources
did not use the same identification system for the AHUs; in these cases,
attempts were made to reconcile differences through discussions with M.
Shamus.
Assignment of AHUs to respondents
A series of maps was prepared from the blueprints contained in the EDA
report identifying the areas served by each air handling unit. Based on these
maps, each respondent was then assigned a unique air handling unit according
to his or her room number as identified on the questionnaire. The final
assignments of AHUs are contained in a SAS file to be made available in the
Public Use Data Tape. The color-coded maps are available for inspection at
the office of the author of this report, Dr. Lance Wallace (703-349-8970).
This information on AHUs was then combined with information on room
location to test two hypotheses: 1) that specific AHUs might be associated
with health symptoms, comfort, or odor complaints; 2) that certain locations
might have higher prevalence rates for these conditions.
Calculation of capacities
Supply air flow rates for 31 AHUs were measured during the week of
monitoring in March of 1989. Supply air flow rates for most of these 31 AHUs
and for a number of additional AHUs were also determined in the 1986 balancing
reports. In most cases, the 1989 flow rates were within 30% of the 1986 flow
rates (Table B-l).
In the absence of measured flow rates, supply air design capacities (in
cubic feet per minute or in tons) have been included. For five cases where
both a measured flow rate and a capacity (in tons) were provided, the
relationship between the two was fairly consistent (one ton corresponded to an
average measured flow rate of about 320 cfm, with an observed range of 240-400
cfm) . This relationship was then used to convert all capacities originally
given only in tons to cfm units.
These measured or calculated supply air capacities were then summed
across all AHUs serving a particular sector, such as the West Tower, 2nd floor
of the Mall, etc. The areas of each of the sectors were then calculated from
the blueprints, and the supply airflow capacity per unit area within each
B-3
-------
cfm). This relationship was then used to convert all capacities originally
given only in tons to cfm units.
These measured or calculated supply air capacities were then-summed
across all AHUs serving a particular sector, such as the West Tower, 2nd floor
of the Mall, etc. The areas of each of the sectors were then calculated from
the blueprints, and the capacity per unit area within each sector was
determined (Table B-2.) A common design goal is 1 cfm/sq.ft., and the table
indicates that some sectors are quite close to that mark. The newest sector,
Northeast Mall, appears to exceed that mark by a factor of two. The garage
and lower garage areas, which include the print shop, appear to fall below
that mark by a factor of two. (It should be noted here that a new AHD has
since been added to supply the print shop with additional fresh air.) The
next lowest sectors were the second and third floors of the Mall, which
reached about 70% and 80% of the design goal, respectively. (Again new AHDs
have been added to both of these sectors since the time of the study.)
To derive a per capita capacity (cfm per person, or cfmp) the number of
people served by each AHD was calculated by using the computerized form of the
EPA Directory for the period of the survey (February-March 1989) . This data
file is also contained in the Public Use Data Tape. This per capita capacity
was used in testing the hypothesis that health or comfort might be affected by
the amount of airflow provided.
Calculation of outside air provided per person
During the monitoring period, the amount of outdoor air supplied by 30
AHUs was calculated (EPA, 1990a). For these cases, it is possible to
calculate the average amount of outdoor air per person (cfmp) by dividing the
measured outdoor air flow rate by the number of people served. (In part of
the second floor of the Mall, this may be a misleading calculation, because
the high ceilings offer increased possibility for short-circuit flow from the
ceiling diffusers to the ceiling returns, thereby not delivering fresh air to
the workers below.) These calculations are summarized in Tables B-3 and B-4.
However, since direct information on outside air delivered is lacking
for the remainder of the AHUs, which constitute about half of the total air
flow in the three buildings, no attempt was made to use a measure of fresh air
delivered per person in the analysis of the questionnaire responses.
B-4
-------
RECORD OF DECISIONS ON IDENTIFICATION OF AHUS
The various sources of information on AHUs sometimes disagreed or were
ambiguous regarding identifications, capacities, or areas served by AHUs.
Some of the decisions made are documented below. The discussion is organized
by the page number of the blueprints incorporated into the 1990 EDA report.
BP Sector/
Pg. Floor Discussion
26 SE2 AHU#3 becomes SE2-1
Unit "A" becomes SE2-2
27 SE3 AHU #1 becomes SE 3-1
AHU #2 becomes SE 3-2
EDG reports three units on SE 3rd floor. From internal
evidence, I associate EDG's "SE3" with AHU#3 on second
floor. Thus McLachlan's SE3 - EDG's SE3 - our SE2-1.
28 M2 Five units (Rooftop 1, 2,and 3; South Unit 2; Split Unit 1)
were installed after the monitoring period. At the time,
these areas were served by AHU #21 in Fan Room #11.
(Personal information from M. Shamus; documented by K.
McLachlan)
Unit A-11-1 serves only TV room--no offices
29 M2 AHU -17 serves non-EPA areas (Shamus)
Unmarked area including rooms 2501-03 served by AHU-3
(Shamus)
30 M2 Two separate unmarked areas (including rooms 2107 and 2416,
respectively) are both assigned to AHU #2 (Shamus)
31 SW "New 15-ton Unit" becomes SW-2
33 M3 AC-2, 9, and 15 assumed to provide no outside air
Unmarked area (Room 3519) not assigned (no questionnaires
back from that room anyway).
34 M3 AC-3, 10, 11, 14, 3100, 3105-7. and BWE09C all assumed to
provide no outside air
AHU-11 assumed to be gone or inoperable
35 M3 AC-5, 6 assumed to provide no outdoor air
B-5
-------
36 M3 AHU-7 serves commercial space (Shamus)
AC-7, 12, 13, 16 and 3706 assumed to provide no outdoor air
37 LG100 One unit
38 G-100 Two units. Areas are marked on blueprint (by Shamus), but
room identifier cannot be discerned.
39 NEB AC-A (Liebert unit) in B029 assumed to provide no outside
air
Two unnamed units are named NEG-1 and NEC-2
40 NEC "AHU-1 from basement" (our NEG-1) serves only corridors.
Outdoor AHU (NEC-3) serves low-numbered rooms (Shamus)
AHU #4 becomes NEG-4
43 NE3 This blueprint is missing. Shamus says there are two AHUs.
No way at present to identify rooms.
B-6
-------
TABLE B-l. COMPARISON OF MEASURED SUPPLY AIRFLOW RATES (cfm) FOR SELECTED
AIR HANDLING UNITS IN WATERSIDE MALL
AHU ID 19891 19862 89/86
1
4
5
6
7
9
10
12
13
14
15
16
18
19
20
SE2-2
SE3-1
15809
16105
9636
28440
12930
17635
7644
11422
12036
10282
27868
10780
14250
7562
8921
3659
27350
14300
13507
21522
24277
11305
12640
5390
11575
12755
7385
17790
8940
10285
5755
6745
3605
25325
1.11
1.19
0.45
1.17
1.14
1.40
1.42
0.99
0.94
1.39
1.57
1.21
1.43
1.31
1.32
1.02
1.08
1 Measurements March 6-10, 1989
by McLachlan Associates
2 Results from 1986 balancing report,
as shown in 1990 report by
Engineering Design Group (EDG)
B-7
-------
Table B-2. ESTIMATED SUPPLY AIRFLOW RATES PER SQUARE FOOT BY SECTOR
AREA (ft2) FLOW RATE CFM
SECTOR (OOP) CFM (OOP) ft2
Northeast Mall 52 1P1 1.9
West Tower 18P 181 1.0
East Tower 195 164 P.9
Southeast Mall 59 52 P. 9
3rd Floor Mall 157 122 P.8
2nd Floor Mall 2P5 148 P.7
Garage & Lower Garage 44 18 P.4
B-8
-------
Fable B-3. Outside Air Delivered per Person by'*W Handling Units at Waterside Mall,
Capacities in Cubic Feet per Minute per Person (CFMP)
Areas Served,''
NO.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
W 15
«o 16
17
19
22
ID
1
4
5
6
8
9
12
13
2
3
10
14
16
18
19
20
21
B
E
Service
Sector
M3
M3
M3
M3
M3
M3
M3
M3
M2
M2
M2
M2
M2
M2
M2
M2
M2
M2
M2
Sample
Room
3402
3234
3201
3002
3702A
3702C
3600
3100
2107
2501
2701
2102
2817
2821
2603
2632
2902
2123
2435
Capacity*
(CPM)
15809
16105
9636
28440
11329
17635
11422
12036
10095
12566
7644
10282
10780
14400
7562
8921
15574
2255
3285
Occupants*
108
85
142
139
60
73
57
59
31
24
26
109
68
92
19
32
49
8*
24
Occ.
146
189
68
205
189
242
200
204
326
524
294
94
159
157
398
279
318
282
137
CFM
Air (%)'
27
18
22
26
16
23
15
20
25
24
13
60
58
58
23
17
20 1 49*
19
16
Fresh Air
(CFMP)
39
34
15
53
30
56
30
41
82
126
38
56
92
91
92
47
64; 156
54
22
26
29
31
42
15
SE2-2
SE3-2
ETB
SW,M2
SE
SE
ET
256
312
55
27868
3605
12700
7219
114
30
83
38
244
120
153
190
52
18
45
43
127
22
69
82
• Supply air capacities and % outside air provided from measurements in March, 1989.
* No. of occupants from EPA Computerized Directory for March, 1989.
• Number of people returning questionnaires, when this number is larger than the number given by the EPA
Directory.
d Measurements made 3/7/89 and 3/8/89.
-------
Table B-4. Capacities, Number of Occupants Served, and Outside Air Delivered
by Air Handling Units on Bach Floor of the Towers
EL.
Number of Occupants on Bach Floor*
Bast Tower West Tower
ET1 ET2 BT3 WT1 WT2 WT3
Measured
Capacities* (cfm)
BT1 BT2
1
2
3
4
5
6
7
e
9
10
11
12
14
30
52
40
38
28
41
35
17
30
31
15
30
36
37
30
24
19
29
42
30
18
35
12
19
32
30
40
17
49
32
27
19
20
27
39
38
20
33
23
25
34
23
16
27
35
33
34
42
41
33
42
37
30
18
14
25
41
33
26
17
5
4
0
5000
5890
7670
4630
4540
4320
4840
4290
4360
4690
4100
4380
4750
5090
6855
4213*
4131*
3931*
4404*
3904*
3968*
4268*
3870
3890
Tot 367 342
265
298 370 165
58710 53470
Average Capacities by Floor" (cfm) and Floors Served by West Tower, Bast
Tower-3
WTl 6564 (1-12)
WT2 5062 (1-12)
WT3 4121 (4-11)
BT3 3002 (4-11)
Per Capita Capacities (cfmp)
WTl WT2 WT3
1
2
3
4
5
6
7
8
9
10
11
12
357
196
148
116
119
154
118
123
256
156
132
292
158
141
185
140
172
207
152
93
132
237
111
324
158
94
100
75
176
61
94
111
158
328
274
168
173
328
199
285
263
193
285
410
187
145
153
149
120
123
153
121
137
169
281
294
165
100
125
159
242
824
999
B-10
-------
Table B-4 (Continued)
Outside Air Delivered (cfmp)
ZLu ETl EE2. BT3 WTl WT2 WT3
(23)' (41) (83) (53) (36) (30)
1 82 65 ---
2 45 58 --- 173 67
3 34 76 145 52
4 27 57 131 89 55 88
5 27 71 78 92 54 50
6 35 85 83 174 43 30
7 27 62 62 105 44 38
8 28 38 146 151 55 48
9 59 54 51 139 44 73
10 36 97 78 102 49 247
11 30 46 92 151 61 300
12 67 133 131 217 101
* Calculated values
" From EPA computerized Directory (about March 1989)
* From 1986 balancing reports
' From March 1989 measurements: total capacity/(no. of floors)
4 From March 1989 measurements: percent outside air
B-ll
-------
APPENDIX C
METHODS AND RESULTS OF PRINCIPAL COMPONENTS ANALYSIS OF
HEALTH, COMFORT, ODOR, AND PSYCHO-SOCIAL QUESTIONS
C-l
-------
METHODS AND RESULTS OF PRINCIPAL COMPONENTS ANALYSIS OF
HEALTH, COMFORT, ODOR, AND PSYCHO-SOCIAL QUESTIONS
Three types of dependent variables were included in the
questionnaire: health symptoms, comfort concerns, and odors. Each
category had multiple questions, some of which might be highly
correlated (e.g., dry throat with sore throat). Running separate
regressions on highly correlated dependent variables would be
redundant and costly. Therefore a principal components analysis
(PCA) was carried out on each of these three types of dependent
variables to identify the major factors within each category. The
Varimax rotation method was used to provide the maximum amount of
differentiation of the factors. Factors are listed in order of the
variance explained by each factor. (See Tables 7-10 for the amount
of the variance.)
Health symptoms
There were 32 questions on health symptoms. One symptom was
problems with contact lenses, which applies only to the subgroup of
persons wearing contact lenses. Therefore this symptom was
eliminated from PCA. The PCA on the remaining 31 symptoms (Table
C-l) identified a total of nine factors. The symptoms associated
with each factor are identified by underlining. For example, the
major components of factor 9 are headache and nausea. Two symptoms
— dizziness and dry skin — did not load heavily on any factor.
Thus, in all succeeding linear and logistic regressions, a
total of 12 health symptom factors were employed as dependent
variables.
Comfort concerns
Ten questions on perceived air movement, temperature, and
humidity were included in the questionnaire. The PCA identified
four factors (Table c-2). The first factor included perceptions of
too little air movement and air that was too hot and too stuffy.
Other factors included perceptions of dry air, cold drafty air, and
C-2
-------
humid air. All four factors were employed in all succeeding
regressions.
Odors noticed
Fifteen question on odors were reduced to six factors by PCA
(Table C-3). The first factor was chemical odors, including odors
of fresh paint and carpet cleaning. The second factor included
cosmetics, body odor, and food smells (other than fishy odors).
Tobacco smoke odor did not correlate with any of the other odors.
All six factors were employed in the succeeding analysis.
Psycho-social factors
Besides the three dependent variable categories described
above, one set of independent variables was also subjected to PCA.
This group consisted of all questions in Part IV of the
questionnaire: the so-called "psycho-social" questions dealing with
job satisfaction, perceived workload, etc. These questions were
drawn from a larger set of questions that have often been used in
workplace investigations by industrial hygienists and
psychologists. The parent set of questions has been repeatedly
validated; however, it was felt to be desirable to test whether the
reduced set of questions would retain the qualities of the full
set.
For the most part, the PCA (Table C-4) showed the expected
relationships. Six factors (role ambiguity, workload, job
satisfaction, influence over job, role conflict, and skill
utilization) were identified in accordance with expectation. A
seventh factor dealt with satisfaction with salary and advancement
opportunities. Since this was the weakest of the factors, it was
not employed in the regressions.
C-3
-------
Table C-l. PRINCIPAL COHPONENTS ANALYSIS OF SYHPTOH VARIABLES
NFACTORS = 9
Orthogonal Transformation Matrii
3456
1 0
2 -0
3 0
4 0
5 0
6 -0
7 -0
8 -0
9 -0
.41558 (
.42568 -i
.29310 -(
.33962 I
.61535 -(
.18413 -i
.02081 -I
.17809 1
.01163 1
). 41589 0
). 50565 0
1.17496 -0
L04051 -0
!. 59947 0
i. 30754 0
i.08949 -0
9.21394 0
5.17712 0
.39013 0
.07468 0
.62954 -0
.32649 0
.42393 -0
.19030 -0
.00990 0
.33133 0
.11043 0
.34041 0
.59082 0
.12051 0
.46818 -0
.09604 -0
.30948 -0
.44286 0
.01257 -0
.00747 0
Rotated Factor
II7A1
II7B1
JI7C1
II7D1
II7E1
II7F1
11761
II7H1
II7I1
II7J1
II7K1
II7L1
1I7H1
II7N1
II701
II7P1
II701
II7R1
II7S1
II7T1
II7U1
II7U1
II7X1'
II7Y1
II721
II7AA1
II7BB1
1I7CC1
II7D01
II7EE1
II7FF1
FACTOR1
0.14413
0.11454
0. 09923
0.U060
0. 09668
0.88302
0.102E3
0.1670S
0.14H0
0.14524
0.15509
0.15045
0.14987
0.12361
0.14474
0.20292
0.18883
0.15784
0.20597
0.13536
0.59816
0.16362
0. 19963
0.18528
0.20308
0.16390
0.24415
0.75705
0.69851
0.76979
0.61844
FACTOR2
0.15119
0.13875
0.12904
0.11364
0.06106
0.07116
0.07291
0.13623
0.15584
0.14027
0.20534
0.10137
0.06910
0.16503
0.10828
0.11673
0.34435
0.30296
0.08259
0.21327
0.21065
0.59489
0.35354
0.73639
0.76628
0.76463
0.40980
0.19844
0.23314
0.15977
0.03323
FACTORS
0.22328
0.14453
0.13463
0.19745
0.14653
0.04232
0.08313
0.15439
0.16396
0.71251
0.74441
0.58419
0.76930
0.12078
0.13101
0.26114
0.23590
0.17013
0.09435
0.04422
0.12453
0.16855
0.25582
0.06310
0.08992
0.18132
0.36971
0.18107
0.13981
0.12182
0.09937
FACTOR4
0.21304
0.06173
0.80077
0.68636
0.74599
0.52309
0.23426
0.11926
0.06051
0.29954
0.11860
-0.01221
0.15107
0.26228
0.12583
0.17635
0.15966
0.19873
0.16066
0.12701
0.12703
0.06695
0.02050
0.07305
0.09670
0.11334
0.20504
0.11268
0.12636
0.04769
0.07925
FACTORS
0.07730
0.25641
.07129
.12849
.12337
.25076
.73508
0.78902
0.77901
0.12116
0.07456
0. 22121
0.14487
0.12740
0.26181
0.11083
0.13406
0.03240
0.04179
0.13057
0.13450
0.12676
0.27315
0.15109
0.12365
0.07322
-0.00153
0.08187
0.08054
0.11776
0.17264
FACTOR6
0.15432
0.08930
.08462
.16775
.13967
.41487
.19856
.09991
.15021
.14757
.14986
-0.02137
0.16105
0.67600
0.74270
0.66411
0.20830
0.19002
0.08309
0.25848
0.13016
0.05245
0.07061
0.06684
0.12247
0.10588
0.22972
0.11310
0.14026
0.10335
0.03857
.32397 0
.21692 0
.49346 0
.70231 0
.00681 -0
.29844 0
.12527 -0
.02660 0
.06895 -0
Pattern
FACTOR7
0.05173
0.36561
.06667
.01781
.13043
.20657
.08869
.08737
.05309
.00543
.03654
0.21204
0.06180
0.20802
0.13726
0.03530
0.07044
0.00999
0.74592
0.71678
0.32470
0.20315
0.34446
0.09949
0.07878
0.07974
0.07993
0.06358
0.04778
0.06446
0.17921
.33388 0
.35824 -0
.13872 0
.14464 0
.10194 -0
.11796 0
.78662 0
.06235 0
.27028 0
FACTORS
0.12333
0.22610
0.04110
•0.01339
0.16867
.12661
.00127
.13008
.02698
.03874
.07023
0.31504
0.08665
0.01428
0.10119
0.18348
0.57888
0.67461
0.13440
-0.11084
0.10961
0.30565
0.29496
0.13973
-0.01601
0.20231
-0.09559
0.01435
-0.02195
0.08259
0.32330
.26028 0.
.04241 -0.
.39818 -0.
.13319 -0.
.14614 -8.
.72436 0.
.28691 0.
.32081 -0.
.15121 -0.
FACTOR9
,0.71051
0.53707
.05835
.24177
.00166
.00742
.04760
.07163
.09323
.13469
.14452
-0.05376
0.12577
0.18061
-0.01432
0.13897
0.28387
0.18937
0.05161
0.10691
0.02742
-0.23405
0.20514
0.19318
0.19170
0.02875
-0.02230
0.14334
0.11803
0.06465
-0.11094
24132
15760
16138
14951
15896
19894
26673
34534
78308
U7el
i!7bl
U7cl
ii7dl
ii7el
ii7fl
U7«l
iI7hl
ii7il
117J1
i!7kl
ii711
ii7il
ii7nl
117ol
Ii7pl
117Q1
ii7rl
ii7sl
ii7tl
ii7ul
ii7ul
ii7il
i!7rl
ii7zl
ii7aal
ii7bbl
117ccl
ii7ddl
ii7eel
Ii7ffl
0.20936
0.07418
-0.15905
-0.04091
-0.12640
0.26882
-0.08604
-0.76530
0.49230
HEADACHE
NAUSEA
RUNNY NOSE
STUFFY NOSE
SNEEZING
COU6H
UHEEZIN6
SHORTNESS OF BREATH
CHEST TI6HTNESS
DRY/TEARING EYES
SORE EYES
BLURRY VISION
BURNING EYES
SORE THROAT
HOARSENESS
DRY THROAT
FATIGUE
SLEEPINESS
CHILLS
FEVER
ACHING MUSCLES/JOINTS
DIFFICULTY REHEHBERIH6
DIZZINESS
DEPRESSED
TENSION
DIFFICULTY CONCENTRATE
DRY SKIN
UPPER BACK PAIN
LOUER BACK PAIN
SHOULDER/NECK PAW
HAND/URIST PAIN
C-4
-------
Table C-2,
Rotation Method: Varir»>
PRINCIPAL COMPONENTS ANALYSIS OF PHYSICAL ENVIRONMENT VARIABLES
NFACTOR = 4
Orthogonal Transformation Matrix
1234
8.16775
8.88918
8.62884
8.76858
niiAi
II11E1
1II1C1
III1D1
iniEi
III1F1
I1I1G1
III1H1
mm
mui
FACTOR!
-8.19615
8.88812
8.76888
8.71782
8.11125
8.67353
8.17353
8.25548
8.27676
8.66718
1
2
3
4
FACTOR2
8.88578
8.27638
8.27845
8.89324
8.84121
8.17881
8.87883
8.98676
8.88846
8.48768
8.78384
-8.13684
8.38377
-8.46994
Rotated
FACTORS
8.78815
-8.28494
8.23921
-8.13619
8.81525
8.47867
8.88759
8.85528
8.86158
-8.88668
8.56892
-8.82598
-8.68326
8.43888
8.18981
8.98624
-8.82884
-8.12257
Factor Pattern
FACTOR4
8.14878
8.84829
8.83717
0. 28381
-8.84137
-8.82496
8.95897
-8.82637
8.129U
8.11827
III1A1
I1I1B1
III1C1
III1D1
1II1E1
III1F1
III161
III1H1
mm
mui
TOO MUCH AIR MOVEMENT
TOO LITTLE AIR MOVEMENT
UANT TO ADJUST AIR MOVEMENT
TOO HOT
TOO COLD
UANT TO ADJUST TEMPERATURE
TOO HUMUD
TOO DRY
WANT TO ADJUST HUMIDITY
TOO STUFFY
C-5
-------
Table C-3 PRINCIPAL OPPONENTS ANALYSIS OF ODOR VARIABLES
NFACTOR : 6
Rotation Method: Varirs>
III2A
III2B
III2C
III20
III2E
III2F
III26
III2H
III2I
I1I2J
III2K
III2L
III2M
III2N
III20
Orthogonal Transformation Hatrii
12345
FACTOR1
•8.87342
8.14899
8.15985
8.12278
8.25166
8.87896
8.31312
8.12997
8.23363
8.22181
8.21638
8.69795
8.68683
8.62385
8.66829
1
2
3
4
5
6
FACTOR2
8.
T.
T.
8.
8.
e!
e.
8.
-e.
e.
e.
e.
-8.
8.
8.
61626
79287
17767
39888
69586
89828
87144
89479
88279
26868
86334
27359
81788
87836
13898
8.54341
-8.34345
-8.39354 -
-8.42841 •
8.58518 -
8.88198 •
FACTORS
8.
8.
8.
-8.
-8.
8.
8.
8.
8.
8.
T.
T
e.
e.
-e.
25794
15785
88627
86952
88586
25641
85198
29324
44912
78684
75932
28881
38797
13348
82279
8.38586
8.78917
•8.28889
•8.21878
•8.27215
•8.17584
Rotated
FACTOR4
8.14821
8.87332
8.85833
8.89752
8.81919
8.31868
8.78598
8.77266
8.84223
8.21854
8.13684
8.87289
8.83695
8.38672
8.39457
8.42664 8.42898 8.38489
-8.28611 -8.34689 8.J7163
8.68948 -8.35825 8.37555
-8.38858 8.54183 8.56393
-8.38521 -8.51958 8.46683
-8.82834 8.89823 -8.38232
Factor Pattern
FACTORS
8.13154
•8.84218
1.14764
8.69941
8.35282
8.59419
8.12798
8.14956
8,58963
8.81598
8.22831
8.83464
8.23879
8.18113
8.89236
8.28485
8.25888
8.16515
8.12769
8.18882
8.98136
FACTOR6
8.
8.
8.
"e"
-e.
8.
8.
8.
8.
8.
8.
-8.
8.
e.
e.
27829
88898
92188
13236
86858
18789
83324
83493
82843
84195
88427
81788
12187
89324
88765
1II2A
III2B
III2C
III2D
III2E
III2F
III26
III2H
III2I
III2J
III2K
III2L
III2H
III2N
IIJ20
BODY ODOR
COSMETICS ODORS
TOBACCO SMOKE ODORS
FISHY SHELLS
OTHER FOOD SHELLS
NUSTY/DAHP SHELLS
NEW CARPET ODORS
NEW DRAPES ODORS
DIESEL/ENGINE EXHAUST
PHOTOCOPYING MACHINE
ODORS
ODORS
PRINTING PROCESSING ODORS
OTHER CHEMICAL ODORS
PESTICIDE ODORS
CARPET CLEANING ODORS
PAINT ODORS
C-6
-------
Table C-4
Rotation Method:
PRINCIPAL COMPONENTS ANALYSIS OF JOB SATISFACTION VARIABLES
NFACTORS = 7
Orthogonal Transformation Hatrii
FACTOR!
RIV1A
RIV1B
FIV1C
RIV1D
RIV2
RIV3
IV4A
IV4E
IV4C
IV5A
IV5B
IV5C
IV5D
IV6A
IV6B
IV6C
IV6D
IV6E
IV6F
IV6G
IV6H
IV6I
IV6J
IV6K
8.
8.
8.
8.
e.
e.
-8.
-8.
-8.
8.
8.
8.
8.
8.
8.
-8.
8.
8.
8.
8.
8.
I
19621
19881
84981
18188
84334
13896
11463
17618
14115
88286
21298
14259
86752
86829
86217
85398
18352
28599
28798
22941
88184
81263
8.82482
V i
£6182
1
2
3
4
5
6
7
FACTOR2
8.86899
8.82482
8.82275
-8.88387
-e.ee54e
8.88189
8.89281
e.i26se
8.34288
-8.86145
-8.89858
8. 25288
8.15782
8.82852
8.86946
8.84938
8.81451
8.12229
8.16928
8.18983
8.84788
8.87117
8.82587
8.83838
8.56288 8.17734 8.48223 8.37234 -
-8.88544 8.84764 -8.18291 8.86359
-8.59255 8.84736 8.63887 -8.28966
-8.39118 -8.19544 -8.18435 8.88817
8.26983 -8.43188 8.21319 -8.88297
8.38362 8.18441 -8.14149 8.88485
8.88129 8.11879 8.52286 8.15719 -
Rotated Factor Pattern
FACTORS
8.71292
8.78291
8.69655
8.75745
8.89295
8.35835
-8.12856
-8.12478
-8.11489
8.18991
8.89178
8.14922
8.88927
8.88371
8.83772
-8.88427
8.86987
8.17126
8.29249
8.21883
8.14968
8.14648
8.15188
8.15756
FACTOR4
8.11326
8.18229
8.83471
8.13462
8.85381
8.11876
-8.84277
-8.87655
-8.87845
8.72756
8.72353
8.78114
8.63259
8.87739
8.85985
8.83848
8.83567
8.88988
8.24649
8.28831
8.16191
8.19622
8.89829
8.12447
FACTORS
-8.13419
-8.12432
-8.83376
-8.12554
-8.89838
-8.86921
8.83181
8.84786
8.71296
-8.82582
-8.88591
-8.81942
-8.11634
8.13846
8.86731
8.18894
8.88814
-8.84984
-8.88717
-8.83336
-8.13139
-8.86972
-8.15261
-8.13617
FACTOR6
8.23987
8.12495
8.11258
8.18619
8.86524
8.87947
-8.82618
-8.83223
-8.88788
8.19598
.83782
.17883
.82786
.84321
.13881
.84536
8.16234
8.82878
8.68279
8.74518
8.19471
8.14868
8.17447
8.13882
8.24524
8.48156
8.84388
8.14368
8.8168?
8.13472
8.88844
8.41282 8.22354
8.14524 -8.88115
8.81932 8.43962
8.88329 8.85854
8.14296 -8.82644
-8.64629 8.65868
-8.68825 -8.55952
FACTOR?
8.
8.
-8.
8.
8.
E
«r
-8.
-8.
8.
8.
8.
8.
-8.
-8.
8.
8.
8.
8.
8.
8.
8.
8.
8.
22641
16342
83849
16866
68518
71715
87341
88819
81278
82878
84279
89589
83521
88735
88948
81765
88278
88551
12327
18823
83386
84131
88113
86885
RIV1A
RIV1B
RIV1C
RIV10
RIV2
RIV3
IV4A
IV4B
IV4C
IV5A
IV5B
IV5C
IV5D
IV6A
IV6B
IV6C
IV6D
IV6E
IV6F
IV6G
IV6H
IV6I
IV6J
IV6K
SATISFIED UI
UOULD TAKE J
TYPE OF JOB
RECOMMEND JO
SATISFIED UI
SATIS. U/ OP
PEERS REQUEST
SUPVS REQUEST
PRTY REQUEST
INFLUENCE - A
INFLUENCE - A
INFLUENCE - P
INFLUENCE - F
HOU OFTEN HAV
HOU OFTEN HAV
HOU OFTN LITT
HOU OFTEN GRE
HOU OFTEN USE
HOU OFTEN DO
HOU OFTN USE
HOU OFTEN CLE
HOU OFTEN PRE
HOU OFTEN UOR
HOU OFTN CLR
C-7
-------
APPENDIX D
NUMBER OF MEN AND WOMEN AT EACH LEVEL OF THE HEALTH, COMFORT AND ODOR INDICES
AND ASSOCIATED BINARY VARIABLES EMPLOYED IN THE LINEAR AND LOGISTIC
REGRESSIONS
D-l
-------
APPENDIX D
NUMBER OF MEN AND WOMEN AT EACH LEVEL OF THE HEALTH, COMFORT AND ODOR INDICES
AND ASSOCIATED BINARY VARIABLES EMPLOYED IN THE LINEAR AND LOGISTIC '
REGRESSIONS
The dependent variables for the linear regressions were the indices
constructed from the factors identified by Principal Components Analysis
(Appendix C). Each factor consisted of responses to one or more questions on
health, comfort or odor symptoms. Each response was graded on a scale of 0 to
3, with 0 used for "never" or "rarely" during the past year, 1 for
"sometimes," 2 for "often" and 3 for "always." Since the number of questions
contained in a factor varied from 1 to 5, the high scores for the indices
range from 3 to 15. Tables D-l through D-22 document the number of men and
women at each level of the 12 health, 4 comfort and 6 odor indices.
For the logistic regressions, the indices had to be reduced to binary
(0,1) variables. It was decided to select the cutoff point at the same index
value for both men and women, and to have at least 10% of respondenst of each
gender in the "positive" category. (Since women consistently reported more
health symptoms than men, this effectively resulted in having more women than
men in the "positive" category for every health variable.) The number of men
and women in the (0,1) categories for each health variable is shown in the
sections of Tables D-l through D-22 marked "BINARY VARIABLE FREQUENCIES." The
percentages associated with the "positive" category for each sex are related
to the values of p occurring in the expression for the odds ratio for that
condition, as listed in Appendix E.
An alternative two-part criterion for defining a "positive" health
variable has been used in previous studies: requiring that (1) all symptoms be
"work-related" in some sense, and that (2) at least one symptom included in
the health variable occur sufficiently often. As a test of how this criterion
might affect our results, we chose the eye irritation factor to run a separate
set of logistic regressions on. If the respondent reported that the symptom
improved after work, and happened "sometimes," "often," or "always" within the
D-2
-------
last year, the response was scored as work-related. Since there were four
symptoms within this category, a respondent could score up to four work-
related symptoms. The scores using this criterion are given in Table D-23.
The second requirement was met by requiring that at least one of the
four symptoms occur either "often" or "always". Using this criterion in place
of our first approach of selecting a cutoff point on the index resulted in a
different (and larger) number of men and women categorized as "positive" (last
part of Table D-23.) The accompanying linear and logistic regressions using
this criterion are presented in Appendix F.
D-3
-------
Table D-l. HEADACHE AND NAUSEA
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
5
6
Total
F
537
13.75
37.32
26.65
736
18.85
52.50
36.53
K -_-_.-_ _-J
497
12.73
66.09
24.67
188
4.81
77.69
9.33
47
1.20
82.46
2.33
8
0.20
72.73
0.40
2
0.05
100.00
0.10
-
2015
51.60
M |
902
23.10
62.68
47.72
666
17.06
47.50
35.24
255
6.53
33.91
13.49
54
1.38
22.31
2.86
h. ....... .4
10
0.26
17.54
0.53
h. ...... .4
3
0.08
27.27
0.16
0
0.00
0.00
0.00
4
1890
48.40
Total
1439
36.85
1402
35.90
752
19.26
242
6.20
57
1.46
11
0.28
2
0.05
3905
100.00
Frequency Missing - 50
(Continued)
D-4
-------
Table D-l (continued). HEADACHE AND NAUSEA
BINARY VARIABLE FREQUENCIES
Frequency|
Percent j
Row Pet j
Col Pet |F |M
0
1
Total
h
1273
32.60
44.81
63.18
K---.-....-J
742
19.00
69.74
36.82
2015
51.60
h .. ...-I
1568
40.15
55.19
82.96
322
8.25
30.26
17.04
1890
48.40
| Total
2841
72.75
1064
27.25
3905
100.00
Frequency Missing - 50
D-5
-------
Table D-2. STUFFY NOSE/SINUS CONGESTION, RUNNY NOSE, SNEEZING, COUGH
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
5
6
F
357
9.20
42.86
17.81
213
5.49
46.41
10.62
248
6.39
46.10
12.37
299
7.70
54.76
14.91
330
8.50
52.88
16.46
193
4.97
59.20
9.63
h ...«„_.. J
144
3.71
61.80
7.18
|M |
476
12.26
57.14
25.36
246
6.34
53.59
13.11
290
7.47
53.90
15.45
247
6.36
45.24
13.16
294
7.57
47.12
15.66
133
3.43
40.80
7.09
89
2.29
38.20
4.74
Total
833
21.46
459
11.82
538
13.86
546
14.06
624
16.07
326
8.40
233
6.00
(Continued)
D-6
-------
Table D-2 (continued).
STUFFY NOSE/SINUS CONGESTION, RUNNY NOSE, SNEEZING,
COUGH
Frequency|
Percent
Row Pet
Col Pet
7
8
9
10
...... J
11
12
F
116
2.99
69.46
5.79
55
1.42
61.11
2.74
26
0.67
81.25
1.30
h.....*..J
15
0.39
71.43
0.75
3
0.08
50.00
0.15
6
0.15
85.71
0.30
|M |
f. ...... --4
51
1.31
30.54
2.72
35
0.90
38.89
1.86
H. ...... -4
6
0.15
18.75
0.32
L. .__... .-*
6
0.15
28.57
0.32
3
0.08
50.00
0.16
h 4
1
0.03
14.29
0.05
Total
2005
51.65
Frequency Missing - 73
1877
48.35
Total
167
4.30
90
2.32
32
0.82
21
0.54
6
0.15
7
0.18
3882
100.00
(Continued)
D-7
-------
Table D-2 (continued). STUFFY NOSE/SINUS CONGESTION, RUNNY NOSE, SNEEZING,
COUGH
BINARY VARIABLE FREQUENCIES
Frequency|
Percent |
Row Pet j
Col Pet |F |M
0
1
1447 |
37.27 |
48.23 |
72.17 |
558 |
14.37 |
63.27 |
27.83 |
1553
40.01
51.77
82.74
324
8.35
36.73
17.26
Total
2005
51.65
1877
48.35
Total
3000
77.28
882
22.72
3882
100.00
Frequency Missing - 73
D-8
-------
Table D-3. WHEEZING, SHORTNESS OF BREATH, CHEST TIGHTNESS
Frequency|
Percent
Row Pet
Total
Col Pet
0
1
2
3
4
F
1494
38.37
49.26
74.25
H ..........
210
5.39
54.69
10.44
142
3.65
62.28
7.06
82
2.11
67.21
4.08
38
0.98
69.09
1.89
|M |
f ........ -1
1539
39.52
50.74
81.77
K. ...... .4
174
4.47
45.31
9.25
86
2.21
37.72
4.57
40
1.03
32.79
2.13
h •
17
0.44
30.91
0.90
3033
77.89
384
9.86
228
5.86
122
3.13
55
1.41
(Continued)
D-9
-------
Table D-3 (continued). WHEEZING, SHORTNESS OF BREATH, CHEST TIGHTNESS
Frequency
Percent
Row Pet
Col Pet
5
6
7
8
9
Total
F
16
0.41
59.26
0.80
21
0.54
70.00
1.04
5
0.13
71.43
0.25
2
0.05
50.00
0.10
2
0.05
50.00
0.10
2012
51.67
M |
11
0.28
40.74
0.58
9
0.23
30.00
0.48
2
0.05
28.57
0.11
2
0.05
50.00
0.11
2
0.05
50.00
0.11
1882
48.33
Frequency Missing - 61
Total
27
0.69
30
0.77
7
0.18
4
0.10
4
0.10
3894
100.00
(Continued)
D-10
-------
Table D-3 (continued). WHEEZING, SHORTNESS OF BREATH, CHEST TIGHTNESS
BINARY VARIABLE FREQUENCIES
Frequency|
Percent
Row Pet
Col Pet
|M | Total
+ +
1704 | 1713 3417
43.76 | 43.99 87.75
49.87 | 50.13
84.69 | 91.02
308 | 169 477
7.91 | 4.34 12.25
64.57 | 35.43
15.31 j 9.98
Total 2012 1882 3894
51.67 48.33 100.00
Frequency Missing - 61
D-ll
-------
Table D-4. DRY, ITCHING OR TEARING EYES, SORE EYES, BURNING EYES, BLURRY
VISION
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
5
6
F
515
13.24
42.32
25.63
291
7.48
46.12
14.48
h _.._.--. -
283
7.28
50.36
14.09
K.....-.-H
264
6.79
56.29
13.14
K---..-.-J
200
5.14
62.11
9.96
142
3.65
61.21
7.07
124
3.19
64.92
6.17
M |
702
18.05
57.68
37.34
340
8.74
53.88
18.09
279
7.17
49.64
14.84
h- ...... .4
205
5.27
43.71
10.90
U. ...... .4
122
3.14
37.89
6.49
90
2.31
38.79
4.79
67
1.72
35.08
3.56
Total
1217
31.29
631
16.23
562
14.45
469
12.06
322
8.28
232
5.97
191
4.91
(Continued)
D-12
-------
Table D-4 (continued). DRY, ITCHING OR TEARING EYES, SORE EYES, BURNING EYES,
BLURRY VISION
Frequency
Percent
Row Pet
Col Pet
7
8
9
10
11
12
Total
F
64
1.65
73.56
3.19
59
1.52
71.08
2.94
24
0.62
70.59
1.19
13
0.33
61.90
0.65
Lm m m m J
14
0.36
66.67
0.70
H
16
0.41
84.21
0.80
2009
51.66
M |
h. ...... .4
23
0.59
26.44
1.22
24
0.62
28.92
1.28
10
0.26
29.41
0.53
8
0.21
38.10
0.43
k *• •» -4
7
0.18
33.33
0.37
h •
3
C.08
15.79
0.16
1880
48.34
Frequency Missing - 66
Total
87
2.24
83
2.13
34
0.87
21
0.54
21
0.54
19
0.49
3889
100.00
(Continued)
D-13
-------
Table D-4 (continued).
DRY, ITCHING OR TEARING EYES, SORE EYES, BURNING EYES,
BLURRY VISION
BINARY VARIABLE FREQUENCIES
Frequency
Percent
Row Pet
Col Pet
0
1
Total
F
1553
39.93
48.52
77.30
456
11.73
66.28
22.70
^........^
2009
51.66
M |
1648 |
42.38 |
51.48 |
87.66 j
232 |
5.97 |
33.72 |
12.34 |
H. ...... .4-
1880
48.34
Total
3201
82.31
688
17.69
3889
100.00
Frequency Missing - 66
D-14
-------
Table D-5. SORE THROAT, HOARSENESS, DRY THROAT
Frequency|
Percent
Row Pet
Col Pet
0
1
2
3
« . . * * m ~*J
4
F
827
21.21
44.51
40.96
418
10.72
53.32
20.70
h ........ *l
285
7.31
57.93
14.12
U. ...... .-J
265
6.80
61.20
13.13
L m „ m J
105
2.69
66.88
5.20
K H
IN
1031
26.44
55.49
54.84
h. ...... .4
366
9.39
46.68
19.47
h. ...... .4
207
5.31
42.07
11.01
k. ...... .4
168
4.31
38.80
8.94
52
1.33
33.12
2.77
h H
Total
1858
47.65
784
20.11
492
12.62
433
11.11
157
4.03
(Continued)
D-15
-------
Table D-5 (continued). SORE THROAT, HOARSENESS, DRY THROAT
Frequency
Percent
Row Pet
Col Pet
5
6
7
8
9
Total
F
51
1.31
62.96
2.53
48
1.23
76.19
2.38
9
0.23
56.25
0.45
6
0.15
85.71
0.30
5
0.13
62.50
0.25
2019
51.78
M |
30
0.77
37.04
1.60
h. ...... .4
15
0.38
23.81
0.80
7
0.18
43.75
0.37
H. ...... .4
1
0.03
14.29
0.05
I- -»
3
0.08
37.50
0.16
K. ...... .4
1880
48.22
Frequency Missing - 56
Total
81
2.08
63
1.62
16
0.41
7
0.18
8
0.21
3899
100.00
(Continued)
D-16
-------
Table D-5 (continued). SORE THROAT, HOARSENESS, DRY THROAT
BINARY VARIABLE FREQUENCIES
Frequency
Percent
Row Pet
Col Pet
0
1
F
1530
39.24
48.82
75.78
h H
489
12.54
63.92
24.22
M |
1604
41.14
51.18
85.32
h 4
276
7.08
36.08
14.68
i- •{
Total 2019 1880
51.78 48.22
Frequency Missing - 56
Total
3134
80.38
765
19.62
3899
100.00
D-17
-------
Table D-6. SLEEPINESS, UNUSUAL FATIGUE
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
5
6
F
H........^
518
13.25
41.74
25.64
360
9.21
47.37
17.82
550
14.07
53.50
27.23
220
5.63
67.90
10.89
255
6.52
63.43
12.62
37
0.95
71.15
1.83
1- H
80
2.05
78.43
3.96
M
723
18.50
58.26
38.27
400
10.23
52.63
21.18
478
12.23
46.50
25.30
104
2.66
32.10
5.51
147
3.76
36.57
7.78
15
0.38
28.85
0.79
1- 4
22
0.56
21.57
1.16
Total
2020
51.68
1889
48.32
Total
1241
31.75
760
19.44
1028
26.30
324
8.29
402
10.28
52
1.33
102
2.61
3909
100.00
Frequency Missing - 46
(Continued)
D-18
-------
Table D-6 (continued). SLEEPINESS, UNUSUAL FATIGUE
BINARY VARIABLE FREQUENCIES
Frequency
Percent
Row Pet
Col Pet
0
1
Total
F
1648
42.16
49.15
81.58
372
9.52
66.91
18.42
........ .J
2020
51.68
|N |
1705
43.62
50.85
90.26
184
4.71
33.09
9.74
W. ....... -4
1889
48.32
Total
3353
85.78
556
14.22
3909
100.00
Frequency Missing - 46
D-19
-------
Table D-7. CHILLS AND FEVER
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
6
Total
F
1370
35.03
46.44
67.72
365
9.33
65.53
18.04
h_---..._J
214
5.47
68.15
10.58
55
1.41
79.71
2.72
18
0.46
94.74
0.89
1
0.03
50.00
0.05
2023
51.73
M |
1580
40.40
53.56
83.69
192
4.91
34.47
10.17
100
2.56
31.85
5.30
14
0.36
20.29
0.74
1
0.03
5.26
0.05
1
0.03
50.00
0.05
H
1888
48.27
Total
2950
75.43
557
14.24
314
8.03
69
1.76
19
0.49
2
0.05
3911
100.00
Frequency Missing - 44
(Continued)
D-2C
-------
Table D-7 (continued). CHILLS AND FEVER
BINARY VARIABLE FREQUENCIES
Frequency
Percent
Row Pet
Col Pet
0
1
Total
F
K .........
1370
35.03
46.44
67.72
653
16.70
67.95
32.28
2023
51.73
M
h ..........
1580
40.40
53.56
83.69
1- H
308
7.88
32.05
16.31
K ........ J
1888
48.27
Total
2950
75.43
h
961
24.57
3911
100.00
Frequency Missing - 44
D-21
-------
Table D-8. ACHING MUSCLES OR JOINTS, PAIN IN BACK, SHOULDER, NECK, WRISTS
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
5
6
7
1
1
1
F
710
18.26
43.64
35.36
h ..........
270
6.94
47.45
13.45
251
6.45
56.53
12.50
209
5.37
59.89
10.41
169
4.35
55.05
8.42
K ........ ..^
117
3.01
61.58
5.83
79
2.03
67.52
3.93
62
1.59
69.66
3.09
|M |
^ ........ 4
917
23.58
56.36
48.75
I-. ----- . . .{
299
7.69
52.55
15.90
193
4.96
43.47
10.26
H«
140
3.60
40.11
7.44
138
3.55
44.95
7.34
H. ...... .4,
73
1.88
38.42
3.88
38
0.98
32.48
2.02
27
0.69
30.34
1.44
Total
1627
41.84
569
14.63
444
11.42
349
8.97
307
7.89
190
4.89
117
3.01
89
2.29
(Continued)
D-22
-------
Table D-8 (continued). ACHING MUSCLES, PAIN IN BACK, NECK, WRISTS
Frequency
Percent
Row Pet
Col Pet
8
9
10
11
12
13
14
H
15
Total
1
1
1
IF
39
1.00
66.10
1.94
30
0.77
66.67
1.49
32
0.82
76.19
1.59
11
0.28
73.33
0.55
13
0.33
81.25
0.65
7
0.18
87.50
0.35
5
0.13
83.33
0.25
• H
4
0.10
66.67
0.20
2008
51.63
M |
^. .»-..- .4
20
0.51
33.90
1.06
15
0.39
33.33
0.80
10
0.26
23.81
0.53
4
0.10
26.67
0.21
3
0.08
18.75
0.16
1
0.03
12.50
0.05
1
0.03
16.67
0.05
h •
2
0.05
33.33
0.11
1881
48.37
Frequency Missing - 66
Total
59
1.52
45
1.16
42
1.08
15
0.39
16
0.41
8
0.21
6
0.15
6
0.15
3889
100.0
D-23
-------
Table D-8 (continued). ACHING MUSCLES, PAIN IN BACK, SHOULDER, NECK, WRISTS
BINARY VARIABLE FREQUENCIES
Frequency|
Percent j
Row Pet j
Col Pet |F
|M
0
1
Total
1609
41.37
48.82
80.13
399
10.26
67.28
19.87
2008
51.63
1687
43.38
51.18
89.69
194
4.99
32.72
10.31
1881
48.37
| Total
3296
84.75
593
15.25
3889
100.00
Frequency Missing - 66
D-24
-------
Table D-9. DIFFICULTY CONCENTRATING, REMEMBERING; DEPRESSION, TENSION
Frequency
Percent
Row Pet
Col Pet
1
2
3
4
5
6
1
F
634
16.31
45.55
31.62
339
8.72
50.82
16.91
h ...... --J
305
7.85
54.66
15.21
230
5.92
52.87
11.47
200
5.15
56.50
9.98
h
109
2.80
56.19
5.44
h H
75
1.93
60.48
3.74
|M
h. ...... .4
758
19.51
54.45
40.30
K. ...... .4
328
8.44
49.18
17.44
K~ ...... .4
253
6.51
45.34
13.45
205
5.28
47.13
10.90
154
3.96
43.50
8.19
h •
85
2.19
43.81
4.52
^
49
1.26
39.52
2.60
Total
1392
35.82
667
17.16
558
14.36
435
11.19
354
9.11
194
4.99
124
3.19
(Continued)
D-25
-------
Table D-9 (continued).
DIFFICULTY CONCENTRATING, REMEMBERING; DEPRESSION,
TENSION
Frequency|
Percent
Row Pet
Col Pet
7
8
9
10
. •> *••»«».. *4
11
12
H
Total
F
f.. ...... _^
41
1.06
57.75
2.04
40
1.03
83.33
2.00
16
0.41
72.73
0.80
5
0.13
55.56
0.25
L «. •. • J
6
0.15
85.71
0.30
5
0.13
100.00
0.25
1- 4
2005
51.60
M
h. ...... .4
30
0.77
42.25
1.59
8
0.21
16.67
0.43
6
0.15
27.27
0.32
4
0.10
44.44
0.21
h- ...... -4
1
0.03
14.29
0.05
0
0.00
0.00
0.00
1- -t
1881
48.40
Total
71
1.83
48
1.24
22
0.57
9
0.23
7
0.18
5
0.13
3886
100.00
Frequency Missing - 69
(Continued)
D-26
-------
Table D-9 (continued). DIFFICULTY CONCENTRATING, REMEMBERING; DEPRESSION,
TENSION
BINARY VARIABLE FREQUENCIES
Frequency|
Percent j
Row Pet
Col Pet
0
1
Total
F
1508
38.81
49.41
75.21
K ........ J
497
12.79
59.59
24.79
2005
51.60
|M |
1544 |
39.73 |
50.59 |
82.08 j
H- ...... .4-
337 |
8.67 |
40.41 I
17.92 |
1881
48.40
Total
3052
78.54
834
21.46
3886
100.00
Frequency Missing - 69
D-27
-------
Table D-10. DIZZINESS
Frequency|
Percent
Row Pet
Col Pet
0
1
2
3
F
1449
37.04
47.14
71.73
H ........ -J
449
11.48
65.74
22.23
1- H
106
2.71
78.52
5.25
K H
16
0.41
80.00
0.79
M |
H. ...... --I
1625
41.54
52.86
85.89
234
5.98
34.26
12.37
29
0.74
21.48
1.53
h ^
4
0.10
20.00
0.21
Total
2020
51.64
1892
48.36
Total
3074
78.58
683
17.46
135
3.45
20
0.51
3912
100.00
Frequency Missing - 43
(Continued)
D-28
-------
Table D-10 (continued). DIZZINESS
BINARY VARIABLE FREQUENCIES
Frequency
Percent
Row Pet
Col Pet
0
1
Total
F
h...-....-J
- 1449
37.04
47.14
71.73
571
14.60
68.14
28.27
2020
51.64
M |
K. ...... .4
1625
41.54
52.86
85.89
267
6.83
31.86
14.11
1892
48.36
Total
3074
78.58
838
21.42
3912
100.00
Frequency Missing - 43
D-29
-------
Table D-ll. DRY OR ITCHY SKIN
Frequency|
Percent
Row Pet
Col Pet
0
1
2
3
Total
F
964
24.67
45.71
47.75
523
13.38
54.31
25.90
360
9.21
60.40
17.83
K ........ J
172
4.40
71.67
8.52
2019
51.66
M
1145
29.30
54.29
60.61
440
11.26
45.69
23.29
236
6.04
39.60
12.49
68 |
1.74
28.33 |
3.60 |
k- -.._.. .-*
1889
48.34
Frequency Missing - 47
Total
2109
53.97
963
24.64
596
15.25
240
6.14
3908
100.00
(Continued)
D-30
-------
Table D-ll (continued). DRY OR ITCHY SKIN
BINARY VARIABLE FREQUENCIES
Frequency
Percent
Row Pet
Col Pet
0
1
Total
F
1487
38.05
48.40
73.65
532
13.61
63.64
26.35
2019
51.66
|M |
H. ...... . J
1585
40.56
51.60
83.91
304
7.78
36.36
16.09
1889
48.34
Total
3072
78.61
836
21.39
3908
100.00
Frequency Missing - 47
D-31
-------
Table D-12. PROBLEMS WITH CONTACT LENSES
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
Total
F
130
16.62
51.79
27.25
171
21.87
62.18
35.85
123
15.73
68.72
25.79
53
6.78
68.83
11.11
477
61.00
M
h. ...... .4
121
15.47
48.21
39.67
104
13.30
37.82
34.10
56
7.16
31.28
18.36
h •»
24
3.07
31.17
7.87
305
39.00
Frequency Missing - 3173
Total
251
32.10
275
35.17
179
22.89
77
9.85
782
100.00
(Continued)
D-32
-------
Table D-12 (continued). PROBLEMS WITH CONTACT LENSES
BINARY VARIABLE FREQUENCIES
Frequency
Percent
Row Pet
Col Pet
0
1
Total
1
F
301
38.49
57.22
63.10
176
22.51
68.75
36.90
477
61.00
IN
K -.--.-.. .4
225
28.77
42.78
73.77
80
10.23
31.25
26.23
305
39.00
| Total
526
67.26
256
32.74
782
100.00
Frequency Missing - 3173
D-33
-------
Table D-13. HOT STUFFY AIR
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
5
6
7
F
60
1.60
36.36
3.15
48
1.28
44.86
2.52
70
1.87
39.11
3.68
95
2.54
45.89
4.99
K. .._..-•«
145
3.88
45.17
7.62
205
5.48
45.66
10.78
177
4.73
54.46
9.31
169
4.52
53.48
8.89
M |
105
2.81
63.64
5.71
K- ...... .4
59
1.58
55.14
3.21
109
2.91
60.89
5.93
112
2.99
54.11
6.09
h. ...... --4
176
4.70
54.83
9.57
244
6.52
54.34
13.27
148
3.96
45.54
8.05
147
3.93
46.52
7.99
Total
165
4.41
107
2.86
179
4.78
207
5.53
321
8.58
449
12.00
325
8.69
316
8.45
D-34
-------
Table D-13 (continued). HOT STUFFY AIR
Frequency
Percent
Row Pet
Col Pet
8
9
10
11
» ^
12
13
14
15
H
Total
1
1
IF
161
4.30
53.49
8.46
161
4.30
50.79
8.46
175
4.68
49.16
9.20
128
3.42
62.75
6.73
100
2.67
58.82
5.26
74
1.98
55.64
3.89
72
1.92
69.23
3.79
62
1.66
71.26
3.26
H
1902
50.84
|M
140
3.74
46.51
7.61
156
4.17
49.21
8.48
181
4.84
50.84
9.84
76
2.03
37.25
4.13
70
1.87
41.18
3.81
59
1.58
44.36
3.21
32
0.86
30.77
1.74
25
0.67
28.74
1.36
1- 4
1839
49.16
Total
301
8.05
317
8.47
356
9.52
204
5.45
170
4.54
133
3.56
104
2.78
87
2.33
3741
100.00
Frequency Missing - 214
(continued)
D-35
-------
Table D-13 (continued). HOT STUFFY AIR
BINARY VARIABLE FREQUENCY
Frequency
Percent
Row Pet
Col Pet
0
1
Total
F
K- .......
1291
34.51
48.05
67.88
K. .......
611
16.33
57.97
32.12
1902
50.84
|M
•
1
1
1
1
•
1
1
1
1
1396
37.32
51.95
75.91
443
11.84
42.03
24.09
1839
49.16
Total
2687
71.83
L
1054
28.17
3741
100.00
Frequency Missing - 214
D-36
-------
Table D-14.
DRY AIR
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
.*.«•*•«. J
4
5
6
F
432
11.38
40.79
22.34
192
5.06
52.89
9.93
391
10.30
47.98
20.22
210
5.53
56.45
10.86
326
8.59
51.26
16.86
106
2.79
62.72
5.48
211
7.30
72.32
14.32
IN |
K. ...... -4
627
16.51
59.21
33.66
171
4.50
47.11
9.18
424
11.17
52.02
22.76
162
4.27
43.55
8.70
L «,„•.• .4
310
8.16
48.74
16.64
H. ...... .4
63
1.66
37.28
3.38
h. ...... .4
106
2.79
27.68
5.69
Total
1934
50.93
1863
49.07
Total
1059
27.89
363
9.56
815
21.46
372
9.80
636
16.75
169
4.45
383
10.09
3797
100.00
Frequency Missing - 158
(continued)
D-37
-------
Table D-14 (continued).
DRY AIR
BINARY VARIABLE FREQUENCY
Frequency|
Percent j
Row Pet
Col Pet
0
1
Total
F
K--..-.-.J
1225
32.26
46.95
63.34
709
18.67
59.68
36.66
1934
50.93
M
K--.-...--J
1384
36.45
53.05
74.29
h _._----. -J
479
12.62
40.32
25.71
U. -*..._. ,4
1863
49.07
Total
2609
68.71
1188
31.29
L
3797
100.00
Frequency Missing - 158
D-38
-------
Table D-15. COLD DRAFTY AIR
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
5
6
F
437
11.45
41.23
22.48
611
16.01
47.62
31.43
469
12.29
55.24
24.13
219
5.74
67.38
11.27
133
3.48
64.56
6.84
35
0.92
71.43
1.80
40
1.05
88.89
2.06
M |
623
16.32
58.77
33.26
672
17.61
52.38
35.88
380
9.96
44.76
20.29
t». ..--.-. .-4
106
2.78
32.62
5.66
73
1.91
35.44
3.90
14
0.37
28.57
0.75
5
0.13
11.11
0.27
Total
1944
50.93
1873
49.07
Total
1060
27.77
1283
33.61
849
22.24
325
8.51
206
5.40
49
1.28
45
1.18
3817
100.00
Frequency Missing - 138
(continued)
D-39
-------
Table D-15 (continued). COLD DRAFTY AIR
BINARY VARIABLE FREQUENCY
Frequency
Percent
Row Pet
Col Pet
0
1
Total
F |M | Total
1048 | 1295 | 2343
27.46 | 33.93 | 61.38
44.73 | 55.27 |
53.91 | 69.14 |
I----.....-J--.--.....1.
896 | 578 | 1474
23.47 | 15.14 | 38.62
60.79 | 39.21 |
46.09 | 30.86 |
1944 1873 3817
50.93 49.07 100.00
Frequency Missing - 138
D-40
-------
Table D-16. HUMID AIR
Frequency|
Percent j
Row Pet j
Col Pet
0
1
2
3
IF
1360
35.57
48.36
69.49
388
10.15
53.96
19.83
149
3.90
67.12
7.61
60
1.57
85.71
3.07
|M |
1452
37.98
51.64
77.81
331
8.66
46.04
17.74
73
1.91
32.88
3.91
10
0.26
14.29
0.54
Total
1957
51.19
1866
48.81
Total
2812
73.55
719
18.81
222
5.81
70
1.83
3823
100.00
Frequency Missing - 132
(continued)
D-41
-------
Table D-16 (continued). HUMID AIR
BINARY VARIABLE FREQUENCY
Frequency
Percent
Row Pet
Col Pet
0
1
j
Total
F |M
1360 |
35.57 |
48.36 |
69.49 |
597 |
15.62 |
59.05 |
30.51 |
1957
51.19
1452
37.98
51.64
77.81
414
10.83
40.95
22.19
1866
48.81
Total
2812
73.55
1011
26.45
3823
100.00
Frequency Missing - 132
D-42
-------
Table D-17. ODOR OF FRESH PAINT, CARPET CLEANING, OTHER CHEMICALS
Frequency|
Percent
Row Pet
Total
2821
73.93
Col Pet
0
1
2
3
*• *4
4
5
F
(•.._..._..
1350
35.38
47.86
68.74
h .........
336
8.81
60.76
17.11
157
4.11
61.33
7.99
53
1.39
59.55
2.70
40
1.05
65.57
2.04
11
0.29
91.67
0.56
|M
f. ...... .4
1471
38.55
52.14
79.43
h. ...... .4
217
5.69
39.24
11.72
99
2.59
38.67
5.35
36
0.94
40.45
1.94
21
0.55
34.43
1.13
1
0.03
8.33
0.05
553
14.49
256
6.71
89
2.33
61
1.60
12
0.31
(Continued)
D-43
-------
Table D-17 (continued). ODOR OF FRESH PAINT, CARPET CLEANING, OTHER CHEMICALS
Frequency
Percent
Row Pet
Col Pet
6
7
8
10
12
Total
F
6
0.16
75.00
0.31
4
0.10
57.14
0.20
3
0.08
75.00
0.15
2
0.05
100.00
0.10
2
0.05
66.67
0.10
1- H
1964
51.47
M |
2
0.05
25.00
0.11
h- ...... .4
3
0.08
42.86
0.16
H. ...... .4
1
0.03
25.00
0.05
0
0.00
0.00
0.00
1
0.03
33.33
0.05
1- 4
1852
48.53
Total
8
0.21
7
0.18
4
0.10
2
0.05
3
0.08
3816
100.00
Frequency Missing - 139
(continued)
D-44
-------
Table D-17 (continued). ODOR OF FRESH PAINT, CARPET CLEANING, OTHER CHEMICALS
BINARY VARIABLE FREQUENCY
Frequency
Percent
Row Pet
Col Pet
0
1
F
1350
35.38
47.86
68.74
1-- ....... J
614
16.09
61.71
31.26
h H
IN |
1471
38.55
52.14
79.43
H. ...... .4
381
9.98
38.29
20.57
H
Total
2821
73.93
995
26.07
Total 1964 1852 3816
51.47 48.53 100.00
Frequency Missing - 139
D-45
-------
Table D-18. ODOR OF COSMETICS, BODY ODOR, ENGINE EXHAUST
Frequency|
Percent
Row Pet
Total
Col Pet
0
1
2
3
4
F
640
16.65
43.18
32.50
449
11.68
48.18
22.80
K......-.-1
389
10.12
56.38
19.76
249
6.48
61.33
12.65
125
3.25
70.22
6.35
M |
842
21.91
56.82
44.93
483
12.57
51.82
25.77
H. ...... .4
301
7.83
43.62
16.06
157
4.09
38.67
8.38
53
1.38
29.78
2.83
1482
38.56
932
24.25
690
17.95
406
10.56
178
4.63
(Continued)
D-46
-------
Table D-18 (continued). ODOR OF COSMETICS, BODY ODOR, ENGINE EXHAUST
Frequency
Percent
Row Pet
Col Pet
5
6
7
8
9
j
Total
F
k- ...... --I
64
1.67
72.73
3.25
32
0.83
78.05
1.63
K ...... - --I
9
0.23
75.00
0.46
6
0.16
85.71
0.30
6
0.16
85.71
0.30
1969
51.24
M |
K. ...... .-4
24
0.62
27.27
1.28
9
0.23
21.95
0.48
H. ...... --4
3
0.08
25.00
0.16
1
0.03
14.29
0.05
1
0.03
14.29
0.05
L J
1874
48.76
Frequency Missing - 112
Total
88
2.29
41
1.07
12
0.31
7
0.18
7
0.18
3843
100.00
(continued)
D-47
-------
Table D-18 (continued). ODOR OF COSMETICS, BODY ODOR, ENGINE EXHAUST
BINARY VARIABLE FREQUENCY
Frequency|
Percent |
Row Pet I
Col Pet |F
|M
1478 | 1626
38.46 j 42.31
47.62 | 52.38
75.06 I 86.77
| Total
'+
3104
80.77
491 | 248 739
12.78 | 6.45 19.23
66.44 | 33.56
24.94 I 13.23
Total 1969 1874 3843
51.24 48.76 100.00
Frequency Missing - 112
D-48
-------
Table D-19. ODOR OF PHOTOCOPIES, PRINTING PROCESS
Frequency
Percent
Row Pet
Col Pet
0
1
2
* «J
3
4
5
6
F
1758
45.56
50.65
88.88
133
3.45
54.07
6.72
55
1.43
58.51
2.78
hj
16
0.41
59.26
0.81
7
0.18
63.64
0.35
2
0.05
100.00
0.10
7
0.18
87.50
0.35
M
K. ...... -4
1713
44.39
49.35
91.07
113
2.93
45.93
6.01
39
1.01
41.49
2.07
11
0.29
40.74
0.58
4
0.10
36.36
0.21
h- ........ .4
0
0.00
0.00
0.00
K. ...... .4
1
0.03
12.50
0.05
Total
1978
51.26
1881
48.74
Total
3471
89.95
246
6.37
94
2.44
27
0.70
11
0.29
2
0.05
8
0.21
3859
100.00
Frequency Missing - 96
(continued)
D-49
-------
Table D-19 (continued). ODOR OF PHOTOCOPIES, PRINTING PROCESS
BINARY VARIABLE FREQUENCY
Frequency
Percent
Row Pet
Col Pet
0
1
Total
F
1758
45.56
50.65
88.88
220
5.70
56.70
11.12
1978
51.26
M |
K. ...... .4.
1713 |
44.39 |
49.35 |
91.07 |
168 |
4.35 j
43.30 |
8.93 |
1881
48.74
Total
3471
89.95
388
10.05
3859
100.00
Frequency Missing - 96
D-50
-------
Table D-20.
ODOR OF NEW CARPETS, DRAPES
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
•» «• J
5
6
F
1640
42.58
51.19
83.00
186
4.83
48.95
9.41
105
2.73
54.40
5.31
24
0.62
60.00
1.21
k_ ........ -J
13
0.34
61.90
0.66
2
0.05
66.67
0.10
6
0.16
54.55
0.30
IN |
1564
40.60
48.81
83.37
h- ...... .4
194
5.04
51.05
10.34
88
2.28
45.60
4.69
16
0.42
40.00
0.85
K* ...... .4
8
0.21
38.10
0.43
H- ...... .4
1
0.03
33.33
0.05
h. ....... 4
5
0.13
45.45
0.27
Total
1976
51.30
1876
48.70
Total
3204
83.18
380
9.87
193
5.01
40
1.04
21
0.55
3
0.08
11
0.29
3852
100.00
Frequency Missing - 103
(continued)
D-51
-------
Table D-20 (continued).
ODOR OF NEW CARPETS, DRAPES
BINARY VARIABLE FREQUENCY
Frequency |
Percent j
Row Pet j
Col Pet |F
0 | 1640
| 42.58
| 51.19
| 83.00
1 | 336
| 8.72
| 51.85
| 17.00
Total 1976
51.30
M |
1564 |
40.60 |
48.81 |
83.37 |
312 |
8.10 |
48.15 |
16.63 |
h +
1876
48.70
Total
3204
83.18
648
16.82
3852
100.00
Frequency Mi**ing - 103
D-52
-------
Table D-21. DAMP/MUSTY ODOR; FISHY SMELLS; ENGINE EXHAUST
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
1
I
I
I
IF
1498
38.91
50.44
75.73
287
7.45
52.09
14.51
108
2.81
55.96
5.46
54
1.40
65.06
2.73
}.
17
0.44
60.71
0.86
h
|M |
1472
38.23
49.56
78.63
264
6,86
47.91
14.10
h. ...... mA
85
2.21
44.04
4.54
29
0.75
34.94
1.55
K •>
11
0.29
39.29
0.59
>• -»
Total
2970
77.14
551
14.31
193
5.01
83
2.16
28
0.73
(Continued)
D-53
-------
Table D-21 (continued). DAMP/MUSTY ODOR; FISHY SMELLS; ENGINE EXHAUST
Frequency
Percent
Row Pet
Col Pet
5
6
7
j
8
9
Total
F
4
0.10
40.00
0.20
6
0.16
75.00
0.30
2
0.05
66.67
0.10
h_-to-_--- J
0
0.00
0.00
0.00
2
0.05
66.67
0.10
1978
51.38
IN |
h ........ 4
6
0.16
60.00
0.32
2
0.05
25.00
0.11
1
0.03
33.33
0.05
hi
1
0.03
100.00
0.05
1
0.03
33.33
0.05
1872
48.62
Frequency Missing - 105
Total
10
0.26
8
0.21
3
0.08
1
0.03
3
0.08
3850
100.00
(continued)
D-54
-------
Table D-21 (continued). DAMP/MUSTY ODOR; FISHY SMELLS; ENGINE EXHAUST
BINARY VARIABLE FREQUENCY
Frequency
Percent
Row Pet
Col Pet
0
H
1
Total
F |M
1498 |
38.91 |
50.44 |
75.73 |
480 |
12.47 |
54.55 |
24.27 |
1978
51.38
1472
38.23
49.56
78.63
400
10.39
45.45
21.37
1872
48.62
Total
2970
77. 14
•
880
22.86
L
3850
100.00
Frequency Missing - 105
D-55
-------
Table D-22. ODOR OF TOBACCO SMOKE
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
•• "1
Total
F
h_ _ _ _ . . J
1687
43.56
50.62
84.77
205
5.29
53.66
10.30
63
1.63
56.25
3.17
35
0.90
76.09
1.76
1990
51.38
M |
H. ...... .4
1646
42.50
49.38
87.41
177
4.57
46.34
9.40
49
1.27
43.75
2.60
11
0.28
23.91
0.58
1883
48.62
Frequency Missing - 82
Total
3333
86.06
382
9.86
112
2.89
46
1.19
3873
100.00
(continued)
D-56
-------
Table D-22 (continued). ODOR OF TOBACCO SMOKE
BINARY VARIABLE FREQUENCY
Frequency|
Percent j
Row Pet |
Col Pet |F
IN
0
1
H
1687
43.56
50.62
84.77
303
7.82
56.11
15.23
h
1646
42.50
49.38
87.41
237
6.12
43.89
12.59
H
Total 1990 1883
51.38 48.62
Frequency Missing - 82
| Total
3333
86.06
540
13.94
3873
100.00
D-57
-------
Table D-23. DRY, ITCHING OR TEARING EYES, SORE EYES, BURNING EYES, BLURRY
VISION
FREQUENCIES USING WORK-RELATED CRITERION
(Synptomi improve after leaving work)
Frequency
Percent
Row Pet
Col Pet
0
1
2
3
4
Total
F
104B
26.55
49.13
51.40
437
11.07
55.39
21.43
316
8.00
53.56
15.50
177
4.48
55.14
8.68
61
1.55
53.04
2.99
2039
51.65
M |
1085
27.48
50.87
56.84
352
8.92
44.61
18.44
274
6.94
46.44
14.35
144
3.65
44.86
7.54
54
1.37
46.96
2.83
1909
48.35
Frequency Minting - 7
Total
2133
54.03
789
19.98
590
14.94
321
8.13
115
2.91
3948
100.00
(Continued)
D-58
-------
Table D-23 (continued). DRY, ITCHING OR TEARING EYES, SORE EYES, BURNING EYES,
BLURRY VISION
BINARY VARIABLE FREQUENCIES USING ALTERNATIVE CRITERION
(At leaat one symptom must have occurred "often" or "alwaya" last year)
Frequency|
Percent
Row Pet
Col Pet
|M
| Total'
1485 | 1437 | 2922
37.61 | 36.40 | 74.01
50.82 | 49.18 |
72.83 75.28
554 | 472 | 1026
14.03 | 11.96 | 25.99
54.00 | 46.00 |
27.17 | 24.72 |
Total
2039 1909 3948
51.65 48.35 100.00
Frequency Mlasing - 7
D-59
-------
APPENDIX E
RESULTS OF LOGISTIC REGRESSIONS
E-l
-------
APPENDIX E. Results of Logistic Regressions
This appendix presents the results of the logistic regressions carried
out on the health, comfort, and odor factors identified in the principal
components analysis (PCA). The first 48 tables provide the results of
individual regressions on the 12 health symptom factors. There are four
tables on each health factor since regressions were run separately for men and
women, and also with and without the comfort and odor indices as independent
variables. The final 20 tables provide the results of regressions on the
four comfort factors and 6 odor factors, again separately for men and women.
The variables included in each logistic regression were those that had
achieved at least the 0.05 level of significance for either men or women in
the second linear regression.
Definitions and sources of the independent variables are provided in
Table 3. Definitions of the dependent variables are found in Appendix B and
Appendix D. The maximum number of persons potentially in each cell of the
binary dependent variable in each regression is provided in Table 11.
Definitions of the combination air handling unit (AHU) and location variables
may be found in Tables 5 and 6. Thus AHU WT2 Fl 46 stands for AHU # WT2
serving floors 4-6 of the West Tower.
Interpretation of Logistic Regression Results
Logistic regressions are often employed in models comparing
prevalence rates. The form of the model is
P
- exp
1 P
where p - prevalence rate (fraction of persons with the given symptom)
PQ - intercept (prevalence rate when all independent variables are
zero, or at their lowest level)
0. - coefficient of the i.th independent variable
Xj - value of the i.th independent variable (0 or 1 if binary, other
values if continuous or multilevel)
n - number of independent variables
E-2
-------
The term p/(l-p) is called the "odds" (Hosmer and Lemeshow, 1989) and
represents the likelihood that the health or comfort complaint will be present
in a member of the population of interest. For example, if 75% of male
respondents reported a headache in the last year, the odds would be 0.75/0.25
- 3, meaning that men having headaches last year are three times as numerous
as those not having headaches last year. Taking the logarithm of both sides
of the equation results in a linear equation:
P
In
1 - p
The effect of any variable x{ on the prevalence ratio p is given by the
coefficient p^. For example, if the variable is a binary (0,1) variable such
as a dusty office, we can evaluate the above equation for Tt-Atat - 1 (dusty
office) and again for x(fcjgt - 0 (clean office) . The difference between the two
values is the effect of the dusty office. This difference is simply fl^t, as
can be seen by noticing that all the other terms on the right-hand side remain
unchanged, and therefore their difference is zero. Since the difference of
two logarithms is the ratio of the antilogarithm, and we have seen earlier
that the antilogarithm is the odds, we can exponentiate the difference (l^t)
to obtain the "odds ratio": exp ftAat- This odds ratio describes the effect
of the dusty office on the prevalence p of the health symptom compared to the
effect of the clean office:
Odds ratio - exp
The significance of the relationship between variables is likely to be
more stable than the numerical magnitude of the odds ratio. This is because
the odds ratio depends so strongly on the choices made in creating indices and
cutoff points for the dependent variables. As can be seen from the above
discussion, these choices are somewhat arbitrary.
As an example of how to interpret the odds ratios listed in Appendix E,
we note that the first table shows that the odds ratio for reporting headache
and nausea among women with dusty offices is 1.31. This means that the odds
that women with dusty offices will report headache and nausea are about 31%
E-3
-------
higher than the odds for women with clean offices. However, the lower 99%
confidence limit is 0.97. This means that we cannot say with 99% confidence
that women with dusty offices will have increased episodes of headache and
nausea compared to women with clean offices.
When the comfort and odor factors were included as independent variables
in the logistic regressions, they retained their original indices, which
varied between 0 and 3, 6, 9, 12, or 15 as described above. Thus these 10
variables were not binary, but rather multilevel. For a multilevel variable,
exp (0.) describes the relative effect of one level of the variable compared
to the next lower level. This has the effect of producing odds ratios that
appear smaller than those for binary variables. Thus it will be noticed in
Table 3 of Appendix E that the odds ratio for headache and nausea among women
reporting hot stuffy air is only about 1.07. This means that each step along
the 15-point index for hot stuffy air is about 7% more likely than the
previous step to be associated with headache and nausea. Thus there is
actually about a (1.07)u - 2.58 ratio, or a 158% difference, in the
prevalence of headache and nausea between women who reported hot stuffy air
the most (15 on the index) and those who reported it the least.
E-4
-------
Response Variable: Headache and Nausea
Number of Observations: 1652
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2198.248
2203.658
2196.248
•
Intercept
and
Covariates
2083.105
2272.446
2013.105
•
Chi-Square for Covariates
183.143 with 34 OF (p=0.0001)
174.601 with 34 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Lamp
New Carpet/Drapes
Glare
Noise
Dust
High Pay
Job Satisfaction
Role Conflict
Workload
Role Clarity
External Stress
College Grad
Allergy to Pollen
Allergy to Animals
Allergy to Dust
Allergy to Molds
Chcm. Sensitivity
Nonsmoker
NEG3
FCFL2
FCFL8
ET1FL89
ET2FL13
UT1FL13
UT1FL67
WT1FL89
UT2FL45
UT2FL67
WT3FL47
WT3FL811
Glued Carpet, Hall
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-2.3542
0.0858
0.0809
0.3839
0.3352
0.2720
-0.3026
-0.2625
0.2187
0.1142
-0.0048
0.0919
-0.3162
0.3208
-0.1847
-0.0796
0.2584
0.4220
0.3831
1.0559
-0.3846
-0.1619
-0.3476
-0.9029
-0.9085
-1.3857
-0.5469
-0.1418
-0.3476
-1.0051
-1.2550
-0.0636
-0.1361
-0.1587
-0.0723
Standard
Error
0.5498
0.1100
0.1429
0.1125
0.1161
0.1173
0.1442
0.1198
0.0921
0.0664
0.0656
0.0450
0.1399
0.1314
0.1624
0.1442
0.1470
0.1199
0.1520
0.5427
0.5244
0.4081
0.4344
0.4797
0.5497
0.6569
0.5469
0.4728
0.4911
0.5939
1.1415
0.1436
0.2054
0.1743
0.1345
Uald
Chi-Sq
18.3314
0.6083
0.3203
11.6474
8.3352
5.3780
4.4048
4.8002
5.6325
2.9551
0.0054
4.1796
5.1067
5.9630
1.2929
0.3047
3.0910
12.3965
6.3564
3.7852
0.5380
0.1575
0.6401
3.5433
2.7313
4.4494
0.9999
0.0900
0.5009
2.8635
1.2089
0.1958
0.4393
0.8285
0.2895
Pr >
Chi-Sq
0.0001
0.4354
0.5714
0.0006
0.0039
0.0204
0.0358
0.0285
0.0176
0.0856
0.9417
0.0409
0.0238
0.0146
0.2555
0.5810
0.0787
0.0004
0.0117
0.0517
0.4633
0.6915
0.4237
0.0598
0.0984
0.0349
0.3173
0.7642
0.4791
0.0906
0.2716
0.6581
0.5075
0.3627
0.5905
Odds
Ratios
0.09497
1.08959
1.08426
1.46800
1.39822
1.31259
0.73889
0.76913
1.24446
1.12098
0.99521
1.09626
0.72891
1.37823
0.83135
0.92349
1.29486
1.52501
1.46682
2.87456
0.68072
0.85053
0.70638
0.40539
0.40313
0.25015
0.57874
0.86779
0.70638
0.36601
0.28508
0.93838
0.87276
0.85325
0.93025
99
Percent
Confidence Intervals
Low Limit
0.02304
0.82073
0.75035
1.09867
1.03679
0.97029
0.50964
0.56490
0.98162
0.94474
0.84048
0.97627
0.50835
0.98247
0.54714
0.63695
0.88668
1.11979
0.99158
0.71028
0.17632
0.29725
0.23071
0.11782
0.09783
0.04606
0.14146
0.25673
0.19935
0.07926
0.01506
0.64823
0.51417
0.54460
0.65785
Upper Limit
0.3914
1.4465
1.5668
1.9615
1.8857
1.7756
1.0713
1.0472
.5777
.3301
.1784
.2310
.0452
.9334
.2632
.3389
.8909
2.0769
2.1698
11.6336
2.6281
2.4336
2.1628
1 .3949
1.6612
1.3586
2.3677
2.9333
2.5030
1.6901
5.3952
1.3584
1.4814
1.3368
1.3154
Concordant * 69.2X
Discordant - 30.5X
Tied > 0.3X
(643860 pairs)
Somers1
Gamma
Tau-a
c
« 0.387
- 0.389
- 0.183
= 0.694
E-5
-------
Response Variable: Headache and Nausea
Number of Observations: 1657
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1513.790
1519.202
1511.790
Intercept
and
Cova Mates
1394.560
1584.007
1324.560
Chi-Square for Coveriates
187.229 with 34 OF (p=0.0001)
181.963 with 34 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Lamp
New Carpet/Drapes
Glare
Noise
Dust
High Pay
job Satisfaction
Role Conflict
Workload
Role Clarity
External Stress
College Grad
Allergy to Pollen
Allergy to Animals
Allergy to Dust
Allergy to Molds
Chem. Sensitivity
Nonsmoker
NEG3
FCFL2
FCFL8
ET1FL89
ET2FL13
UT1FL13
WT1FL67
UT1FL89
UT2FL45
WT2FL67
WT3FL47
UT3FL811
Glued Carpet, Hall
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-3.7723
0.3242
0.4123
0.2366
0.4428
0.4586
-0.1070
-0.2181
0.2201
0.3107
-0.1823
0.1119
-0.1268
-0.0770
-0.2989
0.1500
0.5620
0.5766
-0.2134
1.5916
0.2225
0.2893
-0.2485
-0.2037
0.1574
-0.9196
-38.8228 #
-0.0952
-1.0919
-1.1484
1.5133
-0.1148
-0.0537
-0.0333
0.0874
Standard
Error
0.7667
0.1465
0.1736
0.1457
0.1577
0.1482
0.1864
0.1541
0.1185
0.0930
0.0910
0.0620
0.3134
0.1650
0.2010
0.1841
0.1877
0.1569
0.2185
0.6405
0.5389
0.5236
0.6919
0.5202
0.7138
0.6659
m
0.6885
0.7895
0.7777
0.7187
0.2154
0.2534
0.2362
0.2230
Uald
Chi-Sq
24.2087
4.8994
5.6438
2.6355
7.8805
9.5811
0.3296
2.0021
3.4511
11.1531
4.0127
3.2607
0.1636
0.2179
2.2102
0.6631
8.9692
13.5048
0.9535
6.1754
0.1705
0.3053
0.1290
0.1534
0.0486
1.9072
m
0.0191
1.9128
2.1803
4.4338
0.2841
0.0449
0.0198
0.1537
Pr >
Chi-Sq
0.0001
0.0269
0.0175
0.1045
0.0050
0.0020
0.5659
0.1571
0.0632
0.0008
0.0452
0.0710
0.6859
0.6407
0.1371
0.4155
0.0027
0.0002
0.3288
0.0130
0.6797
0.5806
0.7194
0.6953
0.8255
0.1673
B
0.8900
0.1667
0.1398
0.0352
0.5940
0.8321
0.8880
0.6950
Odds
Ratios
0.02300
1.38292
1.51029
1.26693
1 .55706
1.58186
0.89853
0.80405
1.24620
1.36438
0.83335
1.11840
0.88091
0.92589
0.74163
1.16183
1.75418
1.77998
0.80783
4.91160
1 .24920
1 .33549
0.77997
0.81571
1.17046
0.39868
m
0.90919
0.33558
0.31714
4.54169
0.89154
0.94772
0.96725
1.09133
99
Percent
Confidence Intervals
Low Limit
0.00319
0.94821
0.96571
0.87047
1.03724
1.07987
0.55590
0.54061
0.91837
1.07372
0.65921
0.95331
0.39293
0.60529
0.44190
0.72308
1.08165
1.18818
0.46013
0.94334
0.31170
0.34663
0.13123
0.21358
0.18612
0.07172
m
0.15431
0.04391
0.04278
0.71314
0.51188
0.49339
0.52637
0.61444
Upper Limit
0.1657
2.0169
2.3620
1.8440
2.3374
2.3172
1.4523
1.1959
1.6911
1.7337
1.0535
1.3121
1.9749
1.4163
1.2447
1.8668
2.8449
2.6665
1.4183
25.5728
5.0063
5.1454
4.6359
3.1153
7.3607
2.2161
m
5.3568
2.5647
2.3513
28.9240
1.5528
1.8204
1.7774
1.9384
Association of Predicted Probabilities and Observed Responses
Concordant = 74.2X
Discordant = 25.4X
Tied = 0.4X
(387750 pairs)
NOTE:
Somers1 D = 0.488
Gamma = 0.490
Tau-a = 0.138
c = 0.744
Parameter estimates marked with '#' are regarded to be infinite.
E-6
-------
Response Variable: Headache and Nausea (Including Comfort and Odor)
Number of Observations: 1549
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
201.728 with 32 DF (p=0.0001)
190.806 with 32 DF (p=0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2059.642
2064.988
2057.642
•
Intercept
and
Covariates
1921.914
2098.311
1855.914
•
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Glare
Noise
High Pay
Job Satisfaction
Workload
Role Clarity
External Stress
Wear Contact Lens
College Grad
Allergy to Pollen
Allergy to Molds
Chem. Sensitivity
Nonsmoker
Hot Stuffy Air
Dry Air
Paint Odor
Photocopies Odor
New Carpet Odor
AHU20
AHU21
AHU4
CM FL7
CM FL8
ET1FL67
WT1FL13
WT1FL67
UT2FL67
USATOE
Glued Carpet, Hall
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-1.
0.
0.
-0.
-0.
0.
-0.
0.
-0.
-0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
1.
-0.
0.
-0.
0.
-0.
-1.
-0.
0.
-0.
0.
-0.
-0.
8814
2893
2228
2513
2909
1108
0487
0585
0451
3374
2001
1016
3368
2630
0667
0644
1844
1532
0592
5694
1458
1580
1155
1445
1198
9907
2245
1079
5906
1357
0608
2559
1032
Standard
Error
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.5433
.1170
.1198
.1479
.1218
.0648
.0664
.0467
.1323
.1445
.1280
.1392
.1237
.1588
.0194
.0340
.0603
.1067
.0857
.6155
.6890
.6154
.3839
.4601
.4032
.5934
.6785
.4900
.5366
.1338
.2238
.1815
.1411
Uald
Chi-Sq
11
6
3
2
5
2
0
1
0
5
2
0
7
2
11
3
9
2
0
0
2
0
0
0
0
2
3
0
1
1
0
1
0
.9941
.1154
.4592
.8858
.7081
.9228
.5396
.5672
.1163
.4507
.4427
.5334
.4126
.7430
.7500
.5893
.3568
.0627
.4767
.8558
.7652
.0660
.0905
.0986
.0883
.7868
.2568
.0485
.2117
.0281
.0739
.9886
.5350
Pr >
Chi-Sq
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0005
0134
0629
0894
0169
0873
4626
2106
7331
0196
1181
4652
0065
0977
0006
0582
0022
1509
4899
3549
0963
7973
7635
7535
7664
0950
0711
8258
2710
3106
7857
1585
4645
Odds
Ratios
0.15238
1.33549
1.24957
0.77779
0.74759
1.11717
0.95247
1.06024
0.95590
0.71362
.22152
.10694
.40046
.30083
.06897
.06652
1.20250
1.16556
1.06099
1.76721
3.14496
0.85385
1.12243
0.86545
1.12727
0.37132
0.29390
0.89772
1 .80507
0.87310
1.06269
0.77422
0.90195
99 Percent
Confidence Intervals
Low Limit Upper Limit
0
0
0
0
0
0
0
0
0
0
0
0
1
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.03759 0.6176
.98798 1
.91777 1
.53137 1
.54626 1
.94542
.80272 1
.94007
.67984
.49183
.87842
.77338
.01831
.86410
.01687
.97708
.02949
.88545
.85081
.8052
.7013
.1385
.0231
.3201
.1301
1.1958
.3441
.0354
.6986
.5844
.9260
.9583
.1238
.1641
.4046
.5343
.3231
.36199 8.6273
.53309 18.5536
.17495 4.1673
.41752 3.0175
.26455 2.8312
.39898 3.1850
.08052
.05118
1.7124
1.6876
.25407 3.1719
.45308 7.1914
.61856
.59708
.48508
.62708
1.2324
1.8914
1.2357
1.2973
Association of Predicted Probabilities and Observed Responses
Concordant = 70.5X
Discordant = 29.2X
Tied = 0.3X
(565440 pairs)
Somers1
Gamna
Tau-a
c
0.413
0.414
0.195
0.706
E-7
-------
Response Variable: Headache and Nausea (Including Comfort and Odor)
Number of Observations: 1565
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
U12.793
1418.149
1410.793
Intercept
and
Cover fates
1240.435
1417.171
1174.435
Chi-Square for Covariates
236.358 with 32 OF (p=0.0001)
245.274 with 32 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Glare
Noise
High Pay
Job Satisfaction
Workload
Role Clarity
External Stress
Wear Contact Lens
College Grad
Allergy to Pollen
Allergy to Molds
Chetn. Sensitivity
Nonsmoker
Hot Stuffy Air
Dry Air
Paint Odor
Photocopies Odor
New Carpet Odor
AHU20
AHU21
AHU4
CM FL7
CM FL8
ET1FL67
WT1FL13
WT1FL67
WT2FL67
USATOE
Glued Carpet, Hall
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-4.11950
0.08030
0.22500
-0.00546
-0.26460
0.29860
-0.15070
0.10540
-0.23310
-0.19480
-0.14840
0.52630
0.46990
-0.08020
0.10330
0.14280
0.26740
-0.32780
0.31350
-0.18630
0.41240
0.47310
1.17800
0.33760
1.54390
-0.06840
-0.65180
-0.47680
0.74420
0.00086
0.06320
-0.06580
0.11300
Standard
Error
0.7907
0.1563
0.1675
0.2027
0.1593
0.0931
0.0924
0.0661
0.2095
0.3459
0.1675
0.1741
0.1670
0.2411
0.0259
0.0472
0.0891
0.1674
0.1101
1.0771
0.7227
0.5196
0.3687
0.3878
0.5413
0.8407
0.6605
0.6788
0.5569
0.1921
0.2833
0.2541
0.2413
Wald
Chi-Sq
27.1409
0.2635
1.8042
0.0007
2.7577
10.2877
2.6621
2.5445
1.2389
0.3171
0.7856
9.1386
7.9198
0.1106
15.8455
9.1693
9.0124
3.8325
8.1095
0.0299
0.3257
0.8289
10.2086
0.7581
8.1357
0.0066
0.9739
0.4933
1.7858
0.0000
0.0498
0.0670
0.2192
Pr >
Chi-Sq
0.0001
0.6077
0.1792
0.9785
0.0968
0.0013
0.1028
0.1107
0.2657
0.5734
0.3754
0.0025
0.0049
0.7394
0.0001
0.0025
0.0027
0.0503
0.0044
0.8627
0.5682
0.3626
0.0014
0.3839
0.0043
0.9351
0.3237
0.4825
0.1814
0.9964
0.8235
0.7958
0.6397
Odds
Ratios
0.01625
1.08361
1 .25232
0.99455
0.76751
1.34797
0.86011
1.11115
0.79207
0.82300
0.86209
1 .69266
1.59983
0.92293
1.10882
1.15350
1.30656
0.72051
1.36821
0.83002
1.51044
1.60496
3.24787
1.40158
4.68282
0.93389
0.52111
0.62077
2.10476
1.00086
1.06524
0.93632
1.11963
99
Percent
Confidence Intervals
Low Limit
0.00212
0.72446
0.81344
0.59001
0.50918
1.06054
0.67792
0.93719
0.46173
0.33762
0.55997
1.08093
1.04051
0.49595
1.03726
1.02144
1.03860
0.46812
1.03033
0.05177
0.23474
0.42089
1.25637
0.51614
1.16127
0.10709
0.09506
0.10803
0.50139
0.61019
0.51346
0.48658
0.60134
Upper Limit
0.1246
1.6208
1.9280
1.6765
1.1569
1.7133
1.0912
1.3174
1.3588
2.0062
1.3272
2.6506
2.4598
1.7175
.1853
.3026
.6437
.1090
.8169
13.3073
9.7189
6.1202
8.3961
3.8060
18.8835
8.1437
2.8566
3.5672
8.8355
1.6417
2.2100
1.8018
2.0846
Association of Predicted Probabilities and Observed Responses
Concordant = 78.5X
Discordant = 21.2X
Tied = 0.4X
(340344 pairs)
Somers' D = 0.573
Gamma = 0.575
Tau-a * 0.159
c = 0.787
E-8
-------
Response Variable: Stuffy Nose/Sinus Congestion, Runny Nose, Sneezing, Cough
Number of Observations: 1683
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
280.598 with 26 DF (p=0.0001)
269.289 with 26 DF (p=0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1987.902
1993.330
1985.902
•
Intercept
and
Covariates
1759.304
1905.869
1705.304
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at VDT
Used Chemicals
Water Leaks Nearby
Glare
Dust
High Pay
Role Conflict
External Stress
Wear Contact Lens
Allergy to Pollen
Allergy to Dust
Allergy to Molds
Chem. Sensitivity
Sole Provider
Under 30
AHU12
AHU13
AHU4
NEG3
SE21
ET1F1011
ET2FL89
USATOE
Tackd Carpet, Room
Private Office
Cubicle or Shared
Parameter
Estimate
-3.6428
0.0694
0.0419
-0.1289
0.2265
0.5452
-0.3561
0.2553
0.0910
-0.1090
0.4417
0.6917
0.3781
0.4046
0.2028
-0.5809
0.6251
0.8217
1.0796
0.6563
0.2704
0.5491
0.4423
1.0931
0.3029
0.2358
0.1486
Standard
Error
0.4871
0.0277
0.1449
0.1820
0.1251
0.1323
0.1531
0.0855
0.0515
0.1409
0.1372
0.1555
0.1499
0.1278
0.1838
0.1605
0.4787
0.4858
0.5231
0.5879
0.3310
0.4019
0.5377
0.5269
0.2234
0.1862
0.1439
Wald
Chi-Sq
55.9341
6.2738
0.0836
0.5017
3.2796
16.9720
5.4071
8.9193
3.1165
0.5987
10.3694
19.7804
6.3584
10.0246
1.2170
13.0927
1.7053
2.8611
4.2587
1.2466
0.6669
1.8662
0.6768
4.3046
1.8376
1.6027
1.0659
Pr >
Chi-Sq
0.0001
0.0123
0.7725
0.4788
0.0701
0.0001
0.0201
0.0028
0.0775
0.4391
0.0013
0.0001
0.0117
0.0015
0.2700
0.0003
0.1916
0.0907
0.0390
0.2642
0.4141
0.1719
0.4107
0.0380
0.1752
0.2055
0.3019
Odds
Ratios
0.02618
1.07186
1.04279
0.87906
1.25420
1.72495
0.70040
1.29085
1.09527
0.89673
1.55535
1.99711
1.45951
1.49870
1.22483
0.55939
1.86843
2.27436
2.94350
1.92765
1.31049
1.73169
1.55628
2.98351
1.35378
1.26592
1.16021
99
Percent
Confidence Intervals
Lou Limit
0.00746
0.99805
0.71794
0.55006
0.90868
1.22678
0.47214
1.03567
0.95919
0.62378
1.09229
1.33794
0.99199
1.07830
0.76287
0.36996
0.54442
0.65069
0.76498
0.42395
0.55864
0.61496
0.38953
0.76782
0.76141
0.78360
0.80085
Upper Limit
0.0918
1.1511
1.5146
1.4049
1.7311
2.4254
1.0390
1.6089
1.2507
1.2891
2.2147
2.9810
2.1474
2.0830
1.9665
0.8458
6.4124
7.9496
11.3261
8.7647
3.0742
4.8763
6.2178
11.5929
2.4070
2.0451
1.6808
Association of Predicted Probabilities and Observed Responses
Concordant =75.2%
Discordant = 24.5X
Tied = 0.3X
(567122 pairs)
Somers1
Gamma
Tau-a
c
D = 0.507
= 0.508
= 0.203
* 0.754
E-9
-------
Response Variable: Stuffy Nose/Sinus Congestion, Runny Nose, Sneezing, Cough
Number of Observations: 1689
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1554.194
1559.626
1552.194
Intercept
and
Covariates
1427.457
1574.118
1373.457
Chi-Square for Covariates
178.737 with 26 Of (p=0.0001)
179.086 with 26 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
I ntercept
Hours at VDT
Used Chemicals
Water Leaks Nearby
Glare
Dust
High Pay
Role Conflict
External Stress
Wear Contact Lens
Allergy to Pollen
Allergy to Dust
Allergy to Molds
Chem. Sensitivity
Sole Provider
Under 30
AHU12
AHU13
AHU4
NEG3
SE21
ET1F1011
ET2FL89
USATOE
Tackd Carpet, Room
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-3.6343
0.0531
0.2998
0.4481
0.1472
0.8477
0.0934
0.2628
0.0254
0.1736
0.0605
0.6699
0.3936
0.3953
-0.2019
0.0607
1.1456
-0.2661
0.2421
0.9558
0.2161
-0.0240
-1.0230
-0.7388
0.1222
-0.0495
-0.0424
Standard
Error
0.5643
0.0342
0.2035
0.1866
0.1446
0.1428
0.2077
0.1021
0.0612
0.1820
0.1588
0.1775
0.1776
0.1511
0.1649
0.2956
0.6359
0.6813
0.5288
0.6871
0.4484
0.4856
0.7768
0.6788
0.2665
0.2288
0.2172
Wald
Chi-Sq
41.4842
2.4140
2.1705
5.7691
1.0366
35.2548
0.2022
6.6252
0.1717
0.9099
0.1453
14.2420
4.9125
6.8423
1.4998
0.0421
3.2459
0.1526
0.2096
1.9351
0.2323
0.0024
1.7346
1.1845
0.2103
0.0469
0.0380
Pr >
Chi-Sq
0.0001
0.1203
0.1407
0.0163
0.3086
0.0001
0.6529
0.0101
0.6786
0.3401
0.7031
0.0002
0.0267
0.0089
0.2207
0.8374
0.0716
0.6961
0.6471
0.1642
0.6299
0.9606
0.1878
0.2764
0.6465
0.8286
0.8454
Odds
Ratios
0.02640
1.05454
1.34959
1.56534
1.15859
2.33427
1.09790
1 .30057
1 .02573
1.18958
1.06237
1.95404
1.48231
1.48483
0.81718
1.06258
3.14433
0.76636
1.27392
2.60075
1.24123
0.97629
0.35951
0.47769
1.12998
0.95171
0.95849
99
Percent
Confidence Intervals
Low Limit
0.00617
0.96561
0.79898
0.96794
0.79829
1.61583
0.64298
0.99979
0.87612
0.74436
0.70570
1.23696
0.93810
1.00608
0.53436
0.49621
0.61111
0.13251
0.32625
0.44301
0.39103
0.27946
0.04860
0.08313
0.56876
0.52788
0.54777
Upper Limit
0.1130
1.1517
2.2796
2.5314
1.6815
3.3722
.8747
.6918
.2009
.9011
.5993
3.0868
2.3422
2.1914
1.2497
2.2754
16.1784
4.4323
4.9743
15.2682
3.9400
3.4107
2.6592
2.7450
2.2450
1.7158
1.6772
Concordant =73.9%
Discordant = 25.7X
Tied = 0.4X
(406818 pairs)
S oners'
Ganvna
Tau-a
c
= 0.482
- 0.484
= 0.138
= 0.741
E-10
-------
Response Variable: Stuffy Nose/Sinus Congestion, Runny Nose, Sneezing, Cough
Number of Observations: 1629
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
(Including Comfort and Odor)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1926.008
1931 .403
1924.008
Intercept
and
Covariates
1693.505
1887.751
1621.505
Chi-Square for Covariates
302.503 with 35 DF (p=0.0001)
284.890 with 35 DF
Chi-Sq
0.0001
0.0282
0.0495
0.0019
0.0002
0.0001
0.0893
0.0809
0.0403
0.0004
0.0008
0.0030
0.1712
0.0046
0.0626
0.3188
0.0204
0.0836
0.4131
0.7580
0.6936
0.9718
0.6423
0.1267
0.1772
0.4418
0.8178
0.6407
0.2593
0.2668
0.1580
0.3544
0.1183
0.3044
0.5378
0.8494
Odds
Ratios
0.03185
1 .06354
1.31956
1.48736
1 .69740
2.16972
1.30044
1.38057
1.04488
1.14179
1.16521
1.17998
1.14591
0.63635
2.75605
2.05958
4.17035
2.74807
1.35161
1.25759
0.76414
1.01562
0.77639
3.11927
2.06886
1.41044
1.10252
0.78325
1.84984
1.79409
2.14063
1.75628
2.23983
0.75014
0.89844
1.02819
99
Percent
Confidence Intervals
Low Limit
0.01623
0.98953
0.91720
1 .07069
1.18196
1 .44537
0.87324
0.85766
0.98884
1.03612
1.03634
1.02200
0.88682
0.42206
0.67820
0.31844
0.85403
0.61002
0.52352
0.18511
0.13158
0.32763
0.19080
0.45783
0.51649
0.44571
0.37024
0.20340
0.45403
0.46232
0.53400
0.36668
0.59239
0.36476
0.57418
0.70480
Upper Limit
0.0625
1.1431
1.8984
2.0662
2.4376
3.2571
1.9366
2.2223
1.1041
1.2582
1.3101
1.3624
1.4807
0.9595
11.2000
13.3208
20.3643
12.3797
3.4896
8.5440
4.4378
3.1484
3.1592
21.2522
8.2870
4.4633
3.2831
3.0161
7.5368
6.9623
8.5811
8.4121
8.4688
1.5427
1.4058
1.5000
Association of Predicted Probabilities and Observed Responses
Concordant = 76.5X
Discordant * 23.3X
Tied - 0.2X
(532004 pairs)
S omens'
Gamma
Tau-a
c
0.532
0.534
0.214
0.766
E-ll
-------
Response Variable: Stuffy Nose/Sinus Congestion, Runny Nose, Sneezing, Cough
Number of Observations: 1642
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
(Including Comfort and Odor)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1473.028
1478.432
1471.028
Intercept
and
Covariates
1357.651
1552.183
1285.651
Chi-Square for Covariates
185.377 with 35 DF (p=0.0001)
184.988 with 35 DF (p=0.0001>
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at VDT
Dust
Wear Glasses
Allergy to Pollen
Allergy to Dust
Allergy to Molds
Sole Provider
Hot Stuffy Air
Dry Air
Cold Drafty Air
Paint Odor
Photocopies Odor
Under 30
AHU13
AHU20
AHU4
NEG3
SE21
SE22
SE31
SE32
AUU15SU2
NEG1NEG2
FCFL6
FCFL8
ET1FL13
ET1FL89
ET2FL89
WT1FL13
WT1FL89
UGHAHU19
USATOE
Glued Carpet, Room
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-3.64940
0.06510
0.68250
0.07210
0.10200
0.61250
0.45020
-0.24530
0.09290
0.05140
0.06630
0.13560
0.16860
-0.12030
0.04820
0.40690
0.07590
0.68210
-0.07180
0.84550
0.00850
-0.16940
0.15960
0.17630
0.52600
0.20260
0.13410
-0.21440
-0.38420
-1.22220
-0.93590
-0.20760
-0.16270
0.00052
0.09600
0.00470
Standard
Error
0.3329
0.0350
0.1552
0.1537
0.1650
0.1855
0.1830
0.1740
0.0245
0.0450
0.0642
0.0768
0.1441
0.2729
0.6420
0.8624
0.5383
0.7386
0.4944
0.7373
0.8359
0.6914
0.4740
0.8003
0.6360
0.5690
0.5302
0.6660
0.6612
1.1018
1.0670
0.6191
0.5869
0.3436
0.2332
0.2290
Uald
Chi-Sq
20.1689
3.4641
19.3279
0.2201
0.3822
10.8992
6.0538
1.9877
14.3370
1.3036
1.0640
3.1165
1.3691
0.1941
0.0056
0.2226
0.0199
0.8527
0.0211
1.3150
0.0001
0.0600
0.1133
0.0485
0.6840
0.1268
0.0640
0.1036
0.3377
1.2305
0.7694
0.1125
0.0768
0.0000
0.1695
0.0004
Pr >
Chi-Sq
0.0001
0.0627
0.0001
0.6390
0.5364
0.0010
0.0139
0.1586
0.0002
0.2536
0.3023
0.0775
0.2420
0.6595
0.9402
0.6371
0.8879
0.3558
0.8846
0.2515
0.9919
0.8064
0.7364
0.8257
0.4082
0.7218
0.8003
0.7475
0.5612
0.2673
0.3804
0.7373
0.7816
0.9988
0.6806
0.9836
Odds
Ratios
0.02601
1.06727
1.97882
1.07476
1.10738
1.84504
1.56863
0.78247
1.09735
1.05274
1.06855
1.14522
1.18365
0.88665
1.04938
1.50215
1.07885
1.97803
0.93072
2.32914
1.00854
0.84417
1.17304
1.19280
1.69215
1.22458
1.14351
0.80703
0.68100
0.29458
0.39223
0.81253
0.84985
1.00052
1.10076
1 .00471
99
Percent
Confidence Intervals
Low Limit
0.01103
0.97525
1.32671
0.72337
0.72394
1.14414
0.97902
0.49981
1 .03024
0.93752
0.90567
0.93966
0.81660
0.43898
0.20077
0.16290
0.26961
0.29507
0.26044
0.34862
0.11709
0.14221
0.34596
0.15179
0.32879
0.28276
0.29180
0.14515
0.12400
0.01724
0.02511
0.16490
0.18739
0.41288
0.60367
0.55699
Upper Limit
0.0613
1.1680
2.9514
1.5968
1.6939
2.9753
2.5133
1.2250
1.1688
1.1821
1.2607
1.3958
1.7157
1.7909
5.4849
13.8522
4.3170
13.2598
3.3260
15.5613
8.6866
5.0111
3.9774
9.3734
8.7088
5.3034
4.4812
4.4871
3.7399
5.0331
6.1269
4.0036
3.8542
2.4245
2.0072
1.8123
Concordant = 75.OX
Discordant = 24.6%
Tied = 0.4X
(371541 pairs)
Sorters' D
Gamma
Tau-a
c
0.503
0.505
0.139
0.752
E-12
-------
Response Variable: Wheezing, Shortness of Breath, Chest Tightness
Number of Observations: 1226
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1268.921
1274.032
1266.921
Intercept
and
Covariates
1071.673
1204.573
1019.673
Chi-Square for Covariates
247.247 with 25 OF (p=0.0001)
245.894 with 25 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Used Chemicals
New Furniture
Walls Painted
Lights Too Dim
Dust
Medium Pay
Role Conflict
Asthma
College Grad
Allergy to Animals
Chem. Sensitivity
Sole Provider
Under 30
G1LG1
CM FL2
FCFL3
ET1FL89
ET1F1011
WT2FL67
ET3FL811
Glued Carpet, Hall
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-3.1938
0.3101
0.0743
0.5243
0.4270
0.4513
•0.2627
0.6456
1.5842
-0.7816
0.1672
0.8739
0.6996
-0.6761
1.7115
0.1467
-0.1683
1.2824
-0.4818
-1.9093
-0.9682
0.0157
0.6958
-0.1179
-0.0310
0.0915
Standard
Error
0.3037
0.1916
0.1651
0.1876
0.2621
0.1698
0.1869
0.1094
0.2310
0.1837
0.2086
0.1646
0.2245
0.2215
0.8318
0.5153
0.8721
0.5378
0.6018
1.0947
1.1027
0.1891
0.2633
0.3584
0.2357
0.1977
Wald
Chi-Sq
10.5561
2.6203
0.2024
7.8087
2.6535
7.0662
1.9757
34.8304
47.0275
18.0925
0.6423
28.1813
9.7134
9.3151
4.2335
0.0810
0.0373
5.6855
0.6409
3.0419
0.7710
0.0069
6.9844
0.1083
0.0173
0.2145
Pr >
Chi-Sq
0.0001
0.1055
0.6528
0.0052
0.1033
0.0079
0.1598
0.0001
0.0001
0.0001
0.4229
0.0001
0.0018
0.0023
0.0396
0.7760
0.8470
0.0171
0.4234
0.0811
0.3799
0.9337
0.0082
0.7421
0.8954
0.6433
Odds
Ratios
0.04102
.36356
.07713
.68928
.53265
.57035
0.76897
1 .90713
4.87539
0.45767
1.18199
2.39624
2.01295
0.50860
5.53726
1.15801
0.84510
3.60528
0.61767
0.14818
0.37977
1.01582
2.00531
0.88878
0.96948
1.09582
99
Percent
Confidence Intervals
Low Limit
0.01876
0.83238
0.70399
1.04190
0.78023
1.01399
0.47514
1.43877
2.68893
0.28513
0.69063
1.56814
1.12895
0.28746
0.64972
0.30706
0.08938
0.90215
0.13107
0.00883
0.02218
0.62411
1.01769
0.35305
0.52826
0.65851
Upper Limit
0.0897
2.2337
1.6481
2.7389
3.0107
2.4320
1.2445
2.5280
8.8397
0.7346
2.0229
3.6616
3.5891
0.8999
47.1916
4.3672
7.9903
14.4078
2.9108
2.4859
6.5036
1.6534
3.9514
2.2375
1.7792
1.8235
Association of Predicted Probabilities and Observed Responses
Concordant = 78.9X
Discordant = 20.8X
Tied = 0.3X
(251160 pairs)
Somens'
Gamma
Tau-a
c
D - 0.580
- 0.582
= 0.194
= 0.790
E-13
-------
Response Variable: Wheezing, Shortness of Breath, Chest Tightness
Nunber of Observations: 1494
Sex: Males
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
121.136 with 25 DF (p=0.0001)
129.063 with 25 OF (p=0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
997.726
1003.035
995.726
Intercept
and
Covariates
926.591
1064.630
874.591
Analysis of Maximum Likelihood Estimates
Variable
I ntcrcept
Used Chemicals
New Furniture
Walls Painted
Lights Too Dim
Dust
Medium Pay
Role Conflict
Asthma
College Grad
Allergy to Animals
Chem. Sensitivity
Sole Provider
Under 30
G1LG1
CM FL2
FCFL3
ET1FL89
ET1F1011
WT2FL67
ET3FL811
Glued Carpet, Hall
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-3.4710
0.4080
0.4001
0.2228
0.2949
0.9034
-0.4499
0.2362
1 .4224
-0.2664
-0.2073
0.6804
0.2002
-0.0337
-37.6174 #
-1.8527
-0.6841
-37.4162 #
-0.5133
0.1954
-0.4532
-0.2408
0.5307
-0.2699
0.1068
0.3043
Standard
Error
0.5023
0.2569
0.1888
0.2213
0.3841
0.1852
0.2923
0.1343
0.2369
0.3800
0.2436
0.1922
0.2014
0.3992
B
1 .0561
1.0895
m
0.8076
0.6689
0.7084
0.2213
0.2990
0.4399
0.3079
0.3050
Wald
Chi-Sq
47.7533
2.5219
4.4890
1.0145
0.5892
23.7969
2.3683
3.0932
36.0381
0.4914
0.7244
12.5316
0.9880
0.0071
m
3.0778
0.3942
^
0.4040
0.0854
0.4092
1.1836
3.1502
0.3763
0.1204
0.9956
Pr >
Chi-Sq
0.0001
0.1123
0.0341
0.3138
0.4427
0.0001
0.1238
0.0786
0.0001
0.4833
0.3947
0.0004
0.3202
0.9328
B
0.0794
0.5301
•
0.5250
0.7701
0.5224
0.2766
0.0759
0.5396
0.7286
0.3184
Odds
Ratios
0.03109
1.50381
1.49197
1.24957
1.34299
2.46798
0.63769
1.26643
4.14706
0.76613
0.81278
1 .97467
1.22165
0.96686
B
0.15681
0.50454
•
0.59852
1.21580
0.63559
0.78600
1 .70012
0.76346
1.11271
1.35568
99
Percent
Confidence Intervals
Low Limit
0.00852
0.77587
0.9I737
0.70662
0.49930
1.53162
0.30033
0.89605
2.25274
0.28786
0.43396
1.20357
0.72716
0.34575
a
0.01032
0.03048
•
0.07474
0.21704
0.10249
0.44447
0.78700
0.24584
0.50341
0.61793
Upper Limit
0.11337
2.91472
2.42650
2.20973
3.61229
3.97679
1.35400
1.78989
7.63432
2.03904
1.52229
3.23979
2.05239
2.70375
m
2.38170
8.35159
.
4.79262
6.81064
3.94180
1.38995
3.67269
2.37093
2.45948
2.97421
Association of Predicted Probabilities and Observed Responses
Concordant = 74.3X
Discordant 25.OX
Tied - 0.7%
(207545 pairs)
NOTE:
Somers'
Gamma
Tau-a
c
D = 0.493
= 0.497
= 0.092
= 0.747
Parameter estimates marked with '#' are regarded to be infinite.
E-14
-------
Response Variable: Wheezing, Shortness of Breath, Chest Tightness
Number of Observations: 1177
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Cover-fates
272.602 with 27 OF (p=0.0001)
274.946 with 27 DF (p=0.0001)
(Including Comfort and Odor)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1187.120
1192.191
1185.120
Intercept
and
Covariates
968.518
1110.499
912.518
•
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Walls Painted
Noise
Dust
Role Conflict
Workload
Wear Contact Lens
Asthma
College Grad
Allergy to Animals
Allergy to Dust
Allergy to Molds
Sole Provider
Hot Stuffy Air
Paint Odor
Photocopies Odor
New Carpet Odor
Musty/Diesel Odor
Under 30
AHU16
AHU9
NE1
NEG3
SE32
G1LG1
WT2FL67
Private Office
Cubicle or Shared
Parameter
Estimate
-4.22190
0.21390
0.07190
0.12690
0.63810
0.19190
0.10020
1.46470
-0.81740
0.48980
-0.63970
0.40280
0.76480
0.10240
0.46360
0.25250
0.09360
0.00488
-0.38280
1.62860
0.50390
-0.25030
1.35530
0.20160
1.89310
-1.38810
-0.26050
-0.07730
Standard
Error
0.4233
0.2031
0.1812
0.1944
0.1287
0.1025
0.2008
0.2598
0.2016
0.2505
0.2362
0.2387
0.2414
0.0248
0.0890
0.1567
0.1217
0.1003
0.2296
0.6123
0.5117
0.5512
0.8675
0.5472
1.0292
1.0988
0.2570
0.2102
Wald
Chi-Sq
99.4699
1.1090
0.1574
0.4259
24.5803
3.5063
0.2488
31.7741
16.4437
3.8234
7.3358
2.8473
10.0354
17.0149
27.1387
2.5957
0.5913
0.0024
2.7798
7.0731
0.9699
0.2063
2.4410
0.1357
3.3831
1.5960
1.0276
0.1353
Pr >
Chi-Sq
0.0001
0.2923
0.6916
0.5140
0.0001
0.0611
0.6179
0.0001
0.0001
0.0505
0.0068
0.0915
0.0015
0.0001
0.0001
0.1072
0.4419
0.9612
0.0955
0.0078
0.3247
0.6497
0.1182
0.7126
0.0659
0.2065
0.3107
0.7130
Odds
Rat i os
0.01467
.23850
.07455
.13530
.89288
.21155
.10539
4.32625
0.44158
1.63199
0.52745
1.49601
2.14856
1.10783
1.58979
1.28724
1.09812
1.00489
0.68195
5.09673
1.65516
0.77857
3.87792
1.22336
6.63992
0.24955
0.77067
0.92561
99
Percent
Confidence Intervals
Low Limit
0.00493
0.73397
0.67377
0.68806
1 .35875
0.93040
0.65898
2.21545
0.26271
0.85600
0.28703
0.80889
1.15368
1.03927
1.26407
0.85971
0.80260
0.77609
0.37748
1.05265
0.44298
0.18821
0.41504
0.29880
0.46855
0.01472
0.39751
0.53860
Upper Limit
0.0437
2.0898
1.7137
1.8732
2.6370
1.5777
1 .8542
8.4481
0.7422
3.1114
0.9692
2.7668
4.0014
1.1809
1.9994
1 .9274
1.5025
1.3012
1.2320
24.6773
6.1845
3.2207
36.2334
5.0088
94.0963
4.2309
1.4941
1.5907
Association of Predicted Probabilities and Observed Responses
Concordant = 81.1X
Discordant = 18.6X
Tied = 0.3X
(223482 pairs)
Somers1
Gamma
Tau-a
c
D - 0.625
= 0.627
= 0.202
= 0.812
E-15
-------
Response Variable: Wheezing, Shortness of Breath, Chest Tightness
Hunter of Observations: 1464
Sex: Males
The LOGISTIC Procedure
Criteria for Assessing Model Fit
(Including Comfort and Odor)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
916.385
921.674
914.385
Intercept
and
Covariates
788.785
936.875
732.785
Chi-Square for Covariates
181.600 with 27 OF (p=0.0001)
219.585 with 27 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Walls Painted
Noise
Dust
Role Conflict
Workload
Wear Contact Lens
Asthma
College Grad
Allergy to Animals
Allergy to Dust
Allergy to Molds
Sole Provider
Hot Stuffy Air
Paint Odor
Photocopies Odor
New Carpet Odor
Musty/Diesel Odor
Under 30
AHU16
AHU9
NE1
NEG3
SE32
G1LG1
WT2FL67
Private Office
Cubicle or Shared
Parameter
Estimate
-5.3178
0.3057
-0.3601
0.5013
0.2309
0.1756
-0.2926
1.2168
0.0271
-0.4072
0.1811
0.6036
0.1595
0.1205
0.2732
0.2189
0.2478
0.3191
-0.4173
0.9325
-1.6160
1.0596
1.6975
-37.5638 #
-37.0511
0.6112
0.1522
0.2996
Standard
Error
0.7323
0.2499
0.2158
0.2167
0.1566
0.1244
0.2882
0.2735
0.4674
0.2753
0.2623
0.2623
0.2237
0.0302
0.1220
0.1942
0.1415
0.1017
0.4298
0.6205
1.1412
0.4928
0.7213
•
*
0.6762
0.3569
0.3517
Wald
Chi-Sq
52.7272
1.4961
2.7843
5.3525
2.1732
1.9938
1.0311
19.7905
0.0034
2.1876
0.4763
5.2938
0.5084
15.9675
5.0138
1.2703
3.0652
9.8372
0.9426
2.2588
2.0052
4.6223
5.5387
»
B
0.8170
0.1819
0.7255
Pr >
Chi-Sq
0.0001
0.2213
0.0952
0.0207
0.1404
0.1579
0.3099
0.0001
0.9538
0.1391
0.4901
0.0214
0.4758
0.0001
0.0251
0.2597
0.0800
0.0017
0.3316
0.1329
0.1568
0.0316
0.0186
.
•
0.3661
0.6698
0.3943
Odds
Ratios
0.00490
1.35757
0.69761
1.65087
1.25973
1.19196
0.74632
3.37637
1.02747
0.66551
1.19854
1.82869
1.17292
1.12806
1.31416
1.24471
1.28120
1 .37589
0.65882
2.54085
0.19869
2.88522
5.46028
.
•
1.84264
1.16439
1 .34932
99
Percent
Confidence Intervals
Low Limit
0.00074
0.71317
0.40012
0.94467
0.84156
0.86515
0.35523
1.66907
0.30823
0.32747
0.60982
0.93045
0.65918
1.04363
0.95976
0.75476
0.88985
1.05878
0.21774
0.51381
0.01051
0.81070
0.85166
.
,
0.32281
0.46432
0.54532
Upper Limit
0.0323
2.5843
1.2163
2.8850
1.8857
1.6422
1.5680
6.8301
3.4251
1.3525
2.3556
3.5941
2.0870
1.2193
1.7994
2.0527
1.8447
1.7880
1.9934
12.5649
3.7574
10.2682
35.0077
•
.
10.5180
2.9200
3.3387
Association of Predicted Probabilities and Observed Responses
Concordant = 80.5%
Discordant = 18.9X
Tied = 0.6X
(182988 pairs)
MOTE:
Somers'
Gamma
Tau-a
c
D = 0.617
= 0.621
- 0.105
= 0.808
Parameter estimates marked with '#' are regarded to be infinite.
E-16
-------
Response Variable: Dry, Itching or Tearing Eyes, Sore Eyes, Burning Eyes, Blurry Vision
Number of Observations: 1762
Sex: Females
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
243.188 with 32 DF (p=0.0001)
230.186 with 32 DF (p=0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1869.798
1875.272
1867.798
Intercept
and
Covariates
1690.610
1871.258
1624.610
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Plants Nearby
Hours at VDT
1 Hour at Copier
Use Heater
Walls Painted
Water Leaks Nearby
Window Nearby
Lights Too Dim
Glare
Noise
Dust
Workload
External Stress
Wear Contact Lens
Wear Glasses
College Grad
Allergy to Animals
Allergy to Molds
Chem. Sensitivity
Sole Provider
AHU12
AHU16
SE31
G1LG1
FCFL2
ET1FL67
ET1F1011
ET2F1011
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-3.9155
-0.3248
-0.0254
0.0619
0.0865
0.3961
0.5118
0.2300
-0.2172
0.3552
0.6890
0.1644
0.3303
0.3415
-0.0253
0.7762
0.2500
-0.3440
-0.1353
0.3972
0.5184
0.1864
0.7096
0.1747
-1.6316
0.9698
0.6999
-0.2347
0.4391
0.2432
-0.5486
-0.0969
0.2213
Standard
Error
0.5560
0.1363
0.1280
0.0286
0.1455
0.1446
0.1424
0.1807
0.1349
0.2110
0.1329
0.1343
0.1386
0.0719
0.0536
0.1377
0.1281
0.1514
0.1715
0.1442
0.1317
0.1920
0.4433
0.5500
1.0444
0.8106
0.5027
0.5609
0.4538
0.5127
0.3007
0.1904
0.1527
Wald
Chi-Sq
49.5913
5.6772
0.0393
4.6785
0.3534
7.5084
12.9206
1.6195
2.5922
2.8348
26.8571
1.4993
5.6772
22.5589
0.2221
31.7892
3.8112
5.1636
0.6218
7.5889
15.4922
0.9431
2.5619
0.1009
2.4406
1.4312
1.9388
0.1752
0.9365
0.2250
3.3280
0.2591
2.1010
Pr >
Chi-Sq
0.0001
0.0172
0.8428
0.0305
0.5522
0.0061
0.0003
0.2032
0.1074
0.0922
0.0001
0.2208
0.0172
0.0001
0.6374
0.0001
0.0509
0.0231
0.4304
0.0059
0.0001
0.3315
0.1095
0.7508
0.1182
0.2316
0.1638
0.6756
0.3332
0.6352
0.0681
0.6107
0.1472
Odds
Ratios
0.01993
0.72267
0.97492
1.06386
1.09035
1.48602
1.66829
1.25860
0.80477
1.42647
1.99172
1.17869
1.39139
1 .40706
0.97502
2.17320
1.28403
0.70893
0.87345
1.48765
1.67934
1.20490
2.03318
1.19089
0.19562
2.63742
2.01355
0.79081
1.55131
1.27532
0.57776
0.90765
1.24770
99
Percent
Confidence Intervals
Low Limit
0.00476
0.50869
0.70108
0.98830
0.74953
1.02389
1.15601
0.79019
0.56853
0.82834
1 .41432
0.83397
0.97362
1.16916
0.84927
1.52422
0.92313
0.47998
0.56153
1.02607
1.19619
0.73477
0.64899
0.28878
0.01327
0.32683
0.55153
0.18645
0.48196
0.34044
0.26628
0.55579
0.84193
Upper Limit
0.0835
1.0267
1 .3557
1.1452
1.5861
2.1567
2.4076
2.0047
1.1392
2.4565
2.8049
1.6659
1.9884
1.6934
1.1194
3.0985
1.7860
1.0471
1.3586
2.1569
2.3576
1.9758
6.3696
4.9111
2.8828
21.2829
7.3511
3.3541
4.9933
4.7775
1.2536
1.4823
1.8490
Association of Predicted Probabilities and Observed Responses
Concordant « 74.4X
Discordant = 25.3X
Tied = 0.3X
(537040 pairs)
Somers1
Gamma
Tau-a
c
" 0.491
- 0.493
« 0.170
= 0.746
E-17
-------
Response Variable: Dry, Itching or Tearing Eyes, Sore Eyes, Burning Eyes, Blurry Vision
Number of Observations: 1732
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
182.336 with 32 DF (p=0.0001)
188.200 with 32 DF (p=0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1285.566
1291.023
1283.566
Intercept
and
Covariates
1167.230
1347.312
1101.230
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Plants Nearby
Hours at VDT
1 Hour at Copier
Use Heater
Walls Painted
Water Leaks Nearby
Window Nearby
Lights Too Dim
Glare
Noise
Dust
Workload
External Stress
Wear Contact Lens
Wear Glasses
College Grad
Allergy to Animals
Allergy to Molds
Chen. Sensitivity
Sole Provider
AHU12
AHU16
SE31
G1LG1
FCFL2
ET1FL67
ET1F1011
ET2F1011
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-5.54390
0.08970
0.23180
0.00266
-0.30630
0.49970
-0.06960
0.15210
-0.55440
0.45510
0.32480
0.36680
0.60520
0.36960
0.18250
0.81770
0.25270
-0.42220
0.11480
0.42740
0.61110
0.03930
1.10110
1.65240
-0.90890
-0.39800
-0.30040
-1.04780
-0.03080
-0.51600
0.62400
-0.23860
0.05790
Standard
Error
0.7469
0.1731
0.1674
0.0402
0.1792
0.2309
0.2029
0.2181
0.1747
0.3002
0.1663
0.1780
0.1682
0.0993
0.0695
0.1922
0.1717
0.3243
0.2074
0.1760
0.1703
0.1815
0.6615
0.4874
1.1012
1.1798
0.7559
1.0660
0.5682
0.6828
0.2948
0.2562
0.2471
Wald
Chi-Sq
55.0945
0.2687
1.9172
0.0044
2.9217
4.6826
0.1175
0.4867
10.0704
2.2991
3.8119
4.2480
12.9406
13.8613
6.9022
18.1070
2.1653
1 .6941
0.3062
5.8977
12.8781
0.0469
2.7704
11.4943
0.6812
0.1138
0.1579
0.9663
0.0029
0.5711
4.4813
0.8673
0.0550
Pr >
Chi-Sq
0.0001
0.6042
0.1662
0.9472
0.0874
0.0305
0.7318
0.4854
0.0015
0.1294
0.0509
0.0393
0.0003
0.0002
0.0086
0.0001
0.1412
0.1931
0.5800
0.0152
0.0003
0.8286
0.0960
0.0007
0.4092
0.7359
0.6911
0.3256
0.9568
0.4498
0.0343
0.3517
0.8146
Odds
Ratios
0.00391
1.09385
1.26087
1.00266
0.73617
1.64823
0.93277
1.16428
0.57442
1 .57633
1.38375
1.44311
1.83162
1.44716
1.20021
2.26528
1.28750
0.65560
1.12165
1.53327
1.84246
1.04008
3.00747
5.21949
0.40297
0.67166
0.74052
0.35071
0.96967
0.59690
1.86638
0.78773
1.05961
99
Percent
Confidence Intervals
Low Limit
0.00057
0.70033
0.81920
0.90403
0.46398
0.90928
0.55307
0.66383
0.36625
0.72745
0.90160
0.91235
1.18758
1.12053
1.00347
1.38070
0.82729
0.28434
0.65740
0.97436
1.18816
0.65165
0.54721
1 .48714
0.02362
0.03216
0.10565
0.02251
0.22436
0.10281
0.87336
0.40715
0.56067
Upper Limit
0.0268
1.7085
1.9406
1.1121
1.1680
2.9877
1.5731
2.0420
0.9009
3.4158
2.1238
2.2826
2.8249
1.8690
1.4355
3.7166
2.0037
1.5116
1.9137
2.4128
2.8571
1.6600
16.5291
18.3191
6.8743
14.0295
5.1903
5.4642
4.1908
3.4656
3.9885
1.5240
2.0026
Association of Predicted Probabilities and Observed Responses
Concordant = 76.4X
Discordant * 23.1X
Tied = 0.5X
(320931 pairs)
Somers1
Gamma
Tau-a
c
0.534
0.536
0.114
0.767
E-18
-------
Response Variable:
Number of Observations:
Sex: Female
1730
Dry, Itching or Tearing Eyes, Sore Eyes, Burning Eyes, Blurry Vision
(Including Comfort and Odor)
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1853.524
1858.980
1851.524
Intercept
and
Covariates
1613.039
1754.892
1561.039
Chi-Square for Covariates
290.485 with 25 DF (p=0.0001)
278.745 with 25 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at VDT
1 Hour at Copier
Use Heater
Walls Painted
Glare
Workload
Wear Contact Lens
Wear Glasses
Allergy to Animals
Allergy to Molds
Chan. Sensitivity
Hot Stuffy Air
Dry Air
Cosmetic, Body Odor
Photocopies Odor
AHU10
AHU12
AHU16
AHU21
NE1
SE32
FCFL2
FCFL6
Private Office
Cubicle or Shared
Parameter
Estimate
-4.9092
0.0573
0.0969
0.3717
0.4622
0.6592
0.2620
0.8314
0.2843
-0.1965
0.3898
0.4022
0.0759
0.0927
0.0596
0.3662
0.3592
0.4950
0.4872
0.4370
0.8155
0.9715
0.8944
1.0423
-0.2038
0.1965
Standard
Error
0.3658
0.0291
0.1496
0.1480
0.1452
0.1357
0.0730
0.1401
0.1306
0.1750
0.1474
0.1362
0.0220
0.0381
0.0408
0.0940
0.8968
0.4784
0.5294
0.7140
0.3622
0.3927
0.5076
0.4738
0.1909
0.1513
Wald
Chi-Sq
80.1091
3.8811
0.4199
6.3080
10.1272
23.5817
12.8904
35.2201
4.7406
1.2606
6.9927
8.7189
11.8445
5.9100
2.1402
15.1789
0.1604
1.0708
0.8468
0.3746
5.0695
6.1190
3.1043
4.8383
1.1399
1.6872
Pr >
Chi-Sq
0.0001
0.0488
0.5170
0.0120
0.0015
0.0001
0.0003
0.0001
0.0295
0.2615
0.0082
0.0031
0.0006
0.0151
0.1435
0.0001
0.6888
0.3008
0.3575
0.5405
0.0244
0.0134
0.0781
0.0278
0.2857
0.1940
Odds
Ratios
0.00738
.05897
.10175
.45020
.58756
.93325
.29953
2.29653
1.32883
0.82160
1.47669
1.49511
1.07885
1.09713
1.06141
1 .44224
1.43218
1 .64050
1.62775
1.54806
2.26031
2.64190
2.44587
2.83573
0.81563
1.21714
99
Percent
Confidence Intervals
Low Limit
0.00288
0.98249
0.74941
0.99050
1.09217
1.36293
1.07676
1.60080
0.94921
0.52346
1.01015
1.05269
1.01941
0.99457
0.95552
1.13208
0.14214
0.47838
0.41622
0.24604
0.88912
0.96070
0.66154
0.83677
0.49880
0.82428
Upper Limit
0.0189
1.1414
1 .6197
2.1233
2.3077
2.7422
1.5684
3.2946
1.8603
1.2896
2.1587
2.1235
1.1418
1.2103
1.1790
1.8374
14.4307
5.6258
6.3658
9.7402
5.7461
7.2652
9.0429
9.6100
1.3337
1.7972
Association of Predicted Probabilities and Observed Responses
Concordant = 76.7X
Discordant = 23.OX
Tied - 0.3%
(524496 pairs)
Somers' D
Gamma
Tau-a
c
0.538
0.539
0.189
0.769
E-19
-------
Response Variable:
Number of Observations:
Sex: Males
1728
Dry, Itching or Tearing Eyes, Sore Eyes, Burning Eyes, Blurry Vision
(Including Comfort and Odor)
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1292.394
1297.849
1290.394
Intercept
and
Covariates
1104.712
1246.535
1052.712
Chi-Square for Covariates
237.682 with 25 DF (p=0.0001)
243.197 with 25 DF (p-0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at VDT
1 Hour at Copier
Use Heater
Walls Painted
Glare
Workload
Wear Contact Lens
Wear Glasses
Allergy to Animals
Allergy to Molds
Chem. Sensitivity
Hot Stuffy Air
Dry Air
Cosmetic, Body Odor
Photocopies Odor
AHU10
AHU12
AHU16
AHU21
NE1
SE32
FCFL2
FCFL6
Private Office
Cubicle or Shared
Parameter
Estimate
-5.6472
0.0446
-0.2703
0.4184
0.0459
0.2237
0.3081
0.7789
0.2883
0.0559
0.3608
0.5677
0.1104
0.1939
0.2074
0.4350
0.8138
1.2297
1.3243
0.0806
0.5606
-1.2345
-1.3004
0.3130
-0.0182
0.1454
Standard
Error
0.5303
0.0404
0.1845
0.2320
0.2024
0.1679
0.0995
0.1934
0.1763
0.2079
0.1778
0.1728
0.0266
0.0502
0.0593
0.1380
0.7623
0.6905
0.4942
0.8482
0.4626
1.0700
1.0583
0.6578
0.2617
0.2521
Wald
Chi-Sq
13.3935
1.2238
2.1463
3.2511
0.0515
1.7751
9.5828
16.2267
2.6731
0.0723
4.1175
10.7946
17.2369
14.9090
12.2255
9.9326
1.1395
3.1719
7.1807
0.0090
1.4686
1.3311
1.5098
0.2265
0.0048
0.3326
Pr >
Chi-Sq
0.0001
0.2686
0.1429
0.0714
0.8205
0.1827
0.0020
0.0001
0.1021
0.7879
0.0424
0.0010
0.0001
0.0001
0.0005
0.0016
0.2858
0.0749
0.0074
0.9243
0.2256
0.2486
0.2192
0.6341
0.9446
0.5641
Odds
Ratios
0.00353
1.04561
0.76315
1.51953
1.04697
1.25070
1.36084
2.17907
1.33416
1.05749
1.43448
1.76420
1.11672
1.21397
1.23047
1.54496
2.25647
3.42020
3.75955
1.08394
1.75172
0.29098
0.27242
1.36752
0.98196
1.15650
99
Percent
Confidence Intervals
Low Limit
0.00090
0.94226
0.47446
0.83591
0.62159
0.81155
1.05315
1.32406
0.84717
0.61900
0.90736
1.13040
1.04277
1.06672
1.05616
1.08276
0.31667
0.57751
1.05258
0.12192
0.53203
0.01848
0.01784
0.25120
0.50040
0.60410
Upper Limit
0.0138
1.1603
1.2275
2.7622
1.7635
1.9275
1.7584
3.5862
2.1011
1.8066
2.2678
2.7534
1.1959
1.3816
1.4336
2.2045
16.0785
20.2555
13.4282
9.6366
5.7676
4.5806
4.1611
7.4446
1.9270
2.2140
Association of Predicted Probabilities and Observed Responses
Concordant = 79.8X
Discordant = 19.7X
Tied = 0.4X
(322695 pairs)
Somens'
Gamma
Tau-a
c
D =
0.601
0.603
0.130
0.800
E-20
-------
Response Variable: Sore Throat, Hoarseness, Dry Throat
Number of Observations: 1721
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1916.967
1922.417
1914.967
I ntercept
and
Covariates
1829.492
2020.265
1759.492
Chi-Square for Covariates
155.475 with 34 DF (p=0.0001)
150.368 with 34 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
New Furniture
Window Nearby
Glare
Noise
Dust
Role Conflict
Workload
External Stress
Allergy to Molds
Chan. Sensitivity
Sole Provider
Nonsmoker
Under 30
AHU20
ETB
NEG3
NEG4
SE22
G1LG1
NEG1NEG2
CM FL2
FCFL6
ET1FL67
ET1F1011
WT1FL89
WT1F1011
WT2F1011
ET3FL811
WT3FL47
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-3.3300
-0.2825
0.3877
-0.1941
0.1722
0.1133
0.4476
0.3662
0.0240
0.0973
0.3609
0.3573
0.4153
0.0981
-0.3300
-0.5153
-1.6670
-1.6080
-1.2119
0.4224
0.3966
-1.6302
-0.5192
-0.1807
-0.4466
-0.0952
-0.6671
-1.2436
0.4120
-0.1197
-0.2121
0.6349
0.2172
-0.1026
-0.0998
Standard
Error
0.5179
0.1273
0.1209
0.1284
0.1236
0.1265
0.1318
0.0882
0.0705
0.0500
0.1313
0.1268
0.1753
0.1616
0.1520
0.7833
1.0737
0.7958
0.8237
0.6802
0.7632
1.0840
0.4764
0.5233
0.4834
0.4285
0.6496
0.6429
0.4421
0.5728
0.5810
0.2244
0.2632
0.1711
0.1407
Uald
Chi-Sq
41.3377
4.9274
10.2908
2.2857
1.9411
0.8022
11.5394
17.2261
0.1154
3.7912
7.5593
7.9414
5.6126
0.3690
4.7113
0.4328
2.4107
4.0824
2.1646
0.3858
0.2701
2.2616
1.1877
0.1192
0.8537
0.0494
1.0545
3.7419
0.8688
0.0437
0.1332
8.0061
0.6808
0.3597
0.5030
Pr >
Chi-Sq
0.0001
0.0264
0.0013
0.1306
0.1636
0.3704
0.0007
0.0001
0.7340
0.0515
0.0060
0.0048
0.0178
0.5436
0.0300
0.5106
0.1205
0.0433
0.1412
0.5345
0.6033
0.1326
0.2758
0.7299
0.3555
0.8241
0.3045
0.0531
0.3513
0.8344
0.7151
0.0047
0.4093
0.5487
0.4782
Odds
Ratios
0.03579
0.75390
1 .47359
0.82358
.18792
.11997
.56455
.44224
.02429
.10219
.43462
.42946
1.51483
1.10307
0.71892
0.59732
0.18881
0.20029
0.29763
1.52562
1.48676
0.19589
0.59500
0.83469
0.63980
0.90919
0.51319
0.28834
1.50983
0.88719
0.80888
1.88683
1.24259
0.90249
0.90502
99
Percent
Confidence Intervals
Low Limit
0.00943
0.54312
1.07924
0.59164
0.86399
0.80851
1.11414
1.14912
0.85418
0.96899
1.02293
1.03114
0.96438
0.72747
0.48600
0.07941
0.01188
0.02578
0.03566
0.26453
0.20817
0.01200
0.17440
0.21681
0.18418
0.30149
0.09628
0.05504
0.48343
0.20286
0.18109
1.05849
0.63078
0.58080
0.62987
Upper Limit
0.1359
1.0465
2.0120
1.1464
1.6333
1.5514
2.1971
1.8101
1.2283
1.2537
2.0120
1.9817
2.3795
1.6726
1.0635
4.4928
3.0007
1.5558
2.4842
8.7986
10.6186
3.1969
2.0299
3.2134
2.2225
2.7418
2.7354
1.5106
4.7155
3.8800
3.6131
3.3634
2.4478
1.4024
1.3004
Association of Predicted Probabilities and Observed Responses
Concordant = 69.4X
Discordant = 30.2X
Tied = 0.4X
(547300 pairs)
Somers1
Gamma
Tau-a
c
0.391
0.393
0.145
0.696
E-21
-------
Response Variable: Sore Throat, Hoarseness, Dry Throat
Member of Observations: 1706
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
U30.683
1436.125
1428.683
Intercept
and
Covariates
1339.740
1530.207
1269.740
Chi-Square for Covariates
158.943 with 34 OF (p=0.0001)
152.983 with 34 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
New Furniture
Window Nearby
Glare
Noise
Dust
Role Conflict
Workload
External Stress
Allergy to Molds
Chem. Sensitivity
Sole Provider
Nonsmoker
Under 30
AHU20
ETB
NEG3
NEG4
SE22
G1LG1
NEG1NEG2
CM FL2
FCFL6
ET1FL67
ET1F1011
WT1FL89
WT1F1011
WT2F1011
ET3FL811
WT3FL47
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-4.6943
-0.0799
0.3490
-0.5158
0.3322
0.2101
0.6459
0.2107
0.1184
0.1177
0.3843
0.4692
-0.0210
0.4330
-0.2375
-0.7722
0.1333
-0.7252
-1.7788
-1.1637
-0.3059
-38.5795 #
-1.9052
-0.7305
-0.3049
-0.5607
-0.8261
-37.2038 #
-0.5424
-1.4415
-0.5169
0.6795
0.5456
0.1781
0.1082
Standard
Error
0.6550
0.1605
0.1503
0.1563
0.1500
0.1600
0.1528
0.1141
0.0914
0.0637
0.1531
0.1584
0.1669
0.2621
0.2985
1.0812
0.8546
0.8363
1.1034
1.0867
1.1600
m
1.0343
0.7646
0.6640
0.5563
1.0747
B
0.7853
0.6631
0.6283
0.2715
0.3062
0.2457
0.2424
Wald
Chi-Sq
51.3592
0.2477
5.3874
10.8981
4.9060
1.7249
17.8693
3.4100
1.6770
3.4062
6.3030
8.7715
0.0158
2.7293
0.6328
0.5101
0.0243
0.7519
2.5989
1.1467
0.0695
*
3.3929
0.9128
0.2109
1.0158
0.5908
m
0.4770
4.7263
0.6768
6.2640
3.1745
0.5252
0.1994
Pr >
Chi-Sq
0.0001
0.6187
0.0203
0.0010
0.0268
0.1891
0.0001
0.0648
0.1953
0.0650
0.0121
0.0031
0.8999
0.0985
0.4263
0.4751
0.8760
0.3859
0.1069
0.2842
0.7920
m
0.0655
0.3394
0.6461
0.3135
0.4421
m
0.4898
0.0297
0.4107
0.0123
0.0748
0.4686
0.6552
Odds
Ratios
0.00915
0.92321
1.41765
0.59702
1.39403
1.23380
1.90770
1.23454
1.12569
1.12491
1.46859
1.59871
0.97922
1.54188
0.78860
0.46200
1.14259
0.48423
0.16884
0.31233
0.73646
m
0.14879
0.48167
0.73720
0.57081
0.43775
m
0.58135
0.23657
0.59637
1.97289
1.72564
1.19494
1.11427
99
Percent
Confidence Intervals
Low Limit
0.00169
0.61058
0.96255
0.39915
0.94724
0.81705
1.28697
0.92015
0.88954
0.95467
0.98997
1.06307
0.63703
0.78492
0.36552
0.02851
0.12642
0.05616
0.00984
0.01901
0.03710
m
0.01036
0.06720
0.13327
0.13619
0.02747
m
0.07689
0.04287
0.11820
0.98031
0.78414
0.63456
0.59677
Upper Limit
0.0494
1.3959
2.0879
0.8930
2.0516
1.8631
2.8278
1.6564
1.4245
1.3255
2.1786
2.4043
1.5052
3.0288
1.7014
7.4855
10.3269
4.1750
2.8967
5.1327
14.6181
•
2.1365
3.4525
4.0778
2.3925
6.9750
,
4.3953
1.3056
3.0090
3.9705
3.7976
2.2502
2.0805
Concordant 73.2X
Discordant = 26.3X
Tied = 0.5X
(366408 pairs)
NOTE:
Somers1 D = 0.469
Gamma = 0.471
Tau-a = 0.118
c = 0.735
Parameter estimates marked with '#' are regarded to be infinite.
E-22
-------
Response Variable: Sore Throat, Hoarseness, Dry Throat
Number of Observations: 1621
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
(Including Comfort and Odor)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1813.464
1818.855
1811.464
Intercept
and
Covariates
1657.160
1818.883
1597.160
Chi-Square for Covariates
214.305 with 29 DF (p=0.0001)
214.702 with 29 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Fabric Partitions
Use Fan
Use Heater
New Furniture
Window Nearby
Glare
External Stress
Wear Contact Lens
Wear Glasses
Allergy to Molds
Chem. Sensitivity
Sole Provider
Nonsmoker
Hot Stuffy Air
Dry Air
Cold Drafty Air
Paint Odor
Cosmetic, Body Odor
New Carpet Odor
AHU21
ETB
NE1
NEG4
SE22
SE32
NEG1NEG2
Tackd Carpet, Room
Private Office
Cubicle or Shared
Parameter
Estimate
-3.4840
-0.0026
0.2095
-0.3440
0.2275
-0.1298
0.1217
0.0652
-0.0659
0.1685
0.3103
0.2271
0.3000
0.0725
0.0259
0.1878
0.1738
0.2835
0.0367
0.0976
0.2278
-1.6206
0.0929
-1.1297
0.1040
1.2651
-0.9088
0.3990
-0.1094
-0.1212
Standard
Error
0.5082
0.1363
0.1348
0.1604
0.1281
0.1341
0.1301
0.0528
0.1486
0.1277
0.1388
0.1359
0.1883
0.1696
0.0223
0.0381
0.0469
0.0592
0.0410
0.0869
0.8169
1.0932
0.3789
0.8311
0.7178
0.3792
1.0983
0.2271
0.1760
0.1503
Wald
Chi -So;
47.0036
0.0004
2.4154
4.5978
3.1553
0.9366
0.8752
1.5257
0.1967
1.7414
4.9993
2.7921
2.5368
0.1827
1.3458
24.2694
13.7488
22.9028
0.7994
1.2616
0.0778
2.1978
0.0601
1.8480
0.0210
11.1295
0.6847
3.0866
0.3866
0.6497
Pr >
Chi-Sq
0.0001
0.9848
0.1201
0.0320
0.0757
0.3332
0.3495
0.2168
0.6574
0.1870
0.0254
0.0947
0.1112
0.6690
0.2460
0.0001
0.0002
0.0001
0.3713
0.2613
0.7803
0.1382
0.8063
0.1740
0.8848
0.0008
0.4080
0.0789
0.5341
0.4202
Odds
Ratios
0.03068
0.99740
1.23306
0.70893
1.25546
0.87827
.12942
.06737
0.93622
.18353
.36383
.25496
.34986
.07519
.02624
.20659
1.18982
1.32777
1.03738
1.10252
1 .25583
0.19778
1.09735
0.32313
1.10960
3.54345
0.40301
1.49033
0.89637
0.88586
99
Percent
Confidence Intervals
Low Limit
0.00829
0.70208
0.87132
0.46898
0.90259
0.62173
0.80780
0.93164
0.63846
0.85176
0.95385
0.88428
0.83105
0.69462
0.96895
1.09380
1.05441
1.13997
0.93341
0.88139
0.15312
0.01184
0.41348
0.03798
0.17464
1.33413
0.02380
0.83027
0.56963
0.60147
Upper Limit
0.1136
1.4170
1.7450
1.0716
1.7463
1.2407
1.5791
1.2229
1.3729
1.6445
1.9500
1.7810
2.1925
1.6643
1.0869
1.3310
1.3426
1.5465
1.1529
1.3791
10.2999
3.3052
2.9123
2.7489
7.0502
9.4114
6.8238
2.6752
1.4105
1.3047
Association of Predicted Probabilities and Observed Responses
Concordant = 73.1X
Discordant = 26.5X
Tied = 0.3X
(488400 pairs)
S oners'
Gamma
Tau-a
c
0.466
0.467
0.173
0.733
E-23
-------
Response Variable: Sore Throat, Hoarseness, Dry Throat
Number of Observations: 1625
Sex * Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
(Including Comfort and Odor)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1348.626
1354.019
1346.626
Intercept
and
Covar fates
1199.520
1361.318
1139.520
Chi-Square for Covariates
207.106 with 29 DF (p=0.0001)
213.024 with 29 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Fabric Partitions
Use Fan
Use Heater
New Furniture
Window Nearby
Glare
External Stress
Wore Contact Lens
Wore Glasses
Allergy to Molds
Chem. Sensitivity
Sole Provider
N on-Smoker
Hot Stuffy Air
Dry Air
Cold Drafty Air
Paint Odor
Cosmetic, Body Odor
New Carpet Odor
AHU21
ETB
NE1
NEG4
SE22
SE32
NEG1NEG2
Tackd Carpet, Room
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-4.8160
-0.2831
-0.1836
-0.4388
0.3593
-0.3816
0.1643
0.1129
-0.5479
0.2482
0.3698
0.3773
-0.0205
0.5015
0.0461
0.2742
-0.0328
0.2297
0.1541
0.0567
0.8031
0.3576
0.8127
-1.1944
-0.8829
-0.5417
37.0158*
0.4632
0.2856
0.2755
Standard Uald
Error
0.6728
0.1699
0.1707
0.2800
0.1599
0.1652
0.1615
0.0666
0.2298
0.1687
0.1623
0.1678
0.1790
0.2714
0.0269
0.0474
0.0714
0.0848
0.0588
0.1085
0.6870
0.8643
0.4109
1.1133
1.1028
0.7949
•
0.2824
0.2619
0.2557
Chi-Sq
51.2436
2.7776
1.1574
2.4553
5.0480
5.3373
1.0341
2.8759
5.6858
2.1642
5.1911
5.0553
0.0131
3.4155
2.9351
33.3970
0.2112
7.3393
6.8654
0.2734
1.3666
0.1712
3.9123
1.1510
0.6409
0.4644
•
2.6911
1.1893
1.1606
Pr >
Chi-Sq
0.0001
0.0956
0.2820
0.1171
0.0247
0.0209
0.3092
0.0899
0.0171
0.1413
0.0227
0.0246
0.9089
0.0646
0.0867
0.0001
0.6459
0.0067
0.0088
0.6011
0.2424
0.6790
0.0479
0.2833
0.4234
0.4956
•
0.1009
0.2755
0.2813
99 Percent
Odds Confidence Intervals
Ratios Lower Limit Upper Limit
0.00810
0.75344
0.83227
0.64481
1.43233
0.68277
.17857
.11952
0.57816
.28172
.44745
.45834
0.97971
.65120
.04718
.31548
0.96773
1.25822
1.16661
1.05834
2.23245
1.42989
2.25399
0.30289
0.41358
0.58176
•
1.58915
1.33056
1.31719
0.00143
0.48638
0.53616
0.31346
0.94876
0.44613
0.77746
0.94303
0.31986
0.82997
0.95286
0.94653
0.61779
0.82068
0.97707
1.16427
0.80515
1.01132
1.00263
0.80028
0.38037
0.15430
0.78210
0.01721
0.02414
0.07507
•
0.76777
0.67770
0.68169
0.0458
1.1671
1.2919
1.3264
2.1624
1.0449
1.7866
1 .3290
1.0451
1.9794
2.1987
2.2469
1 .5536
3.3222
1.1223
1.4863
1.1631
1.5654
1.3574
1.3996
13.1026
13.2505
6.4959
5.3306
7.0845
4.5085
.
3.2893
2.6124
2.5451
Concordant = 76.9X
Discordant * 22.7X
Tied - 0.4X
(327804 pairs)
Somens'
Gamma
Tau-a
c
0.542
0.545
0.135
0.771
NOTE: Parameter estimates marked with '#' are regarded to be infinite.
E-24
-------
Response Variable: Sleepiness, Unusual Fatigue
Number of Observations: 1785
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1713.938
1719.425
1711.938
•
Intercept
and
Covariates
1599.078
1708.821
1559.078
•
Chi-Square for Covariates
152.860 with 19 Of (p=0.0001)
155.174 with 19 DF (p-0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Use Heater
New Furniture
Water Leaks Nearby
Window Nearby
Lights Too Dim
Glare
Noise
Dust
Job Satisfaction
External Stress
College Grad
Allergy to Molds
Chem. Sensitivity
G1LG1
WT1F1011
UGHAHU19
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-1.8151
-0.1882
0.1631
0.1345
0.3114
-0.1020
0.8461
0.2649
0.1779
0.4056
•0.6508
0.0829
•0.3464
0.0738
0.5995
0.6570
0.0547
-0.3295
-0.1784
-0.0387
Standard
Error
0.5251
0.1381
0.1499
0.1306
0.1809
0.1378
0.2037
0.1352
0.1381
0.1446
0.1231
0.0529
0.1509
0.1431
0.1342
0.8295
0.4682
0.7791
0.1983
0.1507
Wald
Chi -Se-
ll. 9487
1.8572
1.1844
1.0618
2.9615
0.5484
17.2582
3.8385
1.6602
7.8656
27.9339
2.4577
5.2689
0.2662
19.9680
0.6273
0.0136
0.1789
0.8091
0.0658
Pr >
Chi-Sq
0.0005
0.1729
0.2765
0.3028
0.0853
0.4590
0.0001
0.0501
0.1976
0.0050
0.0001
0.1169
0.0217
0.6059
0.0001
0.4283
0.9070
0.6723
0.3684
0.7976
Odds
Ratios
0.16282
0.82845
1.17715
1.14396
1.36534
0.90303
2.33054
1.30330
1.19471
1.50020
0.52163
1.08643
0.70723
1.07659
1.82121
1.92900
1.05622
0.71928
0.83661
0.96204
99
Percent
Confidence Intervals
Low Limit
0.04210
0.58045
0.80008
0.81716
0.85676
0.63320
1.37901
0.92001
0.83707
1.03366
0.37988
0.94803
0.47945
0.74466
1.28892
0.22769
0.31620
0.09667
0.50197
0.65253
Upper Limit
0.6297
1.1824
1.7319
1.6015
2.1758
1.2878
3.9386
1.8463
1.7051
2.1773
0.7163
1.2450
1.0432
1.5565
2.5733
16.3429
3.5282
5.3520
1 .3943
1.4184
Concordant * 70.4%
Discordant = 29.2X
Tied *> 0.5%
(481274 pairs)
Somers1
Gamma
Tau-a
c
0.412
0.414
0.125
0.706
E-25
-------
Response Variable: Sleepiness, Unusual Fatigue
Number of Observations: 1785
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1117.231
1122.710
1115.231
Intercept
and
Coveriates
1042.543
1152.118
1002.543
Chi-Square for Covariates
112.688 with 19 DF (p=0.0001)
111.772 with 19 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Use Heater
New Furniture
Water Leaks Nearby
Window Nearby
Lights Too Dim
Glare
Noise
Dust
Job Satisfaction
External Stress
College Grad
Allergy to Molds
Chem. Sensitivity
G1LG1
WT1F1011
UGHAHU19
Private Office
Cubicle or Shared
Parameter
Estimate
-3.2813
0.2004
-0.0653
0.2997
0.5755
-0.5944
0.3499
0.4248
0.3164
0.4464
-0.4398
0.1744
0.1927
0.3622
0.2868
-38.0834 *
-36.5466 *
-0.8225
-0.9382
-0.6947
Standard
Error
0.7360
0.1802
0.2766
0.1737
0.2200
0.1780
0.3115
0.1759
0.1894
0.1781
0.1608
0.0726
0.3524
0.1796
0.1852
•
•
0.8247
0.2378
0.2278
Wald
Chi-Sq
19.8755
1.2368
0.0558
2.9762
6.8417
11.1504
1.2619
5.8310
2.7896
6.2811
7.4818
5.7601
0.2989
4.0646
2.3977
0.9947
15.5664
9.2981
Association of Predicted Probabilities and Observed Responses
Concordant = 72.2%
Discordant = 27.1X
Tied = 0.7X
(270569 pairs)
NOTE:
S oners'
Gamma
Tau-a
c
D =
0.452
0.455
0.078
0.726
Pr >
Chi-Sq
0.0001
0.2661
0.8133
0.0845
0.0089
0.0008
0.2613
0.0157
0.0949
0.0122
0.0062
0.0164
0.5846
0.0438
0.1215
0.3186
0.0001
0.0023
Odds
Ratios
0.03758
1.22189
0.93679
1.34945
1.77802
0.55189
1.41893
1.52928
1.37218
1.56268
0.64417
1.19053
1.21252
1.43649
1.33216
0.43933
0.39133
0.49922
99 Percent
Confidence Intervals
Low Limit Upper Limit
0.00564
0.76813
0.45941
0.86265
1.00882
0.34891
0.63602
0.97208
0.84240
0.98769
0.42570
0.98746
0.48915
0.90443
0.82673
0.05250
0.21208
0.27762
0.25023
1.94370
1.91023
2.11098
3.13372
0.87295
3.16554
2.40589
2.23512
2.47239
0.97475
1.43536
3.00562
2.28154
2.14658
3.67637
0.72207
0.89773
Parameter estimates marked with '#' are regarded to be infinite.
E-26
-------
Response Variable: Sleepiness, Unusual Fatigue
Number of Observations: 1582
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
(Including Comfort and Odor)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1497.243
1502.609
1495.243
Intercept
and
Covariates
1313.465
1415.428
1275.465
Chi-Square for Covariates
219.778 with 18 DF (p=0.0001)
221.116 with 18 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Window Nearby
Glare
Medium Pay
Job Satisfaction
Underut iIi zat i on
External Stress
College Grad
Allergy to Molds
chem. Sensitivity
Hot Stuffy Air
Cold Drafty Air
Paint Odor
Cosmetic, Body Odor
New Carpet Odor
FCFL2
Private Office
Cubicle or Shared
Parameter
Estimate
-3.3999
-0.1918
-0.0301
0.1657
-0.2368
-0.4551
0.1759
0.0612
-0.3057
-0.1251
0.5259
0.1625
0.0675
0.1675
0.1498
0.0385
1.2550
-0.2007
-0.0366
Standard
Error
0.6631
0.1525
0.1507
0.1500
0.1625
0.1534
0.0757
0.0587
0.1675
0.1613
0.1509
0.0226
0.0511
0.0612
0.0434
0.0940
0.5678
0.2190
0.1700
Wald
Chi-Sq
26.2883
1.5804
0.0400
1.2203
2.1239
8.8051
5.3982
1.0850
3.3289
0.6013
12.1498
51.7450
1.7457
7.4823
11.9130
0.1678
4.8846
0.8393
0.0463
Pr >
Chi-Sq
0.0001
0.2087
0.8415
0.2693
0.1450
0.0030
0.0202
0.2976
0.0681
0.4381
0.0005
0.0001
0.1864
0.0062
0.0006
0.6821
0.0271
0.3596
0.8297
Odds
Ratios
0.03338
0.82547
0.97035
1.18022
0.78915
0.63438
1.19232
1.06311
0.73661
0.88241
1 .69198
1.17645
1.06983
1.18235
1.16160
1.03925
3.50784
0.81816
0.96406
99
Percent
Confidence Intervals
Low Limit
0.00605
0.55731
0.65816
0.80196
0.51923
0.42730
0.98108
0.91392
0.47846
0.58239
1.14704
1.10991
0.93788
1.00990
1.03873
0.81575
0.81249
0.46541
0.62218
Upper Limit
0.1842
1.2227
1.4306
1.7369
1.1994
0.9418
.4490
.2367
.1340
.3370
2.4958
.2470
.2203
1.3842
1.2990
1 .3240
15.1447
1.4383
1.4938
Association of Predicted Probabilities and Observed Responses
Concordant = 75.1X
Discordant - 24.5X
Tied = 0.4X
(370656 pairs)
Somers1 D
Gamma
Tau-a
c
= 0.505
= 0.507
= 0.150
* 0.753
E-27
-------
Response Variable: Sleepiness, Unusual Fatigue
Number of Observations: 1625
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
(Including Comfort and Odor)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1011.609
1017.002
1009.609
Intercept
and
Coveriates
883.791
986.263
845.791
Chi-Square for Covariates
163.818 with 18 DF (p=0.0001)
172.386 with 18 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Window Nearby
Glare
Medium Pay
Job Satisfaction
Underutilization
External stress
College Grad
Allergy to Molds
Chem. Sensitivity
Hot Stuffy Air
Cold Drafty Air
Paint Odor
Cosmetic, Body Odor
New Carpet Odor
FCFL2
Private Office
Cubicle or Shared
Parameter
Estimate
-5.1505
0.1064
-0.2565
0.2810
0.2255
-0.2454
0.1432
0.1576
0.2913
0.3488
0.2276
0.1908
0.0851
0.1018
0.0491
0.2286
-37.3697 #
-0.8080
-0.5674
Standard
Error
0.9629
0.1975
0.1949
0.1920
0.2352
0.2085
0.1073
0.0780
0.4138
0.1941
0.2018
0.0288
0.0822
0.0990
0.0705
0.1101
.
0.2712
0.2525
Uald
Chi-Sq
28.6120
0.2899
1.7324
2.1419
0.9197
1.3857
1.7818
4.0869
0.4956
3.2284
1.2721
43.7537
1.0728
1.0573
0.4851
4.3054
8.8768
5.0478
Pr >
Chi-Sq
0.0001
0.5903
0.1881
0.1433
0.3376
0.2391
0.1819
0.0432
0.4814
0.0724
0.2594
0001
3003
3038
4861
0.0380
0.0029
0.0247
Odds
Ratios
0.00580
1.11227
0.77375
1.32445
1.25295
0.78239
1.15396
1.17070
1.33817
1.41737
1.25558
1.21022
1.08883
1.10716
1.05033
1.25684
0.44575
0.56700
99 Percent
Confidence Intervals
Low Limit Upper Limit
0.00049
0.66874
0.46834
0.80768
0.68360
0.45726
0.87529
0.95760
0.46087
0.85967
0.74659
1.12368
0.88104
0.85794
0.87590
0.94647
0.22166
0.29587
0.06925
1.84995
1.27834
2.17188
2.29647
1.33870
1.52136
1.43122
3.88549
2.33684
2.11158
1.30342
1.34561
1.42878
1.25949
1.66899
0.89638
1.08659
Association of Predicted Probabilities and Observed Responses
Concordant = 79.4X
Discordant = 20.IX
Tied = 0.5X
(223896 pairs)
NOTE:
Sotners1 D
Gamma
Tau-a
c
= 0.593
* 0.596
= 0.101
= 0.796
Parameter estimates marked with '#' are regarded to be infinite.
E-28
-------
Response Variable: Chills and Fever
Number of Observations: 1723
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2164.657
2170.109
2162.657
Intercept
and
Covariates
2020.459
2156.754
1970.459
Chi-Square for Covariates
192.198 with 24 DF (p=0.0001)
182.412 with 24 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at Work
>1 Hour at Copier
Use Heater
New Carpet/Drapes
New Furniture
Walls Rearranged
Lights Too Dim
Glare
Noise
Dust
High Pay
Role Conflict
College Grad
Allergy to Molds
Chem. Sensitivity
Sole Provider
AHU21
AUU15SU2
NEG1NEG2
CM FL2
UT2FL67
Tackd Carpet, Room
Private Office
Cubicle or Shared
Parameter
Estimate
-1.46230
0.00656
0.11470
0.28350
-0.47680
0.26120
-0.26000
0.12040
0.18970
0.22440
0.40290
-0.44810
0.30810
-0.43120
0.22090
0.20130
0.23050
0.42750
0.17630
-1.31840
-1.68740
-0.80220
0.22010
-0.41710
-0.33560
Standard
Error
0.3130
0.0365
0.1613
0.1277
0.1639
0.1171
0.1477
0.2012
0.1143
0.1172
0.1200
0.1507
0.0797
0.1366
0.1250
0.1185
0.1653
0.7025
0.4269
1.0719
0.6190
0.5037
0.1991
0.1738
0.1313
Wald
Chi-Sq
21.8229
0.0322
0.5054
4.9319
8.4633
4.9770
3.0989
0.3582
2.7533
3.6652
11.2677
8.8372
14.9624
9.9715
3.1206
2.8878
1.9453
0.3703
0.1705
1.5128
7.4300
2.5365
1.2214
5.7587
6.5371
Pr >
Chi-Sq
0.0001
0.8575
0.4771
0.0264
0.0036
0.0257
0.0783
0.5495
0.0971
0.0556
0.0008
0.0030
0.0001
0.0016
0.0773
0.0893
0.1631
0.5428
0.6796
0.2187
0.0064
0.1112
0.2691
0.0164
0.0106
Odds
Ratios
0.23170
1.00658
1.12154
1.32777
0.62077
1.29849
0.77105
1.12795
1.20889
1.25157
1.49616
0.63884
1.36084
0.64973
1.24720
1.22299
1.25923
1.53342
1.19280
0.26756
0.18500
0.44834
1.24620
0.65896
0.71491
99
Percent
Confidence Intervals
Low Limit
0.10346
0.91625
0.74022
0.95556
0.40697
0.96036
0.52704
0.67174
0.90056
0.92542
1.09832
0.43331
1.10826
0.45700
0.90384
0.90126
0.82258
0.25104
0.39717
0.01691
0.03756
0.12249
0.74619
0.42113
0.50975
Upper Limit
0.51892
1.10582
1 .69929
1.84496
0.94687
1.75567
1.12803
1.89400
1.62277
1 .69267
2.03810
0.94186
1.67097
0.92375
1.72099
1.65957
1.92767
9.36650
3.58226
4.23259
0.91132
1.64104
2.08128
1.03108
1.00263
Association of Predicted Probabilities and Observed Responses
Concordant * 70.3X
Discordant « 29.4X
Tied = 0.3X
(647010 pairs)
Sorters' D
Gamma
Tau-a
c
0.409
0.410
0.178
0.704
E-29
-------
Response Variable: Chills and Fever
Number of Observations: 1724
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1534.980
1540.433
1532.980
Intercept
and
Covariates
1484.565
1620.875
1434.565
•
Chi-Square for Covariates
98.415 with 24 DF (p=0.0001)
102.310 with 24 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at Work
>1 Hour at Copier
Use Heater
New Carpet/Drapes
New Furniture
Walls Rearranged
Lights Too Dim
Glare
Noise
Dust
High Pay
Role Conflict
College Grad
Allergy to Molds
Chem. Sensitivity
Sole Provider
AHU21
AUU15SU2
NEG1NEG2
CM FL2
UT2FL67
Tackd Carpet, Room
Private Office
Cubicle or Shared
Parameter
Estimate
-3.2539
0.1316
0.2896
0.6637
0.0496
0.1982
-0.1534
0.1038
0.3134
0.4020
0.3136
0.0109
0.2772
-0.6427
0.0299
0.3430
0.0201
0.2496
1.1247
-0.2533
-1.0692
-0.1724
0.5383
-0.0447
-0.0895
Standard
Error
0.4574
0.0425
0.3137
0.1986
0.1893
0.1464
0.1967
0.2835
0.1402
0.1502
0.1437
0.1803
0.1020
0.2753
0.1498
0.1527
0.1563
0.7199
0.4024
0.7103
0.7419
0.5199
0.2445
0.2219
0.2121
Uald
Chi-Sq
50.6040
9.5988
0.8521
11.1711
0.0686
1.8323
0.6078
0.1341
4.9985
7.1642
4.7603
0.0037
7.3782
5.4494
0.0399
5.0463
0.0166
0.1203
7.8102
0.1272
2.0771
0.1100
4.8467
0.0406
0.1779
Pr >
Chi-Sq
0.0001
0.0019
0.3560
0.0008
0.7934
0.1759
0.4356
0.7142
0.0254
0.0074
0.0291
0.9517
0.0066
0.0196
0.8416
0.0247
0.8974
0.7287
0.0052
0.7214
0.1495
0.7402
0.0277
0.8403
0.6732
Odds
Ratios
0.03862
1.14065
1.33589
1.94196
1.05085
1.21921
0.85779
1.10938
1.36807
1.49481
1.36834
1.01096
1 .31943
0.52587
1.03035
1.40917
1.02030
1.28351
3.07929
0.77623
0.34328
0.84164
1.71309
0.95628
0.91439
99
Percent
Confidence Intervals
Low Limit
0.01189
1.02237
0.59542
1.16428
0.64530
0.83617
0.51680
0.53446
0.95337
1.01520
0.94500
0.63537
1.01455
0.25876
0.70048
0.95089
0.68213
0.20092
1.09212
0.12455
0.05078
0.22054
0.91253
0.53993
0.52947
Upper Limit
0.12548
1.27262
2.99724
3.23909
1.71127
1.77771
1 .42376
2.30273
1.96316
2.20101
.98134
.60858
.71592
.06873
.51556
2.08831
1.52612
8.19941
8.68226
4.83767
2.32085
3.21190
3.21598
1.69370
1.57912
Association of Predicted Probabilities and Observed Responses
Concordant = 68.OX
Discordant = 31.3%
Tied = 0.7X
(405483 pairs)
Somers1
Gamma
Tau-a
c
= 0.367
* 0.369
- 0.100
= 0.683
E-30
-------
Response Variable: Chills and Fever
Number of Observations: 1640
Sex * Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
(Including Comfort and Odor)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2039.951
2045.354
2037.951
Intercept
and
Covariates
1791.691
1899.740
1751.691
Chi-Square for Covariates
286.260 with 19 DF (p=0.0001)
270.662 with 19 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at Work
Use Heater
Glare
Job Satisfaction
College Grad
Live Alone
Sole Provider
Dry Air
Cold Drafty Air
Paint Odor
Cosmetic, Body Odor
Musty/Diesel Odor
AHU13
AHU21
NEG1NEG2
CM FL2
WT1FL45
UT2FL67
UT3FL811
Association of Predicted Probabilities and Observed Responses
99 Percent
Parameter
Estimate
-1.0802
-0.0275
0.1324
0.2125
-0.4087
-0.4899
0.1607
0.1374
0 0585
0.3865
0.1959
0.1289
0.1576
0.9579
0.8585
-1.5197
-2.3512
-1.3517
-1.0561
37.5946*
Standard
Error
0.3671
0.0388
0.1400
0.1207
0.1145
0.1230
0.1399
0.1830
0.0300
0.0439
0.0530
0.0395
0.0693
0.5314
0.6458
1.0848
1.0231
0.6451
0.6100
f
Wald
Chi-Sq
8.6610
0.5031
0.8933
3.0975
12.7542
15.8726
1.3191
0.5634
3.7896
77.4982
13.6550
10.6261
5.1686
3.2490
1.7675
1.9625
5.2816
4.3910
2.9977
m
Pr >
Chi-Sq
0.0033
0.4781
0.3446
0.0784
0.0004
0.0001
0.2508
0.4529
0.0516
0.0001
0.0002
0.0011
0.0230
0.0715
0.1837
0.1612
0.0216
0.0361
0.0834
m
Odds
Ratios
0.33953
0.97287
1.14156
1.23677
0.66451
0.61269
1.17433
1.14729
1.06024
1.47182
1.21641
1.13758
1.17070
2.60622
2.35962
0.21878
0.09525
0.25880
0.34781
m
Confidence Intervals
Lou Limit
0.13188
0.88034
0.79593
0.90627
0.49478
0.44631
0.81899
0.71605
0.98140
1.31444
1.06117
1.02752
0.97930
0.66300
0.44705
0.01338
0.00683
0.04912
0.07226
•
Upper Limit
0.8741
1.0751
1.6373
1.6878
0.8925
0.8411
.6838
.8382
.1454
.6480
.3943
.2594
.3995
10.2450
12.4545
3.5778
1.3288
1.3635
1 .6741
m
Concordant * 75.3X
Discordant = 24.4X
Tied = 0.3X
(578151 pairs)
Sorters'
Gamma
Tau-a
c
= 0.509
= 0.510
= 0.219
= 0.754
NOTE: Parameter estimates marked with '#' are regarded to be infinite.
E-31
-------
Response Variable: Chills and Fever
Number of Observations: 1685
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
(Including Comfort and Odor)
Intercept
Only
1461.605
1467.035
1459.605
Intercept
and
Covariates
1328.838
1437.429
1288.838
Chi-Square for Covariates
170.767 with 19 OF (p=0.0001)
182.931 with 19 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at Work
Use Heater
Glare
Job Satisfaction
College Grad
Live Alone
Sole Provider
Dry Air
Cold Drafty Air
Paint Odor
Cosmetic, Body Odor
Musty/Diesel Odor
AHU13
AHU21
MEG1NEG2
CM FL2
WT1FL45
WT2FL67
WT3FL811
Parameter
Estimate
-2.90430
0.12420
0.39880
0.37680
-0.21370
-0.72650
0.35470
0.00482
0.13100
0.46500
0.21490
0.00814
0.19740
-0.53940
0.56140
0.19730
-0.58530
-0.48480
-0.13530
0.40400
Standard
Error
0.4878
0.0449
0.2137
0.1480
0.1369
0.2547
0.2004
0.1713
0.0392
0.0591
0.0748
0.0591
0.0853
0.8186
0.7473
0.6967
0.7435
0.7607
0.5491
0.7356
Weld
Chi-Sq
35.4497
7.6379
3.4834
6.4802
2.4386
8.1363
3.1317
0.0008
11.1353
61.9116
8.2526
0.0190
5.3512
0.4342
0.5644
0.0802
0.6196
0.4062
0.0607
0.3017
Pr >
Chi-Sq
0.0001
0.0057
0.0620
0.0109
0.1184
0.0043
0.0768
0.9776
0.0008
0.0001
0.0041
0.8904
0.0207
0.5099
0.4525
0.7770
0.4312
0.5239
0.8054
0.5828
Odds
Ratios
0.05479
1.13224
1.49004
1 .45761
0.80759
0.48360
1 .42575
1.00483
1.13997
1.59201
1.23974
1.00817
1.21823
0.58310
1.75313
1.21811
0.55694
0.61582
0.87345
1.49780
99
Percent
Confidence Intervals
Low Limit
0.01559
1.00857
0.85925
0.99556
0.56759
0.25092
0.85084
0.64633
1.03048
1.36719
1.02246
0.86580
0.97791
0.07078
0.25573
0.20242
0.08204
0.08678
0.21229
0.22517
Upper Limit
0.1925
1.2711
2.5839
2.1341
1.1491
0.9320
2.3891
1.5622
1.2611
1.8538
1.5032
1.1740
1.5176
4.8034
12.0185
7.3302
3.7809
4.3700
3.5937
9.9633
Association of Predicted Probabilities and Observed Responses
Concordant = 74.5X
Discordant = 25.OX
Tied = 0.5X
(373986 pairs)
Sorters'
Gamma
Tau-a
c
0.495
0.497
0.130
0.748
E-32
-------
Response Variable: Aching Muscles or Joints, Pain in Back, Shoulder, Neck, Wrists
Number of Observations: 1717
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1679.694
1685.142
1677.694
Intercept
and
Cover i at es
1519.637
1683.087
1459.637
Chi-Square for Covariates
218.057 with 29 DF (p=0.0001)
215.403 with 29 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
U5 0.00603
Laser Printer Nearby
1 Hour at Copier
Walls Painted
Water Leaks Nearby
Lights Too Dim
Glare
Noise
Dust
High Pay
Role Conflict
Workload
Wear Contact Lens
Allergy to Molds
Chem. Sensitivity
Live Alone
Sole Provider
Discomfort Index
Under 30
Over 40
AHU14
AHU21
AHU9
SE21
G1LG1
CM FL10
FCFL6
Private Office
Cubicle or Shared
0.0471
Parameter
Estimate
-3.52530
0.0164 0
-0.24860
-0.20370
0.15420
0.28730
0.41650
0.54230
0.28860
0.37840
-0.57990
0.26590
0.18510
0.13840
0.42420
0.28870
0.07260
0.37830
0.29130
-0.73910
0.31050
2.04530
0.90240
0.45860
0.15040
1.25290
-0.09370
-0.19780
-0.26600
-0.28900
Standard Wald
Error
0.3888
.8982 1
0.1420
0.1497
0.1584
0.1855
0.2150
0.1418
0.1437
0.1509
0.1700
0.0978
0.0797
0.1557
0.1441
0.1403
0.1596
0.1942
0.0549
0.1957
0.1506
0.6280
0.7373
0.4318
0.3801
0.8540
0.4951
0.5769
0.2105
0.1578
Chi-Sq
82.2006
.00605 0
3.0653
1.8511
0.9478
2.3997
3.7532
14.6218
4.0340
6.2875
11.6409
7.3886
5.3995
0.7903
8.6690
4.2302
0.2066
3.7947
28.1417
14.2687
4.2521
10.6062
1.4980
1.1279
0.1566
2.1527
0.0358
0.1176
1.5963
3.3530
Pr >
Chi-Sq
0.0001
.89110
0.0800
0.1737
0.3303
0.1214
0.0527
0.0001
0.0446
0.0122
0.0006
0.0066
0.0201
0.3740
0.0032
0.0397
0.6494
0.0514
0.0001
0.0002
0.0392
0.0011
0.2210
0.2882
0.6923
0.1423
0.8500
0.7317
0.2064
0.0671
Odds
Ratios
0.02944
1.1358
0.77989
0.81571
1.16672
1.33282
1.51664
1.71996
1.33456
1.45995
0.55995
1.30460
1.20334
1.14843
1.52837
1.33469
1.07530
1.45980
1.33817
0.47754
1.36411
7.73148
2.46551
1.58186
1.16230
3.50048
0.91056
0.82053
0.76644
0.74901
99
Percent
Confidence Intervals
Lou Limit
0.01081
0.54097
0.55470
0.77582
0.82651
0.87167
1.19366
0.92167
0.98973
0.36138
1.01407
0.98000
0.76899
1.05443
0.92987
0.71282
0.88518
1.16169
0.28845
0.92548
1.53353
0.36903
0.52011
0.43660
0.38790
0.25434
0.18565
0.44564
0.49883
Upper Limit
0.0802
1.1243
1.1995
1.7546
2.1493
2.6388
2.4783
1 .9324
2.1536
0.8676
1.6784
1.4776
1.7151
2.2153
1.9158
1 .6221
2.4074
1.5414
0.7906
2.0106
38.9792
16.4724
4.8111
3.0942
31.5888
3.2598
3.6266
1.3182
1.1247
Association of Predicted Probabilities and Observed Responses
Concordant = 74.6X
Discordant - 25.0%
Tied = 0.4X
(456652 pairs)
Somers1 D
Gamma
Tau-a
c
0.497
0.499
0.154
0.748
E-33
-------
Response Variable: Aching Muscles or Joints. Pain In Back, Shoulder, Neck, Wrists
Number of Observations: 1754
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
135.021 with 29 DF (p=0.0001)
142.459 with 29 DF (p=0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1175.428
1180.897
1173.428
.
Intercept
and
Covariates
1098.407
1262.496
1038.407
.
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Shelf Index
Laser Printer Nearby
1 Hour at Copier
Walls Painted
Water Leaks Nearby
Lights Too Dim
Glare
Noise
Dust
High Pay
Role Conflict
Workload
Wear Contact Lens
Allergy to Molds
Chem. Sensitivity
Live Alone
Sole Provider
Discomfort Index
Under 30
Over 40
AHU14
AHU21
AHU9
SE21
G1LG1
CM FL10
FCFL6
Private Office
Cubicle or Shared
Parameter
Estimate
-4.9858
0.0335
0.2127
-0.2233
0.1840
0.0536
0.7803
0.6654
0.3917
0.2320
0.2744
0.2448
0.1332
-0.4884
0.6616
0.4533
0.5219
-0.0613
0.1365
0.2784
0.3488
0.3802
0.7179
0.6626
0.5794
1.8023
0.8812
0.7518
-0.0365
0.0339
Standard
Error
0.6073
0.0604
0.1812
0.1872
0.2010
0.2308
0.2949
0.1750
0.1868
0.1744
0.2586
0.1282
0.1066
0.2536
0.1740
0.1777
0.2256
0.2020
0.0738
0.3855
0.1886
1.0619
0.7369
0.5561
0.4733
0.9661
0.3991
0.5604
0.2833
0.2657
Wald
Chi-Sq
67.3910
0.3075
1.3776
1.4233
0.8373
0.0539
7.0021
14.4609
4.3977
1.7698
1.1254
3.6465
1.5623
3.7103
14.4605
6.5051
5.3503
0.0921
3.4231
0.5217
3.4213
0.1282
0.9490
1.4195
1.4986
3.4800
4.8755
1.8002
0.0166
0.0163
Pr >
Chi-Sq
0.0001
0.5792
0.2405
0.2329
0.3602
0.8163
0.0081
0.0001
0.0360
0.1834
0.2888
0.0562
0.2113
0.0541
0.0001
0.0108
0.0207
0.7615
0.0643
0.4701
0.0644
0.7203
0.3300
0.2335
0.2209
0.0621
0.0272
0.1797
0.8975
0.8983
Odds
Ratios
0.00683
1.03407
1.23701
0.79987
1.20202
1.05506
2.18213
1.94527
1.47949
1.26112
1.31574
1.27737
1.14248
0.61361
1.93789
1.57350
1.68523
0.94054
1.14625
1.32101
1.41737
1.46258
2.05012
1.93983
1.78497
6.06358
2.41379
2.12081
0.96416
1.03448
99
Percent
Confidence Intervals
Low Limit
0.00143
0.88507
0.77564
0.49385
0.71621
0.58220
1.02085
1.23937
0.91439
0.80473
0.67587
0.91811
0.86814
0.31928
1.23785
0.99555
0.94248
0.55897
0.94780
0.48936
0.87194
0.09487
0.30717
0.46305
0.52739
0.50340
0.86340
0.50068
0.46474
0.52176
Upper Limit
0.0327
1.2082
1.9728
1.2955
2.0173
1.9120
4.6644
3.0532
2.3938
1.9764
2.5614
1.7772
1.5035
1.1792
3.0338
2.4869
3.0133
1.5826
1.3863
3.5660
2.3040
22.5482
13.6830
8.1264
6.0413
73.0375
6.7482
8.9835
2.0003
2.0510
Association of Predicted Probabilities and Observed Responses
Concordant = 73.8X
Discordant = 25.5X
Tied = 0.7X
(287493 pairs)
Somens'
Gamma
Tau-a
c
= 0.484
= 0.487
= 0.090
= 0.742
E-34
-------
Response Variable: Aching Muscles or Joints, Pain in Back, Shoulder, Neck, Wrists (Including Comfort and
Odor)
Number of Observations: 1661
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
219.228 with 27 DF (p=0.0001)
219.919 with 27 DF (p>0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1618.421
1623.836
1616.421
Intercept
and
Covariates
1453.193
1604.818
1397.193
•
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Hours at VOT
1 Hour at Copier
>1 Hour at Copier
Water Leaks Nearby
Glare
High Pay
Workload
Wear Contact Lens
Allergy to Molds
Chan. Sensitivity
Live Alone
Sole Provider
Discomfort Index
Cold Drafty Air
Paint Odor
Cosmetic, Body Odor
Photocopies Odor
Tobacco Smoke Odor
Under 30
AHU21
AHU9
G1LG1
CM FL10
FCFL6
Private Office
Cubicle or Shared
Parameter
Estimate
-3.2214
-0.2769
0.0254
-0.3458
-0.3317
0.1814
0.5709
-0.4695
0.2511
0.0805
0.4517
0.1917
0.0894
0.2144
0.2502
0.1930
0.1166
0.1463
0.1014
0.1110
-0.9595
1.2951
0.4145
1.1053
0.0445
-0.3975
-0.3117
-0.2160
Standard
Error
0.3846
0.1473
0.0313
0.1969
0.2610
0.1916
0.1445
0.1751
0.0763
0.1574
0.1487
0.1467
0.1625
0.2039
0.0558
0.0477
0.0557
0.0427
0.1043
0.1122
0.1928
0.6767
0.4391
1.0917
0.5295
0.6418
0.2174
0.1631
Wald
Chi-Sq
70.1621
3.5318
0.6613
3.0837
1.6154
0.8968
15.6196
7.1865
10.8197
0.2617
9.2320
1.7084
0.3027
1.1062
20.1087
16.3866
4.3760
11.7652
0.9452
0.9792
24.7573
3.6627
0.8911
1.0251
0.0071
0.3837
2.0553
1.7550
Pr > Odds
Chi-Sq Ratios
0.0001 0.03990
0.0602 0.75813
0.4161 1.02573
0.0791 0.70765
0.2037 0.71770
0.3436
0.0001
0.0073
0.0010
0.6090
0.0024
0.1912
0.5822
0.2929
0.0001
0.0001
0.0364
0.0006
0.3309
.19889
.76986
.62531
.28544
.08383
.57098
.21131
.09352
.23912
.28428
.21288
.12367
.15754
.10672
0.3224 1.11739
0.0001 0.38308
0.0556 3.65136
0.3452 1.51361
0.3113 3.02013
0.9331 1.04550
0.5357 0.67200
0.1517 0.73220
0.1852 0.80574
99
Percent
Confidence Intervals
Low Limit
0.01481
0.51874
0.94627
0.42613
0.36640
0.73186
1.21978
0.39830
1 .05607
0.72255
1.07106
0.83011
0.71950
0.73283
1.11233
1.07264
0.97347
1.03697
0.84597
0.83692
0.23313
0.63885
0.48840
0.18142
0.26727
0.12863
0.41823
0.52933
Upper Limit
0.1075
1.1080
1.1119
1.1752
1 .4058
1.9640
2.5680
0.9817
1.5646
1.6257
2.3042
1.7676
1.6620
2.0952
1.4828
1.3715
1.2970
1.2921
1.4478
1.4919
0.6295
20.8693
4.6909
50.2756
4.0898
3.5106
1.2819
1.2265
Association of Predicted Probabilities and Observed Responses
Concordant = 75.5X
Discordant = 24.2X
Tied = 0.4X
(425020 pairs)
Somers'
Gamma
Tau-a
c
= 0.513
* 0.515
= 0.158
= 0.756
E-35
-------
Response Variable: Aching Muscles or Joints, Pain in Back, Shoulder, Meek, Wrists (Including Comfort and
Odor)
Number of Observations: 1697
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
127.355 with 27 DF (p=0.0001)
136.150 with 27 DF (p=0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1102.252
1107.689
1100.252
.
Intercept
and
Covariates
1028.898
1181.123
972.898
,
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Hours at VDT
1 Hour at Copier
>1 Hour at Copier
Water Leaks Nearby
Glare
High Pay
Workload
Wear Contact Lens
Allergy to Molds
Chan. Sensitivity
Live Alone
Sole Provider
Discomfort Index
Cold Drafty Air
Paint Odor
Cosmetic, Body Odor
Photocopies Odor
Tobacco Smoke Odor
Under 30
AHU21
AHU9
G1LG1
CM FL10
FCFL6
Private Office
Cubicle or Shared
Parameter
Estimate
•5.0202
0.1976
0.0633
0.0614
0.9785
0.0403
0.6485
0.4899
0.1884
-0.6453
0.7647
0.4861
0.5575
-0.1201
0.1673
0.1372
0.0789
0.1416
0.1781
-0.2533
0.0414
1.1605
0.6779
1.6436
0.9761
0.4808
0.0583
0.0813
Standard
Error
0.5824
0.1860
0.0409
0.2208
0.3867
0.2410
0.1809
0.2708
0.1026
0.2642
0.1796
0.1846
0.2317
0.2129
0.0749
0.0733
0.0939
0.0683
0.1702
0.1926
0.3895
0.7325
0.5693
1.0250
0.4148
0.6053
0.2943
0.2785
Wald
Chi-Sq
74.3105
1.1287
2.3909
0.0775
6.4037
0.0279
12.8488
3.2725
3.3689
5.9666
18.1232
6.9378
5.7893
0.3181
4.9940
3.4997
0.7063
4.2986
1.0950
1.7305
0.0113
2.5096
1.4178
2.5712
5.5359
0.6309
0.0392
0.0852
Pr >
Chi-Sq
0.0001
0.2880
0.1220
0.7808
0.0114
0.8672
0.0003
0.0704
0.0664
0.0146
0.0001
0.0084
0.0161
0.5727
0.0254
0.0614
0.4007
0.0381
0.2954
0.1883
0.9153
0.1132
0.2338
0.1088
0.0186
0.4270
0.8430
0.7704
Odds
Ratios
0.00660
1.21847
1.06535
1.06332
2.66046
1.04112
1.91267
1.63215
1.20732
0.52451
2.14835
.62596
.74630
0.88683
.18211
.14706
.08210
1.15212
1.19494
0.77623
1.04227
3.19153
1 .96974
5.17376
2.65409
1.61737
1.06003
1.08470
99
Percent
Confidence Intervals
Low Limit
0.00147
0.75462
0.95881
0.60207
0.98252
0.55961
1.20022
0.81247
0.92691
0.26557
1 .35263
1.01063
0.96141
0.51246
0.97468
0.94969
0.84960
0.96624
0.77079
0.47263
0.38215
0.48364
0.45447
0.36906
0.91172
0.34012
0.49668
0.52934
Upper Limit
0.0296
1.9674
1.1837
1.8780
7.2040
1.9370
3.0480
3.2788
1.5725
1 .0359
3.4122
2.6160
3.1720
1.5347
1 .4337
1.3854
1.3782
1.3737
1.8525
1.2749
2.8427
21.0609
8.5371
72.5300
7.7263
7.6910
2.2624
2.2227
Association of Predicted Probabilities and Observed Responses
Concordant = 74.7X
Discordant = 24.6X
Tied = 0.7%
(258232 pairs)
Sorters'
Gamma
Tau-a
c
= 0.502
= 0.505
= 0.090
= 0.751
E-36
-------
Response Variable: Difficulty Concentrating, Remembering; Depression, Tension
Number of Observations: 1713
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1921.386
1926.832
1919.386
Intercept
and
Covariates
1742.712
1889.754
1688.712
Chi-Square for Covariates
230.674 with 26 DF (p=0.0001)
223.299 with 26 DF
Chi-Sq
0.0004
0.6312
0.5633
0.5176
0.2346
0.3805
0.0301
0.0130
0.0001
0.0001
0.0180
0.0004
0.0207
0.3685
0.0005
0.1469
0.7451
0.3384
0.0256
0.0810
0.1299
0.8830
0.1749
0.7231
0.2540
0.8009
0.5761
Odds
Ratios
0.12168
0.94016
1.07605
1.10032
1.23281
0.89235
1.31614
1.38140
1.74037
0.47816
1.25936
1 .30656
0.84772
1.04666
1.53849
1.22814
1.25772
0.34614
2.03318
2.15114
2.16236
1.07069
1.77855
1.21689
1.84154
0.95638
1.08329
99
Percent
Confidence Intervals
Low Limit
0.02644
0.67505
0.77621
0.75211
0.78323
0.63857
0.94988
0.98803
1.23329
0.34051
0.97965
1.07674
0.70530
0.91851
1.11753
0.85255
0.20443
0.01992
0.89622
0.69429
0.58246
0.32410
0.59604
0.29191
0.46379
0.60651
0.74929
Upper Limit
0.55998
.30941
.49173
.60974
.94048
.24698
1.82362
1 .93139
2.45594
0.67147
1.61892
1.58543
1 .01890
1.19268
2.11802
1.76919
7.73807
6.01394
4.61249
6.66496
8.02771
3.53710
5.30715
5.07293
7.31212
1.50808
1.56616
Association of Predicted Probabilities and Observed Responses
Concordant = 73.3X
Discordant = 26.4X
Tied = 0.3X
(547400 pairs)
Somers1
Gamma
Tau-a
c
0.469
0.470
0.175
0.734
E-37
-------
Response Variable: Difficulty Concentrating, Remembering; Depression, Tension
Number of Observations: 1728
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
253.663 with 26 DF (p=0.0001)
247.504 with 26 DF (p=0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1621.880
1627.335
1619.880
Intercept
and
Covariates
1420.217
1567.494
1366.217
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
New Furniture
Walls Painted
Water Leaks Nearby
Window Nearby
Glare
Noise
Dust
Job Satisfaction
Role Conflict
Workload
Role Clarity
External Stress
Chem. Sensitivity
Live Alone
AHU20
ETB
SE21
CM FL10
ET2FL67
ET2F1011
WT1FL45
ET3FL47
UGHAHU19
Private Office
Cubicle or Shared
Parameter
Estimate
-3.15290
-0.23260
0.00707
0.34160
0.47690
-0.06890
0.26970
0.45770
0.20950
-0.95060
0.35080
0.31790
-0.27920
0.18650
0.47980
0.39290
-0.32810
-0.68740
0.35760
-0.73090
0.35890
-0.61450
-0.40160
-0.68920
-0.43640
0.05020
0.06520
Standard
Error
0.6714
0.1535
0.1486
0.1768
0.1902
0.1466
0.1429
0.1544
0.1452
0.1501
0.1139
0.0902
0.0864
0.0620
0.1512
0.1914
1.0797
0.8752
0.3913
0.4858
0.7854
0.5753
0.7650
0.5591
0.7595
0.2177
0.2159
Wald
Chi-Sq
22.0507
2.2956
0.0023
3.7329
6.2860
0.2209
3.5609
8.7926
2.0814
40.0982
9.4912
12.4113
10.4296
9.0517
10.0656
4.2117
0.0923
0.6169
0.8355
2.2639
0.2088
1.1409
0.2756
1.5192
0.3302
0.0532
0.0911
Pr >
Chi-Sq
0.0001
0.1297
0.9621
0.0534
0.0122
0.6383
0.0592
0.0030
0.1491
0.0001
0.0021
0.0004
0.0012
0.0026
0.0015
0.0401
0.7612
0.4322
0.3607
0.1324
0.6477
0.2855
0.5996
0.2177
0.5656
0.8176
0.7628
Odds
Ratios
0.04273
0.79247
1.00710
1.40720
1.61107
0.93342
1.30957
1.58043
1.23306
0.38651
1.42020
1.37424
0.75639
1.20502
1.61575
1.48127
0.72029
0.50288
1.42989
0.48148
1.43175
0.54091
0.66925
0.50198
0.64636
1.05148
1.06737
99
Percent
Confidence Intervals
Low Limit
0.00758
0.53365
0.68679
0.89240
0.98703
0.63984
0.90628
1.06180
0.84829
0.26256
1.05907
1.08931
0.60546
1 .02715
1.09451
0.90470
0.04463
0.05276
0.52184
0.13775
0.18932
0.12289
0.09327
0.11890
0.09137
0.60014
0.61204
Upper Limit
0.2409
1.1768
1.4768
2.2190
2.6297
1.3617
1.8923
2.3524
1.7924
0.5690
1.9045
1.7337
0.9449
1.4137
2.3852
2.4253
11.6256
4.7928
3.9180
1.6829
10.8275
2.3809
4.8020
2.1192
4.5726
1.8423
1.8615
Association of Predicted Probabilities and Observed Responses
Concordant = 77.OX
Discordant = 22.6%
Tied = 0.3X
(437360 pairs)
Somers1
Gamma
Tau-a
c
D = 0.544
= 0.546
= 0.160
= 0.772
E-38
-------
Response Variable: Difficulty Concentrating, Remembering; Depression, Tension
Number of Observations: 1611
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
(Including Comfort and Odor)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1812.194
1817.579
1810.194
Intercept
and
Covar fates
1632.982
1794.520
1572.982
Chi-Square for Covariates
237.212 with 29 DF (p=0.0001)
233.309 with 29 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Fan
New Furniture
Walls Painted
Water Leaks Nearby
Glare
Job Satisfaction
Role Conflict
Workload
Role Clarity
Chem. Sensitivity
Hot Stuffy Air
Paint Odor
Cosmetic, Body Odor
Tobacco Smoke Odor
SE21
SE22
CM FL10
FCFL7
ET1F1011
ET2FL67
ET2F1011
WT1FL13
WT1FL45
WT2FL13
Glued Carpet, Room
Glued Carpet, Hall
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-1.7555
-0.2825
0.1108
-0.0265
0.2315
0.3812
-0.7577
0.2200
0.3006
-0.1960
0.4939
0.0433
0.1553
0.1098
-0.2198
0.8487
1.0844
0.7173
0.0867
0.7446
0.8515
0.5332
0.8725
0.4939
0.4260
-0.4023
-0.1041
-0.1809
0.0577
0.2387
Standard
Error
0.4350
0.1338
0.1315
0.1555
0.1823
0.1323
0.1355
0.1026
0.0786
0.0737
0.1315
0.0191
0.0539
0.0415
0.1138
0.3551
0.6945
0.4532
0.5379
0.4234
0.5122
0.4811
0.5403
0.4727
0.3966
0.3047
0.1555
0.2704
0.1843
0.1505
Wald
Chi-Sq
16.2898
4.4560
0.7104
0.0291
1.6119
8.3017
31.2555
4.5985
14.6456
7.0761
14.1086
5.1721
8.3116
6.9929
3.7274
5.7117
2.4382
2.5046
0.0260
3.0928
2.7640
1.2283
2.6077
1.0915
1.1542
1.7439
0.4482
0.4475
0.0980
2.5158
Pr >
Chi-Sq
0.0001
0.0348
0.3993
0.8646
0.2042
0.0040
0.0001
0.0320
0.0001
0.0078
0.0002
0.0230
0.0039
0.0082
0.0535
0.0169
0.1184
0.1135
0.8720
0.0786
0.0964
0.2677
0.1063
0.2961
0.2827
0.1866
0.5032
0.5035
0.7542
0.1127
Odds
Ratios
0.17282
0.75390
1.11717
0.97385
1.26049
1.46404
0.46874
1.24608
1 .35067
0.82201
1.63869
1.04425
1.16801
1.11605
0.80268
2.33661
2.95766
2.04889
1.09057
2.10560
2.34316
1.70438
2.39289
1.63869
1.53112
0.66878
0.90114
0.83452
1 .05940
1.26960
99
Percent
Confidence Intervals
Low Limit
0.05636
0.53410
0.79617
0.65242
0.78812
1.04122
0.33063
0.95667
1.10310
0.67987
1.16784
0.99412
1.01659
1.00290
0.59873
0.93610
0.49429
0.63754
0.27282
0.70746
0.62630
0.49356
0.59493
0.48492
0.55121
0.30507
0.60371
0.41584
0.65898
0.86158
Upper Limit
0.5300
1.0641
1.5676
1.4536
2.0160
2.0586
0.6645
1.6230
1.6538
0.9939
2.2994
1.0969
1.3420
1.2420
1.0761
5.8325
17.6977
6.5847
4.3594
6.2669
8.7664
5.8856
9.6245
5.5377
4.2531
1.4661
1.3451
1 .6747
1.7031
1.8708
Association of Predicted Probabilities and Observed Responses
Concordant = 74.8X
Discordant - 24.9X
Tied = 0.3X
(486018 pairs)
S omens'
Gamma
Tau-a
c
0.499
0.501
0.187
0.750
E-39
-------
Response Variable: Difficulty Concentrating, Remembering; Depression, Tension (Including Comfort and Odor)
Number of Observations: 1597
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
257.788 with 29 DF (p-0.0001>
253.022 with 29 DF (p=0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
U97.087
1502.463
U95.087
»
Intercept
and
Covariates
1297.299
1458.575
1237.299
m
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Fan
New Furniture
Walls Painted
Water Leaks Nearby
Glare
Job Satisfaction
Role Conflict
Workload
Role Clarity
Chem. Sensitivity
Hot Stuffy Air
Paint Odor
Cosmetic, Body Odor
Tobacco Smoke Odor
SE21
SE22
CM FL10
FCFL7
ET1F1011
ET2FL67
ET2F1011
WT1FL13
WT1FL45
WT2FL13
Glued Carpet, Room
Glued Carpet, Hall
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-1.
-0.
-0.
0.
0.
0.
-1.
0.
0.
-0.
0.
0.
0.
0.
-0.
0.
0.
-0.
-0.
-1.
0.
-0.
0.
0.
0.
0.
-0.
-0.
0.
0.
9989
3597
0964
2107
3974
1096
0365
3141
3263
2459
5075
1061
1392
1181
0162
2710
8877
9739
3501
0194
3830
6145
7854
0257
3762
0583
2036
0764
0846
1451
Standard
Error
0.5051
0.1598
0.1570
0.1882
0.2038
0.1532
0.1583
0.1206
0.0954
0.0895
0.1601
0.0227
0.0794
0.0569
0.1453
0.4302
0.7467
0.5317
0.6194
0.8020
0.8037
0.5913
0.7098
0.7821
0.7137
0.3435
0.1809
0.3059
0.2278
0.2288
Wald
Chi-Sq
15.
5.
0.
1.
3.
0.
42.
6.
11.
7.
10.
21.
3.
4.
0.
0.
1.
3.
0.
1.
0.
1.
1.
0.
0.
0.
1.
0.
0.
0.
6594
0655
3769
2534
8036
5112
8785
7819
7011
5488
0416
8102
0760
3136
0125
3969
4134
3546
3195
6156
2271
0799
2244
0011
2778
0288
2662
0623
1380
4022
Pr >
Odds
Chi-Sq Ratios
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0001 0.13548
0244 0.69789
5393 0.90810
2629
0511
4746
0001 (
0092
0006
0060 (
0015
0001
0795
0378
.23454
.48795
.11583
). 35469
.36903
.38583
). 78200
.66113
.11193
.14935
.12536
9111 0.98393
5287 1
.31128
2345 2.42954
0670 0.37761
5719 0.70462
2037 0.36081
6337 1
.46668
2987 0.54091
2685 2.19328
9738 1
5981 1
8652 1
.02603
.45674
.06003
2605 0.81579
8029 0.92645
7102 1
5260 1
.08828
.15616
99
Percent
Confidence Intervals
Low Limit
0
0
0
0
0
0
0
1
1
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.03688
.46239
.60603
.76025
.88022
.75198
.23592
.00344
.08388
.62098
.09975
.04878
.93675
.97193
.67672
.43292
.35494
.09599
.14289
.04571
.18501
.11793
.35238
.13683
.23170
.43755
.51191
.42130
.60519
.64128
Upper Limit
0
1
1
2
2
1
0
1
1
0
2
1
1
1
1
3
16
1
3
2
11
2
13
7
9
2
1
2
1
2
.4977
.0533
.3607
.0047
.5153
.6557
.5333
.8678
.7719
.9848
.5091
.1789
.4102
.3030
.4306
.9717
.6298
.4855
.4746
.8478
.6270
.4811
.6514
.6936
.1586
.5681
.3000
.0372
.9570
.0844
Association of Predicted Probabilities and Observed Responses
Concordant = 77.8X
Discordant = 21.8X
Tied = 0.3X
(372892 pairs)
Somers1
Gamma
Tau-a
c
= 0.560
= 0.562
= 0.164
* 0.780
E-40
-------
Response Variable: Dizziness
Member of Observations: 1689
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2012.654
2018.066
2010.654
Intercept
and
Cover iates
1878.285
2204.198
1758.285
Chi-Square for Covarfates
252.369 with 59 DF (p=0.0001)
242.088 with 59 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Hours at VDT
>1 Hour at Copier
New Carpet/Drapes
Walls Painted
Window Nearby
Dust
High Pay
Role Conflict
Workload
Underutilization
Wear Glasses
College Grad
Allergy to Pollen
Allergy to Molds
Allergy: Other
Chem. Sensitivity
Live Alone
Under 30
AHU10
AHU12
NEG3
NEG4
SE21
SE22
SE32
NEG1NEG2
FCFL2
FCFL3
FCFL8
ET1FL13
ET1FL45
ET1FL89
ET1F1011
ET2FL13
ET2FL45
ET2FL67
ET2FL89
ET2F1011
WT1FL13
WT1FL45
WT1FL67
WT1FL89
WT1F1011
Parameter
Estimate
-2.9775
-0.1733
0.0474
-0.1472
0.0856
0.3688
-0.0685
0.5731
-0.3230
0.2597
0.2431
0.1260
0.1090
-0.5422
-0.1306
0.0906
0.1372
0.8111
0.0157
-0.2595
1.8909
0.4786
0.5155
-0.6715
0.9681
2.0117
0.3939
-0.3093
-0.8780
-0.8477
-0.5390
-0.6687
-0.1849
-1.3710
-0.2999
-0.3184
0.7527
0.5819
-0.2898
0.0679
-0.6873
-0.9982
-0.1964
0.0763
-0.5848
Standard
Error
0.3510
0.1293
0.0276
0.1817
0.1738
0.1533
0.1321
0.1291
0.1690
0.0923
0.0762
0.0617
0.1230
0.1537
0.1374
0.1461
0.1881
0.1290
0.1419
0.1547
0.8206
0.4651
0.6059
0.7414
0.3248
0.7664
0.4027
0.7692
0.7028
0.7605
0.5532
0.5528
0.5563
0.6378
0.4740
0.5479
0.5464
0.5762
0.6245
0.5019
0.6544
0.5964
0.6030
0.5891
0.5253
Wald
Chi-Sq
71.9603
1.7968
2.9516
0.6559
0.2425
5.7856
0.2685
19.6992
3.6528
7.9228
10.1857
4.1721
0.7847
12.4434
0.9038
0.3845
0.5322
39.5471
0.0123
2.8149
5.3100
1.0588
0.7240
0.8203
8.8840
6.8897
0.9569
0.1617
1.5606
1.2423
0.9492
1.4631
0.1105
4.6211
0.4005
0.3377
1.8978
1.0198
0.2154
0.0183
1.1028
2.8015
0.1061
0.0168
1.2391
Pr >
Chi-Sq
0.0001
0.1801
0.0858
0.4180
0.6224
0.0162
0.6044
0.0001
0.0560
0.0049
0.0014
0.0411
0.3757
0.0004
0.3418
0.5352
0.4657
0.0001
0.9118
0.0934
0.0212
0.3035
0.3948
0.3651
0.0029
0.0087
0.3280
0.6876
0.2116
0.2650
0.3299
0.2264
0.7396
0.0316
0.5268
0.5612
0.1683
0.3126
0.6426
0.8924
0.2936
0.0942
0.7446
0.8969
0.2657
Odds
Ratios
0.05092
0.84089
1.04854
0.86312
1.08937
1.44600
0.93379
1.77376
0.72397
1.29654
1.27520
1.13428
1.11516
0.58147
0.87757
1.09483
1.14706
2.25038
1.01582
0.77144
6.62533
1.61381
1.67448
0.51094
2.63294
7.47602
1.48275
0.73396
0.41561
0.42840
0.58333
0.51237
0.83119
0.25385
0.74089
0.72731
2.12272
1.78944
0.74841
1.07026
0.50293
0.36854
0.82168
1.07929
0.55722
99
Percent
Confidence Intervals
Low Limit
0.02062
0.60268
0.97658
0.54050
0.69621
0.97424
0.66445
1.27193
0.46844
1.02218
1.04792
0.96760
0.81233
0.39136
0.61598
0.75145
0.70656
1.61413
0.70480
0.51788
0.80014
0.48700
0.35159
0.07567
1.14044
1.03816
0.52547
0.10119
0.06799
0.06040
0.14029
0.12335
0.19831
0.04910
0.21851
0.17732
0.51953
0.40560
0.14979
0.29376
0.09320
0.07930
0.17382
0.23664
0.14400
Upper Limit
0.1258
1.1732
1.1258
1.3783
1.7046
2.1462
1.3123
2.4736
1.1189
1.6445
.5518
.3297
.5309
0.8639
.2502
.5951
1.8622
3.1374
1.4641
1.1491
54.8590
5.3479
7.9749
3.4499
6.0787
53.8362
4.1839
5.3237
2.5406
3.0385
2.4255
2.1283
3.4838
1.3126
2.5121
2.9832
8.6731
7.8947
3.7394
3.8993
2.7140
1.7128
3.8842
4.9225
2.1563
E-41
-------
WT2FL13
WT2FU5
WT2FL67
UT2FL89
WT2F1011
ET3FL47
ET3FL811
UT3FL47
WT3FL811
UGHAHU19
Glued Carpet, Hall
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
0.05H
0.0932
-0.8252
-0.6001
-0.7640
-0.2493
-1.4490
-1.1220
-38.0551
-0.0570
0.5804
0.5027
0.2697
-0.2880
-0.0154
0.4478
0.5637
0.5983
0.5253
0.5669
0.6367
0.8334
0.7401
(K6539
0.3535
0.2880
0.3515
0.1948
0.1474
0.0132
0.0274
1.9022
1.3049
1 .8165
0.1533
3.0230
2.2984
0.9086
0.8686
0.1678
0.2533
0.1777
0.6954
0.0821
0.1295
1.05274
1.09768
0.43815
0.54876
0.46580
0.77935
0.23480
0.32563
0.33216
0.25695
0.09382
0.14181
0.10814
0.15116
0.02744
0.04839
0.0076
2.6961
3.0456
0.5890
2.1871
0.0109
0.9305
0.1006
0.0810
0.4428
0.1392
0.9167
0.94459
1.78675
1.65318
1.30957
0.74976
0.98472
0.17527
0.71877
0.78727
0.52953
0.45393
0.67361
3.3365
4.6893
2.0463
2.1235
2.0064
4.0182
2.0094
2.1913
5".09086
4.44161
3.47150
3.23867
1.23838
1.43951
Association of Predicted Probabilities and Observed Responses
Concordant = 73.5X
Discordant = 26.2X
Tied = 0.3X
(578124 pairs)
NOTE:
Sorters'
Gamma
Tau-a
c
0.472
0.474
0.192
0.736
Parameter estimates marked with '#' are regarded to be infinite.
E-42
-------
Response Variable: Dizziness
Number of Observations: 1705
Sex: Males
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1394.854
1400.295
1392.854
.
Intercept
and
Covariates
1373.085
1699.564
1253.085
.
Chi-Square for Covariates
139.769 with 59 OF (p=0.0001)
133.412 with 59 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Hours at VDT
>1 Hour at Copier
New Carpet/Drapes
Walls Painted
Window Nearby
Dust
High Pay
Role Conflict
Workload
Underutilization
Wear Glasses
College Grad
Allergy to Pollen
Allergy to Molds
Allergy: Other
Chem. Sensitivity
Live Alone
Under 30
AHU10
AHU12
NEG3
NEG4
SE21
SE22
SE32
NEG1NEG2
FCFL2
FCFL3
FCFL8
ET1FL13
ET1FL45
ET1FL89
ET1F1011
ET2FL13
ET2FL45
ET2FL67
ET2FL89
ET2F1011
WT1FL13
WT1FL45
WT1FL67
WT1FL89
WT1F1011
WT2FL13
WT2FL45
Parameter
Estimate
-3.3602
0.0448
0.0252
-0.6539
0.4547
-0.1488
-0.5530
0.4171
0.1119
0.3900
0.1423
0.1718
0.4093
-0.4793
-0.1536
0.2896
0.4857
0.4297
0.2392
-0.0769
0.7181
-1.4010
-1.3443
-1.2573
0.1840
0.3170
-1.6945
-0.8066
-1.0105
-1.4566
-2.0097
-1.3692
-1.4987
-2.1030
-0.2770
-1.5600
-0.9989
-0.5190
-2.1632
-0.9270
-0.9572
-1.9533
-2.2135
-0.3109
-38.4547 #
-39.5523 #
-0.7441
Standard
Error
0.5461
0.1630
0.0363
0.4326
0.2069
0.2055
0.1709
0.1512
0.2241
0.1205
0.0991
0.0810
0.1631
0.3202
0.1667
0.1739
0.2250
0.1666
0.2026
0.3267
0.7584
1.0810
1.1318
1.1080
0.4016
0.7131
1.0808
0.8487
0.7210
0.8970
0.8826
0.7759
0.8803
1.1236
0.5813
0.7182
0.7447
0.8320
0.8887
0.6542
0.9221
1.1169
0.8837
0.8753
m
f
0.7434
Wald
Chi-Sq
37.8652
0.0755
0.4791
2.2850
4.8277
0.5247
10.4665
7.6074
0.2494
10.4699
2.0635
4.5015
6.2968
2.2413
0.8491
2.7730
4.6592
6.6532
1.3938
0.0554
0.8963
1.6796
1.4107
1.2875
0.2099
0.1976
2.4583
0.9032
1.9646
2.6371
5.1843
3.1143
2.8985
3.5029
0.2271
4.7177
1.7990
0.3892
5.9245
2.0077
1.0775
3.0588
6.2740
0.1261
B
a
1.0019
Pr >
Chi-Sq
0.0001
0.7835
0.4888
0.1306
0.0280
0.4688
0.0012
0.0058
0.6175
0.0012
0.1509
0.0339
0.0121
0.1344
0.3568
0.0959
0.0309
0.0099
0.2378
0.8139
0.3438
0.1950
0.2349
0.2565
0.6469
0.6566
0.1169
0.3419
0.1610
0.1044
0.0228
0.0776
0.0887
0.0613
0.6337
0.0299
0.1798
0.5327
0.0149
0.1565
0.2992
0.0803
0.0123
0.7225
•
m
0.3169
Odds
Ratios
0.03473
1.04582
1.02552
0.52001
1.57570
0.86174
0.57522
1.51755
1.11840
1.47698
1.15292
1.18744
1.50576
0.61922
0.85761
1.33589
1.62531
1.53680
1.27023
0.92598
2.05053
0.24635
0.26072
0.28442
1.20202
1.37300
0.18369
0.44637
0.36404
0.23303
0.13403
0.25431
0.22342
0.12209
0.75805
0.21014
0.36828
0.59512
0.11496
0.39574
0.38397
0.14181
0.10932
0.73279
m
•
0.47516
99
Percent
Confidence Intervals
Low Limit
0.00851
0.68723
0.93397
0.17063
0.92471
0.50755
0.37038
1.02799
0.62790
1.08285
0.89317
0.96381
0.98921
0.27141
0.55821
0.85354
0.91037
1.00054
0.75375
0.39913
0.29068
0.01521
0.01412
0.01638
0.42719
0.21872
0.01135
0.05014
0.05682
0.02312
0.01380
0.03446
0.02314
0.00676
0.16958
0.03304
0.05408
0.06979
0.01165
0.07337
0.03570
0.00798
0.01122
0.07687
m
f
0.07001
Upper Limit
0.1418
1.5915
1.1260
1.5848
2.6850
1.4631
0.8934
2.2403
1.9921
2.0146
1.4882
1.4630
2.2920
1.4127
1.3176
2.0908
2.9017
2.3605
2.1406
2.1483
14.4651
3.9895
4.8125
4.9377
3.3822
8.6188
2.9732
3.9735
2.3322
2.3492
1 .3020
1.8767
2.1575
2.2065
3.3886
1.3365
2.5079
5.0745
1.1344
2.1345
4.1294
2.5189
1.0649
6.9858
m
m
3.2249
E-43
-------
WT2FL67
UT2FL89
WT2F1011
ET3FU7
ET3FL811
UT3FL47
UT3FL811
UGHAHU19
Glued Carpet, Hall
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
-1.1584
-0.3002
-1.0248
-1.2486
-1.0339
-1.6182
0.1513
-1.0919
1.0553
0.3933
1.0708
-0.1061
-0.1668
0.7157
0.8078
0.8672
0.6931
0.6424
0.7779
0.8487
0.8084
0.4789
0.3372
0.4847
0.2499
0.2371
2.6197
0.1381
1.3965
3.2458
2.5905
4.3270
0.0318
1.8243
4.8568
1.3601
4.8797
0.1801
0.4950
0.1055
0.7101
0.2373
0.0716
0.1075
0.0375
0.8586
0.1768
0.0275
0.2435
0.0272
0.6713
0.4817
0.31399
0.74067
0.35887
0.28691
0.35562
0.19826
1.16335
0.33558
2.87284
1.48186
2.91771
0.89933
0.84637
0.04969
0.09245
0.03844
0.04812
0.06797
0.02673
0.13069
0.04182
0.83665
0.62168
0.83712
0.47244
0.45952
Association of Predicted Probabilities and Observed Responses
Concordant = 71.7X
Discordant = 27.8X
Tied = 0.5X
(354046 pairs)
NOTE:
Somers1
Gamma
Tau-a
c
0.438
0.441
0.107
0.719
1.9843
5.9340
3.3505
1.7106
1.8606
1.4706
10.3558
2.6927
9.8645
3.5322
10.1694
1.7120
1.5589
Parameter estimates marked with 'it' are regarded to be infinite.
E-44
-------
Response Variable: Dizziness (Including Comfort and Odor
Number of Observations: 1632
Sex: Female The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1924.105
1929.503
1922.105
*
Intercept +
Covariates
1758.721
1985.419
1674.721
•
Chi -Square for Covariates
247.384 with 41 DF (p=0.0001)
241.104 with 41 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Use Fan
Walls Painted
Water Leaks Nearby
Dust
High Pay
Workload
College Grad
Chem. Sensitivity
Over 40
AHU10
AHU18
SE22
FC FL2
FC FL3
FC FL8
ET1FL13
ET1FL45
ET1FL89
ET2FL13
ET2FL89
ET2F1011
WT1FL45
Hot Stuffy Air
Paint Odor
New Carpet Odor
WT1FL67
WT1F1011
WT2FL13
WT2FL67
WT2F1011
ET3FL47
ET3FL811
WT3FL47
UGHAHU19
Glued Carpet, Room
Glued Carpet, Corridor
Tacked Carpet, Room
Tacked Carpet, Corridor
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-2.4929
-0.1702
-0.1741
0.2418
-0.3046
0.4338
-0.2721
0.2283
-0.5142
0.6328
-0.3683
2.0984
0.7197
1.8664
-1.3559
-0.6775
-0.4648
-0.3364
-0.0191
-1.2127
-0.1038
-0.0176
0.2595
-1.5340
0.0626
0.2340
0.1849
0.0207
-0.2103
-0.1553
-0.7937
-0.5862
-0.3093
-1.0100
-0.8124
-0.1172
-0.0265
0.4857
0.5394
0.3100
-0.3316
0.0360
Standard
Error
0.3015
0.1319
0.1292
0.1461
0.1926
0.1357
0.1650
0.0697
0.1529
0.1255
0.1370
0.9338
0.5030
0.8001
0.6860
0.6868
0.4851
0.4519
0.4876
0.5856
0.4563
0.5730
0.4469
0.6479
0.0185
0.0597
0.0881
0.5408
0.4503
0.3865
0.5333
0.5223
0.6115
0.8014
0.6767
0.6463
0.2371
0.1934
0.2289
0.2795
0.1959
0.1506
Wald
Chi-Sq
68.3803
1.6652
1.8139
2.7389
2.5026
10.2224
2.7184
10.7442
11.3126
25.4085
7.2253
5.0503
2.0476
5.4407
3.9072
0.9733
0.9180
0.5543
0.0015
4.2883
0.0517
0.0009
0.3371
5.6064
11.4685
15.3615
4.4018
0.0015
0.2181
0.1614
2.2153
1.2596
0.2557
1.5883
1.4415
0.0329
0.0125
6.3086
5.5530
1.2300
2.8646
0.0570
Pr >
Chi-Sq
0.0001
0.1969
0.1780
0.0979
0.1137
0.0014
0.0992
0.0010
0.0008
0.0001
0.0072
0.0246
0.1524
0.0197
0.0481
0.3239
0.3380
0.4566
0.9688
0.0384
0.8201
0.9755
0.5615
0.0179
0.0007
0.0001
0.0359
0.9695
0.6405
0.6878
0.1366
0.2617
0.6131
0.2076
0.2299
0.8561
0.9110
0.0120
0.0184
0.2674
0.0905
0.8113
Odds
Ratios
12.0963
1.1855
1.1902
0.7852
1.3561
0.6480
1.3127
0.7959
1.6723
0.5311
1.4453
0.1227
0.4869
0.1547
3.8803
1.9689
1.5917
1.3999
1.0193
3.3626
1.1094
1.0178
0.7714
4.6367
0.9393
0.7914
0.8312
0.9795
1.2340
1.1680
2.2116
1.7971
1.3625
2.7456
2.2533
1.1243
1.0269
0.6153
0.5831
0.7334
1.3932
0.9646
99 Percent
Confidence Intervals
Low Limit Up Limit
5.56355 26.2999
0.84402 1.6652
0.85324 1.6602
0.53894 1.1440
0.82569 2.2272
0.45687 0.9192
0.85818 2.0080
0.66508 0.9524
1.12787 2.4795
0.38439 0.7338
1.01551 2.0569
0.01107 1.3594
0.13326 1.7790
0.01969 1.2149
0.66283 22.7152
0.33565 11.5501
0.45620 5.5534
0.43706 4.4839
0.29027 3.5793
0.74393 15.1986
0.34245 3.5939
0.23260 4.4533
0.24397 2.4393
0.87372 24.6060
0.89560 0.9852
0.67856 0.9229
0.66243 1.0429
0.24322 3.9448
0.38687 3.9364
0.43157 3.1611
0.55985 8.7362
0.46802 6.9009
0.28198 6.5832
0.34840 21.6370
0.39425 12.8788
0.21274 5.9421
0.55751 1.8913
0.37385 1.0126
0.32334 1.0515
0.35701 1.5068
0.84111 2.3077
0.65446 1.4218
Concordant = 73.3X
Discordant = 26.4X
Tied = 0.3X
(531900 pairs)
Sorters'
Gamma
Tau-a
c
0.469
0.470
0.187
0.734
E-45
-------
Response Variable: Dizziness (Including Comfort and Odor)
Number of Observations: 1637
Sex: Male The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1330.824
1336.225
1328.824
.
Intercept +
Cover i at es
1271.192
1498.018
1187.192
m
Chi-Square for Covarlates
141.632 with 41 DF (p=0.0001)
153.885 with 41 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Use Fan
Walls Painted
Water Leaks Nearby
Dust
High Pay
Workload
College Grad
Chem. Sensitivity
Over 40
AHU10
AHU18
SE22
FC FL2
FC FL3
FC FL8
ET1FL13
ET1FL45
ET1FL89
ET2FL13
ET2FL89
ET2F1011
WT1FL45
Hot Stuffy Air
Paint Odor
New Carpet Odor
WT1FL67
WT1F1011
WT2FL13
WT2FL67
WT2F1011
ET3FL47
ET3FL811
WT3FL47
UGHAHU19
Glued Carpet, Room
Glued Carpet, Corridor
Tacked Carpet, Room
Tacked Carpet, Corridor
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
99 Percent
Parameter
Estimate
-3.0260
-0.0052
0.0108
-0.1352
0.1402
0.1919
0.2239
0.1261
-0.2563
0.5515
-0.0262
1.0102
1.0734
0.6964
-0.5394
-0.9775
-1.0800
-0.4798
-0.9087
-1.7878
-0.8821
-1.7090
-0.3810
-1.4765
0.0905
0.1128
0.4160
-1.3001
-37.4769
-37.8149
-0.5742
-0.9342
-0.7945
-0.6696
-1.0339
-0.3540
-0.3861
0.4009
-0.0734
0.6576
-0.2737
-0.1489
Standard
Error
0.5122
0.1677
0.1644
0.1980
0.2190
0.1619
0.2126
0.0917
0.3277
0.1622
0.1623
0.7263
0.4423
0.7751
0.6652
0.7943
0.7816
0.6028
0.7951
1.0611
0.5901
0.8581
0.5700
1.0569
0.0241
0.0869
0.1105
0.7819
.
a
0.6065
0.7933
0.6121
0.5430
0.6685
0.6859
0.3168
0.2650
0.2931
0.3567
0.2466
0.2389
Wald
Chi-Sq
34.9010
0.0010
0.0043
0.4660
0.4097
1.4040
1.1090
1.8935
0.6114
11.5618
0.0260
1.9347
5.8903
0.8071
0.6575
1.5143
1.9095
0.6336
1.3060
2.8389
2.2349
3.9658
0.4469
1.9515
14.0702
1.6841
14.1767
2.7644
•
m
0.8961
1.3867
1.6850
1.5206
2.3916
0.2663
1.4853
2.2890
0.0628
3.3978
1.2316
0.3883
Pr >
Chi-Sq
0.0001
0.9752
0.9477
0.4948
0.5221
0.2361
0.2923
0.1688
0.4342
0.0007
0.8720
0.1642
0.0152
0.3690
0.4174
0.2185
0.1670
0.4260
0.2531
0.0920
0.1349
0.0464
0.5038
0.1624
0.0002
0.1944
0.0002
0.0964
.
•
0.3438
0.2390
0.1943
0.2175
0.1220
0.6058
0.2229
0.1303
0.8022
0.0653
0.2671
0.5332
Odds
Confidence
Intervals
Ratios Low Limit Upper Limit
20.6146
1.0052
0.9893
1.1448
0.8692
0.8254
0.7994
0.8815
1.2921
0.5761
1.0265
0.3641
0.3418
0.4984
1.7150
2.6578
2.9447
1.6158
2.4811
5.9763
2.4160
5.5234
1.4637
4.3776
0.9135
0.8933
0.6597
3.6697
B
•
1.7757
2.5452
2.2133
1.9535
2.8120
1.4248
1.4712
0.6697
1.0762
0.5181
1.3148
1.1606
5.51005
0.65260
0.64772
0.68739
0.49443
0.54392
0.46229
0.69606
0.55552
0.37934
0.67578
0.05607
0.10940
0.06767
0.30908
0.34348
0.39321
0.34198
0.31999
0.38845
0.52835
0.60564
0.33712
0.28763
0.85849
0.71416
0.49626
0.48965
.
•
0.37227
0.32978
0.45737
0.48231
0.50250
0.24344
0.65052
0.33840
0.50579
0.20671
0.69660
0.62719
77.1250
.5484
.5109
.9065
.5280
.2525
.3823
.1164
3.0055
0.8749
1.5594
2.3649
1.0682
3.6702
9.5158
20.5655
22.0520
7.6340
19.2378
91 .9453
11.0474
50.3735
6.3555
66.6246
0.9720
1.1175
0.8769
27.5024
.
m
8.4701
19.6434
10.7110
7.9119
15.7361
8.3385
3.3274
1 .3254
2.2897
1.2986
2.4817
2.1475
Concordant = 72.9X Somers1
Discordant = 26.5X Gamma
Tied 0.6% Tau-a
(323610 pairs) c
NOTE: Parameter estimates marked with
D = 0.464
= 0.467
= 0.112
= 0.732
#' are regarded to be infinite.
E-46
-------
Response Variable: Dry or Itchy Skin
Number of Observations: 1828
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2154.861
2160.372
2152.861
Intercept
and
Covariates
2053.672
2235.535
1987.672
Chi-Square for Covariates
165.189 with 32 DF (p=0.0001)
160.093 with 32 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Plants Nearby
Use Heater
Glare
Dust
Medium Pay
Allergy to Molds
Chem. Sensitivity
Eczema
Discomfort Index
AHU20
ETB
SE32
CM FL2
ET1FL13
ET1FL67
ET1FL89
ET1F1011
ET2FL45
ET2FL67
ET2F1011
UT1FL13
WT1FL45
WT1FL67
WT2FL13
UT2FL45
WT2FL89
WT2F1011
ET3FL47
ET3FL811
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-2.16780
0.07060
0.43900
0.08350
0.59080
0.29640
0.44170
0.15040
0.91760
0.09520
-0.44990
0.99030
-0.21900
-1.96730
-0.04190
-0.00467
1.09600
0.55590
0.02780
-0.05380
0.73700
0.24470
0.48960
0.53030
0.79230
-0.17570
0.14320
0.74080
-0.67840
0.69240
-0.88760
0.16940
0.28000
Standard
Error
0.1714
0.1120
0.1288
0.1141
0.1198
0.1200
0.1215
0.1183
0.1844
0.0476
0.6859
0.5617
0.3942
0.7374
0.3780
0.4994
0.4236
0.4293
0.4927
0.5804
0.4336
0.4771
0.3510
0.4676
0.3323
0.4553
0.3540
0.4090
0.6585
0.5746
0.3042
0.1525
0.1303
Uald
Chi-Sq
59.9703
0.3976
11.6181
0.5354
24.3313
6.0985
13.2236
1.6153
24.7594
4.0003
0.4301
3.1084
0.3088
7.1172
0.0123
0.0001
6.6955
1.6771
0.0032
0.0086
2.8897
0.2631
1.9453
1.2861
5.6837
0.1488
0.1637
3.2816
1.0613
1 .4525
8.5101
1.2345
4.6222
Pr >
Chi-Sq
0.0001
0.5283
0.0007
0.4643
0.0001
0.0135
0.0003
0.2038
0.0001
0.0455
0.5119
0.0779
0.5784
0.0076
0.9117
0.9925
0.0097
0.1953
0.9550
0.9261
0.0891
0.6080
0.1631
0.2568
0.0171
0.6997
0.6858
0.0701
0.3029
0.2281
0.0035
0.2665
0.0316
Odds
Ratios
0.11443
1.07315
1.55116
1.08709
1.80543
1.34501
1.55535
1.16230
2.50328
1.09988
0.63769
2.69204
0.80332
0.13983
0.95897
0.99534
2.99217
1 .74351
1.02819
0.94762
2.08966
1.27724
1.63166
1.69944
2.20847
0.83887
1.15396
2.09761
0.50743
1.99851
0.41164
1.18459
1.32313
99
Percent
Confidence Intervals
Low Limit
0.07358
0.80420
1.11317
0.81024
1.32604
0.98736
1.13737
0.85698
1.55673
0.97295
0.10896
0.63341
0.29099
0.02092
0.36217
0.27496
1.00482
0.57696
0.28898
0.21248
0.68389
0.37370
0.66062
0.50954
0.93828
0.25962
0.46361
0.73141
0.09304
0.45486
0.18802
0.79976
0.94587
Upper Limit
0.1779
1.4321
2.1615
1 .4585
2.4581
1.8322
2.1269
1.5764
4.0254
1.2434
3.7321
11.4414
2.2177
0.9345
2.5391
3.6031
8.9102
5.2687
3.6583
4.2262
6.3850
4.3654
4.0300
5.6680
5.1982
2.7106
2.8723
6.0158
2.7674
8.7808
0.9012
1.7546
1.8509
Association of Predicted Probabilities and Observed Responses
Concordant = 68.3X
Discordant = 31.3X
Tied = 0.4X
(667296 pairs)
Somers'
Gamma
Tau-a
c
D =
0.370
0.371
0.148
0.685
E-47
-------
Response Variable: Dry or Itchy Skin
Number of Observations: 1764
Sex: Males
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1581.886
1587.362
1579.886
.
Intercept
and
Covariates
1508.647
1689.333
1442.647
•
Chi-Square for Covariates
137.239 with 32 DF
Chi-Sq
0.0001
0.3972
0.3212
0.0191
0.0003
0.2562
0.3350
0.0145
0.0001
0.9110
0.2932
p
0.2876
0.1530
0.9997
0.1498
0.2205
0.2032
0.5026
0.1438
0.0769
0.7481
0.4771
0.1658
0.1587
0.1446
0.1501
0.5163
0.0242
0.0673
0.0692
0.1757
0.7918
Odds
Ratios
0.12457
1.12502
1.24123
1.38805
1.66496
0.80501
1.15235
1 .45005
4.26525
1.00728
0.32046
„
0.43960
0.41115
1.00019
0.30425
0.38393
0.48167
0.64895
0.19869
0.30495
0.78005
1.43061
0.41908
0.22668
0.22277
0.22036
0.64856
0.17103
0.38236
1.97684
0.75059
0.94667
99
Percent
Confidence Intervals
Low Limit
0.06867
0.78602
0.70825
0.96804
1.15996
0.49218
0.78889
0.98025
2.61245
0.85220
0.01970
.
0.06003
0.08284
0.28964
0.03622
0.05132
0.10977
0.12323
0.01152
0.05410
0.10636
0.39095
0.08321
0.01504
0.01572
0.01470
0.11629
0.02273
0.09876
0.75220
0.43485
0.55442
Upper Limit
0.22598
1.61023
2.17527
1.99029
2.38981
1.31668
1.68326
2.14503
6.96371
1.19057
5.21237
.
3.21906
2.04054
3.45390
2.55587
2.87218
2.11358
3.41734
3.42728
1.71887
5.72096
5.23504
2.11065
3.41551
3.15780
3.30315
3.61723
1.28677
1.48036
5.19532
1.29557
1.61645
Association of Predicted Probabilities and Observed Responses
Concordant = 70.1X
Discordant =29.0%
Tied = 0.8X
(428643 pairs)
NOTE:
Sotners1 D
Gamma
Tau-a
c
0.411
0.414
0.113
0.705
Parameter estimates marked with '#' are regarded to be infinite.
E-48
-------
Response Variable: Dry or Itchy Skin (Including Comfort and Odor)
Number of Observations: 1746
Sex: Females
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
only
2061 .388
2066.853
2059.388
Intercept
and
Covariates
1902.485
2000.856
1866.485
Chi-Square for Covariates
192.903 with 17 OF (p=0.0001)
186.702 with 17 Of (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Heater
Use Lamp
Dust
Medium Pay
Eczema
Dry Air
Humid Air
CM FL10
ET1FL89
ET1F1011
ET2F1011
WT1FL67
WT2F1011
ET3FL811
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-2.2720
0.5171
-0.1773
0.5224
0.3382
0.9022
0.2530
-0.1738
0.5130
0.8779
0.7365
0.8018
0.5622
0.8791
0.4940
-0.9612
-0.0448
0.1512
Standard
Error
0.1623
0.1324
0.1182
0.1243
0.1234
0.1891
0.0299
0.0776
0.4076
0.4392
0.4381
0.4442
0.4817
0.4206
0.6331
0.2908
0.1579
0.1347
Wald
Chi-Sq
95.8589
15.2465
2.2485
17.6643
7.5064
22.7529
71.4911
5.0136
1.5842
3.9954
2.8256
3.2585
1.3622
4.3677
0.6088
10.9213
0.0805
1.2583
Pr >
Chi-Sq
0.0001
0.0001
0.1337
0.0001
0.0061
0.0001
0.0001
0.0251
0.2082
0.0456
0.0928
0.0711
0.2432
0.0366
0.4353
0.0010
0.7767
0.2620
Odds
Ratios
0.10311
1 .67716
0.83753
1.68607
1.40242
2.46502
1.28788
0.84046
1.67029
2.40584
2.08861
2.22955
1.75453
2.40873
1.63886
0.38243
0.95619
1.16323
99
Percent
Confidence Intervals
Low Limit
0.06788
1.19248
0.61768
1.22410
1.02053
1.51449
1.19241
0.68818
0.58451
0.77610
0.67567
0.71002
0.50730
0.81516
0.32082
0.18081
0.63664
0.82219
Upper Limit
0.15662
2.35882
1.13563
2.32239
1 .92722
4.01213
1.39100
1.02644
4.77301
7.45794
6.45623
7.00104
6.06818
7.11756
8.37177
0.80888
1.43613
1.64574
Association of Predicted Probabilities and Observed Responses
Concordant - 70.3X
Discordant = 29.2X
Tied = 0.5X
(610029 pairs)
Somers1
Gamma
Tau-a
c
0.410
0.412
0.164
0.705
E-49
-------
Response Variable: Dry or Itchy Skin (Including Comfort and Odor)
Number of Observations: 1719
Sex: Males
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1542.956
1548.406
1540.956
Intercept
and
Covar iates
1384.756
1482.847
1348.756
Chi-Square for Covarfates
192.200 with 17 OF (p=0.0001)
205.630 with 17 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Heater
Use Lamp
Dust
Medium Pay
Eczema
Dry Air
Humid Air
CM FL10
ET1FL89
ET1F1011
ET2F1011
WT1FL67
WT2F1011
ET3FL811
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-2.7208
0.0498
0.1487
0.2581
-0.2870
1.5875
0.3604
-0.1014
-0.8204
-0.6977
0.0679
-0.8309
-0.7570
-0.0196
-0.4188
0.1961
-0.1845
0.1249
Standard
Error
0.2468
0.2246
0.1456
0.1472
0.1967
0.1959
0.0393
0.1195
0.5198
0.7782
0.5564
0.6610
0.6485
0.6628
0.5117
0.3212
0.2208
0.2147
Weld
Chi-Sq
21.5066
0.0492
1.0436
3.0744
2.1299
65.6600
84.2839
0.7188
2.4914
0.8038
0.0149
1.5799
1.3628
0.0009
0.6699
0.3726
0.6983
0.3384
Pr >
Chi-Sq
0.0001
0.8245
0.3070
0.0795
0.1445
0.0001
0.0001
0.3965
0.1145
0.3699
0.9028
0.2088
0.2431
0.9764
0.4131
0.5416
0.4033
0.5608
Odds
Ratios
0.06582
1.05106
1.16032
1.29447
0.75051
4.89150
1.43390
0.90357
0.44026
0.49773
1.07026
0.43566
0.46907
0.98059
0.65784
1.21665
0.83152
1.13304
99
Percent
Confidence Intervals
Low Limit
0.03486
0.58933
0.79743
0.88596
0.45217
2.95312
1.29585
0.66416
0.11539
0.06705
0.25528
0.07937
0.08825
0.17782
0.17606
0.53189
0.47082
0.65170
Upper Limit
0.12430
1 .87455
.68837
.89134
.24570
8.10222
.58667
.22928
1.67968
3.69486
4.48701
2.39130
2.49312
5.40742
2.45798
2.78295
1 .46856
1.96987
Association of Predicted Probabilities and Observed Responses
Concordant = 74.1X
Discordant = 25.1X
Tied = 0.8X
(407540 pairs)
Somers1
Gamma
Tau-a
c
= 0.491
= 0.495
= 0.135
= 0.745
E-50
-------
Response Variable: Problems with Contact Lenses
Number of Observations: 426
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
•2 LOG L
Score
Intercept
Only
562.766
566.821
560.766
Intercept
and
Covariates
550.192
615.064
518.192
Chi-Square for Covariates
42.574 with 15 DF
40.806 with 15 OF (p=0.0003)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at Work
Use Lamp
New Carpet/Drapes
Dust
Job Satisfaction
Workload
External Stress
Allergy to Dust
AHU14
AHU3
AUU15SU2
CM FL2
CM FL10
Private Office
Cubicle or Shared
Parameter
Estimate
-0.5677
-0.0994
0.5746
0.5028
0.6743
-0.2008
-3.0558
-0.0239
0.6073
0.1010
1.7218
0.7984
-0.6673
-0.3124
0.4632
0.3734
Standard
Error
1.0856
0.0713
0.2151
0.2795
0.2348
0.2167
0.1225
0.0895
0.2197
1.4945
1.2712
0.7510
0.7279
0.9123
0.3148
0.2688
Wald
Chi-Sq
0.2735
1.9428
7.1363
3.2364
8.2476
0.8587
0.2077
0.0715
7.6417
0.0046
1.8348
1.1302
0.8404
0.1173
2.1642
1.9291
Pr >
Chi-Sq
0.6010
0.1634
0.0076
0.0720
0.0041
0.3541
0.6485
0.7891
0.0057
0.9461
0.1756
0.2877
0.3593
0.7320
0.1413
0.1649
Odds
Ratios
0.56683
0.90538
1.77642
1.65334
1.96266
0.81808
0.94573
0.97638
1 .83547
1.10628
5.59459
2.22198
0.51309
0.73169
1.58915
1 .45267
99
Percent
Confidence Intervals
Low Limit
0.03459
0.75347
1.02071
0.80478
1.07192
0.46813
0.68980
0.77534
1.04222
0.02355
0.21165
0.32104
0.07868
0.06977
0.70629
0.72685
Upper Limit
9.289
1.088
3.092
3.397
3.594
1.430
1.297
1.230
3.232
51.979
147.881
15.379
3.346
7.673
3.576
2.903
Association of Predicted Probabilities and Observed Responses
Concordant = 68.OX
Discordant = 31.7X
Tied = 0.3%
(42233 pairs)
Somers1
Gamma
Tau-a
c
0.364
0.365
0.170
0.682
E-51
-------
Response Variable: Problems with Contact Lenses
Number of Observations: 290
Sex: Males
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
339.692
343.362
337.692
Intercept
and
Covariates
329.104
387.822
297.104
Chi-Square for Covariates
40.588 with 15 DF (p=0.0004)
38.348 with 15 DF (p=0.0008)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at Work
Use Lamp
New Carpet/Drapes
Oust
Job Satisfaction
Workload
External Stress
Allergy to Dust
AHU14
AHU3
AUU15SU2
CM FL2
CM FL10
Private Office
Cubicle or Shared
Parameter
Estimate
-5.2081
0.2291
0.2202
0.2623
0.6010
-0.9035
0.4903
0.2591
0.2642
-1.0026
-0.7004
2.3571
-37.6177 #
-0.0690
-0.0943
-0.6131
Standai
Error
1.5690
0.0939
0.3022
0.3674
0.3028
0.3103
0.1815
0.1410
0.3001
1.1700
1.2834
1.2484
*
1.2071
0.4117
0.4404
Wald
Chi-Sq
11.0181
5.9582
0.5310
0.5097
3.9390
8.4782
7.2971
3.3756
0.7747
0.7344
0.2978
3.5651
0.0033
0.0525
1.9375
Pr >
Chi-Sq
0.0009
0.0146
0.4662
0.4753
0.0472
0.0036
0.0069
0.0662
0.3788
0.3915
0.5853
0.0590
0.9544
0.8188
0.1639
Odds
Ratios
0.0055
1.2575
1.2463
1.2999
1.8239
0.4051
1.6328
1.2958
1.3024
0.3669
0.4964
10.5603
0.9333
0.9100
0.5417
99 Percent
Confidence Intervals
Low Limit Upper Limit
0.00010
0.98729
0.57220
0.50453
0.83610
0.18217
1.02302
0.90112
0.60118
0.01802
0.01820
0.42368
0.04165
0.31511
0.17420
0.311
1.602
2.715
3.349
3.979
0.901
2.606
1.863
2.821
7.473
13.540
263.217
20.915
2.628
1.684
Association of Predicted Probabilities and Observed Responses
Concordant = 72.7X
Discordant - 27.OX
Tied = 0.3X
(16536 pairs)
NOTE:
Somers1
Gamma
Tau-a
c
0.457
0.459
0.180
0.729
Parameter estimates marked with '#' are regarded to be infinite.
E-52
-------
Response Variable: Problems with Contact Lenses (Including Comfort and Odor)
Nunfcer of Observations: 447
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
590.690
594.792
588.690
Intercept
and
Covariates
569.770
668.231
521.770
Chi-Square for Covariates
66.920 with 23 OF (p=0.0001)
63.154 with 23 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Lamp
Allergy to Dust
Cosmetic, Body Odor
NEG1NEG2
CM FL10
FCFL8
ET1FL67
ET1F1011
ET2FL13
ET2FL89
ET2F1011
WT1FL13
UT1FL67
UT1FL89
WT1F1011
UT2FL45
UT2FL89
ET3FL47
WT3FL811
Glued Carpet, Hall
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-1.9716
0.6647
0.7165
0.2986
1.6728
-0.7298
0.9511
1.7707
2.3428
1.0160
1.5758
1.9606
0.2741
0.7218
1.2046
0.8941
0.1309
-0.2096
1.4090
0.6136
-0.6919
-1.8424
0.6388
0.4029
Standard
Error
0.3213
0.2193
0.2182
0.0739
1.2993
0.8663
0.7113
0.9480
1.1577
0.8503
0.9702
0.8506
1.2988
0.9992
0.9500
0.8764
0.8605
0.8641
0.7757
1.1438
0.3189
0.7196
0.3082
0.2727
Weld
Chi-Sq
37.6525
9.1878
10.7830
16.3108
1.6577
0.7098
1.7879
3.4889
4.0957
1.4278
2.6378
5.3125
0.0446
0.5219
1.6078
1.0406
0.0232
0.0589
3.2992
0.2878
4.7071
6.5548
4.2948
2.1826
Pr >
Chi-Sq
0.0001
0.0024
0.0010
0.0001
0.1979
0.3995
0.1812
0.0618
0.0430
0.2321
0.1043
0.0212
0.8328
0.4700
0.2048
0.3077
0.8791
0.8083
0.0693
0.5917
0.0300
0.0105
0.0382
0.1396
Odds
Ratios
0.1392
1.9439
2.0473
1.3480
5.3271
0.4820
2.5886
5.8750
10.4103
2.7621
4.8346
7.1036
1.3153
2.0581
3.3354
2.4451
1.1399
0.8109
4.0919
1.8471
0.5006
0.1584
1.8942
1.4962
99
Percent
Confidence Intervals
Low Limit
0.06085
1.10493
1.16698
1.11431
0.18746
0.05175
0.41428
0.51102
0.52759
0.30901
0.39715
0.79410
0.04635
0.15690
0.28863
0.25576
0.12421
0.08755
0.55477
0.09702
0.22016
0.02482
0.85631
0.74113
Upper Limit
0.319
3.420
3.592
1.631
151.381
4.490
16.174
67.542
205.416
24.689
58.853
63.545
37.330
26.997
38.544
23.376
10.460
7.511
30.181
35.164
1.138
1.011
4.190
3.020
Association of Predicted Probabilities and Observed Responses
Concordant = 71.4%
Discordant - 27.7X
Tied = 0.8X
(46530 pairs)
Sorters'
Gamma
Tau-a
c
D =
0.437
0.440
0.204
0.718
E-53
-------
Response Variable: Problems with Contact Lenses (Including Comfort and Odor)
Number of Observations: 285
Sex: Males
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
330.510
334.163
328.510
Intercept
and
Covariates
336.048
423.708
288.048
Chi-Square for Covariates
40.462 with 23 DF (p=0.0136)
32.085 with 23 DF (p=0.0984)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Lamp
Allergy to Dust
Cosmetic, Body Odor
NEG1NEG2
CM FL10
FCFL8
ET1FL67
ET1F1011
ET2FL13
ET2FL89
ET2F1011
UT1FL13
UT1FL67
UT1FL89
UT1F1011
WT2FL45
WT2FL89
ET3FL47
WT3FL811
Glued Carpet, Hall
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter Standard
Estimate Error
0.8924
0.1076
0.6453
0.1753
0.2949
38.2100 it
0.5663
0.3371
0.2050
37.4111 #
0.8698
37.6704 #
1.9408
1.0650
37.2378 it
37.9149 *
37.4807 #
1.3995
37.7074 #
37.6848 it
0.4534
0.6255
0.2189
0.6268
0.4563
0.3068
0.2988
0.1177
1.4691
m
1.2988
1.3318
1.4103
•
1.4686
m
1.3195
1.5352
9
m
m
1.0956
•
•
0.3714
0.8039
0.4225
0.4571
99 Percent
Uald Pr > Odds Confidence Intervals
Chi-Sq Chi-Sq Ratios Low Limit Upper Limit
1.327
1.979
4.117
1.614
59.104
49.999
43.285
46.429
104^890
0.1413 6.96432 0.23265 208^478
0.4879 2.90084 0.05559 151.362
3.8246
0.1231
4.6638
2.2171
0.0403
0.1901
0.0641
0.0211
0.0505
0.7257
0.0308
0.1365
0.8409
0.6628
0.8002
0.8844
0.40967
0.89799
1.90656
1.19160
1.34299
1.76174
1.40088
1.22753
0.12646
0.40742
0.88302
0.87995
0.03052
0.06208
0.04534
0.03245
0.3508 0.5537 2.38643 0.05430
2.1633
0.4813
1.6316 0.2015 4.05317 0.24105
1.4907 0.2221
0.6054 0.4365
0.2685 0.6043
1.8806 0.1703
Association of Predicted Probabilities and Observed Responses
Concordant = 70.4X
Discordant = 28.U
Tied = 1.6X
(15750 pairs)
NOTE:
Somens'
Gamma
Tau-a
c
0.423
0.430
0.165
0.712
68.154
0.63546 0.24411 1.654
0.53499 0.06745 4.243
0.80340 0.27056 2.386
0.53430 0.16459 1.734
Parameter estimates marked with '#' are regarded to be infinite.
E-54
-------
Response Variable: Hot Stuffy Air
Number of Observations: 1742
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2201.640
2207.102
2199.640
Intercept
and
Covar fates
2008.870
2216.456
1932.870
Chi-Square for Covariates
266.770 with 37 DF (p=0.0001)
243.775 with 37 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Photocopier Nearby
Use Fan
Window Nearby
Lights Too Dim
Glare
Noise
Dust
Workload
Underutilization
Allergy to Molds
Chem. Sensitivity
Discomfort Index
AHU13
AHU16
NEG4
NEG1NEG2
FC FL2
FC FL6
FC FL8
FCNOFL
ET1FL67
ET1F1011
ET2FL45
ET2FL89
ET2F1011
WT1FL13
WT1FL67
WT1F1011
WT2FL67
WT2F1011
ET3FL47
ET3FL811
Glued Carpet, Hall
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-2.8844
-0.1102
0.7350
-0.0818
0.5534
0.3248
0.2671
0.6590
0.2512
-0.1345
0.3304
0.2851
0.1048
-0.8849
0.1304
1.1551
-0.9064
-0.7824
0.1035
0.1626
1 .2876
0.1837
-0.6927
-0.7940
0.9468
-0.5671
0.2732
-0.4629
-0.9980
0.1580
0.4256
0.1672
-0.4684
-0.2863
0.1426
-0.6696
0.4023
0.2319
Standard
Error
0.3042
0.1659
0.1144
0.1229
0.2033
0.1160
0.1179
0.1231
0.0658
0.0563
0.1236
0.1198
0.0483
0.5606
0.5285
0.6211
0.8355
0.6182
0.4761
0.4304
0.6790
0.4587
0.4910
0.6514
0.5066
0.5442
0.5394
0.5606
0.5675
0.4821
0.4438
0.5925
0.6971
0.1571
0.2307
0.3279
0.1573
0.1328
Wald
Chi-Sq
89.8847
0.4414
41.2666
0.4436
7.4130
7.8467
5.1344
28.6584
14.5792
5.6985
7.1414
5.6637
4.6975
2.4921
0.0609
3.4587
1.1770
1.6016
0.0472
0.1427
3.5956
0.1604
1.9903
1.4857
3.4928
1.0861
0.2565
0.6818
3.0931
0.1074
0.9195
0.0796
0.4516
3.3241
0.3821
4.1697
6.5393
3.0513
Pr >
Chi-Sq
0.0001
0.5064
0.0001
0.5054
0.0065
0.0051
0.0235
0.0001
0.0001
0.0170
0.0075
0.0173
0.0302
0.1144
0.8051
0.0629
0.2780
0.2057
0.8280
0.7056
0.0579
0.6888
0.1583
0.2229
0.0616
0.2973
0.6125
0.4090
0.0786
0.7431
0.3376
0.7778
0.5016
0.0683
0.5365
0.0412
0.0106
0.0807
Odds
Ratios
0.05589
0.89565
2.08548
0.92146
1.73916
.38375
.30617
.93286
.28557
0.87415
.39152
.32990
1.11049
0.41276
1.13928
3.17434
0.40398
0.45731
1.10905
1.17657
3.62408
1.20166
0.50022
0.45203
2.57745
0.56717
1.31416
0.62946
0.36862
1.17117
1.53051
1.18199
0.62600
0.75104
1.15327
0.51191
1 .49526
1.26099
99
Percent
Confidence Intervals
Low Limit
0.02553
0.58417
1.55318
0.67140
1.03014
1.02632
0.96405
1 .40761
1.08513
0.75614
1.01208
0.97677
0.98057
0.09739
0.29200
0.64092
0.04695
0.09303
0.32532
0.38825
0.63033
0.36865
0.14121
0.08442
0.69893
0.13960
0.32749
0.14852
0.08545
0.33828
0.48791
0.25690
0.10392
0.50108
0.63656
0.21997
0.99710
0.89566
Upper Limit
0.1224
1.3732
2.8002
1.2646
2.9362
1.8657
1.7697
2.6541
1.5230
1.0106
1.9132
1.8107
1.2576
1.7493
4.4452
15.7219
3.4759
2.2481
3.7808
3.5655
20.8365
3.9169
1.7720
2.4206
9.5048
2.3043
5.2735
2.6677
1.5902
4.0548
4.8010
5.4384
3.7710
1.1257
2.0894
1.1913
2.2423
1.7753
Concordant » 72.7X
Discordant = 27.1X
Tied * 0.3X
(666832 pairs)
Sorters'
Gamma
Tau-a
c
0.456
0.457
0.201
0.728
E-55
-------
Response Variable: Hot Stuffy Air
Number of Observations: 1723
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1897.510
1902.962
1895.510
Intercept
and
Cover i at es
1659.054
1866.223
1583.054
Chi-Square for Covariates
312.456 with 37 DF (p=0.0001)
287.886 with 37 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Photocopier Nearby
Use Fan
Window Nearby
Lights Too Dim
Glare
Noise
Dust
Workload
Underutilization
Allergy to Molds
Chem. Sensitivity
Discomfort Index
AHU13
AHU16
NEG4
NEG1NEG2
FC FL2
FC FL6
FC FL8
FCNOFL
ET1FL67
ET1F1011
ET2FL45
ET2FL89
ET2F1011
WT1FL13
WT1FL67
WT1F1011
WT2FL67
WT2F1011
ET3FL47
ET3FL811
Glued Carpet, Hall
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-3.3746
0.0562
0.9725
-0.6047
0.7993
0.3019
0.4183
0.8046
0.2831
0.0683
0.2177
0.0953
0.1273
-1.2515
-0.1804
0.1427
-1.0226
-1.5907
-1.2290
0.1039
0.0022
-37.9532 #
-0.5150
-0.4071
-0.1888
-0.8752
-2.0713
-0.5958
-1.1599
-0.0787
-0.2224
-0.8268
-1.0522
0.2374
0.2766
0.0072
-0.1064
-0.0014
Standard
Error
0.4067
0.1997
0.1277
0.1418
0.2476
0.1304
0.1372
0.1316
0.0788
0.0662
0.1365
0.1440
0.0581
0.6703
0.5013
0.6978
0.8268
0.7697
0.8084
0.4796
1.2097
•
0.5668
0.5849
0.5658
0.6736
1.1487
0.4885
1.0716
0.4789
0.6391
0.6191
0.5988
0.1801
0.2793
0.4069
0.2037
0.2019
Wald
Chi-Sq
68.8541
0.0791
57.9792
18.1821
10.4231
5.3570
9.2905
37.3616
12.8972
1.0638
2.5431
0.4379
4.8049
3.4864
0.1294
0.0418
1.5300
4.2707
2.3117
0.0469
0.0000
*
0.8254
0.4845
0.1113
1.6879
3.2515
1.4874
1.1716
0.0270
0.1211
1.7838
3.0875
1.7372
0.9803
0.0003
0.2728
0.0000
Pr >
Chi-Sq
0.0001
0.7785
0.0001
0.0001
0.0012
0.0206
0.0023
0.0001
0.0003
0.3023
0.1108
0.5081
0.0284
0.0619
0.7190
0.8380
0.2161
0.0388
0.1284
0.8285
0.9985
•
0.3636
0.4864
0.7387
0.1939
0.0714
0.2226
0.2791
0.8694
0.7278
0.1817
0.0789
0.1875
0.3221
0.9858
0.6014
0.9946
Odds
Ratios
0.03423
1.05781
2.64455
0.54624
2.22398
1.35243
1.51938
2.23580
1 .32724
1.07069
1 .24321
1.09999
1.13576
0.28608
0.83494
1.15338
0.35966
0.20378
0.29259
1.10949
1.00224
•
0.59750
0.66558
0.82795
0.41678
0.12602
0.55112
0.31352
0.92432
0.80060
0.43745
0.34917
1.26795
1.31864
1.00726
0.89906
0.99864
99
Percent
Confidence Intervals
Lou Limit
0.01201
0.63240
1.90322
0.37909
1.17525
0.96656
1.06702
1.59297
1.08341
0.90282
0.87466
0.75908
0.97788
0.05088
0.22952
0.19112
0.04275
0.02806
0.03646
0.32253
0.04443
•
0.13875
0.14752
0.19276
0.07351
0.00654
0.15658
0.01983
0.26919
0.15432
0.08878
0.07467
0.79729
0.64219
0.35312
0.53199
0.59366
Upper Limit
0.0976
1.7694
3.6746
0.7871
4.2085
1.8923
2.1635
3.1381
1.6259
1.2698
1.7671
1.5940
1.3191
1.6083
3.0372
6.9604
3.0260
1.4800
2.3477
3.8166
22.6108
•
2.5730
3.0030
3.5562
2.3631
2.4297
1.9398
4.9557
3.1739
4.1534
2.1555
1.6328
2.0164
2.7076
2.8731
1.5194
1.6799
Concordant = 77.6X
Discordant = 22.2X
Tied = 0.3X
(540132 pairs)
NOTE:
Somers1 D = 0.554
Gamma = 0.556
Tau-a = 0.202
c = 0.777
Parameter estimates marked with '#' are regarded to be infinite.
E-56
-------
Response Variable: Dry Air
Number of Observations: 1809
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2383.609
2389.109
2381.609
Intercept
and
Covariates
2221.669
2326.179
2183.669
Chi-Square for Covariates
197.940 with 18 DF
188.849 with 18 DF (p»0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Heater
New Furniture
Lights Too Dim
Glare
Noise
Oust
Role Conflict
Role Clarity
Allergy to Molds
Chem. Sensitivity
Discomfort Index
AHU16
AHU9
G1LG1
NEG1NEG2
UT2FL13
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-2.4203
-0.0123
-0.0870
0.5013
0.2453
0.3034
0.8401
0.0825
0.0819
0.3452
0.1448
0.1275
1.4609
0.6871
-0.3641
-0.8210
0.4673
0.7667
0.2999
Standard
Error
0.3301
0.1248
0.1051
0.1885
0.1079
0.1087
0.1132
0.0803
0.0585
0.1156
0.1115
0.0450
0.5102
0.3938
0.8803
0.7998
0.3404
0.1398
0.1199
Uald
Chi-Sq
53.7650
0.0097
0.6862
7.0690
5.1688
7.7930
55.1153
1.0535
1.9597
8.9243
1.6872
8.0162
8.1983
3.0440
0.1711
1.0537
1.8849
30.0819
6.2523
Pr >
Chi-Sq
0.0001
0.9217
0.4074
0.0078
0.0230
0.0052
0.0001
0.3047
0.1615
0.0028
0.1940
0.0046
0.0042
0.0810
0.6791
0.3047
0.1698
0.0001
0.0124
Odds
Ratios
0.08889
0.98778
0.91668
1.65087
1.27800
1.35446
2.31660
1.08600
1.08535
1.41227
1.15581
1.13598
4.30984
1.98794
0.69482
0.43999
1.59568
2.15265
1.34972
99
Percent
Confidence Intervals
Low Limit
0.03798
0.71621
0.69926
1.01585
0.96788
1.02366
1.73065
0.88307
0.93352
1.04856
0.86725
1.01165
1.15792
0.72085
0.07195
0.05606
0.66394
1.50167
0.99108
Upper Limit
0.2081
1.3623
1.2017
2.6828
1.6875
1.7921
3.1009
1.3356
1 .2619
1.9022
1.5404
1.2756
16.0414
5.4823
6.7097
3.4531
3.8350
3.0858
1.8382
Concordant = 69.1X
Discordant « 30.6X
Tied - 0.3X
(761714 pairs)
Somers1
Gamma
Tau-a
c
D *
0.385
0.386
0.179
0.693
E-57
-------
Response Variable: Dry Air
Number of Observations: 1799
Sex: Males
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2051.756
2057.251
2049.756
Intercept
and
Covariates
1846.398
1950.803
1808.398
Chi-Square for Covariates
241.358 with 18 DF (p»0.0001)
238.787 with 18 DF (p-0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Heater
New Furniture
Lights Too Dim
Glare
Noise
Dust
Role Conflict
Role Clarity
Allergy to Molds
Chem. Sensitivity
Discomfort Index
AHU16
AHU9
G1LG1
NEG1NEG2
WT2FL13
Private Office
Cubicle or Shared
Parameter
Estimate
-2.7627
0.4116
0.1575
0.7682
0.3171
0.3891
0.8146
0.3511
-0.0221
0.2737
0.2788
0.1584
2.2871
0.8933
-37.3187 #
-1.6586
0.5346
-0.0509
-0.1461
Standard
Error
0.3902
0.1867
0.1205
0.2399
0.1201
0.1266
0.1208
0.0935
0.0704
0.1255
0.1319
0.0542
0.4728
0.4266
m
1.0504
0.5558
0.1830
0.1834
Uald
Chi-Sq
50.1284
4.8604
1.7071
10.2534
6.9645
9.4415
45.4660
14.0893
0.0986
4.7554
4.4704
8.5551
23.4034
4.3849
2.4934
0.9250
0.0774
0.6347
Association of Predicted Probabilities and Observed Responses
Concordant = 73.1X
Discordant = 26.5X
Tied = 0.3X
(617694 pairs)
NOTE:
Somens'
Gamna
Tau-a
c
= 0.466
= 0.467
= 0.178
= 0.733
Pr >
Chi-Sq
0.0001
0.0275
0.1914
0.0014
0.0083
0.0021
0.0001
0.0002
0.7535
0.0292
0.0345
0.0034
0.0001
0.0363
0.1143
0.3362
0.7809
0.4256
Odds
Ratios
0.06312
1.50923
1.17058
2.15588
1.37314
1.47565
2.25827
1.42063
0.97814
1.31482
1.32154
1.17163
9.84634
2.44318
0.19041
1.70677
0.95037
0.86407
99 Percent
Confidence Intervals
Low Limit Upper Limit
0.02310
0.93301
0.85821
1.16209
1.00775
1.06500
1.65437
1.11655
0.81591
0.95162
0.94085
1.01896
2.91295
0.81414
0.1725
2.4413
1.5967
3.9996
1.8710
2.0446
3.0826
1.8075
1.1726
1.8166
1.8563
1.3472
33.2825
7.3318
0.01272
0.40773
0.59315
0.53873
2.8497
7.1445
1.5227
1.3859
Parameter estimates marked with '#' are regarded to be infinite.
E-58
-------
Response Variable: Cold Drafty Air
Number of Observations: 1764
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2436.300
2441.776
2434.300
•
Intercept
and
Covar fates
2203.251
2411.314
2127.251
•
Chi-Square for Covariates
307.050 with 37 DF (p=0.0001)
278.424 with 37 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Fabric Partitions
Laser Printer Nearby
Hours at VDT
Use Fan
Use Heater
New Furniture
Water Leaks Nearby
Window Nearby
Noise
Oust
High Pay
Role Conflict
Workload
Allergy to Dust
Allergy: Other
Nonsmoker
AHU14
AHU16
AHU3
AHU9
NE32
NEG3
AWU15SW2
CM FL2
CM FL7
CM FL8
CM FL10
CM FL11
FCFL6
FCFL7
FCNOFL
ET2FL67
WT1FL13
WT1FL67
WT1F1011
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-0.87410
-0.25480
-0.17120
0.02830
-0.43400
1.09000
0.00901
0.33010
0.12580
0.43280
0.01260
-0.25280
0.24160
0.05040
0.24710
-0.10860
0.08600
2.47850
0.24960
-0.50630
1.48050
0.87760
0.81060
0.51850
-1.42600
-1.29960
-1.49800
-1.88520
-0.66790
-1.55670
-2.03640
-1.22750
-0.42320
-0.47810
1.08110
-0.02690
-0.37270
-0.14400
Standard
Error
0.2875
0.1168
0.1138
0.0248
0.1125
0.1348
0.1084
0.1603
0.1147
0.1112
0.1142
0.1317
0.0836
0.0620
0.1111
0.1733
0.1428
0.7950
0.4926
0.7739
0.4810
0.4261
0.5706
0.4187
0.4644
0.4202
0.5526
0.5564
0.4406
0.5607
0.5512
0.4122
0.4877
0.4938
0.5336
0.3639
0.1699
0.1306
Wald
Chi-Sq
9.2432
4.7600
2.2627
1.2998
14.8828
65.4088
0.0069
4.2405
1.2033
15.1473
0.0122
3.6833
8.3503
0.6599
4.9448
0.3928
0.3630
9.7198
0.2567
0.4280
9.4714
4.2429
2.0185
1.5333
9.4285
9.5650
7.3485
11.4783
2.2979
7.7092
13.6472
8.8686
0.7530
0.9375
4.1057
0.0055
4.8130
1.2150
Pr >
Chi-Sq
0.0024
0.0291
0.1325
0.2543
0.0001
0.0001
0.9338
0.0395
0.2727
0.0001
0.9120
0.0550
0.0039
0.4166
0.0262
0.5308
0.5468
0.0018
0.6124
0.5130
0.0021
0.0394
0.1554
0.2156
0.0021
0.0020
0.0067
0.0007
0.1295
0.0055
0.0002
0.0029
0.3855
0.3329
0.0427
0.9411
0.0282
0.2703
Odds
Ratios
0.4172
0.7751
0.8427
1.0287
0.6479
2.9743
.0091
.3911
.1341
.5416
.0127
0.7766
.2733
.0517
.2803
0.8971
1.0898
11.9234
1.2835
0.6027
4.3951
2.4051
2.2493
1.6795
0.2403
0.2726
0.2236
0.1518
0.5128
0.2108
0.1305
0.2930
0.6549
0.6200
2.9479
0.9735
0.6889
0.8659
99
Percent
Confidence Intervals
Low Limit
0.19895
0.57368
0.62854
0.96504
0.48491
2.10172
0.76321
0.92051
0.84395
1.15760
0.75459
0.55319
1.02659
0.89645
0.96166
0.57406
0.75438
1.53815
0.36083
0.08210
1.27309
0.80249
0.51723
0.57117
0.07264
0.09236
0.05385
0.03621
0.16482
0.04973
0.03155
0.10133
0.18646
0.17375
0.74568
0.38125
0.44470
0.61852
Upper Limit
0.8750
1.0472
1.1297
1.0966
0.8657
4.2091
1.3341
2.1023
1.5239
2.0529
1.3590
1.0903
1.5793
1.2338
1.7045
1.4019
1.5744
92.4271
4.5655
4.4250
15.1736
7.2083
9.7812
4.9385
0.7948
0.8048
0.9282
0.6364
1.5953
0.8937
0.5398
0.8473
2.3005
2.2121
11.6540
2.4856
1.0671
1.2122
Concordant * 72.7%
Discordant = 27.1X
Tied - 0.2X
(773024 pairs)
Semens'
Gamma
Tau-a
c
0.456
0.458
0.227
0.728
E-59
-------
Response Variable: Cold Drafty Air
Number of Observations: 1780
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2194.995
2200.479
2192.995
Intercept
and
Covariates
2012.338
2220.744
1936.338
Chi-Square for Covarfates
256.657 with 37 DF (p=0.0001)
231.551 with 37 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Fabric Partitions
Laser Printer Nearby
Hours at VDT
Use Fan
Use Heater
New Furniture
Water Leaks Nearby
Window Nearby
Noise
Dust
High Pay
Role Conflict
Workload
Allergy to Dust
Allergy: Other
Nonsmoker
AHU14
AHU16
AHU3
AHU9
NE32
NEG3
AWU15SW2
CM FL2
CM FL7
CM FL8
CM FL10
CM FL11
FCFL6
FCFL7
FCNOFL
ET2FL67
WT1FL13
WT1FL67
WT1F1011
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-1.4934
0.1085
-0.1161
-0.0380
-0.2981
0.9817
-0.0779
0.0684
0.2999
0.4534
0.3854
0.1950
0.1952
-0.1330
0.1241
-0.1472
0.0486
1.4540
1.3136
-1.8745
1.1439
0.1927
0.1382
0.9500
-1.9816
-1.0463
-1.7205
-0.8379
-1.7828
-1.7751
-38.7898 *
-1.6641
0.6936
1.1617
1.0947
1.7816
0.3224
0.0862
Standard
Error
0.3620
0.1285
0.1267
0.0290
0.1224
0.1799
0.1184
0.1623
0.1210
0.1197
0.1187
0.1545
0.0922
0.0689
0.1156
0.1921
0.1787
0.5505
0.4468
1.0571
0.4411
0.4530
0.6208
0.3668
0.7377
0.4221
0.4386
0.4009
0.7476
0.7477
m
0.5348
0.5872
0.5443
0.3952
0.5931
0.2066
0.1913
Wald
Chi-Sq
17.0206
0.7130
0.8389
1.7234
5.9348
29.7743
0.4327
0.1776
6.1437
14.3453
10.5397
1.5928
4.4821
3.7308
1.1514
0.5871
0.0738
6.9772
8.6452
3.1445
6.7239
0.1811
0.0496
6.7079
7.2155
6.1441
15.3901
4.3688
5.6864
5.6357
•
9.6822
1.3953
4.5558
7.6727
9.0248
2.4338
0.2030
Pr >
Chi-Sq
0.0001
0.3985
0.3597
0.1893
0.0148
0.0001
0.5107
0.6734
0.0132
0.0002
0.0012
0.2069
0.0343
0.0534
0.2833
0.4435
0.7858
0.0083
0.0033
0.0762
0.0095
0.6705
0.8238
0.0096
0.0072
0.0132
0.0001
0.0366
0.0171
0.0176
•
0.0019
0.2375
0.0328
0.0056
0.0027
0.1187
0.6523
Odds
Ratios
0.22461
1.11460
0.89039
0.96271
0.74223
2.66899
0.92506
1.07079
1.34972
1.57365
1.47020
1.21531
1.21555
0.87547
1.13213
0.86312
1.04980
4.28020
3.71954
0.15343
3.13899
1.21252
1.14821
2.58571
0.13785
0.35123
0.17898
0.43262
0.16817
0.16947
.
0.18936
2.00091
3.19536
2.98829
5.93935
1.38044
1.09002
99
Percent
Confidence Intervals
Lou Limit
0.08840
0.80050
0.64244
0.89342
0.54151
1.67913
0.68188
0.70491
0.98827
1.15610
1.08289
0.81628
0.95858
0.73309
0.84056
0.52621
0.66250
1.03656
1.17662
0.01008
1.00766
0.37748
0.23201
1.00514
0.02061
0.11841
0.05783
0.15403
0.02451
0.02469
.
0.04775
0.44086
0.78630
1.07968
1.28888
0.81075
0.66592
Upper Limit
0.5707
1.5520
1.2340
1.0374
1.0174
4.2424
1.2550
1.6266
1.8434
2.1420
1.9961
1.8094
1.5414
1.0455
1.5248
1.4157
1.6635
17.6739
11.7583
2.3363
9.7784
3.8947
5.6824
6.6517
0.9219
1.0419
0.5540
1.2151
1.1538
1.1630
•
0.7509
9.0814
12.9853
8.2708
27.3694
2.3504
1.7842
Concordant =72.1%
Discordant = 27.6X
Tied = 0.3X
(673075 pairs)
Somers1 D
Gamma
Tau-a
c
0.446
0.447
0.189
0.723
NOTE: Parameter estimates marked with '#' are regarded to be infinite.
E-60
-------
Response Variable: Hunid Air
Number of Observations: 1709
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2053.653
2059.096
2051.653
Intercept
and
Covariates
1836.383
1956. 143
1792.383
Chi-Square for Covariates
259.270 with 21 OF (p=0.0001)
250.096 with 21 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Fan
Noise
Dust
Medium Pay
High Pay
Role Conflict
External Stress
Allergy: Other
Chem. Sensitivity
Nonsmoker
Discomfort Index
AHU10
AHU2
AHU20
AHU21
AHU6
G1LG1
CM FL7
Tackd Carpet, Room
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-1.6673
0.6817
0.0278
0.4589
-0.7749
-1.2377
0.1490
0.0875
-0.1665
0.4576
-0.4524
0.0545
3.2910
0.4113
0.8684
1.8274
0.3078
1.6259
0.5532
0.3894
-0.2582
-0.3105
Standard
Error
0.4481
0.1172
0.1235
0.1268
0.1486
0.1739
0.0847
0.0478
0.1987
0.1223
0.1478
0.0501
1.1214
0.5237
0.6103
0.7001
0.2903
0.9350
0.3795
0.1947
0.1921
0.1411
Wald
Chi -So;
13.8424
33.8204
0.0507
13.0896
27.1774
50.6647
3.0965
3.3505
0.7022
13.9875
9.3710
1.1837
8.6132
0.6167
2.0249
6.8131
1.1243
3.0242
2.1244
4.0009
1.8059
4.8456
Pr >
Chi-Sq
0.0002
0.0001
0.8219
0.0003
0.0001
0.0001
0.0785
0.0672
0.4020
0.0002
0.0022
0.2766
0.0033
0.4323
0.1547
0.0090
0.2890
0.0820
0.1450
0.0455
0.1790
0.0277
Odds
Ratios
0.1888
1.9772
1.0282
1.5823
0.4607
0.2901
1.1607
1.0914
0.8466
1.5803
0.6361
1.0560
26.8697
1.5088
2.3831
6.2177
1.3604
5.0830
1.7388
1.4761
0.7724
0.7331
99
Percent
Confidence Intervals
Low Limit
0.05951
1.46198
0.74801
1.14141
0.31421
0.18532
0.93315
0.96499
0.50745
1.15322
0.43469
0.92815
1.49522
0.39151
0.49474
1.02423
0.64403
0.45719
0.65417
0.89391
0.47093
0.50968
Upper Limit
0.599
2.674
1.413
2.194
0.676
0.454
1.444
1.234
1.412
2.165
0.931
1.201
482.860
5.814
11.479
37.745
2.874
56.512
4.622
2.437
1.267
1.054
Concordant * 73.8X
Discordant • 25.9X
Tied « 0.3X
(598764 pairs)
Somers1
Gamma
Tau-a
c
D *
0.480
0.481
0.197
0.740
E-61
-------
Response Variable: Humid Air
Number of Observations: 1701
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1801.552
1806.991
1799.552
.
Intercept
and
Covariates
1703.494
1823.151
1659.494
.
Chi-Square for Covariates
140.058 with 21 DF (p=0.0001)
148.349 with 21 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Use Fan
Noise
Dust
Medium Pay
High Pay
Role Conflict
External Stress
Allergy: Other
Chem. Sensitivity
Nonsmoker
Discomfort Index
AHU10
AHU2
AHU20
AHU21
AHU6
G1LG1
CM FL7
Tackd Carpet, Room
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-1.91620
0.64250
0.33090
0.43210
-0.57390
-0.87550
0.10990
0.04520
-0.47750
0.27240
-0.00064
-0.01470
0.85970
1.79310
2.21510
2.22170
0.79500
1.25640
0.36270
0.23900
0.00745
0.19540
Standard
Error
0.5202
0.1238
0.1341
0.1279
0.2673
0.2496
0.0947
0.0545
0.2271
0.1407
0.1935
0.0596
0.6242
0.6703
0.6930
0.6891
0.3461
0.8654
0.3472
0.2190
0.2083
0.1958
Wald
Chi-Sq
13.5712
26.9220
6.0925
11.4168
4.6096
12.2978
1.3453
0.6868
4.4232
3.7490
0.0000
0.0611
1.8968
7.1560
10.2180
10.3944
5.2762
2.1074
1.0917
1.1909
0.0013
0.9962
Pr >
Chi-Sq
0.0002
0.0001
0.0136
0.0007
0.0318
0.0005
0.2461
0.4072
0.0355
0.0528
0.9973
0.8048
0.1684
0.0075
0.0014
0.0013
0.0216
0.1466
0.2961
0.2751
0.9715
0.3182
Odds
Ratios
0.14717
1.90123
1 .39222
1.54049
0.56332
0.41665
1.11617
1.04624
0.62033
1.31311
0.99936
0.98541
2.36245
6.00805
9.16233
9.22300
2.21444
3.51275
1.43720
1.26998
1.00748
1.21580
99
Percent
Confidence Intervals
Low Limit
0.03853
1.38208
0.98556
1.10808
0.28296
0.21905
0.87455
0.90920
0.34559
0.91389
0.60708
0.84516
0.47320
1.06866
1.53715
1.56296
0.90796
0.37800
0.58761
0.72242
0.58912
0.73419
Upper Limit
0.5620
2.6154
1.9667
2.1416
1.1215
0.7925
1 .4245
1.2039
1.1135
1.8867
1.6451
1.1489
11.7946
33.7774
54.6128
54.4249
5.4008
32.6443
3.5152
2.2325
1.7229
2.0133
Concordant = 67.8X
Discordant = 31.6X
Tied = 0.5X
(499148 pairs)
Sorters'
Gamma
Tau-a
c
0.362
0.364
0.125
0.681
E-62
-------
Response Variable: Odor of Paint, Carpet Cleaning
Number of Observations: 1834
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
2278.368
2283.882
2276.368
Intercept
and
Covad ates
2040.555
2255.611
1962.555
•
Chi-Square for Covariates
313.812 with 38 OF (p-0.0001)
297.955 with 38 DF (p-0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
1 Hour at Copier
Used Chemicals
Walls Painted
Walls Rearranged
Water Leaks Nearby
Window Nearby
Glare
Noise
Dust
Role Conflict
Role Clarity
Wear Glasses
Allergy to Molds
Chem. Sensitivity
AHU16
AHU5
NE1
NE31
NE32
AWU15SW2
G1LG1
CM FL2
CM FL8
FCFL2
ET1FL45
ET1FL67
ET1FL89
ET1F1011
ET2FL67
ET2FL89
ET2F1011
WT1FL67
WT2FL89
WT2F1011
ET3FL47
ET3FL811
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-2.55430
-0.18640
0.70720
0.81780
0.37320
0.15230
-0.24850
0.21130
0.35400
0.73690
0.14760
0.06940
-0.03170
0.11840
0.67490
-0.41420
0.09420
0.22860
-0.89480
-0.36390
1.77420
-0.50210
-0.22280
0.08140
-0.92550
-1.18880
-0.39610
-0.49640
-0.74370
0.26780
-1.32680
0.10210
0.23120
0.38250
-0.10870
-1.23490
-0.91240
0.00182
-0.08290
Standard
Error
0.3693
0.1245
0.1293
0.1402
0.1489
0.1679
0.1187
0.1162
0.1176
0.1227
0.0849
0.0629
0.1129
0.1255
0.1187
0.5299
0.3649
0.3301
0.6761
0.4339
0.4578
0.9439
0.4258
0.4337
0.5990
0.5868
0.4270
0.4700
0.4263
0.4850
0.7660
0.4080
0.4888
0.3622
0.4142
0.7773
0.6482
0.1533
0.1310
Wald
Chi-Sq
47.8435
2.2433
29.8997
34.0470
6.2835
0.8225
4.3848
3.3049
9.0559
36.0865
3.0201
1.2172
0.0789
0.8903
32.3100
0.6111
0.0666
0.4795
1.7519
0.7034
15.0185
0.2830
0.2739
0.0352
2.3872
4.1047
0.8605
1.1157
3.0430
0.3050
3.0008
0.0627
0.2237
1.1150
0.0689
2.5241
1.9813
0.0001
0.4004
Pr >
Chi-Sq
0.0001
0.1342
0.0001
0.0001
0.0122
0.3644
0.0363
0.0691
0.0026
0.0001
0.0822
0.2699
0.7789
0.3454
0.0001
0.4344
0.7963
0.4886
0.1856
0.4016
0.0001
0.5947
0.6008
0.8512
0.1223
0.0428
0.3536
0.2908
0.0811
0.5808
0.0832
0.8024
0.6362
0.2910
0.7930
0.1121
0.1593
0.9906
0.5269
Odds
Ratios
0.07775
0.82994
2.02830
2.26551
1.45237
1.16451
0.77997
1.23528
1.42476
2.08945
1.15905
1.07186
0.96880
1.12569
1.96384
0.66087
1.09878
1.25684
0.40869
0.69496
5.89556
0.60526
0.80027
1.08480
0.39633
0.30459
0.67294
0.60872
0.47535
1.30709
0.26532
1.10749
1.26011
1.46594
0.89700
0.29086
0.40156
1.00182
0.92044
99
Percent
Confidence Intervals
Low Limit
0.03003
0.60223
1.45372
1.57877
0.98969
0.75562
0.57449
0.91573
1.05239
1.52322
0.93137
0.91153
0.72431
0.81474
1.44647
0.16877
0.42922
0.53701
0.07162
0.22727
1.81286
0.05321
0.26723
0.35494
0.08471
0.06718
0.22401
0.18139
0.15852
0.37473
0.03688
0.38716
0.35774
0.57665
0.30861
0.03927
0.07561
0.67497
0.65681
Upper Limit
0.2013
1.1437
2.8300
3.2510
2.1314
1.7947
1.0589
1.6663
1 .9289
2.8662
1.4424
1.2604
1.2958
1.5553
2.6662
2.5878
2.8128
2.9416
2.3323
2.1251
19.1728
6.8853
2.3966
3.3155
1.8543
1.3810
2.0215
2.0428
1 .4254
4.5593
1.9087
3.1680
4.4386
3.7267
2.6072
2.1542
2.1326
1.4869
1.2899
Concordant » 74.9X
Discordant • 24.8X
Tied « 0.3X
(721864 pairs)
Somers1 D
Gamma
Tau-a
c
0.500
0.502
0.215
0.750
E-63
-------
Response Variable: Odor of Paint, Carpet Cleaning
Nunber of Observations: 1788
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1817.195
1822.683
1815.195
Intercept
and
Cover iates
1586.689
1800.754
1508.689
Chi-Square for Covariates
306.506 with 38 Of (p=0.0001)
299.092 with 38 DF (p-0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
1 Hour at Copier
Used Chemicals
Walls Painted
Walls Rearranged
Water Leaks Nearby
Window Nearby
Glare
Noise
Dust
Role Conflict
Role Clarity
Wear Glasses
Allergy to Molds
Chem. Sensitivity
AHU16
AHU5
NE1
NE31
NE32
AWU15SW2
G1LG1
CM FL2
CM FL8
FCFL2
ET1FL45
ET1FL67
ET1FL89
ET1F1011
ET2FL67
ET2FL89
ET2F1011
WT1FL67
WT2FL89
WT2F1011
ET3FL47
ET3FL811
Private Office
Cubicle or Shared
Association of Predicted Probabilities and Observed Responses
Parameter
Estimate
-1.7999
0.0824
0.5170
0.8961
0.2067
0.3818
-0.4968
0.5277
0.5025
0.5732
0.2037
-0.1969
-0.3515
0.2776
0.5786
0.7151
-0.3868
-0.6832
-0.4286
-0.1871
1.2885
1.4872
-0.7790
-0.3218
-0.4178
0.0796
-0.7195
-1.0015
-0.9658
-1.1504
-0.3851
0.4042
-0.5633
-0.2766
-0.6758
-0.4058
-1.7077
-0.1558
-0.3436
Standard
Error
0.4512
0.1539
0.1895
0.1696
0.1792
0.1847
0.1422
0.1349
0.1432
0.1354
0.1034
0.0781
0.1350
0.1411
0.1449
0.4403
0.4021
0.5137
0.5330
0.6499
0.4024
0.8964
0.5739
0.3697
0.5597
0.5479
0.6468
0.7798
0.6452
1.0641
0.5859
0.4546
0.5322
0.7806
0.7821
0.4921
0.6242
0.1994
0.2002
Wald
Chi-Sq
15.9138
0.2869
7.4399
27.9324
1.3314
4.2747
12.2048
15.3025
12.3105
17.9317
3.8787
6.3524
6.7818
3.8680
15.9493
2.6376
0.9254
1.7686
0.6466
0.0829
10.2503
2.7525
1.8429
0.7575
0.5574
0.0211
1.2376
1.6496
2.2407
1.1688
0.4320
0.7903
1.1205
0.1256
0.7467
0.6798
7.4856
0.6103
2.9443
Pr >
Chi-Sq
0.0001
0.5922
0.0064
0.0001
0.2486
0.0387
0.0005
0.0001
0.0005
0.0001
0.0489
0.0117
0.0092
0.0492
0.0001
0.1044
0.3361
0.1836
0.4213
0.7734
0.0014
0.0971
0.1746
0.3841
0.4553
0.8845
0.2659
0.1990
0.1344
0.2796
0.5110
0.3740
0.2898
0.7230
0.3875
0.4096
0.0062
0.4347
0.0862
Odds
Ratios
0.16532
1.08589
1.67699
2.45003
1.22961
1.46492
0.60847
1.69503
1.65285
1.77393
1.22593
0.82127
0.70363
1.31996
1.78354
2.04439
0.67923
0.50500
0.65142
0.82936
3.62734
4.42469
0.45886
0.72484
0.65849
1.08285
0.48700
0.36733
0.38068
0.31651
0.68038
1.49810
0.56933
0.75836
0.50875
0.66644
0.18128
0.85573
0.70921
99
Percent
Confidence Intervals
Low Limit
0.05171
0.73049
1.02927
1.58282
0.77498
0.91029
0.42185
1.19745
1.14295
1.25158
0.93926
0.67160
0.49695
0.91771
1.22794
0.65763
0.24108
0.13446
0.16503
0.15548
1.28649
0.43958
0.10463
0.27967
0.15574
0.26400
0.09203
0.04928
0.07223
0.02041
0.15041
0.46447
0.14453
0.10153
0.06785
0.18760
0.03631
0.51199
0.42345
Upper Limit
0.5286
1.6142
2.7323
3.7924
1.9510
2.3575
0.8777
2.3994
2.3902
2.5143
1.6001
1.0043
0.9963
1.8985
2.5905
6.3555
1.9136
1.8967
2.5713
4.4240
10.2275
44.5373
2.0125
1.8786
2.7843
4.4415
2.5771
2.7381
2.0062
4.9073
3.0777
4.8319
2.2426
5.6645
3.8148
2.3675
0.9051
1 .4303
1.1878
Concordant = 77.8X
Discordant = 22.OX
Tied - 0.3X
(521507 pairs)
Somers1
Gamma
Tau-a
c
D = 0.558
= 0.560
= 0.182
» 0.779
E-64
-------
Response Variable: Odor of Cosmetics, Body Odor, Food
Number of Observations: 1739
ex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
•2 LOG L
Score
Intercept
Only
1969.253
1974.714
1967.253
Intercept
and
Covariates
1769.311
1922.221
1713.311
Chi-Square for Covariates
253.942 with 27 OF (p=0.0001)
237.856 with 27 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at Work
Photocopier Nearby
Plants Nearby
Use Fan
Use Lamp
Water Leaks Nearby
Window Nearby
Glare
Noise
Dust
Medium Pay
High Pay
Role Conflict
College Grad
Eczema
Under 30
AHU12
AHU21
AHU3
AHU5
ETB
CM FL2
CM FL7
ET1F1011
ET3FL811
Private Office
Cubicle or Shared
Parameter
Estimate
-1.4691
-0.0795
0.3223
0.0338
0.3887
-0.0366
0.1503
-0.2023
0.4287
0.2821
0.3527
-0.3978
-0.6863
0.4548
-0.6570
-0.1873
0.2334
-2.2270
-38.8452 *
0.1708
0.5296
-1.4776
0.9369
0.8179
-0.3901
-0.5140
-0.2059
-0.0616
Standard
Error
0.3632
0.0406
0.1604
0.1239
0.1276
0.1279
0.1782
0.1327
0.1256
0.1295
0.1305
0.1656
0.2198
0.0857
0.1598
0.2221
0.1473
1.0796
.
0.8256
0.3993
1.0686
0.3863
0.3936
0.4589
0.6740
0.2043
0.1466
Wald
Chi-Sq
16.3638
3.8283
4.0370
0.0744
9.2776
0.0819
0.7117
2.3237
11.6494
4.7457
7.3064
5.7714
9.7491
28.1651
16.8935
0.7111
2.5122
4.2556
•
0.0428
1.7593
1.9122
5.8820
4.3166
0.7225
0.5815
1.0163
0.1765
Pr >
Chi-Sq
0.0001
0.0504
0.0445
0.7850
0.0023
0.7747
0.3989
0.1274
0.0006
0.0294
0.0069
0.0163
0.0018
0.0001
0.0001
0.3991
0.1130
0.0391
•
0.8361
0.1847
0.1667
0.0153
0.0377
0.3953
0.4457
0.3134
0.6744
Odds
Ratios
0.23013
0.92358
1.38030
1.03438
1.47506
0.96406
1.16218
0.81685
1.53526
1.32591
1.42290
0.67180
0.50344
1.57586
0.51840
0.82919
1.26289
0.10785
.
1.18625
1.69825
0.22818
2.55206
2.26574
0.67699
0.59810
0.81391
0.94026
99
Percent
Confidence Intervals
Low Limit
0.09029
0.83186
0.91312
0.75174
1.06184
0.69345
0.73437
0.58034
1.11088
0.94981
1.01667
0.43850
0.28579
1.26369
0.34347
0.46793
0.86412
0.00668
m
0.14143
0.60714
0.01455
0.94345
0.82200
0.20758
0.10538
0.48086
0.64453
Upper Limit
0.58655
1 .02540
2.08650
1.42328
2.04909
1.34027
1.83922
1.14974
2.12176
1.85094
1.99146
1.02920
0.88684
1.96514
0.78243
1.46936
1.84567
1.74029
.
9.94973
4.75025
3.57911
6.90336
6.24520
2.20787
3.39473
1.37765
1 .37169
Association of Predicted Probabilities and Observed Responses
Concordant = 74.3X
Discordant = 25.4X
Tied = 0.3X
(571560 pairs)
NOTE:
S oners'
Gamma
Tau-a
c
D =
0.489
0.491
0.185
0.745
Parameter estimates marked with '#' are regarded to be infinite.
E-65
-------
Response Variable: Odor of Cosmetics, Body Odor, Food
Number of Observations: 1762
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1355.773
1361.248
1353.773
Intercept
and
Covariates
1269.129
1422.407
1213.129
Chi-Square for Covariates
140.645 with 27 Of (p=0.0001)
141.348 with 27 DF (p-0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at Work
Photocopier Nearby
Plants Nearby
Use Fan
Use Lamp
Water Leaks Nearby
Window Nearby
Glare
Noise
Dust
Medium Pay
High Pay
Role Conflict
College Grad
Eczema
Under 30
AHU12
AHU21
AHU3
AHU5
ETB
CM FL2
CM FL7
ET1F1011
ET3FL811
Private Office
Cubicle or Shared
Parameter
Estimate
-2.1670
-0.0306
0.2486
0.2403
•0.0619
0.3211
0.6036
-0.0867
0.3577
0.4928
0.6936
-0.4186
-0.8489
0.3715
-0.5101
0.1425
-0.1674
-37.6905 #
-0.7523
1.3876
-0.7833
-36.5162 #
0.6364
0.3728
-0.4343
-0.0153
-0.2600
-0.0666
Standard
Error
0.5076
0.0463
0.2212
0.1551
0.1592
0.1583
0.1944
0.1620
0.1547
0.1694
0.1579
0.3128
0.3204
0.1102
0.2881
0.2598
0.2827
m
1.0814
0.6571
0.6134
B
0.4669
0.4158
0.6361
0.4680
0.2433
0.2216
Wald
Chi -So;
18.2219
0.4372
1.2624
2.3995
0.1511
4.1149
9.6369
0.2866
5.3477
8.4631
19.2939
1.7902
7.0206
11.3734
3.1348
0.3009
0.3507
•
0.4839
4.4598
1.6310
•
1.8580
0.8038
0.4662
0.0011
1.1412
0.0902
Pr >
Chl-Sq
0.0001
0.5085
0.2612
0.1214
0.6974
0.0425
0.0019
0.5924
0.0207
0.0036
0.0001
0.1809
0.0081
0.0007
0.0766
0.5833
0.5537
.
0.4866
0.0347
0.2016
m
0.1729
0.3700
0.4947
0.9739
0.2854
0.7639
Odds
Ratios
0.11452
0.96986
1.28223
1.27163
0.93998
1.37864
1.82869
0.91695
1.43004
1.63689
2.00091
0.65797
0.42789
1.44991
0.60044
1.15315
0.84586
•
0.47128
4.00523
0.45690
m
1.88967
1.45179
0.64772
0.98482
0.77105
0.93557
99
Percent
Confidence Intervals
Low Limit
0.03097
0.86082
0.72527
0.85279
0.62375
0.91697
1.10830
0.60410
0.96001
1.05805
1 .33222
0.29394
0.18745
1.09158
0.28586
0.59052
0.40835
•
0.02907
0.73706
0.09410
•
0.56760
0.49743
0.12582
0.29497
0.41200
0.52864
Upper Limit
0.4234
1.0927
2.2669
1.8962
1 .4165
2.0728
3.0173
1.3918
2.1302
2.5324
3.0052
1.4728
0.9767
1 .9259
1.2612
2.2518
1.7521
•
7.6399
21.7647
2.2185
•
6.2911
4.2372
3.3344
3.2880
1.4430
1 .6557
Association of Predicted Probabilities and Observed Responses
Concordant 72.9X
Discordant - 26.6%
Tied = 0.6X
(348445 pairs)
NOTE:
Somers' D - 0.463
Gamma = 0.465
Tau-a - 0.104
c = 0.731
Parameter estimates marked with '#' are regarded to be infinite.
E-66
-------
Response Variable: Odor from Photocopies, Printing
Number of Observations: 1749
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1187.150
1192.617
1185.150
Intercept
and
Covariates
1058.861
1299.400
970.861
•
Chi-Square for Covariates
214.289 with 43 DF
Chi-Sq
0.0001
0.5608
0.2963
0.0001
0.9730
0.0062
0.0620
0.3778
0.9170
0.8014
0.0001
0.0047
0.4007
0.0158
0.0009
0.0286
0.0795
0.6149
*
•
a
0.5102
0.2177
0.6560
0.6225
0.5427
0.9756
0.0419
0.1436
0.8901
0.9303
0.2586
0.8357
0.4720
0.3574
•
0.9139
0.6455
0.5815
0.3986
0.2083
0.6673
0.0055
0.1874
Odds
Ratios
0.03866
1.05475
0.82152
4.10005
1.00602
0.60242
1 .42661
0.85078
0.98138
1.04707
2.85337
1 .37851
0.86424
1.78265
1.80381
1.81031
0.68557
0.62481
.
.
•
0.49169
0.27557
1.26377
1.50652
0.49728
0.98255
6.55547
2.35279
0.85667
1.06951
0.30298
1.12098
0.61656
0.37516
•
1.06791
0.74416
1 .42761
0.38240
1.51513
1.16125
0.47260
0.76982
99
Percent
Confidence Intervals
Low Limit
0.01240
0.83305
0.50578
2.52229
0.63717
0.37376
0.87357
0.53072
0.61646
0.65333
1 .65565
1.02877
0.55262
0.96165
1.14098
0.90046
0.39372
0.05621
.
.
^
0.03060
0.01864
0.32641
0.17640
0.02587
0.22300
0.60626
0.52134
0.04797
0.14779
0.01990
0.27133
0.10908
0.02413
•
0.22302
0.14226
0.27047
0.02035
0.64720
0.47418
0.23568
0.46177
Upper Limit
0.1205
1.3354
1.3344
6.6648
1.5884
0.9710
2.3298
1.3639
1.5623
1.6781
4.9175
1.8471
1.3516
3.3046
2.8517
3.6395
1.1938
6.9449
•
m
f
7.9013
4.0738
4.8929
12.8659
9.5578
4.3293
70.8836
10.6181
15.2999
7.7395
4.6123
4.6311
3.4851
5.8331
,
5.1136
3.8926
7.5352
7.1850
3.5470
2.8439
0.9477
1.2834
E-67
-------
Association of Predicted Probabilities and Observed Responses
Concordant * 78.9X Somers' 0 * 0.583
Discordant = 20.6X Gamma * 0.586
Tied * 0.5X Tau-a » 0.111
(290718 pairs) c - 0.791
NOTE: Parameter estimates marked with 'Id1 are regarded to be infinite.
E-68
-------
Response Variable: Odor from Photocopies, Printing
Nuaber of Observations: 1724
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1016.374
1021.827
1014.374
•
Intercept
and
Covariates
873.950
1113.856
785.950
Chi-Square for Covariates
228.424 with 43 DF (p=0.0001)
229.242 with 43 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Wood Index
Fabric Partitions
Photocopier Nearby
Plants Nearby
1 Hour at Copier
Use Heater
Window Nearby
Glare
Noise
Oust
Role Conflict
Wear Glasses
Allergy: Other
Chem. Sensitivity
Eczema
Nonsmoker
AHU14
AHU16
AHU18
AHU20
AHU4
AHU8
NE1
NEG3
NEG4
SE32
G1LG1
FCFL2
FCFL3
FCFL7
ET1FL45
ET1F1011
ET2FL13
ET2FL45
ET2FL67
ET2F1011
WT1FL45
WT1FL67
ET3FL811
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Parameter
Estimate
-2.9061
-0.1615
-0.2807
1.4827
0.5035
0.4646
0.5100
-0.6493
0.6318
0.3580
0.7431
0.5459
-0.5828
0.1657
0.5282
0.1776
-0.9105
-37.5719 #
-0.6037
-1.3685
-36.2326 *
-37.9683 *
-37.5594 #
-1.2772
-39.2688 *
-37.7690 #
-1.6674
0.9214
-0.5394
-37.4971 #
-0.8847
-0.4670
-1.1820
-0.6600
-37.4315 #
-1.1358
-1.3705
-0.5327
-0.9272
-0.7257
0.0246
0.9532
-0.5817
-0.3969
Standard
Error
0.5425
0.0963
0.2137
0.2311
0.2028
0.2425
0.2769
0.2090
0.2000
0.2187
0.2066
0.1361
0.1952
0.2912
0.2093
0.3162
0.2672
.
0.9103
1.0502
.
•
•
1.0412
m
m
1.1224
1.0927
0.7933
•
1.0569
0.8191
0.8390
0.6525
.
1.1443
0.8516
1.0545
0.7816
0.6359
0.4613
0.3981
0.2802
0.2673
Uald
Chi-Sq
28.6938
2.8147
1.7243
41.1675
6.1632
3.6717
3.3936
9.6534
9.9785
2.6801
12.9340
16.0965
8.9146
0.3240
6.3719
0.3157
11.6085
•
0.4398
1.6979
•
.
*
1.5049
.
.
2.2068
0.7110
0.4624
.
0.7007
0.3251
1.9848
1.0232
.
0.9853
2.5900
0.2552
1.4073
1 .3021
0.0028
5.7319
4.3087
2.2054
Pr >
Chi-Sq
0.0001
0.0934
0.1891
0.0001
0.0130
0.0553
0.0655
0.0019
0.0016
0.1016
0.0003
0.0001
0.0028
0.5692
0.0116
0.5742
0.0007
*
0.5072
0.1926
•
.
*
0.2199
.
.
0.1374
0.3991
0.4965
•
0.4026
0.5686
0.1589
0.3118
•
0.3209
0.1075
0.6134
0.2355
0.2538
0.9574
0.0167
0.0379
0.1375
Odds
Ratios
0.05469
0.85087
0.75525
4.40482
.65450
.59138
.66529
0.52241
.88099
.43047
2.10244
1.72616
0.55833
1.18022
1.69588
1.19435
0.40232
•
0.54678
0.25449
*
.
B
0.27882
.
•
0.18874
2.51281
0.58310
•
0.41284
0.62688
0.30666
0.51685
.
0.32117
0.25398
0.58702
0.39566
0.48399
1.02491
2.59400
0.55895
0.67240
99
Percent
Confidence Intervals
Lou Limit
0.01352
0.66394
0.43553
2.42877
0.98126
0.85208
0.81604
0.30493
1.12367
0.81434
1.23479
1.21568
0.33769
0.55742
0.98910
0.52891
0.20214
»
0.05241
0.01701
•
.
.
0.01908
.
•
0.01048
0.15056
0.07555
•
0.02713
0.07600
0.03532
0.09625
•
0.01685
0.02832
0.03881
0.05283
0.09406
0.31233
0.93025
0.27158
0.33774
Upper Limit
0.2212
1.0904
1.3097
7.9886
2.7896
2.9721
3.3984
0.8950
3.1487
2.5127
3.5798
2.4510
0.9231
2.4989
2.9077
2.6970
0.8008
•
5.7044
3.8069
•
.
•
4.0752
.
*
3.4004
41.9381
4.5003
•
6.2832
5.1707
2.6625
2.7755
•
6.1222
2.2778
8.8790
2.9630
2.4902
3.3633
7.2334
1.1504
1.3387
E-69
-------
Association of Predicted Probabilities and Observed Responses
Concordant = 83.6X Somers1 D = 0.676
Discordant = 16.OX Gairma = 0.679
Tied = 0.4X Tau-a = 0.107
(234675 pairs) c = 0.838
NOTE: Parameter estimates marked with '#' are regarded to be infinite.
E-70
-------
Response Variable: Odor from New Carpet or Drapes
Number of Observations: 1766
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1601.937
1607.413
1599.937
Intercept
and
Covariates
1349.835
1678.424
1229.835
Chi-Square for Covariates
370.102 with 59 DF (p=0.0001)
400.402 with 59 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Used Chemicals
New Carpet/Drapes
Walls Painted
Walls Rearranged
Water Leaks Nearby
Dust
Role Conflict
Role Clarity
Wear Glasses
Asthma
Allergy to Pollen
Allergy to Molds
Chem. Sensitivity
Under 30
AHU12
AHU14
AHU2
AHU20
AHU3
AHU8
ETB
NEG3
NEG4
NEG1NEG2
CM FL2
CM FL7
CM FL8
FCFL2
FCFL3
FCFL7
FCFL8
ET1FL13
ET1FL45
ET1FL67
ET1FL89
ET1F1011
ET2FL13
ET2FL45
ET2FL67
ET2F1011
WT1FL13
UT1FL45
WT1FL67
WT1FL89
Parameter
Estimate
-3.5247
0.3312
1.5898
0.4040
0.2469
0.0661
0.5522
0.2678
-0.0205
0.0808
0.4397
-0.0329
0.2027
0.6838
-0.3934
1.1407
0.1251
-0.1917
-0.8747
1.1164
-0.0394
1.1634
-1.0682
-1.1292
-38.2011 #
-1.4896
-1.6409
-1.2318
-0.7411
-0.7527
-1.6143
-0.0321
-0.7025
-0.8178
-1.5990
-0.9331
-0.5228
-0.3393
-0.2906
0.5145
-0.5693
-0.6705
-0.3813
0.5113
-0.8641
Standard
Error
0.4798
0.1765
0.1903
0.1913
0.1974
0.2186
0.1660
0.1104
0.0814
0.1532
0.2436
0.1660
0.1763
0.1620
0.1930
0.5336
0.7695
0.7411
0.8001
0.7871
0.6496
0.6465
0.8181
0.7583
m
1.0354
0.7870
0.7967
0.7777
0.8736
0.7086
0.6195
0.6510
0.7147
0.8580
0.6724
0.5250
0.6732
0.7297
0.6212
0.5903
0.8432
0.6121
0.6685
0.8338
Wald
Chi-Sq
53.9668
3.5191
69.8245
4.4592
1.5643
0.0915
11.0654
5.8856
0.0631
0.2783
3.2575
0.0392
1.3230
17.8278
4.1553
4.5699
0.0264
0.0669
1.1953
2.0117
0.0037
3.2389
1.7048
2.2174
•
2.0698
4.3467
2.3905
0.9081
0.7424
5.1900
0.0027
1.1642
1.3090
3.4727
1.9255
0.9914
0.2541
0.1586
0.6860
0.9301
0.6323
0.3881
0.5850
1.0740
Pr >
Chi-Sq
0.0001
0.0607
0.0001
0.0347
0.2110
0.7623
0.0009
0.0153
0.8016
0.5978
0.0711
0.8430
0.2501
0.0001
0.0415
0.0325
0.8709
0.7959
0.2743
0.1561
0.9516
0.0719
0.1917
0.1365
•
0.1502
0.0371
0.1221
0.3406
0.3889
0.0227
0.9587
0.2806
0.2526
0.0624
0.1653
0.3194
0.6142
0.6905
0.4075
0.3348
0.4265
0.5333
0.4443
0.3000
Odds
Ratios
0.0295
1.3926
4.9028
1.4978
1.2801
1.0683
1.7371
1 .3071
0.9797
1.0842
1.5522
0.9676
1.2247
1.9814
0.6748
3.1290
1.1333
0.8256
0.4170
3.0538
0.9614
3.2008
0.3436
0.3233
•
0.2255
0.1938
0.2918
0.4766
0.4711
0.1990
0.9684
0.4953
0.4414
0.2021
0.3933
0.5929
0.7123
0.7478
1.6728
0.5659
0.5115
0.6830
1.6675
0.4214
99
Percent
Confidence Intervals
Low Limit
0.00856
0.88385
3.00293
0.91504
0.76982
0.60834
1.13268
0.98355
0.79439
0.73063
0.82877
0.63096
0.77767
1.30537
0.41042
0.79148
0.15612
0.12236
0.05309
0.40205
0.18037
0.60533
0.04177
0.04584
•
0.01566
0.02552
0.03747
0.06428
0.04963
0.03208
0.19634
0.09260
0.07003
0.02217
0.06959
0.15332
0.12575
0.11414
0.33766
0.12370
0.05827
0.14113
0.29797
0.04919
Upper Limit
0.101
2.194
8.005
2.452
2.128
1.876
2.664
1.737
1.208
1.609
2.907
1.484
1.929
3.008
1.109
12.370
8.226
5.570
3.275
23.196
5.124
16.925
2.827
2.280
,
3.247
1.472
2.272
3.533
4.471
1.235
4.777
2.650
2.782
1.843
2.223
2.292
4.034
4.899
8.287
2.589
4.489
3.305
9.331
3.610
E-71
-------
WT1F1011
WT2FL45
WT2FL67
WT2FL89
WT2F1011
ET3FL47
ET3FL811
UT3FU7
UT3FL811
Glued Carpet, Room
Glued Carpet, Hall
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
0.4944
0.9679
1.0951
0.4084
1.3826
1.0269
0.3973
39.2120 *
2.7577
0.9010
0.3575
1.3260
0.7890
•0.1401
0.1501
0.6024
0.8266
0.7373
0.6149
0.7486
0.8965
0.7053
•
1.2084
0.2926
0.3830
0.2863
0.4518
0.2141
0.1758
0.6737
1.3711
2.2065
0.4413
3.4109
1.3121
0.3173
•
5.2077
9.4850
0.8713
21.4476
3.0497
0.4284
0.7293
0.4118
0.2416
0.1374
0.5065
0.0648
0.2520
0.5732
•
0.0225
0.0021
0.3506
0.0001
0.0808
0.5128
0.3931
0.6099
0.3799
0.3345
0.6647
0.2509
0.3581
0.6721
m
15.7635
2.46206
1.42975
3.76595
2.20119
0.86927
1.16195
0.12923
0.04517
0.05007
0.13637
0.03648
0.03557
0.10925
2.879
3.194
2.235
3.240
1.726
3.606
4.135
0.70108
1.15866
0.53307
1.80127
0.68740
0.50076
0.73878
354.440
5.23171
3.83476
7.87355
7.04865
1.50896
1.82752
Association of Predicted Probabilities and Observed Responses
Concordant = 81.8X
Discordant = 17.9X
Tied = 0.3X
(436293 pairs)
NOTE:
Somers1 D
Gamma
Tau-a
c
0.639
0.641
0.179
0.820
Parameter estimates marked with '#' are regarded to be infinite.
E-72
-------
Response Variable: Odor from New Carpet or Drapes
Number of Observations: 1737
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1565.639
1571.099
1563.639
Intercept
and
Cover iates
1271.478
1599.073
1151.478
Chi-Square for Covariates
412.161 with 59 DF (p=0.0001)
435.701 with 59 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Used Chemicals
New Carpet/Drapes
Walls Painted
Walls Rearranged
Water Leaks Nearby
Dust
Role Conflict
Role Clarity
Wear Glasses
Asthma
Allergy to Pollen
Allergy to Molds
Chem. Sensitivity
Under 30
AHU12
AHU14
AHU2
AHU20
AHU3
AHU8
ETB
NEG3
NEG4
NEG1NEG2
CM FL2
CM FL7
CM FL8
FCFL2
FCFL3
FCFL7
FCFL8
ET1FL13
ET1FL45
ET1FL67
ET1FL89
ET1F1011
ET2FL13
ET2FL45
ET2FL67
ET2F1011
WT1FL13
WT1FL45
WT1FL67
Parameter
Estimate
-2.3469
0.1622
1.4285
0.6643
0.1033
0.3583
0.9110
0.2297
-0.2612
-0.3148
0.4227
-0.4085
0.3843
0.3271
0.1830
1.2461
-1.3271
0.2369
-0.4487
-0.0889
-37.4838 #
0.9810
- .7105
- .0364
-2.2830
- .8022
- .6112
- .8370
-2.0122
-0.8181
-0.9342
-1.1102
0.5874
-1.2500
-1.2130
-0.8427
-0.5990
-1.1539
-1.5905
-1.2846
-0.6971
-0.7541
-0.0774
-1.7925
Standard
Error
0.5175
0.2414
0.1930
0.2103
0.2203
0.2161
0.1607
0.1205
0.0914
0.1602
0.2566
0.1770
0.1889
0.1787
0.2873
0.6768
0.8872
0.9749
0.7214
0.8182
•
0.8396
1.1054
0.7260
0.8614
1.0407
0.7670
0.7494
1.1065
0.8720
0.6053
0.8679
0.5769
0.8863
0.8363
0.7656
0.5328
0.7512
1.1127
0.9053
0.6174
0.9202
0.7556
1.1026
Wald
Chi-Sq
20.5688
0.4516
54.8079
9.9729
0.2201
2.7503
32.1476
3.6356
8.1609
3.8609
2.7131
5.3278
4.1381
3.3499
0.4056
3.3896
2.2373
0.0591
0.3868
0.0118
•
1.3651
2.3944
2.0382
7.0239
2.9989
4.4132
6.0083
3.3073
0.8802
2.3822
1.6364
1.0366
1.9891
2.1035
1.2116
1.2640
2.3597
2.0434
2.0136
1.2746
0.6716
0.0105
2.6430
Pr >
Chi-Sq
0.0001
0.5016
0.0001
0.0016
0.6390
0.0972
0.0001
0.0566
0.0043
0.0494
0.0995
0.0210
0.0419
0.0672
0.5242
0.0656
0.1347
0.8080
0.5340
0.9135
•
0.2426
0.1218
0.1534
0.0080
0.0833
0.0357
0.0142
0.0690
0.3481
0.1227
0.2008
0.3086
0.1584
0.1470
0.2710
0.2609
0.1245
0.1529
0.1559
0.2589
0.4125
0.9184
0.1040
Odds
Ratios
0.09567
1.17610
4.17244
1 .94313
1.10882
1 .43089
2.48681
1.25822
0.77013
0.72993
1.52608
0.66465
1.46859
1.38694
1.20081
3.47676
0.26525
1.26731
0.63846
0.91494
•
2.66712
0.18078
0.35473
0.10198
0.16494
0.19965
0.15929
0.13369
0.44127
0.39290
0.32949
1.79930
0.28650
0.29730
0.43055
0.54936
0.31540
0.20382
0.27676
0.49803
0.47043
0.92552
0.16654
99
Percent
Confidence Intervals
Low Limit
0.02522
0.63151
2.53789
1.13040
0.62864
0.82006
1.64384
0.92246
0.60857
0.48313
0.78797
0.42128
0.90275
0.87526
0.57288
0.60815
0.02698
0.10285
0.09956
0.11118
•
0.30672
0.01048
0.05466
0.01109
0.01130
0.02768
0.02311
0.00773
0.04668
0.08262
0.03523
0.40710
0.02921
0.03448
0.05991
0.13925
0.04555
0.01160
0.02687
0.10152
0.04396
0.13215
0.00973
Upper Limit
0.3628
2.1903
6.8597
3.3402
1.9558
2.4967
3.7620
1.7162
0.9746
1.1028
2.9556
1.0486
2.3891
2.1977
2.5170
19.8765
2.6073
15.6152
4.0944
7.5292
•
23.1921
3.1174
2.3020
0.9380
2.4076
1.4399
1.0980
2.3121
4.1711
1.8683
3.0818
7.9526
2.8098
2.5633
3.0941
2.1673
2.1841
3.5816
2.8504
2.4432
5.0346
6.4820
2.8514
E-73
-------
UT1FL89
WT1F1011
WT2FL45
WT2FL67
WT2FL89
WT2F1011
ET3FL47
ET3FL811
UT3FL47
UT3FL811
Glued Carpet, Room
Glued Carpet, Hall
Tackd Carpet, Room
Tackd Carpet, Hall
Private Office
Cubicle or Shared
-1.3743
-0.6081
-0.5680
-0.8590
-1.0770
0.0847
.1335
.4484
.1528
.1459
.1074
.1463
1.2915
1.0557
0.4434
0.3452
1.1157
0.7869
0.7876
0.7332
1.1343
0.6908
0.6640
0.6198
0.7631
1.1775
0.3072
0.4203
0.3050
0.4376
0.2636
0.2602
1.5174
0.5972
0.5200
1.3726
0.9016
0.0150
2.9138
5.4617
2.2823
0.9471
12.9987
0.1212
17.9310
5.8202
2.8294
1.7602
0.2180
0.4396
0.4708
0.2414
0.3424
0.9024
0.0878
0.0194
0.1309
0.3305
0.0003
0.7278
0.0001
0.0158
0.0926
0.1846
0.25302
0.54438
0.56666
0.42359
0.34062
1.08839
0.32190
0.23495
0.31575
0.31794
3.02648
1.15754
3.63824
2.87399
1.55800
1.41227
0.01429
0.07171
0.07451
0.06407
0.01833
0.18364
0.05819
0.04760
0.04422
0.01531
1.37170
0.39204
1.65834
0.93094
0.79007
0.72247
4.4805
4.1328
4.3097
2.8003
6.3278
6.4508
1.7806
1.1598
2.2545
6.6018
6.67752
3.41779
7.98193
8.87251
3.07232
2.76067
Association of Predicted Probabilities and Observed Responses
Concordant = 83.7X
Discordant = 16.OX
Tied = 0.3X
(418472 pairs)
NOTE:
Somens'
Gamma
Tau-a
c
0.676
0.678
0.188
0.838
Parameter estimates marked with '#' are regarded to be infinite.
E-74
-------
Response Variable: Musty/Damp Odor, Diesel Exhaust
Number of Observations: 1760
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1952.340
1957.813
1950.340
Intercept
and
Covariates
1812.263
1965.509
1756.263
Chi-Square for Covariates
194.077 with 27 DF (p=0.0001>
196.612 with 27 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at Work
Use Lamp
Water Leaks Nearby
Glare
Dust
Role Conflict
Underut iIi zat i on
College Grad
Allergy to Molds
Chem. Sensitivity
Eczema
AHU16
AHU2
AHU20
AHU21
AHU5
AHU9
ETB
AWU15SU2
G1LG1
NEG1NEG2
ET1FL67
ET1F1011
UGHAHU19
Tackd Carpet, Room
Private Office
Cubicle or Shared
Parameter
Estimate
•2.80520
0.00277
0.33310
0.30220
0.39360
0.76470
0.22130
-0.00678
-0.27050
0.39460
0.19060
0.50440
1.28930
0.65910
0.60790
0.00537
0.80020
0.72910
1.40460
2.04980
2.79520
1.47090
-0.41250
-0.31200
-0.05940
0.41820
0.16260
0.15940
Standard
Error
0.3551
0.0396
0.1204
0.1719
0.1250
0.1330
0.0851
0.0577
0.1334
0.1325
0.1275
0.1983
0.5647
0.5021
0.6342
0.8153
0.3560
0.3888
0.5590
0.4229
1.1431
0.6653
0.4665
0.4420
0.5947
0.2310
0.1759
0.1441
Uald
Chi-Sq
62.4101
0.0049
7.6588
3.0911
9.9122
33.0498
6.7677
0.0138
4.1108
8.8662
2.2336
6.4719
5.2122
1.7233
0.9187
0.0000
5.0515
3.5159
6.3125
23.4914
5.9795
4.8885
0.7819
0.4983
0.0100
3.2772
0.8545
1.2235
Pr >
Chi-Sq
0.0001
0.9442
0.0056
0.0787
0.0016
0.0001
0.0093
0.9065
0.0426
0.0029
0.1350
0.0110
0.0224
0.1893
0.3378
0.9947
0.0246
0.0608
0.0120
0.0001
0.0145
0.0270
0.3766
0.4802
0.9205
0.0703
0.3553
0.2687
Odds
Ratios
0.0605
1.0028
1.3953
1.3528
1.4823
2.1483
1.2477
0.9932
0.7630
1.4838
1.2100
1.6560
3.6302
1.9331
1.8366
1.0054
2.2260
2.0732
4.0739
7.7663
16.3659
4.3532
0.6620
0.7320
0.9423
1.5192
1.1766
1.1728
99
Percent
Confidence Intervals
Low Limit
0.02424
0.90553
1.02321
0.86882
1.07423
1.52515
1.00208
0.85606
0.54111
1.05472
0.87124
0.99360
0.84758
0.53030
0.35851
0.12309
0.88971
0.76151
0.96523
2.61277
0.86120
0.78434
0.19905
0.23443
0.20365
0.83790
0.74788
0.80913
Upper Limit
0.151
1.110
1.903
2.106
2.045
3.026
1.554
1.152
1.076
2.087
1.680
2.760
15.549
7.046
9.408
8.212
5.569
5.644
17.194
23.085
311.010
24.160
2.202
2.286
4.360
2.755
1.851
1.700
Association of Predicted Probabilities and Observed Responses
Concordant - 70.2X
Discordant = 29.4X
Tied = 0.4X
(569191 pairs)
Somens'
Gamma
Tau-a
c
= 0.408
= 0.410
= 0.150
= 0.704
E-75
-------
Response Variable: Musty/Damp odor, Diesel Exhaust
Number of Observations: 1734
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1784.031
1789.489
1782.031
Intercept
and
Covariate*
1574.255
1727.084
1518.255
Chi-Square for Covariates
263.776 with 27 OF (p=0.0001)
276.906 with 27 OF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at Work
Use Lamp
Water Leaks Nearby
Glare
Dust
Role Conflict
Underutilization
College Grad
Allergy to Molds
Chem. Sensitivity
Eczema
AHU16
AHU2
AHU20
AHU21
AHU5
AHU9
ETB
AWU15SU2
G1LG1
NEG1NEG2
ET1FL67
ET1F1011
UGHAHU19
Tackd Carpet, Room
Private Office
Cubicle or Shared
Parameter
Estimate
-3.5554
0.0511
0.3599
1 .0398
0.3207
0.8282
0.3877
0.1788
-0.1330
0.1736
-0.0281
-0.1199
2.1596
2.4240
0.7696
0.9694
-0.2613
0.6764
1.2630
1.8906
1.7257
0.7950
-0.8392
-2.2904
1.0143
0.5307
-0.1070
-0.0685
Standard
Error
0.4530
0.0406
0.1343
0.1704
0.1333
0.1348
0.0977
0.0656
0.2720
0.1425
0.1513
0.2303
0.5250
0.7417
0.6306
0.6415
0.4286
0.4549
0.6834
0.3896
0.8897
0.6006
0.7589
1.0377
0.5756
0.2647
0.2083
0.2039
Uald
Chi-Sq
61.6016
1.5815
7.1854
37.2177
5.7890
37.7537
15.7602
7.4251
0.2391
1.4832
0.0346
0.2711
16.9189
10.6826
1.4894
2.2832
0.3717
2.2108
3.4157
23.5436
3.7617
1.7523
1.2230
4.8721
3.1052
4.0195
0.2638
0.1128
Pr >
Chi-Sq
0.0001
0.2085
0.0073
0.0001
0.0161
0.0001
0.0001
0.0064
0.6249
0.2233
0.8525
0.6026
0.0001
0.0011
0.2223
0.1308
0.5421
0.1370
0.0646
0.0001
0.0524
0.1856
0.2688
0.0273
0.0780
0.0450
0.6075
0.7369
Odds
Ratios
0.0286
1.0524
1.4332
2.8287
1.3781
2.2892
1.4736
1.1958
0.8755
1.1896
0.9723
0.8870
8.6677
11.2909
2.1589
2.6364
0.7700
1.9668
3.5360
6.6233
5.6165
2.2144
0.4321
0.1012
2.7574
1.7001
0.8985
0.9338
99
Percent
Confidence Intervals
Low Limit
0.00889
0.94792
1.0J404
1.82367
0.97757
1.61762
1.14571
1.00986
0.43445
0.82409
0.65846
0.49010
2.24162
1.67093
0.42536
0.50505
0.25529
0.60931
0.60809
2.42781
0.56770
0.47135
0.06117
0.00699
0.62597
0.85970
0.52541
0.55225
Upper Limit
0.0918
1.1685
2.0256
4.3875
1.9427
3.2396
1.8953
1.4159
1.7642
1.7172
1.4357
1.6054
33.5152
76.2959
10.9575
13.7619
2.3228
6.3485
20.5619
18.0692
55.5658
10.4035
3.0518
1.4663
12.1466
3.3621
1.5366
1.5789
Association of Predicted Probabilities and Observed Responses
Concordant = 75.3X
Discordant = 24.3X
Tied - 0.4X
(498680 pairs)
Somers1 D = 0.511
Gamma = 0.513
Tau-a " 0.170
c = 0.755
E-76
-------
Response Variable: Tobacco Smoke Odor
Number of Observations: 1813
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1517.464
1522.967
1515.464
Intercept
and
Covariates
1414.921
1557.993
1362.921
Chi-Square for Covariates
152.543 with 25 DF
Chi-Sq
0.0001
0.1980
0.0128
0.0010
0.0008
0.0025
0.0011
0.2813
0.0001
0.7224
0.1069
0.0582
0.1246
0.0190
0.0186
0.0118
0.1754
0.8904
0.0045
0.0461
0.3148
0.0003
0.0719
0.2500
0.1219
0.2198
Odds
Ratios
0.02142
.06545
.58613
.84467
.68472
.34084
0.58655
1.19830
2.40008
0.94327
1.39222
2.68264
2.64719
5.65760
8.89065
5.66439
1.92360
1.07885
5.04249
2.91509
1 .61462
5.53173
2.12570
0.29253
1 .34259
1.22079
99
Percent
Confidence Intervals
Low Limit
0.00856
0.93838
0.98435
1.14038
1.12692
1.04465
0.38523
0.77755
1 .62456
0.61777
0.82062
0.70115
0.51715
0.84376
0.81338
0.96015
0.55461
0.26134
1.16374
0.73189
0.47302
1.61807
0.72235
0.01865
0.82212
0.80303
Upper Limit
0.0536
1.2097
2.5558
2.9839
2.5186
1.7210
0.8931
1.8467
3.5458
1.4403
2.3620
10.2640
13.5505
37.9356
97.1794
33.4170
6.6718
4.4537
21.8491
11.6107
5.5114
18.9114
6.2554
4.5883
2.1926
1.8559
Association of Predicted Probabilities and Observed Responses
Concordant = 71.2X
Discordant = 28.3X
Tied = 0.6X
(412782 pairs)
Somers1
Gamma
Tau-a
c
0.429
0.431
0.108
0.714
E-77
-------
Response Variable: Tobacco Smoke Odor
Number of Observations: 1809
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Criterion
AIC
sc
-2 LOG L
Score
Intercept
Only
1368.556
1374.056
1366.556
•
Intercept
and
Cover i at es
1284.153
1427.167
1232.153
•
Chi-Square for Covariates
134.403 with 25 OF (p=0.0001)
156.850 with 25 DF (p=0.0001)
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Shelf Index
Photocopier Nearby
Water Leaks Nearby
Dust
Role Conflict
Allergy to Pollen
Allergy to Molds
Chcm. Sensitivity
Live Alone
Nonsmoker
AHU13
AHU14
AHU3
G1LG1
NEG1NEG2
CM FL11
FCFL7
CM2NOFL
ET2FL67
ET2F1011
WT1FL13
WT2FL13
UGHAHU19
Private Office
Cubicle or Shared
Parameter
Estimate
-3.5818
0.0759
0.0450
0.4175
0.5741
0.3111
-0.1991
0.2213
0.7523
0.3422
0.2594
1.4085
1.4073
-0.1731
1.7852
1.5990
1.3358
0.8395
0.7937
-0.3809
-0.7000
2.8070
0.5423
0.8552
-0.3613
-0.2955
Standard
Error
0.4278
0.0525
0.2417
0.1971
0.1543
0.1078
0.1701
0.1767
0.1639
0.1976
0.2509
0.4908
0.6020
1.0636
0.9061
0.5598
0.4679
0.4616
0.6734
1.0582
0.7511
0.5604
0.6755
0.5889
0.2212
0.2213
Uald
Chi-Sq
70.1065
2.0880
0.0347
4.4860
13.8441
8.3316
1.3693
1.5675
21 .0742
2.9989
1.0687
8.2360
5.4642
0.0265
3.8819
8.1585
8.1493
3.3082
1.3894
0.1296
0.8685
25.0882
0.6445
2.1091
2.6685
1.7839
Pr >
Chi-Sq
0.0001
0.1485
0.8523
0.0342
0.0002
0.0039
0.2419
0.2106
0.0001
0.0833
0.3012
0.0041
0.0194
0.8707
0.0488
0.0043
0.0043
0.0689
0.2385
0.7189
0.3514
0.0001
0.4221
0.1464
0.1024
0.1817
Odds
Ratios
0.0278
1.0789
1.0460
1.5182
1.7755
1.3649
0.8195
1.2477
2.1219
1.4080
1.2962
4.0898
4.0849
0.8411
5.9608
4.9481
3.8030
2.3152
2.2116
0.6832
0.4966
16.5602
1.7200
2.3518
0.6968
0.7442
99
Percent
Confidence Intervals
Lou Limit
0.00924
0.94239
0.56123
0.91372
1.19318
1.03397
0.52873
0.79146
1.39110
0.84635
0.67915
1.15511
0.86636
0.05432
0.57758
1.16994
1.13939
0.70498
0.39025
0.04474
0.07173
3.90950
0.30186
0.51592
0.39412
0.42081
Upper Limit
0.0838
1.2351
1.9496
2.5224
2.6421
1.8018
1.2701
1.9669
3.2365
2.3425
2.4737
14.4805
19.2604
13.0232
61.5170
20.9271
12.6938
7.6033
12.5332
10.4335
3.4378
70.1468
9.8001
10.7210
1.2318
1.3160
Association of Predicted Probabilities and Observed Responses
Concordant = 70.9X
Discordant = 28.4X
Tied = 0.7X
(359114 pairs)
Sorters'
Gamma
Tau-a
c
D = 0.425
= 0.428
= 0.093
= 0.713
E-78
-------
APPENDIX F
RESULTS OF LOGISTIC REGRESSIONS USING ALTERNATIVE CRITERION
F-l
-------
Appendix F. Results of Logistic Regressions Using Alternative Criterion
This appendix presents the results of a set of logistic regressions carried
out on a single health factor (eye irritation) as a test of the effects of
employing a different criterion for identifying a "positive" response. In the
first set of logistic regressions, the criterion to be counted as a positive
response was a sufficiently high score on an index composed of the sum of scores
based on responses to individual symptoms. Previous studies have used different
criteria (e.g., requiring that at least one symptom in a group be experienced
"often" or "always.") Therefore we ran a test set of regressions to determine
whether very different results might be obtained from different criteria for
defining positive responses.
Somewhat surprisingly, we found that this approach, which seemed
potentially more stringent, actually counted many more people as having eye
irritation symptoms than our index-based criterion: more than 400 men and more
than 500 women, compared to 232 men and 456 women. The set of four regressions
using the "often or always" criteria resulted in only one variable (glare) being
identified as significant (p < 0.01) for both men and women with and without the
comfort variables included. The reduced ability to identify significant
variables may be due to the increased number of responders identified as
"positive"; if some of these should have been classified as negative, the
misclassification would have reduced the ability to detect associations.
F-2
-------
Using "often" or "always" Criterion
Response Variable: Dry, Itching or Tearing Eyes, Sore Eyes, Burning Eyes, Blurry Vision
Number of Observations: 1762
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Intercept
Only
1869.798
1875.272
1867.798
Intercept
and
Covariates
1690.610
1871.258
1624.610
Criterion
AIC
SC
-2 LOG L
Score
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Plants Nearby
Hours at VOT
1 Hour at Copier
Use Heater
Walls Painted
Water Leaks Nearby
Window Nearby
Lights Too Dim
Glare
Noise
Dust
Workload
External Stress
Wear Contact Lens
Wear Glasses
College Grad
Allergy to Animals
Allergy to Molds
Chem. Sensitivity
Sole Provider
AHU12
AHU16
SE31
G1LG1
FCFL2
ET1FL67
ET1F1011
ET2F1011
Tackd Carpet, Hall
Private Office
Cubicle or Shared
Chi-Square for Covariates
243.188 with 32 DF (p-0.0001)
230.186 with 32 DF (p=0.0001)
99 Percent
Parameter
Estimate
-1.8188
0.0985
0.0115
-0.0039
0.0814
0.0918
0.0434
-0.3382
-0.0572
0.2356
0.4812
0.1635
0.5987
-0.0614
0.0239
-0.0085
0.1524
0.2290
-0.2805
0.0058
0.0742
-0.0897
-0.6139
-0.9719
0.0421
-37.3537 #
-0.5768
-0.0921
-0.6463
-0.2283
0.1037
-0.1227
-0.1253
Standard
Error
0.4630
0.1179
0.1127
0.0253
0.1265
0.1316
0.1315
0.1746
0.1169
0.1964
0.1146
0.1172
0.1208
0.0598
0.0468
0.1303
0.1119
0.1333
0.1607
0.1323
0.1188
0.1748
0.5130
0.6390
0.4721
•
0.5729
0.4167
0.4569
0.4440
0.2422
0.1652
0.1341
Wald
Chi-Sq
15.4320
0.6980
0.0104
0.0234
0.4140
0.4865
0.1090
3.7517
0.2395
1.4383
17.6377
1.9458
24.5556
1.0564
0.2612
0.0043
1.8546
2.9519
3.0449
0.0020
0.3895
0.2635
1 .4320
2.3132
0.0080
•
1.0134
0.0488
2.0008
0.2645
0.1835
0.5519
0.8730
Pr >
Chi-Sq
0.0001
0.4035
0.9187
0.8785
0.5200
0.4855
0.7413
0.0528
0.6246
0.2304
0.0001
0.1630
0.0001
0.3040
0.6093
0.9478
0.1732
0.0858
0.0810
0.9648
0.5326
0.6078
0.2314
0.1283
0.9289
*
0.3141
0.8251
0.1572
0.6071
0.6684
0.4575
0.3501
Odds
Ratios
0.16222
1.10351
1.01157
0.99614
1.08480
1.09615
1.04436
0.71305
0.94441
1.26567
1.61801
1.17763
1.81975
0.94045
1.02419
0.99150
1.16463
1.25734
0.75541
1.00586
1.07702
0.91421
0.54124
0.37836
1.04300
•
0.56169
0.91201
0.52398
0.79589
1.10927
0.88453
0.88223
Confidence Intervals
Low Limit
0.04922
0.81448
0.75668
0.93329
0.78312
0.78098
0.74428
0.45477
0.69884
0.76313
1.20441
0.87074
1.33312
0.80618
0.90787
0.70879
0.87297
0.89192
0.49934
0.71536
0.79308
0.58276
0.14437
0.07295
0.30912
•
0.12840
0.31176
0.16150
0.25359
0.59440
0.57796
0.62454
Upper Limit
0.53467
1.49513
1 .35231
1.06322
1.50270
1.53849
1.46542
1.11803
1.27626
2.09914
2.17365
1.59266
2.48403
1.09707
1.15541
1.38696
.55372
.77248
.14278
.41431
.46262
.43417
2.02909
1.96239
3.51919
•
2.45712
2.66797
1.70008
2.49788
2.07012
1.35372
1.24625
Association of Predicted Probabilities and Observed Responses
Concordant = 74.4X
Discordant = 25.3X
Tied « Q.3X
(537040 pairs)
NOTE:
Somers1
Gamma
Tau-a
c
0.491
0.493
0.170
0.746
Parameter estimates marked with '#' are regarded to be infinite.
F-3
-------
Using "often" or "always" Criterion
Response Variable: Dry, Itching or Tearing Eyes, Sore Eyes, Burning Eyes, Blurry Vision (Including Comfort
and Odor Variables)
Number of Observations: 1746
Sex: Female
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Squire for Coverfates
64.888 with 25 OF (p-0.0001)
63.978 with 25 OF
Chi-Sq
0.0001
0.9496
0.4782
0.5443
0.6850
0.0006
0.3190
0.9023
0.2910
0.7300
0.8892
0.7787
0.0106
0.0621
0.3313
0.4333
0.0498
0.2779
0.3201
0.2261
0.9312
0.4591
0.3955
0.1445
0.4256
0.4221
Odds
Ratios
0.18968
1.00157
.09352
.08296
.05464
.47757
0.94224
.01603
.12491
0.94724
1.01867
1.03438
1.04970
1.06418
1.03541
0.93314
4.33664
0.57344
0.56637
0.27829
1.02953
1.31075
0.61940
1.83712
0.88012
0.90177
Confidence Intervals
Low Limit
0.09562
0.93935
0.79043
0.77179
0.75219
1.10072
0.80792
0.72801
0.84428
0.63199
0.72410
0.75894
0.99956
0.97670
0.94420
0.74330
0.63193
0.15316
0.12987
0.01830
0.43191
0.51123
0.14495
0.62800
0.58238
0.64714
Upper Limit
0.3763
1.0679
.5128
.5196
.4787
.9834
.0989
.4180
.4988
.4198
.4331
1.4098
1.1023
1.1595
1.1354
1.1715
29.7603
2.1471
2.4700
4.2321
2.4540
3.3606
2.6468
5.3743
1 .3301
1.2566
Concordant = 61.7X
Discordant = 37.7X
Tied = 0.6X
(614673 pairs)
Somers1
Gamma
Tau-a
c
= 0.241
- 0.242
= 0.097
= 0.620
F-4
-------
Using "often11 or "always" Criterion
Response Variable: Dry, Itching or Tearing Eyes, Sore Eyes, Burning Eyes, Blurry Vision
Hunter of Observations: 1758
Sex: Male
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
109.011 with 32 DF (p=0.0001)
107.132 Hith 32 DF (p-0.0001)
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1964.760
1970.232
1962.760
Intercept
and
Covariates
1919.749
2100.323
1853.749
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Laser Printer Nearby
Plants Nearby
Hours at VDT
1 Hour at Copier
Use Heater
Walls Painted
Water Leaks Nearby
Window Nearby
Lights Too Dim
Glare
Noise
Dust
Workload
External Stress
Wear Contact Lens
Wear Glasses
College Grad
Allergy to Animals
Allergy to Molds
Chem. Sensitivity
Sole Provider
AHU12
AHU16
SE31
G1LG1
FCFL2
ET1FL67
ET1F1011
ET2F1011
Tackd Carpet, Hall
Private Office
Cubicle or Shared
99 Percent
Parameter
Estimate
-1.70510
0.20740
-0.09330
0.02720
0.23880
0.26540
-0.04470
0.06150
-0.01870
-0.12580
0.61920
0.33410
0.17730
0.10170
-0.08380
-0.21320
0.10940
-0.06740
0.06160
0.17450
0,32260
0.12560
-0.45280
-0.38650
-0.74510
0.99690
0.50100
-0.35300
-0.11170
0.09330
-0.00533
-0.34510
0.00420
Standard
Error
0.5172
0.1266
0.1225
0.0289
0.1377
0.1887
0.1515
0.1714
0.1235
0.2574
0.1200
0.1251
0.1228
0.0680
0.0523
0.1623
0.1234
0.2502
0.1603
0.1332
0.1317
0.1313
0.6718
0.5322
0.7986
0.8275
0.3884
0.5745
0.4489
0.4501
0.2321
0.1862
0.1782
Wald
Chi-Sq
10.8693
2.6844
0.5801
0.8876
3.0065
1.9782
0.0872
0.1287
0.0230
0.2389
26.6159
7.1358
2.0845
2.2396
2.5710
1.7251
0.7857
0.0725
0.1474
1.7159
6.0036
0.9156
0.4542
0.5273
0.8703
1.4514
1.6638
0.3776
0.0619
0.0430
0.0005
3.4353
0.0006
Pr >
Chi-Sq
0.0010
0.1013
0.4463
0.3461
0.0829
0.1596
0.7678
0.7198
0.8796
0.6250
0.0001
0.0076
0.1488
0.1345
0.1088
0.1890
0.3754
0.7878
0.7010
0.1902
0.0143
0.3386
0.5003
0.4677
0.3509
0.2283
0.1971
0.5389
0.8035
0.8357
0.9817
0.0638
0.9812
Odds
Ratios
0.18175
1.23047
0.91092
1.02757
1.26972
1.30395
0.95628
1.06343
0.98147
0.88179
1 .85744
1.39668
1.19399
1.10705
0.91962
0.80799
1.11561
0.93482
1.06354
1.19065
1.38071
1.13383
0.63585
0.67943
0.47469
2.70987
1.65037
0.70258
0.89431
1.09779
0.99468
0.70815
1.00421
Confidence Intervals
Lou Limit
0.04796
0.88806
0.66441
0.95385
0.89055
0.80196
0.64729
0.68384
0.71403
0.45436
1.36353
1.01191
0.87020
0.92917
0.80370
0.53191
0.81182
0.49070
0.70375
0.84483
0.98348
0.80846
0.11266
0.17248
0.06067
0.32151
0.60682
0.15995
0.28138
0.34433
0.54705
0.43834
0.63455
Upper Limit
0.6888
1.7049
1.2489
1.1070
1 .8103
2.1202
1.4128
1.6537
1.3491
1.7113
2.5303
1.9278
1.6383
1.3190
1.0522
1.2274
1.5331
1.7809
1.6073
1.6780
1.9384
1.5902
3.5886
2.6763
3.7139
22.8406
4.4885
3.0861
2.8425
3.5000
1.8086
1.1440
1.5892
Association of Predicted Probabilities and Observed Responses
Concordant = 66.IX
Discordant = 33.5X
Tied - 0.5X
(573725 pairs)
Somers1
Gamma
Tau-a
c
* 0.326
« 0.328
» 0.121
* 0.663
F-5
-------
Using "often" or "always" Criterion
Response Variable: Dry, Itching or Tearing Eyes, Sore Eyes, Burning Eyes, Blurry Vision (Including Comfort
and Odor Variables)
Number of Observations: 1749
Sex: Hale
The LOGISTIC Procedure
Criteria for Assessing Model Fit
Chi-Square for Covariates
121.356 with 25 DF (p=0.0001)
120.185 with 25 DF (p=0.0001>
Criterion
AIC
SC
-2 LOG L
Score
Intercept
Only
1952.958
1958.424
1950.958
Intercept
and
Covariates
1881.602
2023.738
1829.602
•
Analysis of Maximum Likelihood Estimates
Variable
Intercept
Hours at VDT
1 Hour at Copier
Use Heater
Walls Painted
Glare
workload
Wear Contact Lens
Wear Glasses
Allergy to Animals
Allergy to Molds
Chem. Sensitivity
Hot Stuffy Air
Dry Air
Cosmetic, Body Odor
Photocopies Odor
AHU10
AHU12
AHU16
AHU21
NE1
SE32
FCFL2
FCFL6
Private Office
Cubicle or Shared
Parameter
Estimate
-2.5367
0.0129
0.2457
0.1687
-0.0336
0.4433
0.0687
-0.1621
0.0731
0.0327
0.1259
0.2524
0.0462
0.1141
0.0565
0.1261
0.8171
-0.7357
-0.1695
0.9069
-0.1955
0.1205
0.6335
0.5761
-0.2824
0.1036
Standard
Error
0.3398
0.0292
0.1393
0.1878
0.1518
0.1200
0.0678
0.1613
0.1241
0.1587
0.1330
0.1326
0.0192
0.0370
0.0457
0.1214
0.6353
0.7912
0.4813
0.5747
0.3869
0.5523
0.3955
0.4176
0.1876
0.1781
Wald
Chi-Sq
55.7292
0.1961
3.1126
0.8072
0.0489
13.6579
1.0278
1.0097
0.3476
0.0423
0.8961
3.6225
5.7736
9.4975
1.5256
1.0786
1.6540
0.8647
0.1240
2.4903
0.2553
0.0476
2.5659
1.9027
2.2655
0.3383
Pr > Odds
Chi-Sq Ratios
0.0001 0.07913
0.6579
0.0777
0.3689
0.8251
0.0002
0.3107
0.3150
0.5555
0.8370
0.3438
0.0570
0.0163
.01298
.27852
.18376
.96696
.55784
.07111
.85036
.07584
.03324
.13417
.28711
.04728
0.0021 1.12086
0.2168 1.05813
0.2990 1.13440
0.1984 2.26392
0.3524 0.47917
0.7247 0.84409
0.1145 2.47663
0.6134 0.82242
0.8273 1.12806
0.1092 1.88419
0.1678 1.77909
0.1323 0.75397
0.5608 1.10916
99
Percent
Confidence Intervals
Low Limit
0.03297
0.93958
0.89303
0.72974
0.65401
1.14360
0.89947
0.56124
0.78147
0.68653
0.80517
0.91468
0.99675
1.01897
0.94061
0.82975
0.44068
0.06242
0.24431
0.56353
0.30357
0.27193
0.68024
0.60675
0.46503
0.70104
Upper Limit
0.1899
.0921
.8304
.9203
.4297
2.1221
.2755
.2884
.4811
.5551
.5976
.8112
.1004
.2330
.1903
.5509
11.6305
3.6782
2.9163
10.8844
2.2281
4.6797
5.2190
5.2166
1.2225
1.7549
Association of Predicted Probabilities and Observed Responses
Concordant = 67.2X
Discordant = 32.3X
Tied = 0.5X
(567170 pairs)
Somers1
Gamma
Tau-a
c
0.349
0.351
0.130
0.675
* U.S. GOVERNMENT PRINTING OFFICE: 19 9.1 -261 -72V53176
F-6
------- |