3-EPA
United States
Environmental Protection
Agency
 Estimating Contributions of
  Outdoor Fine Particles to
 Indoor Concentrations and
     Personal Exposures:
  Effects of Household Characteristics
       and Personal Activities

-------
                                             EPA 600/R-06/023
                                                 March 2006
Estimating Contributions of Outdoor Fine
  Particles to Indoor Concentrations and
             Personal Exposures:

  Effects of Household Characteristics and
                Personal Activities
                          By
            Lance Wallace, Ron Williams, Jack Suggs and Paul Jones
             Human Exposure and Atmospheric Sciences Division
                 National Exposure Research Laboratory
                 Research Triangle Park, NC, 27711
                 National Exposure Research Laboratory
                 Office of Research and Development
                 U.S. Environmental Protection Agency
                 Research Triangle Park, NC, 27711

-------
                                           Notice
     The U.S. Environmental Protection Agency through its Office of Research and Development partially
funded  and collaborated  in the  research described here under contract numbers  68-D2-0134 (QST
Environmental), 68-D2-0187 (SRA Technologies, Inc), 68-D-99-012,68-D5-0040 (Research Triangle Institute),
CR-820076 (University of North Carolina-Chapel Hill), and CR-828186-01-0 (Shaw University). It has been
subjected to the Agency's peer and administrative review, and it has been approved for publication as an EPA
document.    Mention  of  trade  names  or commercial  products does  not  constitute endorsement or
recommendation for use.

-------
                                          Abstract
     A longitudinal study of personal, indoor, and outdoor exposures to PM: 5 and associated elements was
earned out involving 37 residents of the Research Tnangle Park area in North Carolina. Participant exposures
were monitored for 7 consecutive days in each of four seasons. A main goal of the study was to estimate the
contribution of outdoor PM; 5 to indoor concentrations and personal exposures  This contribution depends on
the infiltration factor (the fraction of outdoor PM2 5 remaining airborne after penetrating indoors), which can be
estimated using sulfur as a marker for particles of outdoor origin. The annual average infiltration factors ranged
from 0.26 to 0.89, and depended strongly on air exchange rates. The outdoor contnbutions to personal exposure
were then regressed longitudinally on outdoor concentrations measured at a central momtonng station, with a
range of R: values from 0.19 to 0.88. Vanables significantly affecting indoor air PM: 5 concentrations included
smoking and cooking, the number of persons in the household, burned food, use of a kitchen exhaust fan, and
duration of candle use. These findings might have important implications for epidemiological studies
                                                111

-------
                                         Contents
Notice	
Abstract	
Figures
      s.
            	v
 Tables  	vii
 Acknowledgments	ix

 Chapter 1: Introduction	1
 Chapter 2: Description of Study Methods and Database	3
 Chapter 3: Results	4
   Calculation of Fmf Using the Indoor/Outdoor Sulfur Ratio	4
   Calculation of F^f by Regressing Indoor Sulfur on Outdoor Sulfur	8
   Comparison of Methods for Calculating F^	11
   Estimating Indoor and Outdoor Contributions to Indoor PM25	11
   Relationship Between Outdoor Particles and Indoor Particles of Outdoor Origin	15
   Estimating Contributions of Outdoor Air to Indoor Concentrations Using the RCS Model	15
   Estimates of the Contribution of Outdoor Air Particles to Personal Exposure	17
   Outdoor Exposure Factor FpeX Estimated Using PM Measurements	19
   Estimating the Outdoor Exposure Factor FpeX Using Sulfur Measurements	19
   Comparison of Fpex and Finf	19
   Comparison of Indoor-Outdoor Sulfur and Personal Sulfur Measurements	20
   Use of the Outdoor Exposure Factor to Calculate the Contribution to Personal Exposure Made by
   Particles of Outdoor Origin	25
   Relationship Between Outdoor Concentrations and the Contribution to Personal Exposure of Particles of
   Outdoor Origin	28
   Use of Reported Time in Indoor and Outdoor Microenvironments to Predict the Outdoor Exposure
   Factor Fpex from the Infiltration Factor Fmf	28
   Estimating P andk	32
     Calculating Average Values of P and k	32
     Calculating Individual Home Values of P andk	33
     Estimating  F^ from Individual Values of P  andk	36
   Seasonal Analysis	39
   Multivariate Regressions	39
   Variables Affecting Air Exchange and the Infiltration Factor	49
     Variables Affecting Air Exchange	49
     Variables Affecting the  Infiltration Factor	53
Chapter 4: Discussion	55
Chapter 5: Conclusions	58
Chapter 6: References	60
Appendix	64
                                              IV

-------
                                           Figures

Number                                                                                   Page

3-1  Valid pairs of indoor and outdoor 24-h average sulfur measurements (ng/m3)	4
3-2  Indoor/outdoor sulfur ratios (/•",„/) by home averaged across all seasons. Error bars are standard
       errors calculated by propagation of error	8
3-3  Indoor/outdoor sulfur ratios by home and by season	8
3-4  Comparison of the results of regressing indoor sulfur on outdoor sulfur with the simple
       indoor/outdoor ratio averaged over all visits to a home	 11
3-5  Estimates of the fractional contribution of outdoor particles to total indoor PM: 5
       concentrations, averaged over all home visits.  Error bars are standard errors calculated by
       propagation of error	 11
3-6  Comparison of average outdoor-generated and indoor-generated particles based on
       indoor/outdoor sulfur ratios	  11
3-7  Estimates of average outdoor contribution to indoor PM; 5. Error bars are  standard errors
       calculated by propagation of error techniques applied to the three measurements required to
       estimate the outdoor contribution	 13
3-8  Estimates of average indoor-generated PM; 5. Error bars are standard errors calculated by
       propagation of error techniques applied to the four measurements required to estimate
       indoor-generated PM2 5	 13
3-9  Indoor-outdoor average contributions to indoor PM25.  Summer 2000	 13
3-10 Indoor-outdoor average contributions to indoor PM; 5.  Fall 2000	 13
3-11 Indoor-outdoor average contnbutions to indoor PM;5.  Winter 2001	 15
3-12 Indoor-outdoor average contributions to indoor PM: 5.  Spring 2001	 15
3-13 Regression of indoor PM2.5 on residential outdoor PM; 5  data	 17
3-14 Estimates of the infiltration factor F.^from sulfur indoor/outdoor ratios compared to
       regressions of indoor vs. outdoor fine particles. The regression line shown is for the
        18 cases with slopes significantly different from zero	 17
3-15 24-h average fine particle personal exposures vs outdoor  air concentrations	 19
3-16 Personal vs. outdoor PM:5; one outlier removed	19
3-17 Personal vs. outdoor sulfur	 19
3-18 Co-located PEMio and HI; 5 sulfur concentrations outdoors.  Summer 2000	20
3-19 Co-located PEMK> and HI; 5 sulfur concentrations indoors. Summer 2000	20
3-20  Comparison of infiltration factor Fmf and outdoor exposure factor Fpa by participant	20
3-21 Outdoor contributions to personal exposure. Error bars are standard errors calculated by
       propagation of error	25
3-22 Non-outdoor contnbutions to personal exposure.  Error bars are standard errors
       calculated by propagation of error	25
3-23 Outdoor and non-outdoor contributions to personal PM:5  exposure	25
3-24 Outdoor and non-outdoor contributions to personal PM; 5  exposure. The  non-outdoor
       contribution  is divided into indoor-generated PM: 5 encountered while at home and the
       sum of indoor-generated PM;5 while away from home and PM;5 due to the personal cloud	28
3-25 Predicted value of the outdoor exposure factor Fpex using Equation 3-6 compared to
       the measured value using the personal/outdoor sulfur ratio	32
3-26 Predicted personal exposure to PM:5 using only F,nf.	32
3-27 Nonlinear least-squares fit to the indoor/outdoor sulfur ratio vs. the air exchange rate.
       Bounding curves are + 1 SE	32
3-28 Regression of the outdoor/indoor sulfur ratio vs. residence time	33
3-29  Same regression as in Figure 3-28 without four outliers	 33
3-30  Comparison of estimates of P from the linear and nonlinear approaches described in the text.
       Only values  significantly different from zero are plotted (N = 32 homes)	36

-------
3-31  Comparison of the estimates of k from the linear and nonlinear approaches described in the text.
        Only values significantly different from zero are plotted (N = 24 homes)	36
3-32  Comparison of the infiltration factor (F,af) estimates from the simple ratio of indoor sulfur to
        outdoor sulfur by home vs. the nonlinear regression of the same ratio using the
        measured air exchange rates and the linear regression of the inverse ratio (outdoor/indoor)
        against the residence time	36
3-33  Estimates for each home by season of the infiltration factor F^/from regressing indoor
        sulfur on outdoor sulfur (Slope) compared to estimates from the simple ratio of indoor
        sulfur to outdoor sulfur averaged overall visits in a season	39
3-34  Estimates of Fmfb\ home from the indoor/outdoor sulfur ratio	39
3-35  Estimates of Fm/by home from regressions of indoor on outdoor sulfur	39
3-36  Central-site and residential outdoor concentrations averaged over all visits to a home	40
A-1  Adjusted R2 values from regressing the outdoor contribution to personal exposure on outdoor
        PM; 5 measurements just outside the house	64
A-2  Adjusted R2 values from regressing the outdoor contribution to personal exposure on outdoor
        PM;.5 Harvard Impactor (HI) measurements at the central site	64
A-3  Adjusted R2 values from regressing the outdoor contribution to personal exposure on outdoor
        PM:5 Federal Reference Method (FRM) measurements at the central site	65
                                               VI

-------
                                          Tables

Number                                                                                Page

3-1   PM: 5 (ug/m3) and Sulfur Concentrations (ng/m3) Observed in Matched Indoor-Outdoor
        Samples in the RTF Study	5
3-2   Indoor and Outdoor 24-hour Sulfur Measurements (ng/m3) Averaged Over All Visits
        To Each Home	6
3-3   Ratios of the Mean Indoor Sulfur Concentrations to the Mean Outdoor Sulfur Concentrations	7
3-4   Air Exchange Rates (h"1) and Sulfur Indoor/Outdoor Ratios by Season	9
3-5   Results of Regressions of Indoor Sulfur on Outdoor Sulfur by Home, Compared to Average
        Values of the Indoor/Outdoor Sulfur Ratio	 10
3-6   Estimated Annual Average Contnbutions of Outdoor and Indoor-Generated Particles to
       Total Indoor PM25 Concentrations by House (ug/m3)	 12
3-7   Estimated Average Contributions of Outdoor Particles and Indoor-Generated Particles to
       Total Indoor Concentrations (ug/m3) by House and by Season	 14
3-8   Regressions of Indoor PM;5 of Outdoor Origin on Outdoor PM;5 Measured at the Home	  16
3-9   Regressions of Indoor on Outdoor PM:5 by House	 18
3-10 Sulfur Concentrations (ng/m3) and Ratios in Matched Personal, Indoor, and Outdoor Samples	21
3-11 Personal and Indoor Sulfur Concentrations (ng/m3) by Subject 	22
3-12 Ratios of the Mean Personal Sulfur Exposure to the Mean Outdoor Sulfur Concentration
        (Fp^) by Subject  	23
3-13 Companson of Personal/Outdoor, Indoor/Outdoor, and Personal/Indoor Sulfur Ratios by
        House and by Season	  24
3-14 Estimated Contribution of Outdoor Particles to Personal Exposure (ug/m3). Standard
        Deviations and Standard Errors Calculated by Propagation of Error	26
3-15 Contributions to Personal Exposure from PM25 Particles of Outdoor and Non-Outdoor Origin	27
3-16 Regression of Personal Exposure to Particles of Outdoor Origin on Outdoor Concentrations
        Measured Near Residence by Harvard Impactor (HI)	29
3-17 Regression of Personal Exposure to Particles of Outdoor Ongin on Outdoor Concentrations
        Measured at Central Site by Federal Reference Method (FRM)	 30
3-18 Time (in Minutes) Spent in Various Activities/Locations	 31
3-19 Estimates of P and k for Individual Homes Using Nonlinear Fit to the Indoor/Outdoor
        Sulfur Ratio	 34
3-20 Values for k when P is Bound from Above by 1	 35
3-21 Results of Linear Regressions of the Outdoor/Indoor Sulfur Ratio on Residence Time for
        36 Homes	37
3-22 Values for k When P is Bound from Above by 1	38
3-23 Multiple Regression of Outdoor Concentrations on Household Characteristics and Personal
        Activities 	41
3-24 Dependence of Indoor Fine Particle Concentrations on Household Characteristics and Personal
        Activities	 43
3-25 Dependence of Indoor-Generated and Outdoor-Generated Particles on Household Charactenstics
        and Personal Activities	44
3-26 Dependence of Outdoor Sulfur on Outdoor PM: 5 and of Indoor Sulfur on Outdoor
        Sulfur and Household Characteristics and Personal Activities	46
3-27 Regressions of Personal Exposures to PM:5 on Household Charactenstics and Personal Activities .. 47
                                              vn

-------
3-28 Regression of the Non-ambient-related Contribution (Perscontrib) to Personal PM2.s Exposure	 50
3-29 Regression of the Ambient-Related Contribution to Personal PM25 Exposure	 51
3-30 Regressions of Personal Exposure to Sulfur on Indoor and Outdoor Concentrations and
        Questionnaire Variables	 52
3-31 Variables Affecting Air Exchange Rate	54
3-32 Variables Affecting Indoor/Outdoor Sulfur Ratio	54
3-33 Variables Affecting Indoor/Outdoor Sulfur Ratio: Reduced Model	 54
A-l  Values of the Average Sulfur Indoor/Outdoor Ratio (F,n/) and the Air Exchange Rates by
       House and by Season  	66
A-2  Comparison of Seasonal Average Sulfur Indoor/Outdoor Ratios (Su/S^,) with Slopes of
       Regressions of Sin on Sout	69
A-3  Comparison of Seasonal Average Sulfur Personal/Outdoor Ratios (Spera/Sout) with
       Slopes of Regressions of Sperson Sout	72
A-4  Questionnaire Variables and Definitions	75
                                             Vlll

-------
                                 Acknowledgments
    The authors thank Linda Sheldon for many stimulating discussions regarding this report. Robert Kellogg
of Man Tech Environmental performed the X-ray fluorescence analyses that resulted in the excellent sulfur
data. Charles Rodes led the team at Research Triangle Institute International in collecting the field data.  The
Environmental Measurements and Analysis Branch of the National Exposure Research Laboratory  was
responsible for designing and overseeing the study. In particular, we acknowledge the contributions of Carry
Croghan, Anne Rea, Alan Vette, Carvin Stevens, Ten Conner, and Kelly Leovic. We wish to especially thank
the  participants who  earned  the  burden of responsibility  for up  to  28 days over  a year's time.
                                              IX

-------
                                                    Chapter 1
                                                  Introduction
Many studies worldwide in the last decade have documented an
association between health  effects and particle concentrations
measured  at central monitoring sites (Schwartz et al., 1996).
Since the health effects are  presumably related to personal
exposures, an important research need, identified by the National
Academy  of Sciences in 1995, is to determine  how personal
exposures correlate with these outdoor concentrations (NRC-
NAS  1998). A number of  studies (Abt  2000a,b: Allen et al.,
2003; Clayton et al., 1993; Ebelt et al., 2000; Evans et al., 2000:
Hopke et  al., 2003; Howard-Reed et al., 2000; Janssen et al.,
1997, 1998, 1999, 2000; Keeler et al., 2002; Landis et al., 2001;
Liu et al., 2003; Long et  al.,  2000, 2001,  Ozkaynak et al,
1996a,b; Pelhzzan et al., 1992; Rea et al.,  2001; Rojas-Bracho et
al., 2000; Samat et al., 2000,2001; Thomas et al., 1993; US EPA
2002,2003; Vette et al., 2001; Wallace et  al., 2003a; Williams et
al., 2000a,b, 2003a,b) have measured personal exposure directly
using personal  monitors,  and the  correlations of personal
exposure  with outdoor concentrations are  straightforward  to
determine  (Wallace, 2000).   However, the  correlation that
interests epidemiologists is not that between total personal
exposure and outdoor concentrations, but the correlation between
that component of personal exposure due to outdoor particles and
the outdoor concentrations.  This requires the ability to estimate
the contribution to personal  exposures from particles originating
outdoors.  Only a few studies have reported making this estimate
(Ebelt et al., 2000; Allen et al., 2003, 2004).  The goal of this
report is to estimate the contribution of outdoor particles  to
personal exposure for a group of 37 persons monitored one week
per season over four seasons in 2000-2001 (US EPA 2002,2003;
Williams et al., 2003a,b). The correlation between this portion
of personal exposure and PM2 5 outdoor concentrations will then
be calculated for each person. Many of the main findings of this
report appear in Wallace and Williams (2005).

Since people spend on average 89% of their time indoors, the
contribution of outdoor particles to  indoor concentrations will
also  be explored.   For many people, the indoor-outdoor
relationship may be the major determinant of the  personal-
outdoor relationship.

The contribution of outdoor particles to indoor concentrations is
described by the mass balance equation.  The full mass balance
equation includes such phenomena as coagulation, condensation,
and gas-to-particle conversion (Nazaroff and Cass, 1989). We
will consider here a simplified version involving only infiltration,
exfiltration, deposition, and indoor  sources.  The differential
form of this simplified mass balance equation is
           dCm dt = PaCout - (a+k) Cm-SV
(1-1)
where  Cm =  indoor number or mass concentration (cm"3 or
ug/m3),
Cou, = outdoor number or mass concentration
P = penetration coefficient across building envelope
a = air exchange rate (h"1)
(" = volume of building (cm3 or m3)
5 = source strength (h"1 or ug h"1)
k = deposition rate of particles (h"1)

The equation is assumed to be applicable to all particle sizes,
with all terms except air exchange rate  and building volume
considered to be functions of particle size. We assumed that the
entire house  is a single well-mixed zone,  with instantaneous
mixing of particles throughout the house, and that the measured
air exchange rate in one room applies to the entire house.  This
assumption is probably violated by most homes, which are likely
to have different zones on each floor or even on the same floor.
We also  assume that  the deposition rate is constant over the
period of integration. This assumption will not hold if persons
open  windows,  turn  on  fans,  run the  furnace  or the air
conditioner,  or  otherwise  make  changes  in the household
operating characteristics that will affect  particle deposition
during the integration period.   Finally, we assume that the
averaging time over which the equation is to be evaluated is
sufficiently long that transient terms due to short-term changes in
the outdoor concentration are negligible compared to the long-
term average concentrations. Since we are dealing with 24-hour
averages, this assumption is probably a good one.  Under these
assumptions, the solution to the mass balance equation is
        Cm = [Pa'(a-k)] * Cou, - S[V(a-k)]
(1-2)
The coefficient of the outdoor concentration is sometimes called
the infiltration factor Fm/.



The infiltration factor has a major effect on the indoor-outdoor
and the personal-outdoor relationships. This is expected to van-
by household and resident characteristics. For example, a tightly
built  house may have a lower penetration coefficient  than a
drafty house, although almost no data are available to support
this claim.  A house with a large surface area (e.g., many carpets,
rugs,  or fibrous wall hangings) may have higher deposition rates
(L ai and Nazaroff 2000). Use of fans or filters may also increase
particle deposition rates (Howard-Reed et al., 2003; Riley et al.,

-------
2002; Thatcher etal., 2002; Thatcher and Layton, 1995; Wallace
et al., 2004a). Open windows will increase F^ by increasing the
air  exchange rate (Howard-Reed et  al.,  2002; Wallace and
Howard-Reed,  2002) and possibly by redirecting infiltrating
particles through the open window (P = 1) rather than through
the rest of the building envelope (P < 1) (Liu and Nazaroff, 2001;
Mosley  et al., 2001;  Thornburg et  al.,  2001).   Use of  air
conditioning has been shown to lower the  infiltration factor,
either because of  reducing air exchange  rates by  shutting
windows or increasing deposition rates by recirculating indoor
air  through ductwork (Howard-Reed et al., 2003; Sarnat et  al.,
2000; Lai et al., 1999; Thornburg, 2004; Wallace et al., 2002).

Despite  these clear indications that  exposure to outdoor air
particles indoors depends heavily on household and behavioral
characteristics, studies capable of estimating  the infiltration
factor reliably for individual homes are rather few. Even more
rare are  studies capable of estimating the values of P and k for
individual homes (Allen et al., 2004).  In this study we attempt to
estimate both F,nf  and its parameters P  and k for individual
homes, together with an estimate of the uncertainties involved.

Several  investigators  have noted that sulfur has few  indoor
sources  (Ebelt et al., 2000; Samat et al., 2002).  If that is the
case, the source term in Equation 1 -2 above may be ignored,  and
the equation takes the very simple form

                    Stn'Sout = Fmf                      (1-4)

where Sm and S^, are the sulfur concentrations  indoors and
outdoors.

Indoor-outdoor  comparisons   of sulfur  concentrations thus
provide  a direct way to estimate Fmf for each individual home.
Strictly speaking, the sulfur data provides only information  for
particles with similar behavior to that of sulfur with respect to
penetration,  deposition, and reactivity.   Sulfur  particles  are
smaller than most other fine particles.  Since the particles in the
1-2.5  (am range  may  have higher deposition  velocities than
sulfur, the estimates for the sulfur deposition rate ks would be
slight underestimates for the typical aerosol mixture in the PM2 5
category.  To avoid extra notation,  we shall use  P  and k
throughout rather than PS and ks but the reader should remember
that these are values appropriate only for particles in the size
range of sulfur particles.  Later we provide  evidence that this
'•sulfur-related" infiltration factor is actually a very good estimate
of the PM2 5 infiltration factor.

-------
                                                  Chapter 2
                         Description of Study Methods and Database
A full description of the measurement methods used in this study
has been provided in the two preceding reports in this series (US
EPA 2002, 2003) and also in two published articles (Williams et
al, 2003a,b).  The following abbreviated description includes
only the methods discussed in this report.

In  this study, 37 persons  in 36 homes in the Research Triangle
Park, NC area were monitored for up to four seasons, 7 days per
season. (Thirteen of these were monitored for only one, two, or
three seasons; see Appendix A.) Two gravimetric monitor types
were employed for PM? 5 measurements: the Harvard Impactor
(HI), operating at 20  Lpm, and a Personal Exposure Monitor
(PEM), operating at 2  Lpm for a nominal 24-h period. The PEM
was used for personal samples and the HI for indoor-outdoor
samples, the latter measured just outside the home. An HI, a
PEM and a Federal Reference Method (FRM) monitor were also
operated every day of the  study at a central site. There were also
fixed PEM 10 monitors  co-located with the HI2.5 monitors—filters
from these instruments were analyzed for  a suite of elements
using x-ray fluorescence. Since sulfur particles are expected to
be in the size range <0.5  ^m, the PEM]0 filters should have the
same amount of sulfur as the PEMi 5 filters.)

The HI mass measurements had a precision of about 5%; the
PEM measurements had a precision of about 8% (Williams et al.,
2003a). The precision of the sulfur measurements was calculated
to  be about 8% (Kellogg, R, personal communication).

The data analysis concentrated on sulfur as a tracer of outdoor-
generated particles. Indoor-outdoor ratios were used to estimate
the infiltration factor  for PMi.,-; personal-outdoor ratios were
used to estimate the portion of personal exposure due to outdoor-
generated particles. The  estimates of exposure due to outdoor-
generated   particles  were  then   regressed  on  outdoor
measurements, both  at  the home and  at a central  site,  to
determine the relationship between central site measurements
and exposure to particles  of outdoor origin.
Participants filled out activity logs each day, identifying cooking,
cleaning, and other activities that might affect particle exposures.
A questionnaire on household characteristics was also completed
for each house. The full set of questionnaire variables and their
definitions and units is provided in Appendix A (Table A-4).
These variables were then included in multivariate regressions to
determine their influence on air exchange rates, the infiltration
factor, and  the observed PMis and sulfur concentrations.  A
detailed description of this analysis is provided in the section on
multivariate regressions.

-------
                                                     Chapter 3
                                                       Results
 A total of 876 person-days had at least one personal, indoor, or
 outdoor PM: 5 measurement. About 868 days had at least one
 filter analyzed by XRF.  Samples were flagged if they failed any
 of a number of quality control criteria. For example, flow rates
 were required to be within 10% of the target values, filters were
 discarded if seen to be torn or pierced. All outdoor PM2s filters
 collected between April 11 and April 17 were discarded due to
 excessive contamination with pollen. An additional requirement
 for the sulfur measurements  was  that indoor values not be more
 than 1.08 times outdoor values (the 1.08  factor was chosen
 because the estimated uncertainty of the sulfur measurements
 was 8%).  Even  if an indoor/outdoor ratio exceeding 1.08 were
 correct, it  would indicate  an indoor source of sulfur, and
 therefore should not be used in  analyses to determine  the
 infiltration  factor.   Six samples were flagged for exceeding a
 1.08 indoor/outdoor ratio.  Another 21 samples were flagged
 when regressions suggested  an indoor source  of sulfur,  and it
 was discovered that the participant was using a humidifier during
 three seasons  (7 days per season).  Since different combinations
 of variables will have different numbers of flags, the number of
 valid data points sometimes varies in the following tables.

 Calculation of fW Using the Indoor/Outdoor
 Sulfur Ratio
 Table 3-1  lists the distributional  characteristics of all paired
 indoor-outdoor  samples  with validated PM25  and  sulfur
 measurements. Also included in Table 3-1 are estimates of the
 indoor/outdoor sulfur ratio. The average indoor/outdoor sulfur
 ratio was 0.59 (0.16 SD).

 Assuming all the sulfur is in the form of ammonium sulfate, we
 can calculate  the mass  by  multiplying  by  the ratio  of the
 molecular weights (3.5). The  result is an estimate of 8.0 ug/m3
of ammonium  sulfate, or about 41% of the total PM: 5.  Several
Eastern cities included in EPA's speciation network ranged from
26-31% in their sulfate/PM:.3 values (AQCD 2003).
  All the validated indoor and outdoor sulfur concentrations (N:
  775) are displayed in Figure 3-1.
   4500

 _4000
H
 °>3500

 | 3000
 
-------
Table 3-1. PM2.5 (ug/m3) and Sulfur Concentrations (ng/m3) Observed in Matched Indoor-Outdoor Samples in the RTP Study

PM2.5ln
PM2.5 Out
Sulfur In
Sulfur Out
S In/Out
N
774
774
775
775
775
Mean
19.4
19.5
1116
1964
0.59
SD
16
9
653
1123
0.16
Min
2
5
123
404
0.17
10th
7
9
426
765
0.39
25th
10
13
615
1061
0.48
Median
15
19
974
1759
0.59
75th
22
24
1441
2667
0.69
90th
36
32
2034
3610
0.79
Max
119
52
3852
5406
1.06

-------
Table 3-2. Indoor and Outdoor 24-hour Sulfur Measurements (ng/m ) Averaged Over All Visits to Each Home
House
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
28
29
31
32
33
34
36
37
38
Sum/Mean
N
27
26
29
23
27
28
28
6
24
27
8
28
7
27
27
28
26
6
27
27
27
13
13
27
8
24
28
25
34
14
28
23
12
6
20
17
775
S,n
955
1069
1516
882
1199
1293
1476
1270
1087
1049
1661
944
553
701
928
1040
923
1328
1568
1139
1240
1231
1175
909
1827
1829
1335
1059
991
1678
607
522
1746
658
1079
534
1139
SD
372
477
534
359
723
651
901
408
688
558
825
449
224
342
449
554
467
436
769
641
704
607
601
498
679
1009
704
502
446
722
287
322
808
300
327
137
541
SE
72
94
99
75
139
123
170
167
141
107
292
85
85
66
86
105
92
178
148
123
135
168
167
96
240
206
133
100
76
193
54
67
233
123
73
33
126
Sou,
1532
1877
2078
2106
1892
2116
1947
3664
1968
1945
2446
1945
1534
1579
1322
1639
1889
2625
2198
2272
2152
2338
2558
1795
2712
2620
2113
1938
2001
1920
1730
1493
2811
2525
1836
974
2058
SD
612
1030
756
1056
1034
1008
1261
1376
1310
1202
1266
1149
661
774
588
952
1140
996
1218
1229
1322
1175
1508
1153
1292
1457
1276
1147
1269
839
769
978
1139
1130
698
299
1058
SE
118
202
140
220
199
191
238
562
267
231
447
217
250
149
113
180
224
407
234
237
254
326
418
222
457
297
241
229
218
224
145
204
329
461
156
72
252

-------
Table 3-3.  Ratios of the Mean Indoor Sulfur Concentrations to the Mean Outdoor Sulfur Concentrations
House
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
28
29
31
32
33
34
36
37
38
Mean
SJS00,
0.62
0.57
0.73
0.42
0.63
0.61
0.76
0.35
0.55
0.54
0.68
0.49
0.36
0.44
0.70
0.63
0.49
0.51
0.71
0.50
0.58
0.53
0.46
0.51
0.67
0.70
0.63
0.55
0.50
0.87
0.35
0.35
0.62
0.26
0.59
0.55
0.56a
SDca,,"
0.35
0.40
0.37
0.27
0.52
0.42
0.67
0.17
0.51
0.44
0.49
0.37
0.21
0.31
0.46
0.50
0.38
0.25
0.53
0.39
0.48
0.37
0.36
0.43
0.41
0.55
0.51
0.41
0.38
0.54
0.23
0.31
0.38
0.17
0.29
0.22
0.39
SEcalc
0.07
0.08
0.07
0.06
0.10
0.08
0.13
0.07
0.10
0.08
0.17
0.07
0.08
0.06
0.09
0.09
0.08
0.10
0.10
0.08
0.09
0.10
0.10
0.08
0.14
0.11
0.10
0.08
0.07
0.14
0.04
0.07
0.11
0.07
0.06
0.05
0.09
 3 This value obtained by dividing the mean indoor sulfur concentration for all homes by the mean outdoor sulfur concentration.  The arithmetic mean
 of the ratios was 0.59.
 b Values of the standard deviation and standard error are calculated by propagation of error.

-------
where SD<.atc - the calculated standard deviation of the ratio for
agiven house,
 100 ng/m3 and thus may be due to scatter rather than to
 indoor sources of sulfur or sulfates. The scatter is due both to
variations in FlnJ with changing seasons and air exchange rates
                                                           8

-------
Table 3-4.  Air Exchange Rates (h"1) and Sulfur Indoor/Outdoor Ratios by Season
Season
Summer
Fall
Winter
Spring
N
223
187
179
171
airex
0.49
0.61
1.01
0.68
SD
0.57
0.40
0.73
0.49
SJSout
0.50
0.63
0.63
0.62
SDa
0.16
0.14
0.13
0.16
3 Observed standard deviation; SD calculated by propagation of error would be larger.

-------
Table 3-5.  Results of Regressions of Indoor Sulfur on Outdoor Sulfur by Home, Compared to Average Values of the Indoor/Outdoor Sulfur Ratio
House
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
28
29
31
32
33
34
36
37
38
N
27
26
29
23
27
28
28
6
24
27
8
28
7
27
27
28
26
6
27
27
27
13
13
27
8
24
28
25
34
14
28
23
12
6
20
17
Oin/Oout
0.63
0.60
0.74
0.46
0.65
0.61
0.77
0.36
0.58
0.57
0.68
0.54
0.37
0.45
0.69
0.65
0.52
0.52
0.75
0.49
0.61
0.55
0.49
0.55
0.72
0.70
0.66
0.59
0.57
0.89
0.36
0.39
0.62
0.26
0.60
0.56
SE
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.05
0.02
0.02
0.04
0.02
0.04
0.02
0.02
0.02
0.02
0.05
0.02
0.02
0.02
0.03
0.03
0.02
0.04
0.02
0.02
0.02
0.02
0.03
0.02
0.02
0.03
0.05
0.02
0.03
Slope
0.57
0.45
0.68
0.28
0.60
0.63
0.69
0.27
0.49
0.42
0.63
0.36
0.32
0.41
0.72
0.57
0.36
0.42
0.60
0.50
0.47
0.44
0.39
0.40
0.38
0.67
0.53
0.38
0.30
0.84
0.30
0.27
0.64
0.26
0.41
0.40
SE
0.04
0.02
0.04
0.04
0.07
0.03
0.04
0.06
0.04
0.04
0.07
0.03
0.05
0.03
0.05
0.03
0.04
0.06
0.04
0.03
0.05
0.08
0.03
0.03
0.15
0.04
0.03
0.04
0.03
0.06
0.04
0.04
0.09
0.03
0.05
0.06
Inter.
86
231
103
287
70
-46
132
291
115
227
116
238
60
55
-23
111
238
226
239
-8
232
211
182
190
793
83
215
315
398
74
90
119
-65
10
326
140
SE
72
53
81
97
156
60
83
249
90
88
180
63
78
59
74
50
85
173
91
66
127
217
79
69
438
121
74
99
79
124
83
73
283
94
104
57
P
0.24
0.00
0.21
0.01
0.66
0.45
0.12
0.31
0.21
0.02
0.54
0.00
0.48
0.36
0.76
0.04
0.01
0.26
0.01
0.90
0.08
0.35
0.04
0.01
0.12
0.50
0.01
0.00
0.00
0.56
0.28
0.12
0.82
0.92
0.01
0.03
R2
0.86
0.93
0.93
0.67
0.72
0.96
0.93
0.77
0.88
0.82
0.93
0.86
0.88
0.85
0.88
0.94
0.78
0.90
0.91
0.94
0.77
0.69
0.95
0.85
0.45
0.92
0.92
0.76
0.70
0.94
0.63
0.66
0.81
0.92
0.75
0.76
                                                                10

-------
and measurement error.  These have the well-known effect of
lower  slopes  and  higher  intercepts  than the  true  values.
Therefore the calculated slopes are likely to be underestimates of
the true infiltration factor.  Only 4 of the 36 homes had slopes
higher than the mean indoor/outdoor sulfur ratio.  The overall
average indoor/outdoor sulfur ratio is 0.56,  compared to the
overall average slope of 0.49.

Comparison of Methods for Calculating Finf
The  methods  for calculating F,,,/ (regression  vs. the simple
indoor/outdoor ratio) are  compared  in  Figure  3-4.   The
comparison differentiates between the 22 homes with intercepts
not different  from  zero  and the  14  homes with intercepts
significantly different from zero. The latter set of homes cannot
be said with complete confidence to have no indoor sources of
sulfur, although random measurement errors could also cause
nonzero intercepts.  The slopes for these homes are generally
lower, as would be expected if measurement errors are affecting
the regressions.  Home number 25 is an outlier in this graph
(large red diamond), having the highest intercept of all homes,
but with high uncertainty, due to a small number of samples (N =
8). Without home 25, the 21  remaining homes with intercepts
not different from zero have slopes that are related to the mean
indoor/outdoor ratios with a hish R~ of 0.91.
      0.9
      0.8
      0.7
    £ 0.6
    I 0.5
    5 0.3
    1
    0-0-2
                         y = 0.94x - 0.03
                        R2 = 0.79 N = 22
                                              y= 1.10X-0.22
                                              R2 = 0.76 N = 14
                           0.4        06
                         Sulfur Indoor-Outdoor Ratio
                                               0.8
Figure 3-4.  Comparison of the results of regressing indoor sulfur on
outdoor sulfur with the simple indoor/outdoor ratio averaged over all
visits to a home.
their indoor PM: 5 concentrations supplied by outdoor air, and 7
homes had less than 40% supplied by outdoor air.
  1  1
  o
  £ 0.8
  Q.

  •o
  _c
  O 0.4
  c
  o
  « 0.2
	
             It
                             House ID

Figure 3-5.  Estimates of the fractional contribution of outdoor
particles to total indoor PM2 5 concentrations, averaged over all home
visits. Error bars are standard errors calculated by propagation of
error.

The outdoor and indoor-generated contributions to total indoor
PM: 5, averaged overall visits to each house, are shown in Figure
3-6. The relative importance of these two sources varies widely,
with some homes having essentially no indoor contribution and
others having more than 50% indoor contributions.
   60
              n Indoor-generated
              • Outdoor contribution
                                                               E  40
Estimating Indoor and Outdoor
Contributions to Indoor PM2 5
The indoor/outdoor sulfur ratio may be multiplied by the outdoor
air concentration to estimate the  contribution of outdoor air
particles to the total indoor particle level for each home (Table 3-
6).  The  indoor-generated particles are then the difference
between the total indoor PM: 5 and the outdoor contribution to
indoor PM: 5. The fractional contributions of outdoor air particles
to indoor concentrations, averaged over all visits to each home,
are shown in Figure 3-5. Fourteen homes had more than 80% of
Figure 3-6.  Comparison of average outdoor-generated and indoor-
generated particles based on indoor/outdoor sulfur ratios.

The 95% confidence limits are displayed separately for outdoor
and indoor contributions in Figures 3-7  and 3-8. (These values
are taken from Table 3-6.) The  indoor contributions cover a
wider range than the outdoor contributions. For example, three
homes had indoor sources producing an average concentration
over the  four seasons > 25 ug/nr, compared to no  homes with
outdoor contributions that high.  On the other hand. 20 homes
had indoor contributions of 5 u.g/mj or lower compared to only
                                                          11

-------
Table 3-6.  Estimated Annual Average Contributions of Outdoor and Indoor-Generated Particles to Total Indoor PM2 5 Concentrations by House (Mg/m3)
House
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
28
29
31
32
33
34
36
37
38
Sum/ Mean
N Outdoor contribution SD^
to indoor particles
27
26
29
23
27
28
28
6
24
27
8
28
7
27
27
28
26
6
27
27
27
13
13
27
8
24
28
25
34
14
28
23
12
6
20
17
775
11.0
13.8
19.4
9.4
12.2
11.7
18.0
9.6
11.3
10.9
15.5
9.1
4.8
7.4
10.8
10.7
8.7
10.7
17.2
10.8
13.0
10.6
9.9
9.4
21.4
17.6
12.7
10.4
9.7
21.7
5.8
6.4
15.4
6.2
10.4
7.4
11.7
12.9
18.8
16.6
16.8
17.4
15.0
22.6
15.8
20.0
18.9
17.8
16.1
9.6
13.3
11.4
15.1
15.4
11.9
20.4
19.8
20.6
15.6
19.0
16.6
19.1
22.3
18.1
15.8
16.8
20.7
12.3
18.0
16.7
17.4
10.4
8.3
16.5
SEcaic Indoor-generated SDcaic SE^t
particles
2.5
3.7
3.1
3.5
3.4
2.8
4.3
6.5
4.1
3.6
6.3
3.1
3.6
2.6
2.2
2.9
3.0
4.8
3.9
3.8
4.0
4.3
5.3
3.2
6.7
4.5
3.4
3.2
2.9
5.5
2.3
3.7
4.8
7.1
2.3
2.0
3.9
5.0
2.3
1.6
31.0
5.9
5.8
1.6
8.1
4.8
0.2
11.8
3.7
3.8
4.1
7.2
-0.5
4.1
1.4
3.3
-0.3
25.4
14.0
15.1
27.6
32.6
2.2
22.7
1.0
16.3
-0.2
7.3
5.1
1.3
3.0
1.3
0.3
7.8
15.9
21.3
20.0
37.0
20.9
16.5
25.5
16.5
22.0
19.3
22.2
16.9
10.4
14.8
20.3
15.6
18.6
12.2
21.5
20.2
29.6
23.5
33.7
29.1
29.0
24.2
32.1
16.6
22.7
23.2
13.6
20.0
17.3
17.5
10.9
8.8
20.5
3.1
4.2
3.7
7.7
4.0
3.1
4.8
6.7
4.5
3.7
7.8
3.2
3.9
2.8
3.9
3.0
3.6
5.0
4.1
3.9
5.7
6.5
9.4
5.6
10.3
4.9
6.1
3.3
3.9
6.2
2.6
4.2
5.0
7.2
2.4
2.1
4.8
                                                                 12

-------
one home with an outdoor contribution that low.  This reflects
much more variability in indoor particle-generating activities
compared with outdoor particle concentration variability.
Figure 3-7.  Estimates of average outdoor contribution to indoor PM25.
Error bars are standard errors calculated by propagation of error
techniques applied to the three measurements required to estimate the
outdoor contribution.
   35
„— 30

I"
5 20
CL
1 15
i 10
-   0

   -5

   -10
                               House ID
 Figure 3-8. Estimates of average indoor-generated PM35. Error bars
 are standard errors calculated by propagation of error techniques
 applied to the four measurements required to estimate indoor-
 generated  PM25.

 We saw in Figure 3-3 that the indoor/outdoor sulfur ratio varied
 by season, and was much lower in summer than the other three
 seasons.  We next look at the estimates of indoor and outdoor
 contributions by season (Table 3-7; Figures 3-9 through 3-12).
 Table 3-7 shows that indoor-generated particle concentrations
 were 45-46% of the total indoor particle concentrations  averaged
 across all homes in the summer and fall seasons, but only 31% in
 the winter and spring seasons.  This  was  due mostly to  a
 reduction in  indoor sources in the latter two seasons,  since the
 outdoor contribution  did not change  greatly  over  the four
 seasons. Because of the small number of measurements in each
 house and each season (N = 1-7), the uncertainty, particularly in
                                                              the estimates of indoor-generated concentrations, is much larger
                                                              than for the equivalent concentrations  for each home over all
                                                              seasons (as in Table 3-6). The average  standard error was  15%
                                                              for the estimates of the outdoor contributions, but 66% for the
                                                              indoor estimates.
                                                                                            House ID

                                                              Figure 3-9. Indoor-outdoor average contributions to indoor PM2 5.
                                                              Summer 2000.
                                                                   70


                                                                   60
                                                                 _ so
                                                                 "E
                                                                              a Indoor-generated
                                                                              • Outdoor contribution
                                                                                  II  I   ilil  ll
Will
                                                                                           House ID

                                                              Figure 3-10.  Indoor-outdoor average contributions to indoor PM2 5.
                                                              Fall 2000.
                                                             13

-------
Table 3-7.  Estimated Average Contributions of Outdoor Particles and Indoor-Generated Particles to Total Indoor Concentrations (M9/rn3) by House
and by Season
Summer 2000
House
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
28
29
31
32
33
34
36
37
38
Sum/Mean
N
8
6
7
2
6
7
7
6
6
6
6
7
7
7
6
6
7
6
6
7
7
6
7
7
2
6
6
5
13
7
7
5
7
5


216
Out
14.0
14.2
16.2
7.5
12.0
12.9
17.3
9.6
12.4
10.7
17.0
10.0
4.8
5.0
8.7
13.0
9.4
10.7
16.1
13.9
12.1
9.2
11.4
8.6
14.4
18.6
14.2
10.5
9.8
13.9
3.3
2.6
13.1
6.9


11.3
In
1.6
0.5
-0.1
10.9
16.5
8.5
5.6
8.1
3.5
1.8
15.4
5.7
3.8
4.9
1.5
-0.4
5.6
1.4
8.4
1.0
25.8
15.3
5.1
31.4
51.5
4.1
49.9
0.0
23.0
0.8
6.1
2.7
1.6
2.7


9.5
Fall 2000
N
6
4
6
7
6
5
7

5
7

7

6
7
7
5

7
6
6
7
6
7
6
6
7
7
6
7
7
5
3

5
3
186
Out
8.1
14.9
24.3
13.5
8.3
18.7
26.0

6.2
7.5

6.7

6.1
9.8
7.6
8.9

12.3
5.0
17.7
11.7
8.1
10.7
23.8
23.3
9.8
11.7
7.2
29.5
6.9
4.0
21.1

8.7
4.9
12.4
In
8.1
-1.8
11.8
44.8
8.7
3.5
3.1

11.8
1.1

8.3

2.7
18.5
1.3
0.6

5.3
2.6
28.1
13.0
26.9
23.1
26.2
4.2
27.0
1.8
11.3
-1.2
10.7
4.3
-0.7

3.7
1.9
10.0
Winter 2001
N
6
6
7
7
7
7
6

5
7

7

7
7
7
6

7
7
7


6

5
7
6
7

7
6


7
7
171
Out
10.8
14.3
23.6
8.0
16.6
9.3
15.0

15.6
17.3

10.4

10.4
13.1
11.4
9.7

25.2
16.0
13.9


11.9

21.5
15.2
11.9
12.9

6.7
10.1


12.2
8.0
13.5
In
12.5
-1.9
-6.8
43.9
-1.2
8.2
-0.1

8.8
-1.4

1.5

7.7
1.4
-1.7
1.2

-2.4
-4.9
18.6


27.9

-2.4
12.0
3.3
18.1

6.4
3.0


0.7
0.5
5.9
Spring 2001
N
2
7
7
4

7
5

5
5
2
7

7
3
7
1

7
7
6


7

5
7
6
5

7
4


7
5
142
Out
8.7
12.0
12.9
6.9

7.8
14.7

12.0
7.5
11.2
9.4

7.7
12.6
10.9
2.8

15.2
7.5
7.7


6.4

9.2
11.0
8.3
8.9

6.4
9.1


10.5
7.4
9.4
In
0.7
11.3
2.0
11.7

3.2
-1.8

-1.3
0.2
0.9
-0.7

0.9
3.4
-1.0
2.5

2.8
0.5
19.5


28.1

4.1
3.9
-0.9
9.7

6.1
4.2


0.3
-0.1
4.2
                                                                  14

-------
   30
   20
             D Indoor-generated
             • Outdoor contribution
      12345678 9 10 11 12 13 14 15 16 17 18 19 20 21 22 2
                            House ID
Figure 3-11.  Indoor-outdoor average contributions to indoor PM2 5.
Winter 2001.
 40

 35

 30 j

, 25

 20

 15

 10

  5

  0
             D Indoor-generated
             • Outdoor contribution
Figure 3-12.
Spring 2001.
         Indoor-outdoor average contributions to indoor PM:;
Relationship Between  Outdoor Particles
and  Indoor Particles of Outdoor Origin
As  described  in the Introduction,  a quantity  of  interest to
epidemiologists is the relationship between outdoor particles and
indoor particles of outdoor origin. Since epidemiologists work
with measurements made at a central site, we examined how this
relationship changes as we go from the particles measured just
outside the home to those measured at the central site, and as we
go from particles measured by  the same type of equipment to
those  measured  by  different monitors.  We first  performed
regressions for each home of  the estimated  contributions to
indoor PM:5  made by  particles of outdoor origin  vs. the
measured concentrations just outside the home (Table 3-8).
These individual regressions by home have high R: values in
most cases (median = 0.77; range 0.39-0.95).  For all 752 valid
daily measurements, the regression has a slope of 0.56 (0.01 SE)
and a non-significant intercept of 0.31 (0.31) (Table 3-8). The
Pearson correlation coefficient was 0.82, with an R" (adjusted)
value of 0.665.  (A Spearman rank correlation was also 0.82,
indicating that the Pearson coefficient was  not affected by
outliers.)  When the regression was run against the HI at the
central site, the R" value was reduced to 0.58 (N = 736). Against
the FRM at the central site, it was reduced to 0.49 (N = 775).
We caution that these  R~ values must be considered upper
bounds, because the assumptions of our model, which force the
infiltrated particles of outdoor  origin  to be proportional to
outdoor concentrations, virtually ensure that R" values will be
high.

Estimating  Contributions of Outdoor Air to
Indoor Concentrations Using the RCS Model
Another way of estimating an average infiltration factor, this
time using the PM: 5 mass data, is the Random Component
Superposition (RCS) model developed  by Ott and co-workers
(Ott et al., 2000).  The  RCS model simply assumes  that a
regression  of all the  measured  indoor vs. outdoor  mass
concentrations will provide an overall average infiltration factor
together with an estimate of the indoor source strength averaged
across all homes (that being the constant term in the regression).
The overall average infiltration factor estimated by this approach
is 0.60 (0.06 SE, N  = 774) (Figure 3-13), compared with the
average indoor/outdoor sulfur ratio of 0.589 (0.006 SE, N = 775)
from Table 3-1.  The close agreement of these two values for
infiltration factor, one based on particle concentrations alone and
the other based on sulfur concentrations  alone,  is important
evidence that using the sulfur  results to estimate fine particle
parameters is justified, at least for  these  overall averages.  It
should be noted that the RCS model determines only one average
value of Fml for all homes, whereas the methods discussed above
provide estimates for individual homes. The RCS approach as
applied to PM,,, values in Riverside, CA, Philipsburg, NJ, and
Toronto, Canada found similar values of 0.55, 0.60 and 0.61 for
the infiltration factor (Ott et al., 2000) compared to our values
using PM: j sulfur and mass data.  Further  evidence  for the
applicability of the sulfur results to the PM: 5 fraction is the close
agreement between the RCS  estimate of 7.7 ug/nv  for the
indoor-generated PM: ? compared to 7.8 ng/irr estimated using
the sulfur indoor-outdoor ratios (Table 3-6).
                                                         15

-------
Table 3-8. Regressions of Indoor PM2 s of Outdoor Origin on Outdoor PM2 5 Measured at the Home
House
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
28
29
31
32
33
34
36
37
38
Sum/Mean
N
27
26
24
23
27
28
28
6
24
22
8
28
7
27
27
28
26
6
22
22
27
13
13
27
8
21
28
25
34
14
28
23
12
6
20
17
752
Slope
0.53
0.45
0.62
0.35
0.67
0.63
0.81
0.22
0.52
0.37
0.67
0.32
0.31
0.41
0.66
0.60
0.41
0.40
0.57
0.52
0.58
0.38
0.38
0.47
0.78
0.69
0.52
0.37
0.27
0.74
0.31
0.38
0.59
0.26
0.43
0.52
0.56
SE
0.03
0.05
0.04
0.05
0.09
0.03
0.05
0.08
0.06
0.04
0.10
0.04
0.04
0.04
0.07
0.03
0.06
0.10
0.07
0.04
0.07
0.11
0.04
0.06
0.33
0.05
0.04
0.06
0.05
0.05
0.05
0.06
0.15
0.05
0.06
0.05
0.01
Intercept
1.47
3.06
2.34
1.71
-0.56
-0.35
-0.73
3.55
1.05
2.81
0.26
3.26
0.67
0.47
0.55
0.74
1.69
2.32
3.13
-0.30
0.47
2.95
1.88
1.03
-1.59
0.26
2.49
3.58
4.62
2.86
0.72
-0.07
0.53
-0.14
2.86
0.45
0.31
SE
0.65
1.16
0.99
1.18
1.81
0.68
1.25
2.14
1.31
0.86
2.46
0.71
0.64
0.68
1.15
0.52
1.18
2.15
1.40
0.75
1.58
2.20
0.90
1.21
9.98
1.36
0.85
1.24
1.10
1.40
0.92
1.13
3.80
1.23
1.16
0.67
0.31
P
0.03
0.01
0.03
0.16
0.76
0.61
0.56
0.17
0.43
0.00
0.92
0.00
0.34
0.50
0.64
0.17
0.16
0.34
0.04
0.69
0.77
0.21
0.06
0.40
0.88
0.85
0.01
0.01
0.00
0.06
0.44
0.95
0.89
0.91
0.02
0.51
0.32
R2
0.91
0.79
0.93
0.69
0.69
0.93
0.91
0.61
0.76
0.77
0.85
0.75
0.88
0.82
0.77
0.94
0.62
0.75
0.78
0.90
0.74
0.51
0.89
0.68
0.39
0.90
0.86
0.59
0.43
0.95
0.56
0.66
0.58
0.85
0.71
0.89
0.67
                                                             16

-------
   120
                                   = 060 (006 SE)» *
                                      R2 = 0.10: N = 774
                 ?V&": Vil^r
                 .V .I----'--.
      0       10       20      30      40       SO       60
                        Outdoor PM?3 (jig/m3)
Figure 3-13. Regression of indoor PM2.5 on residential outdoor
PM2.5data.

Many studios have PM; 5 measurements indoors and outdoors but
do not have sulfur or sulfate measurements to determine Fml.
The RCS model provides an estimate of the average infiltration
factor across all homes in a given study. The question arises: Is
it possible to apply the RCS model estimate of Fln/ for individual
homes from PM measurements alone?

The regression results show that only half (N =  18) of the homes
had slopes that were significantly different from zero (Table 3-
9). Also, three of the slopes were negative and two others were
greater than 1, both unphysical results.  By contrast, all 36 sulfur
regressions had slopes that were significantly different from zero
Comparing the slopes of the PM: 5 regressions to the estimates of
Fm/ from the sulfur indoor/outdoor ratios resulted in a low R"
value of 0.11 (Figure 3-14).
    1.5
  S 0.5
 a.
 £
   -0.5
            y = 0.69x + 0.15
Significant Slopes

Nonsignificant slopes
                0.2
                          0.4        0.6        0.8
                    l Based on Sulfur Indoor/Outdoor Ratios
Figure  3-14.    Estimates of the  infiltration  factor  F,nf from sulfur
indoor/outdoor ratios compared to regressions of indoor vs. outdoor fine
particles.  The  regression line shown  is for the 18 cases with slopes
significantly different from zero.
                                                   Estimates of the Contribution of Outdoor
                                                   Air Particles to Personal  Exposure
                                                   If total  personal exposure  E to fine  particles  is the sum of
                                                   exposures while indoors and exposures while outdoors, then the
                                                   following equation holds:
                                                                                  =./;„ c,n +./;„„ c.HII
                                                                                                        (3-2)
                                                   where /,„ and _/,'„„ are the fractions of time spent  indoors and
                                                   outdoors.

                                                   Exposure can also be expressed as the sum of outdoor and non-
                                                   outdoor sources:
                                                                    E = E,, +£„„
(3-3)
                                                   where E,, is the exposure due to outdoor sources and Em, is the
                                                   exposure due to non-outdoor sources.  The exposure  due to
                                                   outdoor sources may be written as
                                                                   EH — jm Fn,l Cml, + /„„ C,,iu
(3-4)
                                                   Now we assume that all of the time spent  indoors was spent
                                                   indoors at home where a measurement of outdoor concentration
                                                   is available, and all of the time spent both outdoors and in transit
                                                   is considered to involve exposure to the outdoor concentration.
                                                   This outdoor concentration may be that measured at the home or
                                                   at the central site if that is available.  In this case, the fractions
                                                  /„„ and /„, add up to 1.   We now define a factor /•),„ such that
                                                                   E = />v C,ml + £•„„
(3-5)
                                                   The factor F/F,.V plays a role with respect to personal exposure that
                                                   F,nt played  with respect to indoor concentrations.   Some
                                                   investigators have  labeled  Fpa as  an "attenuation  factor,"
                                                   marking the decrease in the effect, or attenuation, of outdoor
                                                   concentrations on  personal  exposure.  However,  this would
                                                   imply that attenuation increases as F,,t,, increases,  whereas the
                                                   reverse is true. We shall call Fpcx the "outdoor exposure factor."
                                                   indexing the contribution  of outdoor particles  to  personal
                                                   exposure.  From Equations 3-3 and 3-4, we have a relationship
                                                   for the outdoor exposure factor /•),,.,:
                                                                   * pex   Jin*mj  Jottl

                                                   Rewriting the equation as

                                                                   ppa-  = F,n/-+. /;,,,,
-------
Table 3-9. Regressions of Indoor on Outdoor PM2 5 by House
House
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
28
29
31
32
33
34
36
37
38
N
27
26
24
23
27
28
28
6
27
22
9
28
7
27
28
28
27
21
22
22
28
13
13
27
8
21
28
25
34
14
28
23
12
6
20
17
Slope
0.37
0.43
0.61
0.67
0.72
0.44
0.99
-0.22
0.35
0.31
-0.25
0.23
0.51
0.33
0.04
0.48
0.80
0.52
0.66
0.44
0.47
0.91
-0.55
0.42
1.66
0.75
0.24
0.36
1.03
0.69
0.22
0.34
0.48
0.11
0.29
0.44
SE
0.21
0.22
0.23
0.76
0.24
0.17
0.16
0.26
0.19
0.05
0.68
0.11
0.25
0.18
0.51
0.05
0.24
0.17
0.12
0.04
0.38
0.69
0.85
0.62
1.18
0.14
0.60
0.12
0.27
0.11
0.18
0.19
0.16
0.10
0.12
0.06
p-level
0.09
0.07
0.01
0.39
0.01
0.02
0.00
0.45
0.08
0.00
0.73
0.04
0.09
0.08
0.94
0.00
0.00
0.01
0.00
0.00
0.23
0.22
0.53
0.51
0.21
0.00
0.69
0.01
0.00
0.00
0.24
0.08
0.01
0.31
0.03
0.00
Intercept
9.4
6.0
6.4
26.0
4.5
9.0
-3.4
23.9
9.4
4.5
34.4
8.6
1.8
6.0
17.5
2.1
-1.1
5.9
6.4
2.3
28.2
6.6
36.6
29.6
4.6
1.8
30.6
4.8
6.8
4.0
9.6
5.8
4.7
6.4
6.6
1.8
SE
4.1
5.6
6.2
17.8
5.0
3.5
3.9
7.5
4.0
0.9
15.6
2.1
3.5
3.3
8.9
0.9
4.4
3.5
2.6
0.8
8.9
14.5
19.6
11.9
35.8
3.7
13.0
2.4
5.5
3.0
3.2
3.6
4.1
2.4
2.3
0.9
p-level
0.03
0.30
0.31
0.16
0.38
0.02
0.39
0.03
0.03
0.00
0.06
0.00
0.64
0.08
0.06
0.03
0.81
0.11
0.02
0.01
0.00
0.66
0.09
0.02
0.90
0.64
0.03
0.06
0.22
0.22
0.01
0.12
0.28
0.06
0.01
0.05
R2
0.08
0.10
0.21
0.00
0.23
0.18
0.59
0.00
0.08
0.67
0.00
0.12
0.35
0.08
0.00
0.76
0.29
0.30
0.58
0.85
0.02
0.06
0.00
0.00
0.12
0.58
0.00
0.24
0.29
0.76
0.02
0.09
0.43
0.07
0.19
0.78
                                                             18

-------
Several articles have discussed this idea that personal exposure
to particles  of outdoor origin may  be  greater than indoor
concentrations of such particles (Wilson and Suh, 1997; Wilson,
Mage, and Grant 1998).  The central idea is that persons spend
some time outdoors or in vehicles where they are exposed to the
full outdoor concentrations, thus adding to their exposures
indoors where they are "protected" somewhat by the infiltration
factor Finf.

As an example, suppose fin = 0.89, as found by the NHAPS
study, and/™, = {fraction of time outdoors (6%) plus fraction of
time in vehicles (5%)} =0.11. Setting Finf= 0.60 as found by the
RCS model, we find that Fpex would be 0.64, an increase of 7%.

Outdoor Exposure Factor Fpex Estimated
Using PM Measurements
A regression of personal PM2.5 on outdoor PM2.5 concentrations
provides an estimate of Fpex averaged across all participants: 0.64
+ 0.01  SE (Figure 3-15). However, the fit is poor (R2 = 0.09).
The intercept in this case (11.6 + 1.4 ug/m3) is an estimate of £„„,
the average  exposure  due  to  non-outdoor  sources.    The
difference between the indoor source contributions (estimated at
8.0 ug/m3 in Table 3-1) and the non-outdoor source contribution
is sometimes attributed to the "personal cloud," which in this
case would equal 3.6  + 2.0 ug/m3.
    250
    200
  3 150
  :
  I
  | 100

  •
  B.
     50
                     R* = 0.09: N = 776
 .". •:•"•  ''••  ••  t
, ••«.	• • J   _ .-\    *
               10       20       30       40

                         Outdoor PMj , uig/m3)
                                               50
                                                       60
Figure 3-15. 24-h average fine particle personal exposures vs. outdoor
air concentrations.

One outlier in Figure 3-15 (personal PM2.5 > 200 ug/m3) has a
strong  influence on the slope of the line.  If the outlier is
removed, the slope decreases from 0.64 to 0.59 (Figure 3-16),
while the intercept increases from 11.6 to 12.3.  This  would
increase the personal cloud from 3.6 to 4.3 ug/m3. The similarity
of the slopes for the personal vs. outdoor and indoor vs. outdoor
regressions (0.59-0.64 compared to 0.61) suggests that the time
spent indoors drives the relationship between personal exposure
and outdoor concentrations. That is, the infiltration factor F-mf,
which governs the reduction of particle concentrations as they
enter a house, is very similar to the outdoor exposure factor Fpac,
                                           which  governs  the  reduction  in  outdoor  concentrations
                                           contributing to personal exposure.

                                                         10       20      30      40      50       60
                                           Figure 3-16.  Personal vs. outdoor PM2.s; one outlier removed.


                                           Estimating the Outdoor Exposure Factor
                                           Fpex Using Sulfur Measurements
                                           Another estimate of the outdoor exposure factor is given by the
                                           regression of personal sulfur exposures vs. outdoor sulfur
                                           concentrations (Figure 3-17). The slope is 0.49, identical to the
                                           slope of the indoor/outdoor sulfur regression (Figure 3-1), while
                                           the intercept of 90 ng/m3 is less than the indoor/outdoor intercept
                                           of 1 50 ng/m3. The estimate using sulfur (R2=0.78) is preferred to
                                           the estimate using PM2.5 (R2=0.09).
3500


3000


2500 -
                                             =. 2000
                                             ta

                                             | 1500
                                                                1000
                                                                500
                                                        y = 0.49X + 89.5
                                                       Rz = 0.78; N = 750
                                                                                            .•*"••
                                                        1000     2000     3000     4000     5000
                                                             Outdoor Sulfur Concentration (ng/m3)
                                                                                              6000
                                           Figure 3-17. Personal vs. outdoor sulfur.


                                           Comparison of Fpex and Finf
                                           Both the  PM2.5  and sulfur  regressions  fail to support the
                                           argument based on Equation 3-7 that F^ is larger than F-mf. One
                                           possibility is that time spent in office buildings, which usually
                                           recirculate a portion of the air through filters, will involve lower
                                           exposures to sulfur than time spent in homes  without such
                                           recirculation and refiltration.   Some evidence for this can be
                                           found in the PTEAM study, where persons who worked away
                                                         19

-------
from home had lower particle exposures than those who stayed
home (Ozkaynak et al., 1996). In that case, the proper equation
for total exposure would have to take account of the different
concentrations in homes and offices:

           £  Jinhome *~ inhume ~*~ Join *~ oul+J inoffice ^inoffice    (j"°)

where the relationship Cinoffli.e < CMmmt would hold.  This will
reduce the magnitude of the outdoor factor.

Comparison of Indoor-Outdoor Sulfur and
Personal Sulfur Measurements
Since the personal samples were collected using the 2 Lpm PEM,
but the indoor and outdoor samples were collected using the 20
Lpm HI,  \ve need  to address the question of how the  two
samplers  agree.  During the summer season,  166 co-located
indoor and outdoor samples were collected. As showii in Figures
3-18 and 3-19, the agreement was excellent (R: = 0.97 for both).
with the  slopes near 1 for both.  This agreement between
different  monitor  types  is  justification for  comparing  the
indoor/outdoor sulfur ratio (the infiltration factor F,,,,) with the
personal/outdoor sulfur ratio (the outdoor exposure factor Fp^).
     6000
     5000
     4000
   £ 3000
   i
     2000
     1000
y=1.03(0.03SE)x+41 (28)
    R1 = 0.97. N = 166
                        2000     3000     4000

                              HI (ng/mj)
 Figure 3-18.  Co-located PEM2sand HI25 sulfur concentrations
 outdoors.  Summer 2000
                                                  4000 -

                                                      I
                                                  3500 -

                                                      i
                                                  3000


                                                 - 2500
y = 0.98 (0.03 SE) X +47 (33)
    R2 = 0.97. N = 166
                                                            500   1000   1500   2000   2500

                                                                            HI (ng/m1)
                                                                                         3000   3500   4000
                                               Figure 3-19.  Co-located PEM25and HI25 sulfur concentrations
                                               indoors  Summer 2000

                                               The overall comparison shows that personal sulfur exposures
                                               were somewhat smaller  than indoor concentrations  at home
                                               (Table 3-10).  When compared on a home-by-home basis, only
                                               33% of the participants had sulfur exposures higher than indoor
                                               concentrations (Table 3-11; Figure 3-20). The ratio of the mean
                                               personal sulfur concentration averaged over all visits to a home
                                               to the mean outdoor sulfur concentration (Fpfx) is 0.54 (Table 3-
                                                12).
                                                 0.9 .

                                                 0.8

                                                 0.7 -

                                                 0.6 -

                                                 0.5 •

                                                 0.4.

                                                 0.3

                                                 0 2

                                                 a,
                                                  o •	  .
                                                    it ;  ) u 14 A 11 .^ i& s /; i u i. 17 /i f t » it w n ft !• it n » u " " '• i n w u '
                                                                            Participant
                                               Figure 3-20. Comparison of infiltration factor F,n, and outdoor exposure
                                               factor Fpe, by participant.

                                               Finally, only  32% of 117 cases compared by home and season
                                               had higher personal sulfur exposures than indoor sulfur levels
                                               (Table 3-13). Only in the summer  season  was the  average
                                               exposure higher than the average indoor concentration, but only
                                               by  1%.  18  of 35  homes had higher exposures than indoor
                                               concentrations in the summer, compared to only 8,4, and 8 cases
                                               in the other seasons.   In most cases, /•},„  was less than F,,,;,
                                               contrary to the assumption that personal exposures to particles ot
                                               outdoor origin would be larger than indoor concentrations due to
                                               the time  spent outdoors.   This supports the idea that indoor
                                               concentrations in buildings, where our participants spent some ot
                                                           20

-------
Table 3-10.  Sulfur Concentrations (ng/m3) and Ratios in Matched Personal, Indoor, and Outdoor Samples
Concentrations/ratios
Personal
Indoor
Outdoor
Indoor/outdoor
Personal/outdoor
Personal/indoor
N
780
780
750
745
750
780
Mean
1062
1116
1944
0.59
0.55
0.97
SD
632
655
1140
0.17
0.14
0.23
Min
137
123
323
0.17
0.16
0.28
10th
393
423
755
0.38
0.38
0.75
25th
556
621
1037
0.48
0.46
0.83
Median
915
957
1671
0.58
0.54
0.92
75th
1419
1442
2592
0.69
0.63
1.03
90th
1975
2039
3647
0.80
0.73
1.21
Max
3254
3852
5406
1.08
1.08
2.64
                                                               21

-------
Table 3-11.  Personal and Indoor Sulfur Concentrations (ng/m3) by Subject
Subject
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
28
29
31
32
33
34
35
36
37
38
Mean
N
27
24
26
25
26
27
27
7
28
27
10
27
7
26
27
27
26
7
28
28
28
14
13
25
8
24
28
27
28
14
28
25
12
6
6
20
17
780
Personal
854
910
1338
934
1254
1082
1210
1256
1124
940
1671
892
643
707
841
951
878
1398
1464
1104
1369
1329
1209
1012
1686
1562
1101
896
996
1469
811
485
1692
1056
1143
940
511
1101
SE
71
86
108
105
141
117
155
165
141
105
264
85
112
69
80
108
93
189
139
128
149
189
213
119
246
175
106
88
109
161
65
54
224
170
249
82
41
132
Indoor
932
1058
1508
853
1196
1281
1488
1234
1130
1029
1542
959
553
720
950
1028
976
1421
1597
1120
1211
1285
1175
880
1827
1833
1335
1079
989
1678
607
524
1746
997
658
1079
534
1135
SE
74
86
108
72
145
127
176
145
128
106
251
87 '
85
66
87
108
103
177
146
120
134
165
167
98
240
206
133
97
88
193
54
65
233
154
123
73
33
126
Pers/ln
0.91
0.84
0.88
1.08
1.07
0.85
0.79
1.01
0.97
0.90
1.08
0.93
1.16
0.97
0.88
0.91
0.91
0.97
0.91
0.97
1.12
1.01
0.99
1.15
0.91
0.85
0.84
0.82
0.99
0.88
1.43
0.97
0.97
1.05
1.77
0.86
0.96
0.99
SE
0.01
0.02
0.02
0.07
0.04
0.02
0.02
0.04
0.04
0.02
0.06
0.02
0.10
0.02
0.02
0.01
0.04
0.04
0.02
0.02
0.04
0.04
0.04
0.06
0.03
0.02
0.02
0.02
0.06
0.02
0.08
0.04
0.02
0.03
0.17
0.04
0.04
0.04
                                                               22

-------
Table 3-12.  Ratios of the Mean Personal Sulfur Exposure to the Mean Outdoor Sulfur Concentration (F^) by Subject
Subject
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
28
29
31
32
33
34
35
36
37
38
Sum/Mean
N
25
23
21
23
26
27
27
6
25
21
8
27
7
26
27
27
25
6
22
22
28
13
14
23
14
21
28
25
27
14
28
22
13
27
6
21
20
765
Spere/Sout
0.57
0.49
0.64
0.46
0.67
0.51
0.62
0.35
0.57
0.46
0.73
0.45
0.42
0.44
0.63
0.59
0.46
0.50
0.65
0.49
0.66
0.54
0.47
0.55
0.59
0.60
0.52
0.45
0.53
0.77
0.47
0.33
0.63
0.60
0.45
0.52
0.53
0.54a
SDb
0.34
0.33
0.37
0.37
0.54
0.38
0.58
0.18
0.54
0.44
0.53
0.34
0.26
0.30
0.42
0.49
0.38
0.27
0.55
0.45
0.57
0.40
0.44
0.47
0.36
0.50
0.41
0.36
0.47
0.46
0.29
0.29
0.37
0.42
0.32
0.29
0.33
0.40
SEb
0.07
0.07
0.08
0.08
0.11
0.07
0.11
0.08
0.11
0.10
0.19
0.06
0.10
0.06
0.08
0.09
0.08
0.11
0.12
0.10
0.11
0.11
0.12
0.10
0.10
0.11
0.08
0.07
0.09
0.12
0.05
0.06
0.10
0.08
0.13
0.06
0.07
0.09
a Ratio of the mean personal sulfur concentration to the mean outdoor sulfur concentration. The mean of the ratios is also 0.54.
"Standard Deviations (SO) and Standard Errors (SE) Calculated by Error Propagation.
                                                                23

-------
Table 3-13.  Comparison of Personal/Outdoor, Indoor/Outdoor, and Personal/Indoor Sulfur Ratios by House and by Season
Summer
House
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
28
29
31
32
33
34
36
37
38
Sum/Mean
N
6
6
5
2
6
6
6
6
6
5
6
7
7
7
6
6
6
6
6
7
7
6
7
7
7
6
7
5
14
7
7
4
7
6


215
P/Oa
0.50
0.46
0.72
0.41
0.57
0.43
0.54
0.35
0.47
0.45
0.74
0.39
0.42
0.39
0.60
0.57
0.45
0.50
0.59
0.54
0.55
0.46
0.47
0.54
0.51
0.52
0.45
0.39
0.42
0.91
0.49
0.25
0.52
0.45


0.50
I/O"
0.57
0.57
0.83
0.28
0.50
0.57
0.67
0.35
0.48
0.45
0.68
0.40
0.36
0.39
0.58
0.57
0.42
0.51
0.62
0.53
0.43
0.43
0.43
0.41
0.56
0.62
0.54
0.40
0.39
0.96
0.25
0.20
0.54
0.26


0.49
P/lc
0.88
0.81
0.86
1.45
1.13
0.75
0.81
1.02
0.97
1.00
1.09
0.98
1.16
1.02
1.04
0.99
1.08
0.99
0.96
1.01
1.27
1.08
1.10
1.32
0.92
0.84
0.83
0.97
1.07
0.95
1.95
1.27
0.97
1.74


1.01
N
6
6
7
7
6
7
7

7
7

7

6
7
7
6

7
6
7
7
7
6
7
6
7
7
6
7
7
6
6

7
6
205
Fall
P/O
0.58
0.48
0.64
0.46
0.66
0.57
0.70

0.58
0.44

0.50

0.44
0.63
0.65
0.38

0.71
0.44
0.74
0.63
0.48
0.52
0.71
0.69
0.61
0.51
0.62
0.67
0.51
0.39
0.75

0.52
0.54
0.58
I/O
0.63
0.51
0.70
0.47
0.66
0.63
0.84

0.69
0.57

0.56

0.46
0.74
0.72
0.48

0.79
0.43
0.70
0.64
0.58
0.51
0.80
0.78
0.77
0.66
0.59
0.82
0.40
0.38
0.84

0.65
0.54
0.64
P/l
0.93
0.93
0.91
0.98
1.01
0.89
0.82

0.84
0.77

0.89

0.96
0.85
0.91
0.80

0.90
1.01
1.05
0.98
0.83
1.02
0.89
0.88
0.80
0.77
1.06
0.82
1.29
1.01
0.89

0.80
1.00
0.91
N
7
6
7
7
7
7
7

7
7
2
7

6
7
7
7

7
7
7


6

7
7
7
7

7
5


7
7
179
Winter
P/O
0.60
0.48
0.59
0.48
0.66
0.50
0.58

0.57
0.50
0.60
0.56

0.45
0.64
0.55
0.47

0.71
0.41
0.71


0.60

0.56
0.58
0.49
0.56

0.56
0.47


0.54
0.56
0.55
I/O
0.67
0.63
0.67
0.48
0.72
0.59
0.76

0.60
0.59
0.67
0.62

0.47
0.77
0.67
0.56

0.82
0.47
0.68


0.57

0.68
0.66
0.61
0.68

0.48
0.50


0.67
0.63
0.62
P/l
0.89
0.76
0.88
1.00
0.92
0.85
0.77

0.94
0.86
0.90
0.91

0.95
0.82
0.82
0.85

0.86
0.86
1.05


1.06

0.81
0.88
0.80
0.82

1.18
0.95


0.81
0.90
0.89
N
6
5
2
7
7
7
7

5
2

6

7
7
7
6

2
2
7


4

2
7
6
7

7
7


7
7
146
Spring
P/O
0.63
0.59
0.59
0.48
0.80
0.53
0.71

0.70
0.47

0.43

0.46
0.66
0.57
0.57

0.66
0.52
0.75


0.64

0.74
0.54
0.47
0.62

0.37
0.29


0.51
0.51
0.56
I/O
0.64
0.66
0.74
0.39
0.74
0.61
0.85

0.63
0.50

0.46

0.46
0.78
0.62
0.58

0.68
0.51
0.67


0.62

0.78
0.67
0.60
0.69

0.30
0.37


0.53
0.50
0.59
P/l
0.99
0.90
0.79
1.25
1.09
0.87
0.84

1.12
0.93

0.93

0.99
0.85
0.92
0.98

0.97
1.02
1.13


1.04

0.95
0.81
0.78
0.90

1.22
0.80


0.96
1.03
0.95
 a Personal/outdoor sulfur concentration ratio (outdoor exposure factor)
 " Indoor/outdoor sulfur concentration ratio (infiltration factor)
 c Personal/indoor sulfur concentration ratio
                                                                   24

-------
their time, may be lower than in homes, due to recirculation and
filtration of outdoor air. The fact that the summer season gave
similar  estimates   for   personal   exposure   and  indoor
concentrations  of  sulfur  is  additional  evidence  for  this
hypothesis, since air conditioning was in general use in summer
and would provide extra filtration for the indoor sulfur particles.

Use of the Outdoor Exposure Factor Fpex to
Calculate the Contribution to Personal
Exposure Made by Particles of Outdoor
Origin
We next used our estimated outdoor exposure factors Fnex for
each subject (based  on the sulfur personal/outdoor ratio)  to
calculate the contribution  of particles  of outdoor origin  to
personal exposure (Table 3-14; Figures 3-21 through 3-23).  On
average, particles  of outdoor origin contributed 10.9  ug/nv
(47%) of the total personal exposure compared with 12.5 ug/rrr
(53%) for the particles of non-outdoor origin.  The range of the
non-outdoor-generated particle contributions (5-33 ug/mj) was
much  greater  than  the   range  of  the outdoor-generated
contributions (6-19 ug/m').

The contribution of panicles of non-outdoor origin to personal
exposure can be  subdivided  into particles generated indoors
while the person is at home and particles from all other sources.
                                   » ***************
                             Participant
 Figure 3-21.  Outdoor contributions to personal exposure. Error bars
 are standard errors calculated by propagation of error.
                             Participant

Figure 3-22.  Non-outdoor contributions to personal exposure. Error
bars are standard errors calculated by propagation of error.
   70
  01
  d>
  ~ 40

  B.
  •3 30
  c
  o
  at
  | 20


   10
            [3 Non-outdoor contribution
            • Outdoor contribution
                           Participant
Figure 3-23.  Outdoor and non-outdoor contributions to personal PM2s
exposure.

The  particles from  other sources include  particles generated
while the person is indoors at other locations or in a vehicle, as
well as particles from the "personal cloud" that may be generated
throughout the day at all locations (Yakovlcva ct al, 1999). The
indoor-generated particle concentrations estimated from the Flnl
values can be multiplied by the fraction of time spent indoors at
home (from the activity logs) to calculate the contribution of
indoor-generated particles to personal exposure.  The outdoor
contribution  is  calculated  by  multiplying  the   outdoor
concentration times the fraction of time spent outdoors. The
contribution of indoor-generated  particles at all  other indoor
locations as well as the particles from the personal cloud is then
found by subtracting the sum of the particles of outdoor origin
and the indoor-generated particles while at home from the total
personal exposures (Table 3-15; Figure 3-24).
                                                           25

-------
Table 3-14. Estimated Contribution of Outdoor Particles to Personal Exposure (ng/m3). Standard Deviations and Standard Errors Calculated by
Propagation of Error
Subject
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
28
29
31
32
33
34
35
36
37
38
Sum/Mean
^ Outdoor contribution to -p. „_ Nonoutdoor contribution to „_. „_
personal exposure personal exposure
25
23
21
23
26
27
27
6
25
21
8
27
7
26
27
27
25
6
22
22
28
13
14
23
14
21
28
25
27
14
28
22
13
28
6
21
20
766
9.9
11.6
15.5
9.8
12.8
9.8
14.5
9.7
11.2
7.9
16.4
8.4
5.6
7.3
10.3
9.7
8.2
10.5
13.3
9.0
14.2
10.6
9.9
9.8
17.8
14.8
10.5
8.5
9.6
18.8
7.8
5.9
15.5
9.6
10.6
9.0
7.1
10.9
13.3 2.7
17.9 3.7
17.4 3.8
19.4 4.0
17.2 3.4
15.9 3.1
23.8 4.6
16.5 6.7
20.5 4.1
18.9 4.1
17.9 6.3
16.3 3.1
10.0 3.8
13.3 2.6
12.6 2.4
15.8 3.0
16.1 3.2
12.4 5.1
18.7 4.0
19.1 4.1
20.8 3.9
16.3 4.5
21.2 5.7
16.8 3.5
20.1 5.4
22.6 4.9
17.9 3.4
16.3 3.3
17.6 3.4
20.5 5.5
11.8 2.2
17.7 3.8
16.0 4.4
13.7 2.6
18.6 7.6
11.2 2.5
10.1 2.3
16.9 4.0
10.9
7.4
19.0
29.6
16.6
11.7
10.4
13.4
9.0
5.1
13.5
5.5
7.9
5.2
10.0
7.8
12.8
5.9
6.6
6.7
31.9
18.4
15.7
27.1
33.0
19.5
13.7
9.1
9.0
5.9
8.7
9.9
5.4
13.9
7.3
10.3
6.6
12.5
16.2 3.2
20.3 4.2
33.1 7.2
36.1 7.5
29.8 5.8
25.6 4.9
29.7 5.7
17.8 7.3
22.8 4.6
19.7 4.3
21.9 7.8
16.9 3.2
13.7 5.2
14.6 2.9
16.7 3.2
17.8 3.4
23.5 4.7
13.2 5.4
20.1 4.3
20.4 4.3
24.9 4.7
23.5 6.5
26.9 7.2
26.2 5.5
52.7 14.1
27.1 5.9
22.0 4.1
18.7 3.7
19.0 3.7
24.5 6.5
14.3 2.7
21.4 4.6
17.9 5.0
10.4 3.4
19.4 7.9
16.5 3.6
11.6 2.6
21.8 5.2
                                                                   26

-------
Table 3-15.  Contributions to Personal Exposure from PM2 5 Particles of Outdoor and Non-Outdoor Origin
Subject Fraction of time at
home
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
28
29
31
32
33
34
35
36
37
38
Mean
SD
0.90
0.77
0.89
0.83
0.78
0.88
0.84
0.75
0.73
0.90
0.85
0.84
0.89
0.86
0.92
0.79
0.88
0.84
0.87
0.71
0.61
0.88
0.70
0.81
0.69
0.83
0.87
0.84
0.91
0.87
0.67
0.95
0.83
0.97
0.74
0.87
0.68
0.82
0.08
Outdoor contribution to
personal exposure3
9.9
11.6
15.5
10.0
12.8
9.8
14.5
9.7
11.1
7.9
16.4
8.4
5.6
7.3
9.8
9.7
8.2
10.5
13.3
9.0
14.6
10.6
10.0
9.8
19.3
14.6
10.5
8.5
9.6
18.8
7.8
5.9
14.9
9.5
10.6
9.1
7.1
10.9
3.3
Home contribution" Other0 PM2 5 foutd
Personal exposure
5.0
2.1
3.7
26.4
4.7
5.2
1.3
6.1
3.5
0.6
10.0
3.3
3.4
3.6
6.8
-0.4
3.7
1.2
4.3
0.8
15.5
12.4
10.7
21.5
22.6
2.2
19.7
0.9
16.1
-0.2
4.9
5.0
1.0
11.6
2.2
1.1
0.2
6.6
7.0
5.9
5.3
15.2
3.2
12.0
6.5
9.1
7.3
5.5
4.5
3.6
2.3
4.5
1.7
3.2
8.2
9.1
4.7
2.4
5.9
16.5
6.0
5.0
5.6
10.5
17.3
-6.0
8.3
-7.1
6.0
3.8
4.9
4.4
2.3
5.1
9.2
6.4
5.9
4.8
20.8
19.0
34.5
39.6
29.5
21.5
24.9
23.1
20.2
13.0
29.9
14.0
13.5
12.6
19.8
17.5
21.0
16.4
19.9
15.6
46.5
29.0
25.7
36.9
52.4
34.1
24.2
17.6
18.6
24.6
16.5
15.8
20.3
23.4
17.9
19.4
13.6
23.3
9.3
0.48
0.61
0.45
0.25
0.43
0.46
0.58
0.42
0.55
0.61
0.55
0.60
0.41
0.58
0.49
0.55
0.39
0.64
0.67
0.58
0.31
0.37
0.39
0.27
0.37
0.43
0.43
0.48
0.52
0.76
0.47
0.37
0.73
0.41
0.59
0.47
0.52
0.49
0.12
c
0.24
0.11
0.11
0.67
0.16
0.24
0.05
0.26
0.17
0.05
0.33
0.24
0.25
0.29
0.34
-0.02
0.18
0.07
0.22
0.05
0.33
0.43
0.42
0.58
0.43
0.06
0.81
0.05
0.87
-0.01
0.30
0.32
0.05
0.50
0.12
0.06
0.01
0.25
0.22
» '
lolher
0.28
0.28
0.44
0.08
0.41
0.30
0.37
0.32
0.27
0.35
0.12
0.16
0.33
0.13
0.16
0.47
0.43
0.29
0.12
0.38
0.35
0.21
0.19
0.15
0.20
0.51
-0.25
0.47
-0.38
0.24
0.23
0.31
0.22
0.10
0.28
0.47
0.47
0.26
0.18
' product of Fpex and the mean residential outdoor concentration
  product of the indoor-generated concentration and the fraction of time at home
0 sum of particle concentrations indoors at other locations times the fraction of time spent there plus contributions from the personal cloud; obtained
by subtraction of outdoor and home contributions from total personal exposure
" Proportion of personal exposure contributed by outdoor-generated particles
° Proportion of personal exposure contributed by indoor-generated particles while at home
'  Proportion of personal exposure contributed by indoor-generated particles while indoors other than at home plus personal activities throughout the
day
                                                                 27

-------
    so


    so


    40


    30
 I

 ?  20


    10


    0


   -10
D Personal cloud plus particles generated indoors away from home
• Particles generated indoors at home
3 Outdoor-generated contribution
                            Participant
Figure 3-24.  Outdoor and non-outdoor contributions to personal PM25
exposure  The non-outdoor contribution is divided into indoor-generated
PM; 5 encountered while at home and the sum of indoor-generated PM; 5
while away from home and PM25 due to the personal cloud.

From Table 3-15, the contribution  of at-home indoor-generated
particles to personal exposure varied widely, from about 0 to 26
ug/nr (mean 6.6 ug/mj). The contribution of particles generated
while the person was away from home (either from exposures
indoors at other locations or from personal activities throughout
the day) ranged from -7 to 17 u.g/nr (mean 5.9 u,g/m'). Two of
the 37  participants had negative values for this variable due to
uncertainties in estimating the outdoor and home contributions.
On average, the home contribution plus the contribution of the
personal cloud and exposures in locations other than the home
added  up to  a bit more than the  contribution  from  outdoor-
generated particles ( 12.5 ug/m3 vs.  10.9 ug/mj). On an individual
basis,  however, some participants have virtually all  of their
exposures contributed from outdoor particles, while others have
most of their exposure from indoor-generated or personal cloud
particles (Figure 3-24).

Relationship Between Outdoor
Concentrations and the  Contribution to
Personal  Exposure of Particles  of Outdoor
Origin
As described in the Introduction, our ultimate aim is to determine
the  relationship between  personal  exposure to  particles of
outdoor origin and ambient  measurements.  We regressed our
estimate of the outdoor contribution to personal exposure on the
concentrations just outside each home (Table 3-16). Adjusted R~
values ranged from 0.42-0.93 (median of 0.81). The overall R:
was 0.71 (N = 750).  When the regression was run against the
HI at the central site, the overall R~ was reduced to 0.64 (N =
735). The regressions of exposure  to particles of outdoor origin
against central site FRM concentrations is the one that interests
epidemiologists, who use values at the central site to estimate
personal exposure. For this set of regressions, the mean R: was
0.60(N = 743), with a range from 0.19 to 0.88 (median = 0.73).
Again, we caution that our assumptions force high R' values, and
these should be considered upper bounds for the true R: values.

Tables  3-16 and  3-17  provide clear evidence of  some
degradation in the  relationship between personal exposure to
particles of outdoor origin and the outdoor concentrations as we
consider measurements made further from the home. 30 of the
37  participants  showed higher correlations  with  outdoor
concentrations measured just  outside  the  home than  with
concentrations measured at the central  site by the FRM.  The
range in R2 differences was  -0.24 to +0.52, with an  average
differenceofO.il.

Would these regressions show improved results if we considered
each season separately? There would be an advantage perhaps in
that the infiltration factor would probably be less variable within
a season than across seasons.  However, a drawback would be
the limitation to no more than 7 observations per household per
season. After testing this on the seasonal values, we found that
the drawback of fewer observations outweighed the advantage of
reducing cross-season variability.  That is, adjusted R" values by
season included many that were well below the minimum of 0.19
observed for the full set of year-round observations.   The
adjusted R: results  for seasonal regressions with the residential
outdoor and central-site outdoor monitors are shown in Figures
A-l to A-3  in the Appendix.

Use of Reported Time in Indoor and
Outdoor Microenvironments to Predict the
Outdoor Exposure Factor Fpex from the
Infiltration Factor Finf
In some studies, personal exposure is not measured; therefore, a
general theory has been developed using time-activity data and
measurements of indoor and outdoor concentrations as a way of
estimating personal exposure. Equation 3-6 provides a way to
predict the  outdoor exposure factor Fpl,v from the fractions ot
time indoors and outdoors and  the measured infiltration factor
F,m-. Since we also have a measured value of Fpiv, using the
sulfur exposure/outdoor ratio, we can compare values predicted
by Equation 3-6 to measured values.

From the daily time-activity questionnaires, we can find the time
spent in various activities (Table 3-18).  We define the fraction
of time outdoors/',,,, as the time either outdoors or in travel, with
a mean value of 10%. The fraction of time indoors/,, (mean
90°'o) includes indoors at home and all other locations. For each
participant and each day we can calculate Fp^ from Equation 3-
6. Table 3-18 and Figure 3-25 show that the predicted value of
/•),„ exceeds the measured value at nearly all points of the
distribution. Although the R:  is 0.57, we can almost match it
                                                         28

-------
Table 3-16.  Regression of Personal Exposure to Particles of Outdoor Origin on Outdoor Concentrations Measured Near Residence by Harvard
Impactor (HI)
Person
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
28
29
31
32
33
34
35
36
37
38
Sum/Mean
N
25
24
21
23
27
27
27
6
25
21
9
27
7
27
27
27
25
6
22
22
28
13
14
23
14
21
28
25
27
14
28
22
13
27
6
21
21
750
Slope
0.48
0.44
0.50
0.44
0.68
0.56
0.65
0.31
0.53
0.35
0.32
0.42
0.72
0.44
0.67
0.55
0.40
0.45
0.54
0.55
0.55
0.48
0.48
0.52
0.69
0.60
0.40
0.36
0.40
0.58
0.30
0.27
0.55
0.40
0.46
0.43
0.50
0.53
SE
0.04
0.05
0.05
0.05
0.06
0.05
0.07
0.04
0.08
0.03
0.03
0.10
0.12
0.04
0.05
0.03
0.07
0.09
0.04
0.04
0.04
0.10
0.06
0.04
0.11
0.06
0.03
0.04
0.06
0.06
0.07
0.04
0.17
0.06
0.12
0.07
0.05
0.01
Inter
1.47
1.36
3.12
0.46
-0.08
-0.93
-0.81
1.10
0.76
1.83
2.45
0.00
-0.11
-0.12
-0.90
0.59
1.29
1.02
2.43
-1.10
-1.10
1.14
0.02
0.68
-2.20
0.20
2.48
1.90
2.38
4.00
2.77
1.39
1.77
2.56
-0.46
1.61
0.46
0.26
SE
0.83
1.20
1.29
1.21
1.28
0.98
1.66
1.24
1.76
0.66
0.69
1.39
3.11
0.70
0.97
0.55
1.25
2.03
0.95
0.73
0.73
2.19
1.36
0.83
3.21
1.57
0.61
0.75
1.19
1.69
1.17
0.78
4.34
1.22
3.01
1.28
0.76
0.26
P
0.09
0.27
0.03
0.71
0.95
0.35
0.63
0.42
0.67
0.01
0.00
1.00
0.97
0.87
0.36
0.30
0.31
0.64
0.02
0.15
0.15
0.61
0.99
0.42
0.51
0.90
0.00
0.02
0.06
0.04
0.03
0.09
0.69
0.05
0.89
0.23
0.55
0.40
R2
0.84
0.77
0.84
0.76
0.82
0.84
0.79
0.91
0.63
0.84
0.80
0.77
0.74
0.83
0.85
0.93
0.59
0.81
0.88
0.92
0.85
0.63
0.83
0.87
0.76
0.83
0.88
0.78
0.62
0.88
0.42
0.67
0.44
0.60
0.74
0.64
0.81
0.71
                                                               29

-------
Table 3-17.  Regression of Personal Exposure to Particles of Outdoor Origin on Outdoor Concentrations Measured at Central Site by Federal
Reference Method (FRM^
Subject
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
28
29
31
32
33
34
35
36
37
38
Sum/
Mean
N
25
24
21
23
27
27
27
6
25
21
9
27
7
27
27
27
25
6
22
22
28
13
14
23
14
21
28
25
27
14
28
22
13
27
6
21
21
750
Slope
0.58
0.45
0.48
0.40
0.76
0.50
0.57
0.36
0.55
0.46
0.62
0.35
0.54
0.48
0.71
0.58
0.35
0.52
0.47
0.58
0.51
0.43
0.45
0.41
0.68
0.60
0.41
0.34
0.37
0.60
0.31
0.33
0.57
0.23
0.56
0.46
0.53
0.52
SE
0.06
0.06
0.06
0.05
0.10
0.04
0.07
0.06
0.09
0.06
0.18
0.04
0.15
0.05
0.05
0.07
0.10
0.22
0.06
0.07
0.14
0.12
0.06
0.09
0.10
0.07
0.04
0.04
0.06
0.09
0.06
0.06
0.11
0.08
0.09
0.06
0.08
0.02
p (slope)
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.00
0.00
0.00
0.00
0.08
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.00
0.00
0.00
Intercept
0.95
2.71
5.81
1.49
0.24
0.39
2.07
0.49
1.11
1.05
3.08
2.58
-0.92
-0.11
0.25
0.85
2.09
0.08
5.18
-1.47
4.08
1.64
1.39
2.48
-1.56
1.56
3.01
2.23
3.61
5.77
2.66
0.85
0.93
6.06
-2.39
1.99
0.38
1.29
SE
1.08
1.16
1.35
1.19
1.75
0.87
1.77
1.54
1.82
0.97
4.32
0.67
1.88
0.77
0.83
1.11
1.77
4.54
1.22
1.32
2.93
2.64
1.20
1.68
3.12
1.70
0.73
0.83
1.14
2.30
1.09
1.00
2.95
1.40
2.20
1.02
1.10
0.31
P (Int)
0.39
0.03
0.00
0.22
0.89
0.65
0.25
0.77
0.55
0.29
0.50
0.00
0.64
0.88
0.76
0.45
0.25
0.99
0.00
0.28
0.17
0.55
0.27
0.15
0.63
0.37
0.00
0.01
0.00
0.03
0.02
0.41
0.76
0.00
0.34
0.07
0.73
0.00
R2
0.77
0.74
0.76
0.73
0.69
0.84
0.69
0.88
0.59
0.74
0.57
0.77
0.67
0.8
0.87
0.74
0.34
0.47
0.71
0.77
0.33
0.5
0.83
0.49
0.76
0.78
0.82
0.73
0.56
0.76
0.47
0.59
0.68
0.19
0.88
0.73
0.67
0.60
                                                                30

-------
Table 3-18.  Time (in Minutes) Spent in Various Activities/Locations
Activity/Location
Indoors at home
Cooking
Cleaning
Grooming
Other indoor locations
Travel
Outdoors
Unknown
Exposed to smoke
Outdoors + travel
Indoors
fou, (fraction of time outdoors or in travel)
fm (fraction of time indoors)
Infiltration factor Fmf (indoor/outdoor S)
Measured outdoor factor Fpex (personal/outdoor S)
Predicted outdoor factor F^x from Equation 3-6
N
727
727
727
727
727
727
727
727
727
727
727
727
727
727
727
727
Mean
1179
98
39
57
90
78
67
11
14
145
1295
0.10
0.90
0.59
0.55
0.63
SD
181
72
45
47
82
61
63
22
44
103
103
0.07
0.07
0.16
0.14
0.15
Min
495
0
0
0
0
0
0
0
0
0
840
0.00
0.58
0.17
0.16
0.19
10th
915
15
0
15
0
0
0
0
0
30
1155
0.02
0.80
0.38
0.38
0.44
25th
1065
45
0
30
30
30
15
0
0
60
1230
0.04
0.85
0.47
0.45
0.53
Median
1215
75
30
45
75
60
45
0
0
120
1320
0.08
0.92
0.58
0.54
0.63
75th
1320
135
60
75
120
105
105
15
0
210
1380
0.15
0.96
0.68
0.63
0.72
90th
1395
195
105
120
200
165
150
30
45
285
1410
0.20
0.98
0.79
0.73
0.82
Max
1440
480
210
480
540
345
390
135
375
600
1440
0.42
1.00
1.06
1.08
1.05
                                                                 31

-------
by  using only Fmf (not  F,,,.v) to estimate personal exposure
(Figure 3-26). The latter regression has a higher slope (0.85 vs.
0.80),  a lower intercept  (0.12  vs. 0.19) and an R2 of 0.49,
indicating that the extra information from the time-activity diary
was not sufficient or was not precise enough to produce a better
result than simply using the indoor/outdoor sulfur ratio alone.
      1.2
            y = 0.80 (0.03 SE) x t 0.19 (0.01)
  5 £            R; = 0.57: N = 727
         0       02      0.4      0.6       0.8        1
                   Measured ratio of personal vs outdoor sulfur

 Figure 3-25.  Predicted value of the outdoor exposure factor Fpex using
 Equation 3-6 compared to the measured value using the personal/outdoor
 sulfur ratio.
                                      y = 0.667 (0.025)x + 4.0 (0.7)
                                            0.49 N = 753
                      40     60     80     100    120    140    160
                     Observed Personal PM2 5 Exposure (jig/m3)

  Figure 3-26. Predicted personal exposure to PM25 using only Fml.

  Equation 3-2 relates personal exposure to the time-weighted
  averages of the indoor and outdoor microenvironments.  We
  have measurements of tine particle concentrations both indoors
  and outdoors, but they are only 24-h  averages and are  not
  necessarily the actual concentrations experienced by the persons
  at the time they were in those microenvironments.  Also,  fine
  particle concentrations were not measured in some locations such
  as the car and workplace. If we assign the measured residential
  outdoor   value    to  all    the   outdoor    and    transport
  microenvironments, and the measured indoor concentrations in
  the home to all the other indoor locations, we can test how well
  Equation 3-2 does in predicting personal exposure by comparing
  with the measured personal exposure.   Results  (Figure 3-26)
show a moderate ability of Equation 3-2 to predict exposure,
with an adjusted R2 of 0.49 (N=753).   However, the average
personal exposure predicted from Equation 3-2 was only 19.4
ug/nv compared to the observed average exposure of 23.1 ng/m3.
The difference of 3.7 ug/mj provides an independent estimate of
the magnitude of the personal cloud, and this value agrees well
with the earlier estimates above. The inability of Equation 3-2 to
estimate  personal exposure  has   been  shown  previously
(Ozkaynak et al., 1996; Pellizzari et al., 1999).

Estimating P and k
Calculating Average Values ofP and k
Equation 1-3 for the infiltration  factor is nonlinear in the  air
exchange rate a. Therefore if we plot our measured 24-h average
indoor/outdoor sulfur ratios versus  the measured air exchange
rate, we can solve for the overall average P and k by minimizing
some appropriate function. We tried minimizing the squares of
the   differences   between  the   measured  and  modeled
concentrations (ordinary Gaussian least squares approach) and
also tried minimizing the absolute differences (a procedure
giving less weight to outliers).  Both approaches gave almost
identical results, so we report only the results from the ordinary
least squares approach.  Figure 3-27 gives the results for all the
individual 24-h measurements (N = 720). The estimate for the
penetration coefficient P averaged  across all measurements is
0.85 (0.02 SE). The estimate for the average deposition rate k is
0.22 h"' (0.01 SE). The R2 value was moderately  high at 0.45.
                                                                    1.2 -]
                                                                    0.2
                                                                                              P = 0.85 (0.016 SE): * = 0.22 (0.014) h'1

                                                                                                  R- (adj) = 0.446: N = 720
                          Air Exchange Rate (h ')
Figure 3-27.  Nonlinear least-squares fit to the indoor/outdoor sulfur ratio
vs. the air exchange rate. Bounding curves are ± 1 SE.

A second (linear) approach to calculating average values of P
and  k follows from the  simple time-averaged mass  balance
equation for sulfur with no indoor sources:
                   »AIH, = Pa/(a+k)
(3-9)
This equation is not linear for one of the unknowns (k), and it
also mixes the values off and k together in one factor. We can
partially separate P from k by inverting the equation (Long et al.,
2001):
                                                             32

-------
                Suu/Sin=k/(Pa)+l/P
(3-10)
y = 0.278ic » 1.22
RJ = 0.356 N * 716
Note that writing the equation this way isolates P in a single
term, the intercept of the regression. This value can then be used
to calculate k from the other term.  Since the inverse of the air
exchange rate a is the residence time T, we can put this equation
into the form of a linear regression on r:
                 Soll/S,n = (k/P)r
(3-11)
The intercept of the regression will be our best estimate of P and
the slope will lead to an estimate of k (based on our value for P).
One problem with this approach is that the intercept is always an
extension of a line determined by points that may be far from the
intercept. For example, participants 4 and 6 in the following
graphs have no values of r lower than 2 h or 1 h, respectively, so
that the extrapolation of the line to a value of 0 goes well beyond
the domain where all the points lie.  This will lead to more
variation and a greater uncertainty in our estimates of P.

As an initial check on whether the sulfate data provides useful
data for determining P and k we can run the regression on the
full data set (Figure 3-28).
  3
  o>
  5 3
  o

  I2
                     y = 0.228 (0.012 SE) « * 1.328 (0.035)
                           R2 = 0.335: N =720
                                   r/P
                 P = 0.75 (0.04): k = 0.17 (0.05) h''
                               10
                         Residence time T (h)
                                            15
Figure 3-28.  Regression of the outdoor/indoor sulfur ratio vs. residence
time.
However, there are four very influential outliers at unusually
high residence times (air exchange rates < 0.1 h"').  These are
unlikely values given other measurements in the same house. If
we rerun the regression without  these  outliers  we obtain
somewhat larger values for P and k, as well as an improved R2
(Figure 3-29). The overall average deposition rate k= 0.23 (0.01
SE) h"' while the overall average penetration coefficient P =
                                   Residence time i (h)
          Figure 3-29.  Same regression as in Figure 3-28 without four outliers.

          0.81  (0.03  SE).  Since measurement error generally leads to
          lower slopes and higher intercepts, it may be that these estimates
          are lower bounds for the actual average k and P. These linear
          estimates agree well with the nonlinear estimates of 0.22 h"1 for k
          and 0.85 for P.

          Calculating Individual Home  Values ofP
          and k
          In general  the  different ways  of analyzing the data do not
          disagree violently, and therefore we can go ahead and try to
          calculate k and P for the individual homes. We first assume that
          P and k do not change greatly across seasons for any home.  This
          is unlikely to be the case if certain household characteristics such
          as window  opening, use of fans and filters, changes by season.
          Nonetheless, by making this assumption, we can make use of all
          measurements for each house  in a single regression. We can
          employ  a  nonlinear  fit  to  the observed values  of  the
          indoor/outdoor  sulfur  ratio  (see  Equations  1-3  and  1-4),
          calculating  the best value  of P and k for each home averaged
          across all seasons (Table 3-19). The results are mixed, with R2
          values ranging from zero to 0.94. 32 of 37 estimates for P were
          significantly different from zero, but only 16 estimates of A: were
          significantly different from zero (significant results shown in
          boldface).  Of the 32 significant estimates for P, three were
          considerably greater than 1, a physical impossibility. The range
          of significant  P estimates was from 0.52  to  1.40,  but the
          interquartile range was more  tightly clustered between 0.66 and
          0.98.

          There are 5 values for P (3  of them significant) that are  well
          above  1, ranging  from  1.16  to 2.02.    We can  rerun the
          regressions for these 5 cases bounding P from above at 1 (Table
          3-20).  All five of the new estimates  of k are significantly
          different from zero, compared to three when P was unbounded.
          The R2 values are decreased slightly compared to the unbounded
          case,   but   remain   quite    high,  from  0.31   to  0.81.
                                                           33

-------
Table 3-19.  Estimates of P and k for Individual Homes Using Nonlinear Fit to the Indoor/Outdoor Sulfur Ratio
House
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
28
29
32
33
34
37
38
31
36
35
N
22
23
28
21
20
26
25
6
21
25
8
28
7
27
24
28
19
6
27
27
26
13
13
27
8
22
27
24
31
28
20
10
19
15
14
5
0
P
0.85
0.61
0.83
0.61
1.16
0.67
0.88
0.44
0.80
0.71
0.67
0.91
0.30
0.45
1.03
0.67
0.60
1.17
1.00
0.52
0.88
0.97
0.82
0.77
1.37
0.83
0.77
0.75
1.46
0.60
0.55
2.02
0.79
0.68

SE
0.10
0.04
0.04
0.15
0.15
0.07
0.04
0.23
0.09
0.08
0.05
0.12
0.08
0.05
0.18
0.03
0.07
0.86
0.07
0.04
0.08
0.23
0.10
0.10
0.15
0.07
0.04
0.08
0.38
0.20
0.09
2.05
0.06
0.06

P(P)
0.000
0.000
0.000
0.001
0000
0.000
0.000
0.134
0.000
0.000
0.000
0.000
0.013
0.000
0.000
0.000
0.000
0.246
0.000
0.000
0.000
0.001
0.000
0.000
0.000
0.000
0.000
0.000
0.001
0.006
0.000
0.352
0.000
0.000
Did not converge
Did not converge
Same as house 29
k
0.26
-0.01
0.12
0.09
0.41
0.06
0.14
0.06
0.17
0.14
-0.01
0.29
-0.04
0.00
0.23
0.02
0.03
0.37
0.35
0.02
0.21
0.28
0.50
0.18
0.48
0.13
0.15
0.13
0.69
0.18
0.22
0.93
0.12
0.07

SE
0.13
0.04
0.04
0.09
0.12
0.06
0.05
0.16
0.08
0.07
0.03
0.10
0.04
0.04
0.13
0.03
0.02
0.50
0.10
0.03
0.07
0.16
0.15
0.08
0.12
0.07
0.06
0.06
0.31
0.15
0.12
1.36
0.04
0.03

PW
0.049
0.861
0.012
0.326
0.003
0.377
0.013
0.746
0.036
0.065
0.725
0.006
0.358
0.994
0.097
0.557
0.192
0.498
0.002
0.505
0.004
0.106
0.007
0.033
0.007
0.071
0.021
0.046
0.033
0.246
0.086
0.511
0.004
0.048

R2
0.29
0.00
0.27
0.17
0.53
0.04
0.32
0.05
0.35
0.23
0.02
0.51
0.11
0.00
0.25
0.02
0.15
0.40
0.51
0.02
0.53
0.48
0.74
0.28
0.94
0.22
0.27
029
0.53
0.15
0.37
0.39
0.55
0.37

                                                                 34

-------
Table 3-20.  Values for k when P is Bound from Above by 1

  Subject        N          R2           k          SE
    25           8         0.81        0.21        0.03
    29          31         0.48        0.33        0.03
    34          10         0.31        0.26        0.03
    18           6         0.40        0.28        0.02
     5          20         0.50        0.28        0.03
                                                               35

-------
A  similar approach  to  the  above  nonlinear  use of the
indoor/outdoor sulfur ratio is to use the linear equation involving
the inverse of that ratio, as well as the inverse of the air exchange
rate  (Table  3-21).   24 homes  had both slope and  intercept
significantly  different from zero (shown in boldface).

Five values of P were well above 1, so the regressions  were
rerun bounding  P from above by 1 (Table 3-22).  All 5 new
estimates of A" are significantly different from zero, compared to
none when P was  unbounded.  The R" values arc decreased
slightly compared to the unbounded case, but remain quite  high,
from 0.32 to  0.82.

The estimates of P and k from the nonlinear approach  (Table 3-
 19)  are  compared  to  the estimates from the linear  approach
(Table 3-21)  in Figures 3-30 and 3-31. Only values significantly
different from zero are included. Figure 3-30 suggests that the
nonlinear approach gave consistently smaller estimates of P at
the high end  of the range. In fact, of the five estimates exceeding
unity due to  the linear approach, all five were lower, and two
were less than one, in the nonlinear approach.

The median  value  for P was  0.81  (interquartile range 0.66 to
0.90). The median for k was 0.24 h"1 (interquartile range 0.12 to
0.35 h'1).

Having  calculated P  and k  for each  home from the linear
regression of the outdoor/indoor sulfur ratio  vs.  the residence
time, the estimated average infiltration factor for each home can
be calculated from the equation
                    ml = Pa/(a+k)
            (3-12)
 where a in this case is the arithmetic, geometric, or harmonic
 average of the air exchange rates for each home.

      2

     1.8
     16

     1.4

     1.2

   0.  1

     08

     0.6

     0.4

     0.2
• P(lln)
• P(nlin)
        29 25 15 22 21 19 31
 Figure 3-30. Comparison of estimates of P from the linear and nonlinear
 approaches described in the text. Only values significantly different from
 zero are plotted (N = 32 homes).
                          1.2
                           1 •
                          0.3
                        £ 0.6
                          0.1
                          0.2
                                                     •
                                                       k(lm|
                                                       K(nfm)
                                                             I •
       23 26  a 22 H 21  19 16 4 12  1  28 24 5 27  26 7 37 10  9  3 U X 17
                              House ID
Figure 3-31.   Comparison of the estimates of k from the linear and
nonlinear approaches described in the text.  Only values significantly
different from zero are plotted (N = 24 homes).

Estimating Finf from Individual Values ofP
and k
These estimated values of P and k from the two approaches
(linear and nonlinear) can be used to estimate F,n/-for each home.
 These estimates are compared to the observed indoor/outdoor
sulfur ratio in Figure 3-32.  Although the excellent agreement
(R: = 0.96 for the nonlinear estimate and 0.99 for  the linear
estimate) of the F,,,/  estimates may be an artifact due to the
related nature of the three calculations, at least this agreement
suggests that the P and k values used to estimate Fml  from both
the linear and nonlinear regressions (which both involve the air
exchange rate or its  inverse)  are at  least  consistent with the
measured indoor/outdoor sulfur ratios (which do not involve the
air exchange rate).  However, the many cases in which neither
the linear  nor  the  nonlinear approach  gave values for k
significantly different from zero, and the several cases  in both
approaches  in  which values  of P  were  greater than unity,
suggests that the assumption  of constant  values for k and P
across seasons for each home is violated.
       1
      0.9
      0.8
      0.7
      0.6
                         2
                                            y = 0.97x * 0.003
                                            R! = 0.96. N = 37
                                                          y = 0.98x » 0.004
                                                          R' = 0.99, N = 37
                            0.2
                               0      0.2      0.4      0.6      0.8      1
                                      P „- from Sulfur Indoor-Outdoor Ratio

                       Figure 3-32. Comparison of the infiltration factor (F,nf) estimates from the
                       simple ratio of indoor sulfur to outdoor sulfur by home vs. the nonlinear
                       regression of the same ratio using the measured air exchange rates and
                       the linear regression of the inverse ratio (outdoor/indoor) against the
                       residence time.
                                                            36

-------
Table 3-21. Results of Linear Regressions of the Outdoor/Indoor Sulfur Ratio on Residence Time for 36 Homes
House
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
28
29
31
32
33
34
36
37
38
Sum/Mean
N
22
23
28
21
20
26
25
6
21
25
8
28
7
27
24
28
19
6
27
27
26
13
13
27
8
22
27
24
31
14
28
20
10
5
19
15
720
1/P
1.19
1.66
1.24
1.08
1.12
1.47
1.14
2.25
1.45
1.51
1.49
1.13
3.03
2.20
0.85
1.51
1.66
0.99
1.00
1.93
0.96
0.88
1.19
1.23
0.65
1.17
1.28
1.15
0.52
1.05
1.26
2.59
0.58
3.46
1.28
1.56
1.33
SE
0.13
0.11
0.06
0.45
0.21
0.15
0.05
1.36
0.13
0.15
0.10
0.16
0.79
0.24
0.18
0.07
0.20
0.65
0.06
0.14
0.12
0.29
0.17
0.21
0.07
0.10
0.08
0.17
0.25
0.08
0.62
0.35
0.54
1.78
0.10
0.11
0.03
P
0.84
0.60
0.81
0.93
0.90
0.68
0.88
0.44
0.69
0.66
0.67
0.89
0.33
0.45
1.18
0.66
0.60
1.01
1.00
0.52
1.05
1.14
0.84
0.81
1.54
0.85
0.78
0.87
1.91
0.96
0.79
0.39
1.73
0.29
0.78
0.64
0.75
SE
0.09
0.04
0.04
0.39
0.17
0.07
0.04
0.27
0.06
0.06
0.05
0.13
0.09
0.05
0.26
0.03
0.07
0.67
0.06
0.04
0.13
0.37
0.12
0.14
0.16
0.07
0.05
0.13
0.91
0.07
0.39
0.05
1.64
0.15
0.06
0.04
0.00
k/P
0.31
0.00
0.11
0.32
0.25
0.10
0.16
0.14
0.14
0.17
-0.01
0.32
-0.05
0.04
0.30
0.03
0.07
0.29
0.36
0.04
0.35
0.36
0.64
0.29
0.40
0.19
0.24
0.27
0.58
0.16
0.47
0.19
0.44
0.14
0.15
0.09
0.23
SE
0.09
0.07
0.04
0.11
0.09
0.08
0.05
0.33
0.04
0.06
0.04
0.06
0.17
0.07
0.09
0.05
0.03
0.19
0.06
0.04
0.05
0.10
0.12
0.08
0.03
0.07
0.07
0.07
0.09
0.11
0.16
0.07
0.21
0.44
0.03
0.03
0.01
k
0.26
0.00
0.09
0.29
0.22
0.07
0.14
0.06
0.10
0.11
-0.01
0.28
-0.02
0.02
0.35
0.02
0.04
0.29
0.35
0.02
0.37
0.41
0.54
0.23
0.61
0.16
0.18
0.24
1.11
0.15
0.37
0.08
0.77
0.04
0.12
0.05
0.17
R2
0.34
0.00
0.20
0.28
0.25
0.02
0.33
0.00
0.32
0.20
0.00
0.52
0.00
0.00
0.30
0.00
0.22
0.19
0.60
0.00
0.66
0.52
0.72
0.29
0.96
0.25
0.31
0.40
0.55
0.07
0.21
0.27
0.28
0.00
0.53
0.39
0.34
                                                              37

-------
Table 3-22. Values for k when P is Bound from Above by 1
Subject
15
22
25
29
34
N
24
13
8
31
10
k
0.22
0.32
0.26
0.28
0.44
SE
0.02
0.03
0.03
0.03
0.05
R2
0.32
0.55
0.82
0.55
0.34
                                                            38

-------
Seasonal Analysis
Up to this point, the analyses have been carried out using values
averaged across all visits during the year.  However, some
seasonal variation was noted in the values for the infiltration
factor F-mj and the personal exposure factor F/>ex. In this section,
we test whether we can obtain  useful results on  an individual
home or person on a seasonal basis by regressing personal and
indoor sulfur on outdoor sulfur  concentrations by season. The
advantage to this is that the seasonal variations, if any, will be
observed; the disadvantage is that limiting  the regression to a
maximum of 7 values is likely to lead to wider variance  and
more uncertainty in estimating the two quantities  of interest.

Since  air exchange rates are  the only measured  variable
contributing  to the infiltration  factor (the others  being the
unmeasured P and k parameters), we compare the (harmonic)
average air exchange by house and by season to the estimate of
/•"^obtained by taking the average indoor/outdoor  sulfur ratio by
house and by season (Appendix Table A-l). Then we compare
the estimate of Fin/ from the indoor/outdoor sulfur ratio to the
corresponding estimate obtained by regressing indoor on outdoor
sulfur and taking the slope of the regression as an estimate of Fml
for each house and season (Appendix Table A-2).  The estimates
of Fmf from the ratio are compared to the estimates from the
slope in Figure 3-33.  The R" value is 0.62, and  the estimates
from the regression slopes vary more widely (from 0 to >1) than
the estimates from the indoor/outdoor ratios (0.2 to <1).
   1.2
    0.2 4
     0.2
            0.3
                  0.4
                        0.5
                               0.6
                              Ratio
                                     0.7
                                            0.8
Figure 3-33. Estimates for each home by season of the infiltration factor
F,n,from regressing indoor sulfur on outdoor sulfur (Slope) compared to
estimates from the simple ratio of indoor sulfur to outdoor sulfur averaged
over all visits in a season.

Another way of comparing the estimates using the regressions to
the estimates using the ratios is provided by Figures 3-34 and 3-
35. Figure 3-34 presents the homes ordered by season and by the
value of Finj as determined from the ratio; Figure 3-35 presents
the comparable results from the regressions. Comparing the two
figures shows many cases  in which the regression approach
predicts low values for the infiltration factor (< 0.4), as well as a
smaller number of cases in which the approach predicts very
high values, including one impossible result (>1). This analysis
confirms that the ratio provides the more stable estimate of the
infiltration factor.
     1

   0.9

 -- 0.8

 J 0.7

 !o.6

2
6 nil - -
%
•V
^
•

••.
•
IU-J
« 0.2
0.1
n
• Summer;
- Fall \
• Winter '••
• Spring
Figure 3-34.  Estimates of fm, by home from the indoor/outdoor sulfur
ratio.
    1.2	
  a.
  * 0.8
                                                                o °-6
                                                                |

                                                                1 0.4
                                                                                                      • Summer
                                                                                                      • Fall
                                                                                                      • Winter
                                                                                                      • Spring
                                                              Figure 3-35.  Estimates of Fm, by home from regressions of indoor on
                                                              outdoor sulfur.
Multivariate Regressions
All participants filled out a questionnaire on their household
characteristics and daily activities.   A total of 54 questionnaire
variables were included in the dataset (see Table A-4 in the
Appendix for the variable names and definitions). We carried out
a series of multiple regressions on the personal, indoor, and
outdoor  fine  particle  and  sulfur  concentrations vs.  these
questionnaire responses to try to identify sources of increased
exposure.

Our first priority was to investigate the 54 independent variables
for possible collinearity. This procedure is explained fully in the
book  Regression Diagnostics by Belsley, Kuh, and Welsch
(1980). A factor matrix is prepared of the 54 variables and a
"condition number" is calculated for each of the eigenvalues of
the matrix.  Belsley, Kuh. and Welsch (1980) recommend that if
                                                           39

-------
a condition number exceeds 30, then variables that have a heavy
weight on that eigenvalue should  be inspected  and either
combined or a way found to drop one of the variables. Four
eigenvalues did in fact exceed a condition number of 30.  The
pairs of variables with the heaviest loadings on each eigenvalue
were       ROOMS/AREA,       DRYER/DRYER VENT,
DSTFACTORavg/C  FUEL, and TEMPC/TEMPDELTA. Since
AREA seemed the more precise factor, we dropped ROOMS.
Since DRYER seemed the  more fundamental variable,  we
dropped  DRYER_VENT.  And since TEMPDELTA was  a
calculated variable including a number of imputed values, we
dropped it. The choice between DSTFACTORavg and C_FUEL
was more difficult.   The  dust factor as  estimated  by  the
technician has previously (in the PTEAM Study, for example)
been  one of the strongest predictors of airborne  particles.
However, the cooking fuel  (electric  or gas) is  the more
fundamental    variable,   so    we    reluctantly   dropped
DSTFACTORavg  from the  list of variables.   The  final  list
contained 50 variables  dealing with household  characteristics,
personal  activities, and two measured quantities (air exchange
and outdoor temperature).

To include 50 variables in the regression and achieve an overall
significance of p<0.05,  the Bonferroni criterion  was applied by
dividing  the chosen significance level  by the number  of
variables. This gives us a p-value of 0.001  as the value required
to achieve an overall  significance of p<0.05 for the final model.
In all of the following  tables dealing with multiple regression
results, the  variables that meet the Bonferroni criterion for
significance are listed in boldface type.

The first step in the multiple regression approach was to carry
out a  stepwise regression (combined forward  selection and
backward elimination) on all 50 household characteristics and
personal  activities  as well as one  or two measured particulate
matter or sulfur variables using Statistica 6.1 software.  A second
regression was run including only those variables at the p < 0.05
level. It is important  to run the second regression on the smaller
set of variables since  this will include more cases (typically, for
our data set, 20-70 additional records that were dropped because
of the casewise elimination employed for all 50 independent
variables.)  To check our results, all of the initial 50-variable
regressions  were rerun on SAS  statistical software using  a
standard  backward elimination stepwise regression. Normally,
SAS and Statistica identified the  same significant variables, with
minor differences in  the slopes, intercepts, and R: values (the
latter were always within 0-3% of each  other). When a variable
differed,  both final versions were run one more time in Statistica,
and the run with the higher R" value was selected.

Although major collinearities were  avoided, less strong ones
remain.  This could  cause one or two of a collinear  pair of
variables to fail to register as significant in the first run involving
50 variables; they will then have no chance to be considered in
the final  run picking  out only the significant variables from the
first run.  In some situations, we inserted a variable that we
suspected might be relevant after the "final" run and discovered
that it too was significant, but never succeeded in changing the
final R: value by  more  than  1% when adding individual
variables, indicating that the most important variables in each
regression were identified.

In carrying out analyses of how household characteristics and
personal activities affect indoor concentrations of particles, we
need to  carefully examine how the study design may  have
created  unequal  conditions among the homes.  First,  not all
homes were visited for an equal number of seasons. Only 24
homes were visited in all  four seasons.  Second, even within a
season,  only 3-6 homes were visited each week.  Therefore
different homes within any given season encountered different
outdoor conditions. The homes were scrambled each season so
that the  same homes were not visited together in a given week
over two seasons, so this should have evened out  the outdoor
conditions encountered to some extent, at least for the 24 homes
visited all four seasons.  However, for the entire group of 36
homes,  there was  extensive variation  in  the  outdoor  PM:5
concentrations, whether at the central site or outside the homes,
ranging  over a factor of two from the highest  to the  lowest
(Figure 3-36).
   35
   30
    25
 „ 20 -
  I
        v •.
                                ambHI2S
                                FRM2S
                                PM25ui.ii
Figure  3-36.    Central-site and residential  outdoor concentrations
averaged over all visits to a home.

Clearly, if homes at the high end of the outdoor concentrations
happen by chance to have some  characteristics substantially
different  from those at the low end, a  simple regression of
outdoor concentrations at the central site on these household
characteristics may show significant results, even though the
relationship cannot be causal (household  characteristics cannot
affect  outdoor concentrations).    In fact, some  of these
"impossible"   relationships are  observed.  For  example,  a
regression of  the two  central site monitors and the outdoor
residential monitor on various household characteristics showed
a number of artifactual relationships (Table 3-23). The building
age and air exchange rates were among the strongest variables in
                                                          40

-------
Table 3-23.  Multiple Regression of Outdoor Concentrations on Household Characteristics and Personal Activities
Variable
PM25out
Intercept
AGE
Windopen
airex
outdoor
DRYER
ambHI25
Intercept
airex
AGE
C_FUEL
Windopen
FAN
MILDEWavg
spacehtr
VAC
FRM
Intercept
airex
AGE
FAN
C_FUEL
Windopen
MILDEWavg
spacehtr
VAC
Slope

13.78
0.13
3.94
-3.18
0.01
1.84

10.58
-2.82
0.11
2.95
2.45
-3.66
-2.77
-4.63
2.00

10.60
-2.56
0.10
-3.73
2.46
1.97
-2.74
-4.27
1.95
Std.Err.

1.13
0.02
0.65
0.57
0.01
0.87

1.82
0.52
0.02
0.72
0.66
1.00
1.03
1.96
1.02

1.68
0.49
0.02
0.93
0.65
0.60
0.96
1.87
0.95
p-level N R2(adj.)
720 0.12
0.000000
0.000000
0.000000
0.000000
0.003668
0.034377
743 0.10
0.000000
0.000000
0.000004
0.000042
0.000202
0.000264
0.007241
0.018484
0.049494
772 0.10
0.000000
0.000000
0.000002
0.000067
0.000168
0.001016
0.004431
0.022320
0.039863
                                                                41

-------
all three cases,  along with open  windows in two of the three
cases, but  it is  likely  that these variables  simply represent
differential outdoor air  conditions encountered at the different
times that the homes were monitored.  The variables "explain"
only 10-12% of the variance of the outdoor monitors, but they
represent  a  caution  nonetheless in  our  interpretation  of
"significant" variables in all other of the multiple regressions we
will consider. In particular, those regressions that include both
an outdoor measurement and one or more of the variables AGE
and airex (and windopen)  will be putting the variables into
"double jeopardy", examining their effect both explicitly and
implicitly as part of the outdoor particle measurement. Since the
signs of airex and  windopen are opposite, one might think that
because they are correlated they  end up with opposite signs, a
common occurrence  in regressions with correlated variables.
Although it might  seem that air exchange rates should be fairly
well correlated with open windows, the Spearman correlation ot
airex and  windopen was only 0.24.   Nonetheless,  the first
regression in Table 3-23  was  rerun twice, dropping airex from
the first run and then restoring it and dropping windopen from
the second run,   to  test whether some kind of  cross-tenn
relationship was occurring.  However, in both cases the variable
left in continued to be significant with the  sign in the same
direction as before, and the adjusted R" dropped in  both cases,
once to 0.10 and once to  0.08. Therefore, the "best" model for
the residential outdoor  PM: 5 measurements continues to be  the
one listed first in Table 3-23.  Again, for the definitions of the
variables appearing in the next five tables, see Table A-4 in the
Appendix.
 The indoor model is
          PM25in =a+ fi,,n*PM25oitt + fi*x,
(3-13)
 where the x, are the 50 appropriate continuous/categorical
 variables.

 The  model  above can  be  repeated  using  FRM25 as  the
 independent variable:
          PM25m = a + ft,,ul*FRM25 + [i*x,
(3-14)
 The point of using the FRM25 as the outdoor variable is that
 epidemiological studies are often limited  to  the central-site
 monitor.

 The results of these model runs are prov ided in Table 3-24. The
 first regression is a simple regression of indoor PM-> 5 on outdoor
 PM: 3. As can be seen, the relationship is particularly poor, with
 an R" of only 0.09.  Since epidemiologists are often restricted to
 use of a fixed-site urban monitor, the regression is rerun using
 the central-site FRM monitor.   The R: remains low  at 0.11.
 These results are  consistent with other studies showing low
 cross-sectional  relationships between  indoor and outdoor PM
levels.

The R: value is increased to above 0.40 for multiple regressions
including the questionnaire responses and either the residential
outdoor monitor or the central-site monitor. In each case, the
outdoor  PM concentration has the greatest influence on the
indoor PM (as judged by the p-value). Estimates forthe outdoor
contribution to indoor PM ranged from 48% (using residential
outdoor PM) to 57% (using central site FRM). The difference in
these estimates can be attributed to the consistently lower values
returned by the FRM.  Investigators have noted that  the FRM
may underestimate PM levels due to loss of volatile species such
as nitrates.

Burned  food added about 12 ug/nr to the indoor concentration,
while a  nearby dirt road added 8.4 to 8.9 ug/nr, and  use of an
exhaust fan added about 5 ug/nr.  It is likely that the use of the
exhaust fan  because of cooking or burned food added to the
indoor concentration.

Homes  with persons who reported being near cigarette smoke
also had higher indoor fine particle concentrations.  The number
of smokers increased indoor PM concentrations by 4-6 ug/nr,
while time spent near smokers increased concentrations by 0.04
ug/nr per minute exposed.

Cooking increased PM: 5 concentrations by about 0.03 ug/nv per
minute.   Homes with electric stoves had  PM: 5 concentrations
6.6-6.8 ug/m"1 higher than homes with gas stoves, but it should be
remembered that the homes with electric stoves also had higher
outdoor PM: 5 levels (see Table 3-23).  Each additional person
living in the household added 1.3 ug/'nr to the daily average
PM: 5  concentrations.   Vacuuming  appeared  capable of
increasing daily average PM: 5 concentrations by about 4 ug nr.
although the variable was only marginally significant (using the
Bonferroni  criterion) in one of the two regressions.   Finally,
homes with a clothes dryer were associated with a  decrease of
4.7 ug/m' in their daily average concentrations. The presence of
a clothes dryer (nearly all of  which were vented outdoors)
increases air exchange since almost the same volume of air must
enter the house to replace the heated air vented outdoors.

The next series of two multiple regressions  takes advantage of
our ability to split indoor concentrations into indoor-generated
and outdoor-generated fine  particles (Table 3-25).

The model for indoor-generated tine particles is the following:
                            Inconthh = a + ft, *x,
                                                    (3-15)
          "Incontrib" is the contribution of indoor sources after subtracting
          the outdoor contnbution determined by the indoor outdoor sulfur
          ratio; therefore, this model does not  include the outdoor PM^5
          variable. The model was initially run  with this variable included
                                                          42

-------
Table 3-24.  Dependence of Indoor Fine Particle Concentrations on Household Characteristics and Personal Activities
Variable
PM25in
Intercept
PM25out
PM25in
Intercept
FRM25
PM25in
Intercept
PM25out
DIRT_RD
Burning
C_FUEL
Numpeopl
Exhstfan
cooking
numsmok
DRYER
vacuum
smoke
Otherjndoor
Unknown
outdoor
PM25in
Intercept
FRM25
Burning
DIRT_RD
Exhstfan
C_FUEL
Numpeopl
cooking
vacuum
DRYER
Otherjndoor
Unknown
smoke
numsmok
outdoor
Slope

8.47
0.54

7.21
0.67

-10.50
0.48
8.89
11.80
6.77
1.34
5.04
0.03
6.33
-4.66
3.89
0.04
0.09
0.06
-0.02

-11.15
0.57
11.94
8.42
5.34
6.57
1.30
0.03
4.04
-4.65
0.10
0.06
0.04
4.55
-0.02
Std.Err.

1.37
0.06

1.36
0.07

2.91
0.05
1.42
1.93
1.36
0.29
1.11
0.01
1.71
1.40
1.20
0.01
0.03
0.02
0.01

2.91
0.06
1.91
1.40
1.11
1.36
0.28
0.01
1.19
1.40
0.03
0.02
0.01
1.76
0.01
p-level N
775
0.000000
0.000000
760
0.000000
0.000000
762
0.000325
0.000000
0.000000
0.000000
0.000001
0.000003
0.000007
0.000099
0.000225
0.000912
0.001220
0.001506
0.002624
0.012012
0.016669
747
0.000138
0.000000
0.000000
0.000000
0.000002
0.000002
0.000006
0.000045
0.000750
0.000909
0.001069
0.004758
0.006304
0.009757
0.021123
R2(adj.)
0.09


0.11


0.42















0.42















                                                               43

-------
Table 3-25.  Dependence of Indoor-Generated and Outdoor-Generated Particles on Household Characteristics and Personal Activities
Variable
Incontrib
Intercept
Burning
C_FUEL
DIRT_RD
airex
Numpeopl
Exhstfan
S_WIN
cooking
smoke
Otherjndoor
outdoor
Cleaning_1
DRYER
Unknown
Outcontin
Intercept
PM25out
Outcontin
Intercept
PM25out
TempC
AGE
Windopen
cigsmokd
A_C
windowall
Numpeopl
FLRCOVav
vacuum
airex
fry
Slope

-12.03
12.06
8.70
8.01
-4.42
1.43
4.89
-4.56
0.03
0.05
0.10
-0.03
2.60
-3.64
0.05

0.04
0.58

-0.64
0.59
-0.15
0.06
1.68
035
0.39
0.0022
-018
-0.02
0.64
055
-0.47
Std.Err.

2.88
1.97
1.45
1.45
0.83
0.30
1.15
1.12
0.01
0.01
0.03
0.01
1.00
1.45
0.02

0.31
0.01

0.54
0.01
0.02
0.01
0.23
0.06
0.07
0.0006
0.06
0.01
0.25
0.23
0.21
p-level N R2(adj.)
709 0.37
0.000033
0.000000
0.000000
0.000000
0.000000
0.000003
0.000024
0.000055
0.000064
0.000068
0.001016
0.002000
0.009668
0.012452
0.024292
775 0.69
0.889273
0.000000
686 0.82
0.239662
0.000000
0.000000
0.000000
0.000000
0.000000
0.000000
0.000069
0.002316
0.002557
0.011674
0.014491
0.025024
                                                                44

-------
 as an independent variable to confirm that it had been fully
 accounted for. Results showed that it was not significant and
 could be omitted from the final model.

 The indoor-generated contribution (Incontrib) depends on many
 of  the variables appearing in  the Equation 3-13  and 3-14
 regressions for indoor PM:5 with about  the same coefficient
 values, but does not depend on the outdoor concentration. The
 value of R:  is 0.37 and again, burned food,  cooking,  use of
 electric stoves, nearness to a dirt road, use of the exhaust fan, the
 number of people living in the home, and exposure to ETS turn
 up as significant variables affecting indoor-generated particles.
 A new variable, the air exchange rate (airex), reduces indoor-
 generated concentrations as airex increases.

 The model for the outdoor contribution to indoor concentrations
 is the following:
          Outcontin = a + p~uul*PM25out + /3*x,
(3-16)
 "Outcontin'' is the contribution of outdoor sources to indoor
 PM2 5 levels determined by the indoor/outdoor sulfur ratio.

 Results (Table 3-25) show that air exchange (airex, windopen,
 windowall) again significantly influences the results, this time
 increasing the outdoor contribution as their values increase.  The
 additional variables increase R2 to 0.82, compared with an R2 of
 0.69 for outdoor PM alone. The fact that air exchange variables
 appear as highly significant in these two regressions is important
 confirmation  of our assumptions in using sulfur to identify
 indoor-generated and outdoor-generated  PM.  That is, if the
 assumption of no  indoor sulfur source were violated in some
 homes, we would not be able to separate indoor-generated PM
 from outdoor-generated PM in those homes, and the relationship
 with air exchange would  be weakened by  the amount of the
 misclassification. Air exchange variables are not selected as
 significant in regressions of total indoor PM, which lumps both
 indoor-generated  and  outdoor-generated  PM,  because   the
 influence of air exchange can work in two directions depending
 on whether indoor air PM is greater than or less  than outdoor
 PM. But increased air exchange must increase outdoor-generated
 PM and it must decrease  indoor-generated  PM; therefore our
 finding of a strong effect  in the expected directions when we
 analyze the indoor-  and outdoor-generated  concentrations
 separately must reflect a  successful  separation  of the  two
 sources.

 A model for indoor sulfur concentration is the following:

              SH, = a+pou*Sottl+p*xl             (3-17)

Once again we first examine whether the outdoor concentration
shows artifactual dependence on household variables by looking
at outdoor sulfur (S,M) alone  (Table  3-26). Several artifacts
appeared—air exchange, storm windows, time spent outdoors.
However, these variables increased  R2 from 0.59 using only
PM2.5 outdoors as the independent variable, to 0.65 using all
significant variables, a relatively small increase.

The results for indoor sulfur (Sm) suggest that there may be more
sources of indoor sulfur than we  have assumed above.  For
example,  the  number of pilot lights  (pilot) was a  strongly
significant variable, adding  133  ng/m3 to the  indoor  sulfur
concentration  per each  additional pilot light.   Natural  gas
sometimes contains some sulfur, although if the level exceeds a
certain  amount it  must be  scrubbed.  Smoking  was also
significant (cigsmokcf), possibly due to the use of matches. The
effect of open windows was also clear and expected.  Homes
with at least one open window (windopen) had increased indoor
sulfur concentrations of 120  ng/m3.  A quantitative measure
(windowalf) was an increase of 0.32 ng/m3 for every inch-hour a
window was open. Thus a window opened 6 inches wide for a
day would result in an increase of 48 ng/m3 S.  Building age
(AGE) was shown to be artifactually associated with  outdoor
PM; therefore,  its appearance in this multiple regression may
also be artifactual.   Adding  these  variables  to the simple
regression of indoor on outdoor sulfur increased the R: from 0.70
to 0.83.

We  can  estimate  personal  exposure  from   (1)  outdoor
concentrations  alone;  (2) indoor  concentrations alone; (3)
outdoor  and  indoor concentrations  together;  (4)  outdoor
concentrations  and   indoor  concentrations  together  with
household characteristics; and (5) outdoor concentrations alone
together with household characteristics.  These choices result in
the following models of personal exposure:
          PM25pers =a+ ftoul*PM25out

          PM25pers = a+ pm,*FRM25

          PM25pers = a +/?,„* PM25'm
                                                   (3-18)

                                                   (3-19)

                                                   (3-20)
           PM25pers = a + ftmi*PM25ont +p,n*PM25in         (3-21)

           PM25pers = a + P
-------
Table 3-26.  Dependence of Outdoor Sulfur on Outdoor PM2 5 and of Indoor Sulfur on Outdoor Sulfur and Household Characteristics and Personal
Activities
Variable
Soul
Intercept
PM25out
Souf
Intercept
PM25out
airex
S_WIN
outdoor
numsmok
A_C
Sln
Intercept
S0uf
Sm
Intercept
Soul
AGE
PILOT
windowall
TempC
Windopen
clgsmokd
FAN
A_C
FLRCOVav
Numpeopl
fry
Unknown
DIRT_RD
vacuum
Slope

47.09
97.14

594.47
96.98
-354.67
-216.48
-1.50
-209.72
-40.72

151.44
0.49

-19.76
0.51
7.91
133.01
0.32
-9.74
120.11
28.21
-140.52
33.49
-1.98
-20 14
-60.97
1.21
81.66
61.71
Std.Err.

66.44
3.01

85.83
2.85
43.81
54.80
0.41
82.71
18.05

27.27
0.01

48.56
0.01
0.73
13.43
0.05
1.59
23.85
5.66
30.52
7.29
0.51
6.13
20.94
0.47
32.60
25.90
p-level

0.478737
0.000000

0.000000
0.000000
0.000000
0.000086
0.000240
0.011441
0.024394

0.000000
0.000000

0.684261
0.000000
0.000000
0.000000
0.000000
0.000000
0.000001
0.000001
0.000005
0.000005
0.000104
0.001074
0.003705
0.009623
0.012471
0.017422
N R2(adj.)
720 0.59


720 0.65







720 0.70


741 0.83
















                                                              46

-------
Table 3-27.  Regressions of Personal Exposures to PM2 5 on Household Characteristics and Personal Activities
Variable
PM2Spers
Intercept
PM25out
PM25pers
Intercept
FRM25
PM25pers
Intercept
PM25in
PM25pers
Intercept
PM25in
PM25out
PM25pers
Intercept
PM25in
PM25out
grill
cooking
PM25pers
Intercept
numsmok
PM25out
CANDLDUR
Burning
C_FUEL
DIRT_RD
Other_indoor
A_C
Exhstfan
Cleaning_1
cooking
outdoor
PM25pers
Intercept
FRM25
numsmok
Slope

10.78
0.66

10.44
0.73

9.98
0.67

5.66
0.63
0.27

4.32
0.60
0.26
11.54
0.02

-11.15
13.76
0.55
0.05
11.19
7.42
6.05
0.11
1.32
4.29
3.46
0.02
-0.02

-11.81
0.60
13.31
Std.Err.

1.56
0.07

1.53
0.08

0.69
0.03

1.12
0.03
0.05

1.21
0.03
0.05
3.22
0.01

3.63
1.54
0.06
0.01
2.23
1.56
1.69
0.03
0.40
1.36
1.11
0.01
0.01

3.69
0.07
1.61
p-level N R2(adj.)
750 0.10
0.000000
0.000000
734 0.11
0.000000
0.000000

0.000000 727 0.45
0.000000
727 0.47
0.000001
0.000000
0.000001
727 0.49
0.000370
0.000000
0.000001
0.000359
0.003309
737 0.37
0.002239
0.000000
0.000000
0.000000
0.000001
0.000003
0.000363
0.000800
0.001039
0.001652
0.001968
0.009918
0.013482
721 0.39
0.001449
0.000000
0.000000
                                                               47

-------
Table 3-27.  Continued
        Variable
Slope
Std.Err.
 p-level
R(adj.;
CANDLDUR
Burning
C_FUEL
A_C
Other_indoor
Exhstfan
DIRT_RD
Cleaning_1
cooking
outdoor
 0.05
11.71
 7.44
 1.42
 0.11
 4.49
 5.45
 3.52
 0.02
-0.02
 0.01
 2.25
 1.59
 0.41
 0.03
 1.37
 1.70
 1.13
 0.01
 0.01
0.000000
0.000000
0.000003
0.000492
0.000553
0.001106
0.001407
0.001927
0.010664
0.015074
                                                              48

-------
 When only outdoor PM: 5 and the household characteristics were
 included in the regression, the most significant variable other
 than outdoor PM2.s was the number of smokers, adding between
 13.3 and 13.8 ug/m3 to the daily average personal exposure.  The
 use of candles was highly significant, adding 0.05 ug/m  per
 minute burned to daily average personal exposure. Burned food
 added between 11.2  and 11.7 ug/m3 to  personal exposure.
 Electric stoves increased particle concentrations by 7.4 ug/m3.
 Although these household characteristics and personal activities
 were useful in  improving  the  R2  for personal exposure,
 collectively they were not able to improve it as much as simply
 adding in the measured indoor PM25 concentration (R" = 0.37-
 0.39  for the outdoor + household characteristics variables
 compared to 0.47 for the outdoor + indoor concentrations).

 By assuming that no sulfur is created indoors, we can estimate
 the contribution  of personal activities to personal exposure
 (Perscontrib) by multiplying the outdoor PM2 5 by the ratio of
 the  personal  sulfur   concentration to  the  outdoor  sulfur
 concentration, and subtracting that product from the observed
 personal exposure.   As with the  similarly  defined indoor-
 generated particle contribution to total indoor concentrations, we
 can regress Perscontrib on  the household characteristics  and
 personal activities variables without including the  outdoor
 concentration as  an input, since we used  it  to  determine the
 values of Perscontrib. Again, we checked to confirm that, when
 included, the outdoor concentration was not significant.  The
 model is
        Perscontrib = a + ft, *x,
(3-25)
 The focus on the non-outdoor part of personal exposure again
 showed that air exchange was an important variable, tending to
 decrease the personal contribution by 4.1 ug/m3 per unit change
 in the air exchange variable  (Table 3-28).  Smoking, burned
 food, and cooking fuel variables significantly influence personal
 exposure concentration.  Candle  burning increased  personal
 exposure by 0.05 ug/m3 per minute burned.

 The factors affecting the outdoor contribution to PM2.5 personal
 exposure were investigated  using multiple regression (Table 3-
 29). Here the model is
     Outcontribpers = a + /3oul*pm25out + fi,*x,
(3-26)
The most influential factor, as expected, was the outdoor PM2.5
concentration. This is so much stronger than all the rest of the
factors put together (t-value of 45 compared to values <7 for the
other variables)  that it can be considered the single variable
driving personal exposure to particles of outdoor origin.  The
next strongest factor (AGE) may be an artifact due to the
unbalanced design of the study, in which not all homes were
monitored on the same day.  The next three factors have to do
with air exchange, which depends on window opening behavior
          (windopen,  windowall) and  on  indoor-outdoor  temperature
          differences (roughly approximated by TempC). The increased air
          exchange rates due to increases in these variables brings more
          outdoor particles indoors, and thus affects personal exposure to
          the extent the person is at home.  Note that the R2 of 0.82 for this
          model of the influence of outdoor particles on personal exposure
          precisely matches that for the  influence of outdoor particles on
          indoor particle concentrations  (last part of Table 3-25).

          Finally,  personal exposure to  sulfur was  regressed against
          outdoor  sulfur and  the household  characteristics, using the
          following set of models:
                     Spers = a + fi,n*Sin

                     Spers = a+ fioll,*Sout

                     Spers = a + /?„„, *Soitt
                                                    (3-27)

                                                    (3-28)

                                                    (3-29)
            Spers = a + ftuu,*Sout + pn*Sin + b*x,    (3-30)

            Spers = a+pol,,*Sout + P*x,             (3-31)

These correspond to having greater or less information about the
outdoor and indoor sulfur concentrations and the questionnaire
variables.

In general, very high values of R2  are achieved (Table 3-30).
The worst case is having only the outdoor sulfur concentration,
and that alone explains 0.78 of the variance in personal exposure
to sulfur. If the indoor sulfur were known alone, the R" could be
improved to 0.88.  Using both measured indoor and outdoor
concentrations to estimate personal exposure  increases R" to
0.91.  Other influential  variables included time  spent near a
smoker and the number of pilot lights in the house, both possible
sources  of sulfur as mentioned above.

Variables Affecting Air Exchange and the
Infiltration Factor
Our primary focus has been on documenting the influence of
outdoor fine particles on indoor concentrations and personal
exposures. Two of the most influential factors are air exchange
and the infiltration factor.

Variables Affecting Air Exchange
Previous work (Howard-Reed et al., 2002; Wallace et al., 2002)
identified window opening and  the absolute indoor-outdoor
temperature difference as  two main  variables affecting air
exchange.  To  create a variable approximating  the absolute
indoor-outdoor  temperature difference, we took the absolute
value of the  difference between  the  outdoor temperature
(TempC) and a typical indoor temperature of 72 F (22.2 C). The
new variable was called AbsTempDif.
                                                         49

-------
Table 3-28.  Regression of the Non-ambient-related Contribution (Perscontrib) to Personal PM2 5 Exposure
Variable
Perscontrib
Intercept
numsmok
CANDLDUR
C_FUEL
Burning
airex
Cleaning_1
DIRT_RD
Exhstfan
outdoor
A_C
Other indoor
cooking
Slope

-9.68
11.45
0.05
8.44
10.83
-4.12
4.26
5.85
4.61
-0.03
1.31
0.09
0.02
Std.Err.

3.55
1.57
0.01
1.62
2.25
0.97
1.14
1.69
1.38
0.01
0.41
0.03
0.01
p-level N R2(adj.)
677 0.30
0.006489
0.000000
0.000000
0.000000
0.000002
0.000024
0.000194
0.000569
0.000850
0.001324
0.001585
0.005808
0.022302
                                                               50

-------
Table 3-29.  Regression of the Ambient-Related Contribution to Personal PM2 5 Exposure

Intercept
PM25out
AGE
windowall
Windopen
TempC
cigsmokd
MILDEWavg
smoke
cooking
A_C
DIRT_RD
Travel
Pets
dust
FLRCOVav
Grooming
AREA
B
-0.952
0.521
0.044
0.003
1.078
-0.062
0.221
1.113
0.009
0.005
0.234
0.959
0.004
0.754
-0.688
-0.012
-0.006
-0.001
SE
0.588
0.012
0.007
0.000
0.211
0.012
0.056
0.296
0.002
0.001
0.068
0.295
0.002
0.298
0.271
0.004
0.002
0.000
t(691)
-1.6
45.2
6.7
6.4
5.1
-5.0
3.9
3.8
3.6
3.6
3.5
3.3
2.9
2.5
-2.5
-2.8
-3.0
-3.2
p-level N R2(adj.)
0.105780 710 0.82
0.000000
0.000000
0.000000
0.000000
0.000001
0.000101
0.000187
0.000288
0.000311
0.000593
0.001196
0.004144
0.011484
0.011302
0.005730
0.002672
0.001409
                                                            51

-------
Table 3-30.  Regressions of Personal Exposure to Sulfur on Indoor and Outdoor Concentrations and Questionnaire Variables
Variable
Spers
Intercept
Sin
Spers
Intercept
Sin
Sout
Spers
Intercept
Sout
Spers
Intercept
Sin
Sout
DIRT_RD
smoke
windowall
MILDEWavg
C_FUEL
Grooming
outdoor
Otherjoc
TempC
Spers
Intercept
Sout
PILOT
windowall
AGE
AREA
smoke
DIRT_RD
FLRCOVav
Windopen
TempC
MILDEWavg
Travel
BUSY_RD
FAN
cooking
cigsmokd
sweep
Slope

42.50
0.91

-12.71
0.66
0.17

89.52
0.49

-1.59
0.65
0.17
117.68
0.87
0 14
81 04
-55.33
-0.45
0.34
0.22
2.46

28.25
0.50
123.65
036
3.63
-0.10
1.03
111 72
-1.62
72.43
-459
92.35
0.47
70.43
-76.58
0.33
1398
^5.29
Std.Err.

15.67
0.01

14.32
0.02
0.01

21.54
0.01

31.71
0.02
0.01
18.99
0.16
0.03
20.08
15.33
0.14
0.11
0.08
1.07

58.70
0.01
13.27
0.05
074
0.02
0.26
31.23
0.48
22.40
1 51
30.90
0.16
24.83
29.86
0.14
601
21.61
p-level N R2(adj.)
727 0.88
0.006860
0.000000
727 0.91
0.374879
0.000000
0.000000
750 0.78
0.000036
0.000000
722' 0.93
0.959998
0.000000
0.000000
0 000000
0.000000
0.000006
0 000060
0.000327
0.001283
0.002197
0.010413
0.022455
715 0.85
0.630470
0 000000
0.000000
0.000000
0.000001
0.000040
0.000105
0.000372
0.000862
0.001280
0.002480
0.002903
0 003808
0 004686
0.010533
0.017800
0.020369
0.036424

-------
Because we saw increased air exchange rates in summer, we
considered using Season as a separate independent variable. The
reasoning here is that perhaps people would  tend to keep the
windows closed and the air conditioner running in Summer even
on a relatively cool day, whereas on a day with the identical
temperature in another season they might not bother turning on
the air conditioner. Thus, both temperature and season could
have separate effects.   First, Season was  coded by  average
temperature, using a 3-point scale (Winter = 1, Fall and Spring =
2, Summer= 3); however, a regression including either season or
TempC showed that TempC produced slightly higher R".  It was
then thought that a 3-point scale might be too restrictive, so a 5-
point scale was created by month (e.g., Jan-Feb = 1, July-August
=5). There were only 10 months, with March and December
missing. The new variable was called MonthbyTemp. This time
when the  regression was run putting  in either  TempC  or
MonthbvTemp, the R~ using the  MonthbyTemp variable was
slightly higher than that using TempC (Table 3-31).

The two strongest variables are in fact those that we already
know from previous studies: the window opening width and the
absolute indoor-outdoor temperature difference. The latter has a
coefficient of 0.04 ach/°C, which agrees well with the estimate of
0.02 ach/°C in Wallace et al. (2002). The next strongest variable
is the number  of persons in the home.  This variable, as well as
Ihepets variable, was included on the questionnaire because of
results from past studies (e.g., Thatcher and Layton,  1995)
indicating that people and pets can increase particle levels due to
resuspension and going outdoors more often, which will bring in
outdoor particles when the door is open.  The pets variable also
appears, although just missing significance.  The next strongest
variable is age of the home—previous studies have noted that
older  homes are constructed more  loosely.  The  use of the
exhaust fan tends to increase air exchange, as noted in a previous
study  (Wallace et  al.,  2002).  Another  variable   is  the
MonthbyTemp  variable,  reflecting  the  increased use of air
conditioners (and therefore closed windows) in the summer. The
presence of a clothes dryer (nearly all of which were vented
outdoors) increases air exchange since almost the same volume
of air must enter the house to replace the heated air vented
outdoors. We have no explanation for the appearance here of the
VAC and FLRCOVav variables, although their effect on the total
R" value of the model is very small.

Variables Affecting the Infiltration Factor
The multiple regression on the indoor/outdoor sulfur ratio was
run on all variables, including airex (Table 3-32). However, the
five strongest variables contributing to this ratio can all be seen
to be variables already  contributing to the air exchange rate.
Therefore these variables are, in a sense, being double-counted.
They are obscuring and weakening the actual relationship with
the air exchange rate. Therefore the regression was run again
including all variables except those already found to contribute
to the air exchange rate (Table 3-33).  The final model shows
that  the air exchange rate (together with the windows open
variable)  is  the  strongest  variable  affecting  the  sulfur
indoor/outdoor ratio. The air exchange rate is the only one of our
variables  that  explicitly  appears  in  the  equation  for the
infiltration factor:  (Pa/(a+k)). The  three remaining variables
provide very little contribution (about 3%) to the total R2 of the
model. One of these, A/C, appears to have the wrong sign, since
presence of an air conditioner would be expected to reduce the
infiltration factor. However, this variable ranges from 1 to 5 air
conditioners and therefore is measuring the falloff in efficiency
as one goes from central air conditioning to multiple window air
conditioning.
                                                          53

-------
Table 3-31.  Variables Affecting Air Exchange Rate
Variable
a/rex
Intercept
windowall
ABSTempDIF
Numpeopl
AGE
VAC
FLRCOVav
DRYER
MILDEWavg
Monthbytemp
Mealsckd
Exhstfan
Unknown
Pets
Slope Std.Err.

0.5156 0.1355
0.0010 0.0001
0.0434 0.0040
0.0693 0.0094
0.0079 0.0013
-0.3079 0.0496
-0.0043 0.0008
0.2071 0.0458
-0.2406 0.0544
-0.0673 0.0153
-0.1488 0.0390
0.1375 0.0410
0.0025 0.0008
0.1586 0.0523
t(711) p-level N R2(adj.)
744 0.51
3.8 0.00015
13.2 0.000000000000
10.8 0.000000000000
7.3 0.000000000001
6.3 0.00000000062
-6.2 0.00000000093
-5.1 0.00000044
4.5 0.0000073
-4.4 0.000011
-4.4 0.000012
-3.8 0.00014
3.4 0.00085
3.3 0.0012
3.0 0.0025
TableS- 32. Variables Affecting Indoor/Outdoor Sulfur Ratio
Variable
Sinout
Intercept
Monthbytemp
AGE
airex
Windopen
windowall
C_FUEL
DIRT_RD
VAC
DUSTY_RD
Slope Std.Err.

0.68439 0.02916
-0.03656 0.00343
0.00299 0.00037
0.06455 0.00925
0.06168 0.01027
0.00014 0.00002
-0.06117 0.01047
0.06170 0.01362
-0.05484 0.01323
-0.04057 0.01203
Table 3-33. Variables Affecting Indoor/Outdoor Sulfur Ratio:
Variable
Sinout
Intercept
airex
Windopen
A_C
numsmok
BUSY_RD
Slope Std.Err. t(699)

0.409 0.010 40.3
0.121 0.008 15.1
0.090 0.010 9.2
0.015 0.003 4.3
0.055 0.015 3.7
0.036 0.011 3.2
t(638) p-level N R2 (adj.)
659 0.46
23.5 0.00000000000000
-10.7 0.00000000000000
8.1 0.00000000000000
7.0 0.000000000008
6.0 0.0000000031
5.9 0.0000000062
-5.8 0.0000000083
4.5 0.0000070
-4.1 0.000039
-3.4 0.00079
Reduced Model
p-level N R2 (adj.)
704 0.43
0.000000000000
0.000000000000
0.000000000000
0.000018
0.00021
0.0013
                                                                54

-------
                                                   Chapter 4
                                                  Discussion
 To reach our primary goal of estimating the contribution of
 outdoor PMi.s to personal exposures, we set an intermediate goal
 of estimating the contribution of outdoor particle concentrations
 to indoor particle concentrations.   Since personal  exposures
 depend heavily on indoor concentrations, this may be a good first
 approximation.   We calculated the contribution of  outdoor
 particles by multiplying the outdoor concentration by the ratio of
 the indoor to outdoor sulfur concentrations:
   C   =
   *- mu   '
                            1   C
                            olil *- Mill
(4-1)
 where Cmo is the concentration indoors of particles infiltrating
 from outdoors, and where the S,n/So:i, term is an estimate of Fmj.
 We  also  calculated  Fmt  using  a  combination  of  the
 indoor/outdoor sulfur ratios and the air exchange measurements
 and found excellent agreement (R2 = 0.96-0.99), indicating that
 the estimates of Fml for individual homes were quite stable. This
 is one of the  first studies to calculate  infiltration factors and
 outdoor exposure  factors  for  individual  homes and persons,
 respectively, and  is also one of the  first  to  carry  out the
 regressions of that portion of personal exposure due to particles
 of outdoor origin on outdoor concentrations.

 Limitations  of the Regressions and
 Influence of Assumptions
 When we run a regression of  €,„„ on  C,,,,,, we are in fact
 regressing a term on a portion of itself:
p  ,r   = A C   +
1 inj *- out   **• *- out
                                                    (4-2)
The infiltration factor does not have a high degree of day-to-day
variability: Fm, = 0.59 (0.16 SD) for all homes, and an even
smaller  standard deviation within  homes.   Therefore  our
regression is not very different from regressing a term on a
constant fraction of itself, which will result in an R2 of 1.  This is
a result  that is forced by our basic  assumptions   no  indoor
sources of sulfur, no coagulation, condensation,  particle-gas
conversion,  or indoor chemistry, negligible  transient  terms,
instantaneous  perfect  mixing,  penetration  and  deposition
characteristics of PMi.s identical to those of sulfur - leading to a
simple linear relation between the outdoor concentration and the
fraction of outdoor air particles infiltrating the house.  The
regression on outdoor air measured at all the homes resulted  in
an R2 of 0.71, a value that is certainly higher than reality due  to
our assumptions. The individual regressions on 36 homes gave
even higher values of R" (median 0.77). When the regressions
were run on the outdoor PMi.j concentrations measured by the
FRM at the central site, the overall R2 value was lowered to 0.60
(median 0.73), but with the same caveat that our assumptions
force a high correlation and therefore these estimates are not be
considered best estimates, but rather upper bounds.

A regression of personal PM2 5 on outdoor PM: 5 concentrations
provides an estimate of F^ of 0.59 + 0.01  SE, with one
influential outlier  removed.   With  the outlier included, the
estimate was 0.64  + 0.02  SE.   The intercept with the  outlier
excluded (12.3 + 1.4 ug/m3) is an estimate of £„„, the exposure
due to non-outdoor sources. The difference between the  indoor
source contributions estimated at 7.7 ug/m3 in Table 3-5 and the
non-outdoor source contribution is sometimes attributed to the
"personal cloud," which in this case equals 4.6 + 2.0 ug/m3.

The similarity of the slopes for the personal  vs. outdoor and
indoor vs. outdoor PMi, regressions (0.59  compared to 0.60)
suggests that the time spent indoors drives  the relationship
between personal exposure and outdoor concentrations. That  is,
the infiltration factor F,,,/, which governs the reduction of particle
concentrations as they enter a  house, is very similar  to the
outdoor exposure factor Fpcx, which governs  the reduction  in
outdoor concentrations contributing to personal exposure.

A major assumption has been that fine particles in general will
behave like sulfur in terms of penetration and deposition. How
good is this  assumption?  Sulfur is found primarily in the fine
fraction, and within that fraction it typically has diameters <0.5
                                                          55

-------
(.an, smaller than  most other elements.  From theoretical and
experimental studies,  deposition velocities appear to reach a
minimum around  diameters of 0.1-0.2 um (Lai and Nazaroft,
2000).  Deposition  rates for sulfur and other elements  were
estimated in the PTEAM study (Ozkaynak et al, 1996).  The rate
for sulfur was 0.16 h"' (0.02 SE).  Rates for other elements were
generally higher (e.g., iron, 0.70 h"1 (0.30 SE)). Therefore sulfur
will stay elevated  indoors longer than most other elements, and
possibly longer than most of the remainder of the fine particle
mass. This leads to overestimates of the contribution of outdoor
particles to indoor  PM: 5.  Evidence  for  the amount of the
overestimate is provided by the  results for iron concentrations.
The indoor/outdoor ratio for iron was 0.38 (0.18 SD; N =  766).
This was the lowest ratio of all the elements. For Si it was 0.58
(N = 766);  for Mn it was 0.52 (N = 409). These were elements
that appeared  to have few indoor sources, judging from their
high indoor-outdoor product-moment correlations (0.81 for Fe,
0.74 for Si, N = 762 samples for each; 0.58 for Mn, N = 460
samples; these correlations were almost as high as the value of
0.85 for sulfur).  For the other elements, the indoor/outdoor
correlations were lower and the indoor/outdoor ratios higher due
to indoor sources, and cannot  be used to help  estimate the
amount of the overestimate due  to using the sulfur ratios. Iron
itself makes up only a small fraction of PM: 5 mass, but it could
be a marker for the behavior of a larger fraction.

Sulfur accounted for 20% of indoor PM: 5, assuming it is in the
form of ammonium sulfate. If the remaining 80° o  of the indoor
particle mass behaved more like iron, outdoor air particles would
account for only 42% of indoor air PM; 5, rather than 58% as was
found using sulfur alone.   However, it is unlikely  that the
remainder  of  the  indoor air particles  behave more like  iron,
because the RCS  model regression of indoor PM: 5 on outdoor
PM: j (Figure 3-13) resulted in  a slope very close to the  slope
predicted  by  the  sulfur  indoor-outdoor  ratio.    These
considerations indicate that the overestimate of the outdoor
contribution due to relying on the sulfur ratios is not a major one
and does not affect the relative order of the homes in terms of the
fraction of indoor PM: 5 produced by outdoor particles.

Based on previous studies,  we attempted to use the particle
measurements to  estimate an infiltration  factor for each home;
however, we found poor agreement with the estimates in this
study  using the sulfur ratio.  Most other studies of personal
exposure and/or indoor concentrations of fine particles  have
fewer measurements per home than this study.  Therefore our
results  suggest that particle mass measurements alone cannot
provide reliable estimates of /•"„,,.

We  estimated the  outdoor  exposure factor FfH., using  two
methods: personal, outdoor sulfur ratios and a model using time
indoors and outdoors.  The model consistently overestimated the
outdoor factor. We speculate that this  is because people spend
time  in unmonitored  environments, and our assumption that
exposures  are the  same in  those environments as in the
monitored home environment may be incorrect.  We found that
estimating  personal exposure using only the indoor/outdoor
sulfur ratio (F,n/) gave better results than estimates using the
modeled outdoor exposure  factor F/KY based on time spent
indoors and outdoors.

Although several different methods were employed in making
these estimates, the methods ultimately depended on the sulfur
measurements only. Because of the way P and k are related in
the equation for F,,,h an infinite number of solutions are available
for any given infiltration factor. The solution surface for P and k
is very flat (meaning a very wide error associated with all point
estimates); therefore a slight error in any measurement can lead
to  a very large error in the estimation of P and k.  For example,
each of the slightly different methods produced  four or five
physically impossible  estimates for P  Although bounds can be
established to prevent  P from exceeding unity, the existence of
these nonphysical solutions suggests that measurement errors or
violations of our assumptions may have caused large errors in
estimating these parameters.  Since we have no  independent
methods of estimating  these parameters, we are  unable to
validate our estimates of P and k for individual homes.  The
estimates of P and k that were significantly different from zero
included about'24 of the 36 homes. The median value for the
penetration coefficient P was 0.81  (interquartile range 0.66 to
0.90).   The median   for the  deposition  rate k was 0.24 h'1
(interquartile range 0.12  to 0.35 h"'). These estimates arc similar
to those obtained in some other studies (Liu et al., 2003; Wallace
et al., 2002; Thatcher et  al., 2003).

Because not all homes  were done in the same time periods, there
were certain  unavoidable differences in  the outdoor  particle
concentrations encountered at the time they were monitored. We
attempted to identify possible artifacts (significant associations
with no possible causal explanations) by regressing central-site
and  residential outdoor concentrations vs. the questionnaire
variables. We did find some artifacts, and therefore caution that
some apparently significant relationships may be  due to these
temporal variations rather than a true causal relationship.  This
may be particularly  true for  certain unvarying household
characteristics such as building age, location near a road, and
type of cooking fuel (gas vs. electric) because these unvarying
characteristics enter the regressions as a block of entries repeated
up to 28 times per house. Therefore their number of degrees of
freedom is  greatly reduced and any variation due to temporal
heterogeneity  will be  multiplied by this repeated appearance.
This caution  extends to the variables AGE,  DIRT  ROAD.
C_FUEL (cooking fuel). S_WIN (storm windows). VAC. and
AREA among others.

Nonetheless, regressions of indoor concentrations and personal
exposures  showed   greatly   improved   R:  results  when
questionnaire   variables  were   added   to   the   outdoor
concentrations. The improvement was from R: about 0.1 to R:
about 0.4, for  both indoor and personal estimates.  Among the
                                                          56

-------
significant contributors to indoor and personal exposure were
smokers in the household, cooking, and number of people in a
household, all variables that have previously been found  to
contribute to indoor PM concentrations (Ozkaynak et al, 1996b;
Wallace et al., 2004b).  Some variables not previously identified
as contributing to indoor or personal exposures were burning
food, duration of candle use, number of pilot  lights, use of an
exhaust fan, proximity to a  dirt road, and presence of electric
stoves (compared to gas stoves).  The latter two variables are to
be treated with caution since they are unvarying household
characteristics and therefore may have been subject to temporal
unevenness of sampling. Vacuuming was another variable that
achieved significance at times. The presence of a clothes dryer
was significantly associated with reduced indoor concentrations
and personal exposure.

An important result was the appearance of the air exchange
variables (including windows open or closed) in the regressions
on estimated concentrations of indoor-generated and outdoor-
generated particles. The fact that these variables appeared in the
expected directions whereas they did not appear when regressing
on the mixture of both indoor-generated and outdoor-generated
particles  is  confirmation  that  our  division into the  two
components using the sulfur ratios was successful.

We found several variables that were important in affecting air
exchange rates. Opening windows was the single most important
variable, as has been shown in previous studies (Howard-Reed et
al., 2002). The indoor-outdoor temperature difference (which
affects the indoor-outdoor pressure difference and therefore the
driving force  for air exchange)  was the next most powerful
variable.  The coefficient for this variable (0.04 ach/°C) was of
similar magnitude to that found by  Wallace et al. (2002). The
number of people in the home and the presence of pets, both
significant variables in this study but not often found in other
studies,  contribute to  increased  air exchange  through the
increased number of times going in and out of the house. The use
of a kitchen exhaust fan was associated with an increase of 0.14
ach.  This also compares well with the finding of a 0.8 ach
increase associated with use of an attic exhaust fan (Wallace et
al., 2002), given the much longer periods that  an attic exhaust
fan may run compared to a kitchen  exhaust fan.

The single strongest variable affecting the infiltration factor F,nf
is the air exchange rate, as can be seen from the equation for Fm]
and the observed relationship shown in Figure 3-27. All other
variables entering our analysis are either highly correlated with
air exchange rate or appear to  be artifacts arising from the
unequal monitoring of homes on different days.
                                                          57

-------
                                                    Chapter 5
                                                 Conclusions
The sulfur measurements showed excellent agreement between
the personal samplers and the fixed indoor-outdoor Harvard
Impactors.   More than  160 co-located  measurements  both
indoors and outdoors resulted in slopes insignificantly different
from 1  and intercepts insignificantly different from zero, with R"
values  of  0.97  for  both  indoor and  outdoor  locations.
Uncertainties of individual sulfur measurements were estimated
at 8%.

The data appear to be consistent with the hypothesis that there
are few  indoor  sources of  sulfur.   Only  about   1%  of
measurements showed higher levels indoors than out. Although
nearly  all regressions of indoor sulfur vs. outdoor sulfur  gave
positive intercepts, these intercepts were often relatively small
and could be due to measurement error.  Therefore we accepted
the indoor/outdoor sulfur ratio as our best estimate of Fml for
PM:.S.

Estimates of F,nl averaged over all seasons varied over a large
range  by household (0.26-0.87), but half of the homes were
within 20% of the median value of 0.59. The infiltration factor
was significantly lower for many  homes in the summer  season,
presumably  due  to  the  closed  windows  and  increased
recirculation and filtration of air associated with the use of air
conditioners.  Evidence from measurements of iron and silicon
suggested that the estimates  of the infiltration factor using the
sulfur  indoor/outdoor  ratio are  likely to overestimate  the
influence of outdoor air due to the low deposition velocity of
sulfur compared to other likely constituents of PM: 5. However,
other considerations lead us to think that the overestimate is not
large; also, it would not affect the relative ranking of the homes
with respect to the outdoor contributions to indoor PM:  5.

In general, the outdoor exposure factor FpL,x was very similar to
F!nf.  This is expected, since persons spend most of their time
indoors, but the very close agreement is also an indication that
the sulfur measurements were reproducible and agreed well even
though the personal monitor collected 10 times less material than
the indoor/outdoor monitors.

Fpn was usually smaller by a few percent than Fm(, although a
simplistic application of time-activity budgets indicates that it
should be larger by about 6 or 7% for persons spending a typical
amount of time outdoors and in vehicles. This may be due to
time  spent  in  unmonitored  indoor  locations  (e.g., office
buildings, department stores) that have mechanical ventilation,
recirculation, and filtration,  thus lowering exposure to sulfur
while in those locations.  The fact that /•),„ was slightly larger
than F,,,/ during  the summer season, when many homes were
closed and using recirculated and  filtered air, supports this
hypothesis.  We conclude that the model advocated in earlier
publications for determining /•*,,„ by using time-activity budgets
is  not  useful  for  studies  that  do   not  measure   indoor
concentrations in schools, workplaces, shopping malls, and other
locations where persons spend substantial amounts of time. The
preferred way  to estimate Fpcx  is  by using personal sulfur
measurements, but lacking those, it may be just as useful to use
Fm/ as the best estimate of Fpcx.

Regressions of indoor air concentrations due to  particles of
outdoor origin vs. outdoor concentrations had generally high R:
values, as did regressions  of personal exposure to particles of
outdoor origin vs. outdoor concentrations. However, this result
is partially due to the many assumptions in our approach leading
to  a simple linear relation between indoor concentrations of
particles  of  outdoor  origin and outdoor  concentrations.
Therefore these estimates should not be treated as best estimates,
but rather as upper bounds to the actual amount of variance in
personal  exposure  to  PM:5 of outdoor origin that can be
explained by outdoor measurements at the central site.

We investigated whether particle measurements alone could be
used to estimate the infiltration factor.  However, comparisons
with the sulfur ratio method indicated that particle measurements
                                                          58

-------
alone cannot be used to give reliable estimates of Finf.

The two unmeasured parameters contributing to Fln/ (P and k)
were estimated using both linear and nonlinear approaches. The
median  value  for  the penetration  coefficient P was  0.81
(interquartile range 0.66 to 0.90). The median for the deposition
rate k was 0.24 h'1 (interquartile range 0.12 to 0.35 h"1). Recall
that these values  of k and P are for particles in the same size
range as sulfur particles.   However, we  conclude  that  the
unphysical values obtained for P in some cases casts doubt on all
the estimates of P and k for individual homes; we are unable to
validate these estimates.

Air exchange rates were found to depend primarily on opening
windows and  on the absolute  indoor-outdoor temperature
difference. Other contributors included use of a kitchen exhaust
fan, presence of a vented clothes dryer, and number of persons in
a household. The infiltration factor was primarily dependent on
air exchange, as expected from the basic equilibrium equation.

Regressions of indoor concentrations and personal exposures
showed considerably improved R2 estimates by consideration of
questionnaire variables. In particular, smoking, cooking, number
of persons in  a household, and burned food were important
contributors.  Air exchange  rates were also very important
variables, but only after the total indoor particle concentration
had been split into indoor-generated  and  outdoor-generated
portions.  The strong effect of air exchange (increasing outdoor-
generated  particle  concentrations  and decreasing  indoor-
generated particle concentrations) was an important confirmation
of the success of our efforts to resolve these two  contributors to
total indoor particle levels.
                                                          59

-------
                                                    Chapter 6
                                                   References
Abt, E.. Suh, H.H., Allen, G. and Koutrakis, P. (2000a).
  Characterization of indoor particle sources: A study conducted
  in the metropolitan Boston area. Environmental Health
  Perspectives 108( 1 ):35-44.

Abt, E., Suh, H.H., Catalano, P.J. and Koutrakis, P. (2000b).
  Relative contribution of outdoor and indoor particle sources to
  indoor concentrations. Environmental Science ami Technology
  34:3579-3587.

Allen R., Larson, T., Sheppard, L., Wallace, L. and Liu, L-J S.
  (2003).  Use of Real-time Light Scattering Data to Estimate
  the  Contribution of Infiltrated and Indoor-Generated Particles
  to Indoor Air.  Environmental Science and Technology
  37:34X4-3492.

Allen R., Larson, T., Sheppard, L.. Wallace, L. and Liu, L-J S.
  (2004).  Estimated hourly personal exposures to ambient and
  non-ambient paniculate matter among sensitive populations in
  Seattle, Washington. Journal of the Air and Waste
  Management Association 54:1197-1211.

AQCD (2003).  Fourth External Review Draft of Air Quality
  Criteria for Paniculate Matter (June, 2003). National Center
  for  Environmental Assessment, U.S. Environmental Protection
  Agency, Research Triangle Park, NC.

Belsley, D.A., Kuh. E., and Welsch, R.E., Regression
  Diagnostics, John Wilev & Sons, 1980
Clayton. C.A., Perritt. R.L., Pellizzari, E.D., Thomas, K.W.,
  Whitmorc, R.W., Wallace. L.A., Ozkaynak, H.. and Spengler,
  J.D. (1993).  Panicle Total Exposure Assessment
  Methodology (PTEAM) study: distributions of aerosol and
  elemental concentrations in personal, indoor, and outdoor air
  samples in a southern California community. Journal of
  Exposure Analysis and Environmental Epidemiology 3:227-
  250.

Ebelt, ST., Petkau, A.J., Vedal, S., Fisher, T.V., and Brauer, M.
  (2000).  Exposure of chronic obstructive pulmonary disease
  patients to paniculate matter:  relationship between personal
  and ambient air concentrations.  Journal of the Air and Waste
  Management Association 50:1081 -1094.

Evans, G.F., Highsmith, R.V., Sheldon, L.S., Suggs, J.C.,
  Williams, R.W.. Zweidmger, R.B., Creason, J.P., Walsh, D..
  Rodes, C.E. and Lawless^.A. (2000).  The 1999 Fresno
  paniculate matter exposure studies: comparison of community,
  outdoor, and residential PM mass measurements. Journal of
  the Air and Waste Management Association 50( 11): 1887-
  1896.

Highsmith. V.R., Rodes, C.E. and Hardy. R.J. (1988).  Indoor
  panicle concentrations associated with use of tap water in
  portable humidifiers.  Environ Sci Tech 22:1109-1112.

Hopke PK, Ramadan Z, Paatero P, Norris GA, Landis ML,
  Williams RW,  and Lewis CW (2003).  Receptor modeling of
  ambient and personal exposure samples: 1998 Baltimore
  Paniculate Matter Epidemiology-Exposure Study.
  Atmospheric Environment 37(23):3289-3302.
                                                           60

-------
 Howard-Reed, C. Rea, A. W., Zufall, M. J., Burke, J. M,
   Williams, R. W., Suggs, J. C., Sheldon, L. S., Walsh, D., and
   Kwock, R. (2000).  Use of a continuous nephelometer to
   measure personal exposure to particles during the U.S.
   Environmental Protection Agency Baltimore and Fresno panel
   studies. Journal of the Air and Waste Management Association
   50:1125-1132.

 Howard-Reed C, Wallace LA, and Emmerich SJ (2003).  Effect
   of ventilation systems and air filters on decay rates of particles
   produced by indoor sources in an occupied townhouse.
   Atmospheric Environment 37(38):5295-5306.

 Howard-Reed, C.H., Wallace, L.A., and Ott, W.R. (2002). The
   effect of opening windows on air change rates in two homes.
   Journal of the Air and Waste Management Association
   52:147-159.2002.

 Janssen, N. A. H., Hoek, G., Harssema, H., and Brunekreef, B.
   (1997). Childhood exposure to PM]0: relation between
   personal, classroom, and outdoor concentrations. Occupational
   and Environmental Medicine 54:888-894.

 Janssen, N.A.H., Hoek, G., Brunekreef, B., Harssema, H.,
   Mensink, I., and Zuldhof, A. (1998).  Personal sampling in
   adults: relation among personal, indoor, and outdoor air
   concentrations. American Journal of Epidemiology 147:537-
   547.

 Janssen, N. A. H., Hoek, G., Harssema, H., and Brunekreef, B.
  (1999).  Personal exposure to fine particles in children
  correlates closely with ambient fine particles. Archives of
  EnvironmentalHealth 54:95-101.

 Janssen, N. A. H., De Hartog, J. J., Hoek, G., Brunekreef, B.,
  Lanki, T., Timonen, K. L., and Pekkanen, J. (2000). Personal
  exposure to fine particulate matter in elderly subjects: relation
  between personal, indoor, and outdoor concentrations. Journal
  of the Air and Waste Management Association 50:1133-1143.

 Keeler, G. J., Dvonch, J. T., Yip, F. Y., Parker, E. A., Israel, B.
  A., Marsik, F. J., Morishita, M., Barres, J.  A., Robins, T. G.,
  Brakefield-Caldwell, W., and Sam, M. (2002).  Assessment of
  personal and community-level exposures to particulate matter
  among children with asthma in Detroit, Michigan, as part of
  Community Action Against Asthma (CAAA). Environ. Health
  Perspect. 110(suppl. 2):173-181.

Lai, A.C.K., Burne, M.A., and Goddard, A.J.H. (1999).
  Measured deposition of aerosol particles on a two-dimensional
  ribbed surface in a turbulent duct flow. Journal of Aerosol
  Science 30(9): 1201-1214.
Lai, A.C.K. and Nazaroff, W.W. (2000). Modeling indoor
  particle deposition from turbulent flow onto smooth surfaces.
  Journal of Aerosol Science 31:463-476.

LaRosa, L.E., Buckley, T.J., and Wallace, L.A. (2002).  Real-
  time indoor and outdoor measurements of black carbon in an
  occupied house: An Examination of Sources. Journal of the
  Air and Waste Management Association 52:174-185.

Landis, M.S., Morris, GA., Williams, R.W., and Weinstein, J.P
  (2001).  Personal exposures to PM2 5 mass and trace elements
  in Baltimore, Maryland. Atmospheric Environment 35:6511-
  6524.

Liu, L.-J. S., Box, M., Kalman, D., Kaufman, J., Koenig, J.,
  Larson, T., Sheppard, L.,  Slaughter, C., Lewtas, J., and
  Wallace, L.A. (2003).  Exposure assessment of particulate
  matter for susceptible populations in Seattle, WA.
  EnvironmentalHealth Perspectives 111(7):909-918.

Liu, D.L. and Nazaroff, W.W. (2001). Modeling pollutant
  penetration across building envelopes. Atmospheric
  Environment 35:4451-4462.

Long, C.M., Suh, H.H., Catalano, P.J. and Koutrakis, P. (2001).
  Using time- and size-resolved particulate data to quantify
  indoor penetration and deposition behavior. Environmental
  Science & Technology 35(10):2089-2099.

Long, C.M., Suh, H.H. and Koutrakis, P. (2000).
  Characterization of indoor particle sources using continuous
  mass and size monitors. Journal of the Air and Waste
  Management Association  50(7): 1236-1250.

Mosley, R.B., Greenwcll,  D.J., Sparks, L.E., Guo, Z., Tucker,
  W.G., Fortmann, R. and Whitfield, C. (2001). Penetration of
  Ambient Fine Particles into the Indoor Environment. Aerosol
  Science and Technology 34:( 1) 127-136.

National Research Council-National Academy of Science
  (1998).  Research priorities for airborne particulate matter I:
  immediate priorities and a long-range research portfolio.
  National Academy Press,  Washington, DC.

Nazaroff, W.W. and Cass, G.R.  (1989). Mathematical modeling
  of indoor aerosol dynamics. Environmental Science &
  Technology 23:157'-166.

Ott, W., Wallace, L. and Mage D. (2000).  Predicting Particulate
  (PMIO) Personal exposure distributions using a random
  component superposition  statistical model. Journal oj the Air
  and Waste Management Association 50:1390-1406.
                                                          61

-------
Ozkaynak, H., Xue, J., Weker, R., Butler, D., Koutrakis, P., and
  Spengler, J. (1996a).  The Particle TEAM (PTEAM) study-
  analysis of the data. Volume 111. Final Report.  EPA Contract
  #68-02-4544. Research Triangle Park, NC:  US
  Environmental Protection Agency.

Ozkaynak. H., Xue, J., Spengler, J.D., Wallace, L.A., Pellizzari,
  E.D., and Jenkins, P. (1996b).  Personal exposure to airborne
  particles and metals: results from the Particle TEAM Study in
  Riverside, CA. Journal of Exposure Analysis and
  Environmental Epidemiology: 6:57-78.

Pellizzari, E.D., Thomas, K.W., Clayton, C.A., Whitmore, R.W.,
  Shores, R.C., Zelon, H.S., and Perritt, R.L. (1992). Particle
  Total Exposure Assessment Methodology (PTEAM): Riverside.
  California pilot study.  Volume I. Final Report. EPA Contract
  # 68-02-4544, Research Triangle Park, NC:  US
  Environmental Protection Agency.

Rea, A., Zufall, M., Williams, R., Reed, C, and Sheldon, L.
  (2001). The influence of human activity patterns on personal
  PM exposure: a comparative analysis of filter-based and
  continuous particle measurements. Journal of the Air and
  Waste  Management Association 51:1271-1279.

Riley, W.J., McKone, T.E., Lai, A.C. and Nazaroff, W.W.
  (2002). Indoor particulate matter of outdoor origin:
  importance of size- dependent removal mechanisms.
  Environmental Science & Technology 36(2):200-207.

Rojas-Bracho, L., Suh, H.H., and Koutrakis, P. (2000).
  Relationships among personal, indoor, and outdoor fine and
  coarse particle concentrations  for individuals with COPD.
  Journal of Exposure Analysis and Environmental
  Epidemiology: 10:294-306.

Sarnat, J.A., Koutrakis, P. and Suh, H.H. (2000).  Assessing the
  relationship between personal  particulate and gaseous
  exposures of senior citizens living in Baltimore, MD. Journal
  of the Air and Waste Management Association 50(7): 1184-
   1198.

Sarnat, J. A., Schwartz, J.,  Catalano, P. L, and Suh, H. H.
  (2001). Gaseous pollutants in particulate matter
  epidemiology: confounders or surrogates? Environmental
  Health Perspectives 109:1053-1061.

Sarnat, J. A., Long, C. M., Koutrakis, P., Coull, B. A., Schwartz,
  J., and Suh, H. H. (2002). Using sulfur as a tracer of outdoor
  fine particulate matter. Environmental Science and Technology
  36:5305-5314.

Schwartz, J., Dockery. D. W., and Neas, L. M. (1996). Is daily
  mortality associated specifically with fine particles? Journal of
  the Air and Waste Management Association 46:927-939.
Thatcher, T. and Layton, D. (1995). Deposition, resuspension,
  and penetration of particles within a residence. Atmospheric
  Environment 29( 13): 1487-1497.

Thatcher, T. L.. Lai, A.  C. K., Moreno-Jackson, R., Sextro, R.
  G., and Nazaroff, W.  W. (2002). Effects of room furnishings
  and air speed on particle deposition rates indoors. Atmospheric
  Environment 36:\8\\-\8\9.

Thomas, K.W., Pellizzari, E.D., Clayton, C.A., Whitaker, D.A.,
  Shores, R.C., Spengler, J.D., Ozkaynak, H., and Wallace, L.A.
  (1993).  Particle Total Exposure Assessment Methodology
  (PTEAM) 1990 study: method performance and data quality
  for personal, indoor, and outdoor monitoring. Journal of
  Exposure Analysis and Environmental Epidemiology: 3:203-
  226.

Thornburg, J., Ensor, D.S., Rodes, C.E., Lawless, P.A., Sparks,
  L.E. and Mosley, R.B. (2001). Penetration of particles  into
  buildings and associated physical  factors. Part 1: model
  development and computer simulations. Aerosol Science and
  Technology 34:284-296.

Thornburg J.W., Rodes  C.E., Lawless P.A., Stevens C.D., and
  Williams R.W. (2004).  A pilot study of the influence of
  residential HAC duty cycle on indoor air quality. Atmospheric
  Environment 38:1567-1577.

US  EPA (2002).  Preliminary:particulate matter mass
  concentrations associated with  longitudinal panel studies:
  assessing human exposures of high-risk subpopulations to
  particulate matter.  National Exposure Research Laboratory,
  Office of Research and Development, EPA/600/R-01/086.
  Washington, D.C.:  U.S. Environmental  Protection Agency.

US  EPA (2003).  Exposure of High-Risk Subpopulations  to
  Particles. Final Report (APM-21). National Exposure
  Research Laboratory, Office of Research and Development,
  EPA/600/R-03/145. Washington,  D.C.:  U.S. Environmental
  Protection Agency.

Vette, A.F., Rea, A.W.,  Lawless,  P.A., Rodes, C.E., Evans, G.,
  Highsmith, V.R. and Sheldon, L. (2001). Characterization of
  indoor-outdoor aerosol concentration relationships during the
  Fresno PM exposure studies. Aerosol Science and Technology
  34:118-126.

Wallace L.A. and Williams, R. (2005).   Use of personal-indoor-
  outdoor sulfur concentrations to estimate the infiltration factor,
  outdoor exposure factor, penetration coefficient, and
  deposition rate for individual homes.  Environ Sci Tech
  39:1707-1714.
                                                           62

-------
 Wallace L.A., Emmerich S.J., and Howard-Reed C. (2004b).
   Source strengths of ultrafine and fine particles due to cooking
   with a gas stove. Environmental Science & Technology
   38(8):2304-2311.

 Wallace L.A., Emmerich S.J., and Howard-Reed C. (2004a).
   Effect of central fans and in-duct filters on deposition rates of
   ultrafine and fine particles in an occupied townhouse.
   Atmospheric Environment 38(4):405-413.

 Wallace L.A., Mitchell H., O'Connor G.T., Liu L.-J., Neas L.,
   LippmannM., Kattan M., Koenig J., Stout J.W.,  Vaughn B.J.,
   Wallace D., Walter M., and Adams K. (2003). Particle
   concentrations in inner-city homes of children with asthma:
   the effect of smoking, cooking, and outdoor pollution.
   Environmental Health Perspectives 111:1265-1272.

 Wallace, L.A. and Howard-Reed, C.H. (2002). Continuous
   monitoring of ultrafine, fine, and coarse particles in a
   residence for 18 months in 1999-2000.  Journal of the Air and
   Waste Management Association 52(7):828-844.

 Wallace, L.A., Howard-Reed, C.H. and Emmerich, S.J. (2002).
   Continuous measurements of air change rates in  an occupied
   house for one year: the effect of temperature, wind, fans, and
   windows. Journal of Exposure Analysis and Environmental
   Epidemiology 12:296-306.

 Wallace, L.A. (2000a) Correlations of Personal Exposure to
   Particles with Outdoor Air Measurements: A Review of
   Recent Studies. Aerosol Science and Technology 32:15-25.

 Wallace, L.A. (2000b). Real-Time Monitoring of Particles,
  PAH, and CO in an Occupied Townhouse, Applied
   Occupational and Environmental Hygiene 15(1): 39-47.

 Williams, R., Creason, J., Zweidinger, R., Watts, R., Sheldon, L.
  and Shy, C. (2000a). Indoor, outdoor and personal exposure
  monitoring of paniculate air pollution:  the Baltimore elderly
  epidemiology-exposure pilot study. Atmospheric Environment
  34:4193-4204.

 Williams, R., Suggs, J., Zweidinger, R., Evans, G., Creason, J.,
  Kwok, R., Rodes C., Lawless, P., and Sheldon, L. (2000b).
  Comparison of PM2 5 and PM,0 monitors. Journal of Exposure
  Analysis and Environmental Epidemiology 10:497-505.

Williams R.W., Suggs, J, Rea A., Leovic K., Vette A., Croghan
  C., Sheldon L., Rodes C., Thornburg J., Ejire A. et al. (2003a).
  The Research Triangle Park paniculate matter panel study:
  PM mass concentration relationships, Atmospheric
  Environment 37(38):5349-5363.
Williams R, Suggs J, Rea A, Sheldon L, Rodes C, and
  Thornburg J (2003b).  The Research Triangle Park paniculate
  matter panel study: modeling ambient source contribution to
  personal and residential PM mass concentrations, Atmospheric
  Environment 37(38):5365-5378.

Wilson, W. E., Mage, D. T., and Grant, L. D. (2000). Estimating
  separately personal exposure to ambient and nonambient
  paniculate matter for epidemiology and risk assessment: why
  and how. Journal of the Air and Waste Management
  Association 50:1167-1183.

Wilson, W. and Suh, H. (1997). Fine and coarse particles:
  concentration relationships relevant to epidemiological studies.
  Journal o] the Air and Waste Management Association
  47:1238-1249.

Yakovleva, E., Hopke, P. and Wallace, L. (1999).  Positive
  Matrix Factorization in Determining Sources of Particles
  Measured in EPA's  Particle TEAM Study, Environmental
  Science & Technology 33:3645-3652.
                                                          63

-------
                                                     Appendix
In this Appendix we document the seasonal  variation of the
infiltration factor Fmland the ambient exposure factor Fpl,x. Table
A-l  presents the seasonal average of Fml and  the  seasonal
harmonic average of the air exchange rate for each home.

Table A-2 compares the seasonal average indoor/outdoor ratio
for sulfur to the slopes obtained when  regressing indoor on
outdoor  sulfur.  The slope is usually smaller than the ratio,
averaging 91% of its value, and also has a wider  range, with a
standard deviation almost twice that of the ratio (0.26 compared
to 0.14). This is to be expected, since measurement error causes
lower slopes in regressions, and the regressions are based on a
maximum of 7 values (with a few exceptions), and usually only
5, 6, or 7; this leads to more variability than is desirable.

Table A-3 compares the personal/outdoor ratios (/•",,„) averaged
over one season with the slopes determined from regressions of
personal on outdoor sulfur. As with the /•",„/ values in Table A-2,
the slopes are lower than the ratios and once again the standard
deviations are nearly twice as high (0.19  compared to 0.11).

The results of regressing  the outdoor contribution to personal
exposures (by season) on  the outdoor monitors arc provided in
Figures A-l to  A-3, which provide boxplots of the adjusted R"
values.  Note that the definition of the adjusted  R~ parameter
allows negative values. The range of these seasonally calculated
R" values is quite a bit larger than the range of the year-round R:
and suggests that the increased variability due to the smaller
number of observations in each regression outweighs whatever
advantage was  gained in looking at the smaller time period.
    1.2


    1.0


    0.8


    0.6


 H 04
 (N
 ±
    0.2


    00


    -02


    -04
             1 Summer      3 Winter
                     2 Fall        4 Spring
                       Season
                   D2.
 c  Median
I  I 25%-75%
~T 5%-95%
 c  Outliers
Figure A-1.  Adjusted R" values from regressing the outdoor contribution
to personal exposure on outdoor PM25 measurements just outside the
house.
    1.2

    1.0
    0.6
                                                                   0.0
                                                                           1 Summer      3 Winter
                                                                                   2 Fall        4 Spring
                                                                                      Season
                                               c  Median
                                              I  I 25%-75%
                                              ~T 5%-95%
                                               c  Outliers
                                                               Figure A-2.  Adjusted R2 values from regressing the outdoor contribution
                                                               to  personal  exposure  on  outdoor  PM25  Harvard  Impactor  (HI)
                                                               measurements at the central site.
                                                            64

-------
    1.2


    1.0


    0.8


    0.6

 tsi
 5 0.4


    0.2


    0.0


    -0.2


    -0.4
            1 Summer      3 Winter
                     2 Fall        4 Spring

                       Season
T~ 5%-95%
 <-  Outliers
 *  Extremes
Figure A-3. Adjusted R2 values from regressing the outdoor contribution
to personal exposure on outdoor PM2 5 Federal Reference Method (FRM)
measurements at the central site.
                                                                    65

-------
Table A-1.  Values of the Average Sulfur Indoor/Outdoor Ratio (Fml) and the Air Exchange Rates by House and by Season
House
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
6
6
6
6
7
7
7
7
8
9
9
9
9
10
10
10
10
11
11
12
12
12
12
13
14
Season
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Winter
Summer
Fall
Winter
Spring
Summer
Summer
N
8
6
2
6
6
4
7
6
8
6
7
7
3
7
4
7
7
6
7
7
5
7
7
7
7
5
6
6
6
5
5
5
6
7
5
7
6
2
7
7
7
7
7
7
S In/Outa
0.58
0.64
0.76
0.65
0.59
0.53
0.64
0.66
0.78
0.70
0.69
0.76
0.32
0.48
0.51
0.45
0.51
0.65
0.74
0.60
0.65
0.58
0.61
0.67
0.83
0.75
0.86
0.36
0.48
0.55
0.62
0.69
0.46
0.57
0.56
0.64
0.69
0.67
0.40
0.57
0.62
0.54
0.37
0.41
SE
0.03
0.03
0.06
0.03
0.03
0.04
0.03
0.03
0.03
0.03
0.03
0.03
0.05
0.03
0.04
0.03
0.03
0.03
0.03
0.03
0.04
0.03
0.03
0.03
0.03
0.04
0.03
0.03
0.03
0.04
0.04
0.04
0.03
0.03
0.04
0.03
0.03
0.06
0.03
0.03
0.03
0.03
0.03
0.03
Airexb
0.63
0.79
1.29
0.82
0.46
0.81
1.35
0.71
1.05
0.50
0.87
1.07
0.21
0.26
0.34
0.22
0.43
0.36
0.60
0.45
0.57
0.60
0.58
0.56
1.06
1.50
1.94
0.24
0.23
0.40
0.84
0.66
0.30
0.54
0.67
0.66
0.39
4.49
0.27
0.45
0.64
0.33
0.22
0.23
SE
0.10
0.11
0.19
0.11
0.11
0.14
0.10
0.11
0.10
0.11
0.10
0.10
0.16
0.10
0.14
0.10
0.10
0.11
0.10
0.10
0.12
0.10
0.10
0.10
0.10
0.12
0.11
0.11
0.11
0.12
0.12
0.12
0.11
0.10
0.12
0.10
0.11
0.19
0.10
0.10
0.10
0.10
0.10
0.10
                                                              66

-------
Table A-1.  Continued
House
14
14
14
15
15
15
15
16
16
16
16
17
17
17
17
18
19
19
19
19
20
20
20
20
21
21
21
21
22
22
23
23
24
24
24
24
25
25
26
26
26
26
27
27
- 	
Season
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Summer
Fall
Summer
Fall
Winter
Spring
Summer
Fall
Summer
Fall
Winter
Spring
Summer
Fall
N
6
7
7
7
7
3
7
7
7
7
7
7
5
1
6
6
6
7
7
7
7
6
7
7
7
6
6
7
6
7
7
6
7
7
7
6
2
6
6
6
5
5
6
7
S In/Out*
0.46
0.45
0.47
0.61
0.73
0.76
0.77
0.59
0.71
0.68
0.61
0.42
0.53
0.43
0.59
0.52
0.62
0.79
0.84
0.73
0.56
0.43
0.47
0.51
0.43
0.68
0.67
0.67
0.44
0.65
0.43
0.55
0.41
0.55
0.60
0.64
0.46
0.81
0.61
0.78
0.70
0.73
0.54
0.77
SE
0.03
0.03
0.03
0.03
0.03
0.05
0.03
0.03
0.03
0.03
0.03
0.03
0.04
0.08
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.06
0.03
0.03
0.03
0.04
0.04
0.03
0.03
Airex"
0.33
0.51
0.35
0.42
0.49
0.72
0.56
0.41
0.65
1.28
0.95
0.11
0.14
0.51
0.27
0.30
0.61
1.33
1.54
0.93
0.24
0.23
0.52
0.35
0.28
0.53
0.63
0.58
0.27
0.48
0.57
1.02
0.33
0.35
0.51
0.69
0.27
0.67
0.50
0.79
1.15
0.79
0.61
1.25
SE
b~7i
0.10
0.10
0.10
0.10
0.16
0.10
0.10
0.10
0.10
0.10
0.10
0.12
0.27
0.11
0.11
0.11
0.10
0.10
0.10
0.10
0.11
0.10
0.10
0.10
0.11
0.11
0.10
0.11
0.10
0.10
0.11
0.10
0.10
0.10
0.11
0.19
0.11
0.11
0.11
0.12
0.12
0.11
0.10
                                                               67

-------
Table A-1.  Continued
House
27
27
28
28
28
28
29
29
29
29
31
31
32
32
32
32
33
33
33
33
34
34
36
37
37
37
38
38
38


Season
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Summer
Fall
Winter
Spring
Fall
Winter
Spring
Mean
SD
N
7
7
5
7
6
6
13
6
5
7
7
7
7
7
7
7
5
5
4
6
7
3
5
5
7
7
3
5
7
720

S In/Out3
0.66
0.68
0.41
0.65
0.61
0.62
0.40
0.59
0.75
0.70
0.97
0.80
0.25
0.40
0.48
0.31
0.23
0.35
0.49
0.40
0.54
0.74
0.25
0.58
0.66
0.55
0.52
0.64
0.50
0.59
0.14
SE
0.03
0.03
0.04
0.03
0.03
0.03
0.02
0.03
0.04
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.04
0.04
0.04
0.03
0.03
0.05
0.04
0.04
0.03
0.03
0.05
0.04
0.03


Airex"
2.61
0.67
0.28
0.47
0.86
0.37
0.31
0.42
0.61
0.53
2.77
1.11
0.23
0.27
0.33
0.25
0.15
0.25
1.12
0.64
0.36
0.45
0.25
0.27
0.69
0.30
0.18
0.54
0.26
0.64
0.56
SE
0.10
0.10
0.12
0.10
0.11
0.11
0.08
0.11
0.12
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.12
0.12
0.14
0.11
0.10
0.16
0.12
0.12
0.10
0.10
0.16
0.12
0.10

WWHXKiBnHUWeUUAKHaK
 a Average ratio of indoor/outdoor sulfur concentrations
 b Harmonic average of air exchange rate (h"1)
                                                                  68

-------
Table A-2.  Comparison of Seasonal Average
Sulfur indoor/Outdoor Ratios (Sin/Sout) with Slopes of Regressions of Sin on Sout
House
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
6
6
6
6
7
7
7
7
8
9
9
9
9
10
10
10
10
11
11
12
12
12
12
13
14
Season
S
F
W
V
S
F
W
V
S
F
W
V
S
F
W
V
S
F
W
V
S
F
W
V
S
F
W
V
S
S
F
W
V
S
F
W
V
S
W
S
F
W
V
S
S
N
8
6
7
6
6
7
7
6
8
7
7
7
3
7
6
7
7
6
7
7
7
7
7
7
7
7
7
7
6
6
6
7
5
6
7
7
7
6
2
7
7
7
7
7
7
Sin/Sout
(Fir,f)
0.58
0.64
0.66
0.65
0.59
0.51
0.64
0.66
0.78
0.71
0.69
0.76
0.32
0.48
0.50
0.45
0.51
0.65
0.71
0.74
0.60
0.63
0.58
0.61
0.67
0.83
0.76
0.84
0.36
0.48
0.57
0.61
0.69
0.46
0.57
0.60
0.64
0.69
0.67
0.40
0.57
0.62
0.54
0.37
0.41
SE
0.01
0.03
0.06
0.03
0.04
0.02
0.02
0.01
0.03
0.02
0.04
0.02
0.06
0.02
0.02
0.06
0.01
0.04
0.05
0.08
0.03
0.02
0.01
0.01
0.02
0.02
0.02
0.03
0.02
0.02
0.05
0.01
0.10
0.01
0.03
0.04
0.06
0.03
0.01
0.02
0.02
0.03
0.06
0.03
0.04
Intercept
99
96
-64
106
552
18
39
22
537
60
190
365
880
104
45
166
198
-250
-68
-148
164
-100
-63
52
37
-288
-3
-257
291
-14
284
23
263
76
45
192
1002
382

253
69
23
132
60
155
SE
111
101
210
135
222
153
113
77
193
173
88
119
108
157
116
87
105
222
185
403
270
187
47
66
238
54
53
119
247
227
223
26
477
164
94
119
266
368

184
90
57
100
78
113
P (Int)
0.41
0.40
0.77
0.48
0.07
0.91
0.75
0.79
0.03
0.74
0.08
0.03
0.08
0.54
0.72
0.12
0.12
0.32
0.73
0.73
0.57
0.62
0.24
0.47
0.88
0.00
0.96
0.08
0.31
0.95
0.27
0.40
0.62
0.67
0.66
0.17
0.01
0.36

0.23
0.47
0.70
0.24
0.48
0.23
Slope
0.52
0.57
0.72
0.57
0.30
0.51
0.60
0.64
0.48
0.68
0.54
0.62
0.00
0.42
0.47
0.31
0.44
0.83
0.77
0.81
0.54
0.67
0.63
0.57
0.66
0.97
0.76
1.07
0.27
0.49
0.34
0.58
0.53
0.43
0.54
0.37
0.22
0.55

0.31
0.51
0.60
0.41
0.32
0.28
SE
0.06
0.06
0.15
0.09
0.10
0.05
0.09
0.06
0.10
0.07
0.06
0.04
0.03
0.07
0.07
0.04
0.04
0.15
0.13
0.17
0.10
0.06
0.03
0.04
0.07
0.02
0.05
0.10
0.06
0.07
0.17
0.03
0.16
0.06
0.06
0.13
0.09
0.12

0.06
0.05
0.04
0.04
0.05
0.07
p (Slope)
0.0001
0.0008
0.0045
0.0027
0.0447
0.0001
0.0010
0.0005
0.0027
0.0002
0.0002
0.0001
0.9711
0.0014
0.0025
0.0004
0.0001
0.0055
0.0018
0.0050
0.0030
0.0001
0.0001
0.0001
0.0002
0.0001
0.0001
0.0001
0.0141
0.0017
0.1111
0.0001
0.0484
0.0014
0.0003
0.0389
0.0518
0.0098

0.0033
0.0002
0.0001
0.0002
0.0010
0.0111
(adj.)
0.92
0.94
0.79
0.89
0.60
0.95
0.88
0.95
0.77
0.94
0.94
0.97
-1.00
0.87
0.90
0.92
0.96
0.85
0.85
0.78
0.82
0.95
0.99
0.97
0.95
1.00
0.97
0.95
0.77
0.92
0.39
0.99
0.70
0.92
0.93
0.53
0.48
0.80

0.82
0.94
0.97
0.94
0.88
0.71
Ratio3
0.90
0.88
1.09
0.87
0.51
0.99
0.94
0.97
0.62
0.96
0.78
0.81
0.00
0.88
0.94
0.68
0.85
1.29
1.09
1.10
0.89
1.07
1.08
0.93
0.98
1.17
1.00
1.28
0.75
1.01
0.59
0.95
0.77
0.94
0.94
0.61
0.35
0.80

0.78
0.89
0.96
0.76
0.87
0.69
                                                             69

-------
Table A-2. Continued
House
14
14
14
15
15
15
15
16
16
16
16
17
17
17
17
18
19
19
19
19
20
20
20
20
21
21
21
21
22
22
23
23
24
24
24
24
25
25
26
26
26
26
27
27
27
Season
F
W
V
S
F
W
V
S
F
W
V
S
F
W
V
S
S
F
W
V
S
F
W
V
S
F
W
V
S
F
S
F
S
F
W
V
S
F
S
F
W
V
S
F
W
N
6
7
7
7
7
6
7
7
7
7
7
7
6
7
6
6
6
7
7
7
7
6
7
7
7
7
6
7
6
7
7
6
7
7
7
6
2
6
6
6
7
5
7
7
7
Sin/Sout
(F.nf)
0.46
045
0.47
0.61
0.73
0.66
0.77
0.59
0.71
0.68
0.61
0.42
0.52
0.55
0.59
0.52
0.62
0.79
0.84
0.73
0.56
0.43
0.47
0.51
0.43
0.70
0.67
0.67
0.44
0.65
043
0.55
0.41
0.55
0.60
0.64
0.46
0.81
0.61
0.78
0.69
0.73
0.54
0.77
0.66
SE
0.04
0.04
0.02
0.04
002
0.05
0.03
0.02
0.03
0.01
0.02
0.03
0.05
0.06
0.03
0.03
0.03
0.01
0.05
0.02
0.04
0.02
0.02
0.01
0.02
0.03
0.03
0.02
0.03
0.03
0.02
0.02
0.02
0.03
0.04
0.04
0.01
0.02
0.01
0.02
0.02
0.04
0.01
0.03
0.01
Intercept
54
-71
316
85
-15
-212
-116
246
-49
34
-217
4
285
-31
102
226
-118
-19
135
-102
755
-25
-33
-152
-151
-73
173
-55
586
89
88
15
248
125
67
-129

-95
-66
-63
100
-680
-81
-2
-3
SE
111
139
164
116
60
145
147
126
106
15
77
340
198
222
87
173
510
31
115
504
197
108
59
192
167
312
66
91
177
128
168
63
174
127
66
218

173
212
349
87
1169
73
126
50
P (Int)
0.65
063
0.11
0.50
0.82
0.22
034
0.11
0.67
0.08
0.04
0.99
0.22
0.89
0.31
0.26
0.83
0.57
0.29
0.85
0.01
0.82
0.60
0.46
0.41
0.82
0.06
0.58
0.03
0.52
0.62
0.82
0.21
0.37
0.35
0.59

0.62
0.77
0.87
0.30
0.60
0.32
0.99
0.96
Slope
0.42
0.53
0.33
0.55
0.75
0.91
0.88
0.50
0.45
0.64
0.81
042
0.35
0.59
0.51
0.42
0.65
080
0.69
0.77
032
0.45
0.50
0.57
0.48
0.73
0.50
0.72
0.22
0.60
0.40
0.54
0.32
0.48
0.51
0.76

0.86
0.63
0.80
0.57
0.99
0.57
0.77
0.66
SE
0.07
0.12
0.07
0.07
0.04
0.15
0.09
0.04
0.07
0.01
0.06
0.12
0.07
0.19
0.05
0.06
0.14
0.02
0.10
0.18
0.05
0.06
0.05
0.07
0.05
0.11
0.05
0.08
0.06
0.06
004
0.04
0.06
0.05
0.05
0.19

0.08
0.05
0.11
0.10
0.43
0.02
0.08
0.04
p (Slope)
0.0040
0.0062
0.0050
0.0007
0.0001
0.0033
0.0002
0.0001
0.0001
0.0001
0.0001
0.0190
0.0090
0.0250
0.0006
0.0025
0.0086
0.0001
0.0009
0.0084
0.0016
0.0017
0.0001
0.0006
0.0001
0.0012
0.0006
0.0003
0.0247
0.0001
0.0003
0.0002
0.0035
0.0002
0.0002
00150

0.0003
0.0002
0.0016
0.0020
0.1038
0.0001
0.0002
0.0001
R2
(adj.)
0.87
0.77
078
0.90
0.98
0.88
093
0.96
0.96
1.00
0.96
0.64
0.81
0.60
095
0.90
0.82
1.00
0.89
0.74
0.86
0.92
0.95
0.91
0.95
0.88
0.95
093
0.69
0.95
0.93
0.97
0.81
0.94
0.94
0.76

096
0.97
092
0.85
0.52
0.99
0.94
0.98
Ratio3
0.91
1.17
0.69
0.89
1.02
1.38
1.15
0.83
0.63
0.94
1.31
1.00
0.67
1.07
0.88
0.80
1.06
1.02
0.82
1.05
0.58
1.04
1.07
1.12
1.11
1.04
0.75
1.08
0.49
0.92
0.93
0.97
0.77
0.86
0.86
1.20

1.06
1.03
1.03
083
1.34
1.05
1.00
1.00
70

-------
Table A-2.  Continued
House
27
28
28
28
28
29
29
29
29
31
31
32
32
32
32
33
33
33
33
34
34
36
37
37
37
38
38
38
Mean
SD
Season
V
S
F
W
V
S
F
W
V
S
F
S
F
W
V
S
F
W
V
S
F
S
F
W
V
F
W
V


N
7
5
7
7
6
13
7
7
7
7
7
7
7
7
7
5
6
5
7
7
5
6
6
7
7
3
7
7
775

Sin/Sout
(Fmf)
0.68
0.41
0.65
0.62
0.62
0.40
0.61
0.71
0.70
0.97
0.80
0.25
0.40
0.48
0.31
0.23
0.40
0.49
0.42
0.54
0.72
0.26
0.59
0.66
0.55
0.52
0.64
0.50
0.59
0.14
SE
0.04
0.02
0.01
0.02
0.05
0.02
0.05
0.06
0.03
0.02
0.03
0.02
0.02
0.03
0.02
0.04
0.07
0.05
0.04
0.03
0.03
0.02
0.04
0.03
0.03
0.03
0.03
0.01
0.03
0.02
Intercept
257
126
-20
173
980
89
182
259
135
59
-319
2
35
-6
335
95
454
-89
165
-41
-21
10
-152
-94
565
-23
136
96
89
241
SE
273
241
25
68
319
210
311
142
167
106
160
71
91
112
129
31
373
146
83
174
331
93
326
129
113
87
112
59
167
137
P (Int)
0.39
0.64
0.46
0.05
0.04
0.68
0.58
0.13
0.46
0.60
0.10
0.97
0.72
0.96
0.05
0.06
0.29
0.58
0.10
0.83
0.95
0.92
0.66
0.50
0.00
0.84
0.28
0.17
0.47
0.30
Slope
0.56
0.37
0.67
0.42
0.14
0.36
0.47
0.39
0.60
0.92
0.95
0.25
0.38
0.48
0.16
0.12
0.05
0.60
0.29
0.56
0.74
0.26
0.69
0.73
0.30
054
046
0.41
0.54
0.20
SE
0.11
0.07
0.01
0.07
0.15
0.06
0.22
0.15
0.10
0.07
0.06
0.05
0.05
0.07
0.05
0.02
0.27
0.17
0.03
0.06
0.11
0.03
0.21
0.08
0.04
0.07
013
005
0.08
0.06
P (Slope)
0.0040
0.0210
0.0001
0.0024
0.4100
0.0001
0.0840
0.0508
0.0019
0.0001
0.0001
0.0032
0.0005
0.0011
0.0306
0.0078
0.8700
0.0350
0.0003
0.0002
0.0060
0.0017
0.0290
0.0003
0.0011
0.0869
0.0172
0.0006
0.03
0.13
R2
(^j.)
0.80
0.88
1.00
0.84
-0.03
0.73
0.38
0.48
085
0.97
0.97
0.82
0.91
0.88
0.57
0.91
-0.24
0.76
0.93
0.94
0.92
0.92
0.67
0.93
0.88
0.96
0.65
0.90
0.83
0.25
Ratio3
0.82
0.89
1.03
0.67
0.22
0.90
0.77
0.55
0.86
0.95
1.19
0.98
0.94
1.01
052
0.55
0.12
1.23
0.69
1.03
1.02
0.99
1.17
1.10
0.54
1.05
0.73
0.81
0.91
0.26
"Ratio of slope to Sin/Sout.
                                                                 71

-------
Table A-3.  Comparison of Seasonal Average Sulfur Personal/Outdoor Ratios (Spers/Sout)  with Slopes of Regressions of Spers on Sout
Subject
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
6
6
6
6
7
7
7
7
8
9
9
9
9
10
10
10
10
11
11
12
12
12
12
13
Season
S
F
W
V
S
F
W
V
S
F
W
V
S
F
W
V
S
F
W
V
S
F
W
V
S
F
W
V
S
S
F
W
V
S
F
W
V
S
W
S
F
W
V
S
N
6
6
7
6
6
6
6
6
5
7
7
2
2
7
7
7
6
7
7
7
6
7
7
7
6
7
7
7
6
6
7
7
5
5
7
7
2
6
3
7
7
7
6
7
Spers/Sout
(Fpex)
050
058
059
065
047
048
047
059
0.72
064
060
0.59
041
0.48
050
047
059
0.65
066
082
044
055
0.50
054
054
066
056
070
036
047
058
057
070
044
046
051
062
076
062
040
051
056
044
043
SE
002
003
005
0.05
004
0.03
004
002
004
0.04
0.02
001
0.01
0.04
005
006
004
0.05
003
004
0.04
003
002
002
002
005
004
0.04
001
002
007
0,07
005
002
003
003
025
004
005
002
002
0.06
003
004
Slope
052
0.57
066
0.54
0.29
046
0.56
074
045
068
0.50


033
035
0.55
046
0.66
062
079
042
068
055
045
050
099
078
1 00
033
043
056
046
072
048
0.36
048

049
077
0.28
045
060
042
039
SE
009
005
0 14
0 11
012
009
016
009
0.44
0 12
0.03


009
013
0 12
0.10
023
0.07
006
0.13
008
0.04
005
0 11
006
0 12
0 11
004
0.06
030
020
0 14
005
0.05
008

012
013
004
006
009
005
0 10
P
00049
0.0003
0.0051
0.0078
00683
00073
0.0240
0.0012
03799
0.0026
0.0001


00150
0.0440
0.0051
00085
0.0336
00004
0.0001
00327
0.0004
00001
0.0003
00111
00001
00013
00003
00007
00016
0 1230
00660
00128
00031
00008
00023

00148
0 1075
00015
00009
00012
00008
00094
Intercept
-53
17
-76
141
345
55
-103
-173
439
-95
122


282
182
-131
307
0
61
12
21
-341
-76
109
131
-680
-180
-334
77
120
31
100
-54
-91
104
17

746
-118
337
72
-56
10
37
SE
184
80
197
169
254
269
216
110
736
304
53


218
206
282
279
315
105
152
369
254
63
79
429
150
119
137
138
198
400
192
393
161
76
76

370
127
135
105
129
121
159
P
079
084
0.71
045
0.25
0.85
0.66
0.19
059
077
0.07


0.25
042
0.66
0.33
1 00
059
094
0.96
0.24
0.28
023
0.77
001
019
006
0.61
0.58
0.94
0.62
090
061
023
083

011
0.52
0.05
0.52
068
094
082
R2 (adj)
0.86
0.96
0.78
082
051
0.83
070
0.93
0.01
0.83
0.97


0.67
0.51
0.78
0.82
055
0.92
0.96
065
092
0.97
0.93
079
098
087
093
094
0.92
0.29
0.43
0.87
095
0.90
084

076
0.94
087
089
088
094
0.72
Ratio3
1.06
0.98
1.11
083
062
0.96
1 18
1.26
0.63
1 06
084


0.69
070
1.17
078
1 00
0.93
0.97
096
1.24
1 12
0.85
093
1 51
1 38
1 44
093
092
096
080
1 03
1 08
079
0,95

065
1 23
069
087
1 07
095
0.92
                                                              72

-------
Table A-3.  Continued
«• 	 «•*
Subject
	 	
14
14
14
14
15
15
15
15
16
16
16
16
17
17
17
17
18
19
19
19
19
20
20
20
20
21
21
21
21
22
22
23
23
24
24
24
24
25
25
26
26
26
26
27
Season
S
F
W
V
S
F
W
V
S
F
W
V
S
F
W
V
S
S
F
W
V
S
F
W
V
S
F
W
V
S
F
S
F
S
F
W
V
S
F
S
F
W
V
S
N

7
7
6
7
6
7
7
7
6
7
7
7
6
6
7
6
6
6
7
7
2
7
6
7
2
7
7
7
7
6
7
7
7
7
6
6
4
7
7
6
6
7
2
7
Spers/Sout
(F >
	 \rr>ext
0.40
0.43
0.43
0.48
0.59
0.61
0.61
0.64
0.57
0.66
0.57
0.56
0.44
0.43
0.46
0.57
0.51
0.59
0.72
0.72
0.66
0.55
0.45
0.39
0.52
0.55
0.74
0.75
0.73
0.48
0.60
0.47
0.50
0.53
0.53
0.63
0.64
0.52
0.70
0.52
0.70
0.56
0.74
0.46
SE

0.02
0.02
0.04
0.03
0.03
0.02
0.05
0.03
0.01
0.03
0.03
0.03
0.03
0.06
0.06
0.03
0.02
0.03
0.01
0.03
0.06
0.03
0.03
0.02
0.03
0.03
0.01
0.05
0.05
0.04
0.03
0.03
0.02
0.02
0.04
0.10
0.03
0.02
0.02
0.02
0.01
0.02
0.02
0.01
Slope

0.35
0.51
0.65
0.23
0.69
0.67
0.86
0.81
0.53
0.66
0.46
0.90
0.51
0.27
0.47
0.55
0.47
0.61
0.68
0.66

0.42
0.39
0.50

0.54
0.71
0.52
1.21
0.22
0.68
0.49
0.46
0.64
0.48
0.49
0.67
0.49
0.86
0.47
0.65
0.42

0.43
SE
	
0.05
0.05
0.14
0.05
0.08
0.03
0.14
0.05
0.06
0.06
0.04
0.05
0.09
0.08
0.19
0.08
0.07
0.11
0.02
0.07

0.07
0.09
0.04

0.07
0.03
0.08
0.15
0.11
0.05
0.13
0.04
0.04
0.04
0.16
0.16
0.07
0.06
0.05
0.05
0.09

0.03
P
0.0011
0.0001
0.0097
0.0084
0.0011
0.0001
0.0019
0.0001
0.0008
0.0001
0.0001
0.0001
0.0053
0.0293
0.0574
0.0023
0.0028
0.0055
0.0001
0.0002

0.0018
0.0103
0.0001

0.0007
0.0001
0.0015
0.0004
0.1198
0.0001
0.0120
0.0001
0.0001
0.0002
0.0377
0.0547
0.0012
0.0001
0.0006
0.0002
0.0051

0.0001
Intercept
_.

-99
-247
526
-129
-56
-219
-224
107
-16
107
-389
-176
250
6
34
76
-49
45
49

412
69
-102

33
81
197
-505
660
-92
-66
39
-275
71
125
-32
68
-317
187
136
120

90
SE
83
68
178
132
137
45
146
82
177
100
47
59
271
216
228
134
201
417
41
82

265
156
44

265
97
98
167
323
111
473
62
114
99
215
193
278
141
199
157
79

105
P
0.49
0.21
0.24
0.01
0.40
0.27
0.19
0.04
0.58
0.88
0.07
0.00
0.55
0.31
0.98
0.81
0.73
0.91
0.32
0.57

0.18
0.68
0.07

0.91
0.44
0.10
0.03
0.11
0.45
0.89
0.56
0.06
0.52
0.59
0.88
0.82
0.07
0.40
0.43
0.19

0.43
R (adj)
__

0.96
0.80
0.74
0.93
0.99
0.85
0.97
0.94
0.95
0.96
0.98
0.85
0.67
0.46
0.90
0.89
0.85
0.99
0.94

0.85
0.80
0.97

0.90
0.99
0.87
0.92
0.37
0.97
0.70
0.95
0.98
0.97
0.63
0.84
0.88
0.97
0.95
0.97
0.78

0.97
Ratio"
0.87
1.21
1.51
0.49
1 16
1.09
1.41
1.26
0.93
1.01
0.80
1.62
1.17
0.61
1.01
0.95
0.93
1.03
0.95
0.92

0.77
0.88
1.26

0.98
0.95
0.70
1.66
0.47
1.12
1.04
0.92
1.20
0.90
0.79
1.05
0.95
1.23
0.89
0.93
0.75

0.93
                                                                73

-------
Table A-3.  Continued
Subject
27
27
27
28
28
28
28
29
29
29
29
31
31
32
32
32
32
33
33
33
33
34
34
35
36
37
37
37
38
38
38
Mean
SD
Season
F
W
V
s
F
W
V
s
F
W
V
s
F
S
F
W
V
S
F
W
V
s
F
S
S
F
W
V
F
W
V


N
7
7
7
5
7
7
6
7
6
7
7
7
7
7
7
7
7
4
6
5
7
7
6
7
6
7
7
7
7
7
7
750
_™,.,,,,,,,:,
Spers/Sout
(Fp.«)
0.62
0.58
0.54
0.39
0.50
0.50
0.48
0.43
0.61
0.59
0.62
0.91
0.66
0.47
0.53
0.60
0.38
0.30
0.40
0.48
0.32
0.52
0.74
0.41
0.45
0.55
0.54
0.53
0.54
0.57
0.52
0.55
0.11
SE
0.01
0.02
0.03
0.02
0.02
0.02
0.04
0.03
0.06
0.05
0.06
0.02
0.03
0.03
0.04
0.05
0.04
0.08
0.07
0.04
0.03
0.02
0.04
0.04
0.03
0.09
0.02
0.03
0.04
0.04
0.04
0.04
0.03
Slope
0.60
0.54
0.50
036
0.54
0.33
0.10
0.49
0.97
0.28
0.72
0.94
0.76
0.56
0.47
0.38
0.15
0.16
0.15
0.51
0.26
0.54
0.72
0.28
0.51
0.26
0.60
0.32
0.57
0.47
0.36
0.53
0.19
SE
0.04
0.06
0.10
0.06
0.06
0.08
0.13
0.12
0.37
0.11
0.20
0.06
0.07
0.06
0.09
0.09
0.19
0.06
0.30
0.12
0.01
0.04
0.14
0.07
0.09
0.33
0.10
0.07
0.06
0.19
0.18
0.10
0.07
P
0.0001
0.0004
0.0034
0.0089
0.0002
0.0085
04929
0.0090
0.0571
0.0522
0.0152
0.0001
0.0001
0.0003
0.0033
00099
0.4704
0.1091
0.6425
0.0233
0.0001
0.0001
0.0064
0.0122
0.0057
0.4687
0.0016
0.0058
0.0002
0.0566
0.0934
0.03
0.10
Intercept
20
49
98
109
-61
144
785
-156
-503
252
-150
-49
-207
-110
74
267
516
117
318
-35
81
-38
85
313
-133
381
-83
473
-39
79
157
41
231
SE
65
75
235
211
108
72
282
428
541
100
325
87
183
98
170
145
466
86
416
105
37
130
417
219
256
484
151
174
83
160
193
190
125
P
0.77
0.54
0.69
0.64
0.60
0.10
0.05
0.73
0.40
0.05
0.66
0.60
0.31
0.31
0.68
0.13
0.32
0.31
0.49
0.76
0.08
0.78
0.85
0.21
0.63
0.47
0.61
0.04
0.66
0.64
0.45
0.48
0.29
R2(adj)
0.97
0.92
0.81
0.90
0.94
0.73
-0.09
0.73
0.55
0.47
0.67
0.98
0.95
0.93
0.82
0.72
-0.07
0.69
-0.18
0.81
0.98
0.96
0.84
069
0.85
-0.07
0.86
0.77
0.94
0.46
0.35
0.79
0.24
Ratio3
0.97
0.92
0.92
0.92
1.10
0.66
0.20
1.12
1.58
0.47
1.17
1.04
1.15
1.20
0.88
063
039
0.53
0.38
1.08
0.80
1.03
0.97
0.67
1.13
0.46
1.12
0.60
1.06
0.82
0.70
0.96
0.22
' Ratio of slope to Spers/Sout.
                                                               74

-------
Table A-4.  Questionnaire Variables and Definitions
  #  Variable Name   Definition
                                                                                     Units
                                                                                     ug/m
                                                                                     ng/m3
                                                                                     ng/m3
                                                                                     ng/m3
 1   Persdate        Julian date'100+Subject ID
 2   DatelD          Date (YYYYMMDD)*100 + SUBJECT id
 3   EPA ID         epa id
 4   Subject         SUBJECT ID
 5   Cohort          Cohort (Cardiovascular or hypertensive)
 6   House          House ID
 7   Season         Season
 8   Date            Julian date
 9   Month          Month
 10  Day            Day
 11   Year            Year
 12  ambHI25        Central-site Harvard Impactor PM25
 13  FRM25         Central-site Federal Reference Method PM2 5                         ug/m3
 14  Flaghi25        Validity code: 2 = valid; used when central-site HI25 used in regressions
 15  FlagFRM        Validity code. 2 = valid; used when FRM is a variable in regressions
 16  Flagairex        Validity code: 2 = valid; used when airex is a variable in regressions
 17  persS          personal sulfur
 18  Sin            Indoor sulfur
 19  Sout            Outdoor sulfur
 20  Sinout          Indoor/outdoor sulfur ratio
 21   FlagSinout      Validity code: 2 = valid; used for all indoor/outdoor regressions
 22  SpersSout      personal/outdoor sulfur ratio
 23  Flagspersout    Validity code: 2 = valid; used for all personal/outdoor regressions
 24  PM25in         Indoor PM25
 25  PM25out        Outdoor PM2 5
 26  PM25pers      Personal PM25
 27  Outcontin       Outdoor contribution to indoor PM25 concentrations
 28  Outcontpers     Outdoor contribution to personal PM25 concentrations
 29  Incontrib        Indoor-generated contribution to indoor PM25 concentrations
 30  Perscontrib      Non-outdoor contribution to  personal PM25 concentrations
 31   Perscloud       Remainder after accounting for time-weighted indoor-outdoor exposures ug/m3
 32  airex            air exchange rate                                                 h"1
 33  Cleaning        Time spent cleaning                                              minutes
 34  Grooming       Time spent grooming                                              minutes
 35  Otherjndoor    Time in other indoor locations                                      minutes
 36  Otherjoc       Time in other locations                                            minutes
 37  Travel          Time spent in travel                                               minutes
 38  Unknown        Time in Unknown location                                         minutes
 39  cooking         Time spent cooking                                               minutes
 40  outdoor         Time spent outdoors                                              minutes
 41   smoke          Time spent near smokers                                          minutes
_42   TempC	Outdoor Temperature (°C)         	°C	
                                                                                     ug/m3
                                                                                     ug/m3
                                                                                     ug/m3
                                                                                     ug/m3
                                                                                     ug/m3
                                                                                     ug/m3
                                                                                     ug/m3
                                                                75

-------
Table A-4.  Continued
  #  Variable Name  Definition
                                                                                  Units
  43 Tempdelta      Absolute Outdoor-Indoor Temp Difference using thermostat setting     °F
  44 Numpeopl      Number of persons living in house
  45 numsmok      Number of smokers in house
  46 cigsmokd       Number of cigarettes  smoked in house
  47 Mealsckd       Number of meals cooked
  48 Burning        Was food burned today?
  49 Exhstfan       Was exhaust fan used today?
  50 Candles       Were candles used today?
  51  CANDLDUR    Duration of candle use                                          minutes
  52 Incense        Was incense used today?
  53 INCENDUR     Duration of incense use                                         minutes
  54 Windopen      Were windows open today?
  55 windowall      Sum of products of open windows X width opened X duration open     inch-hours
  56 spacehtr       Was a space heater used today?
  57 Cleaning_1     Did cleaning occur today?
  58 Pets           Any dog or cat pets?
  59 broil           broiled food today
  60 fry            fried food today
  61  grill            grilled food today
  62 sautee         sauteed food today
  63 sweep         swept floors today
  64 vacuum        vacuumed today
  65 dust           dusted today
  66 TYPE          Type of building (1 = detached, 2 = duplex, 4 = apartment, 6 = trailer)
  67 AGE           age of building (years)                                          years
  68 BUSY_RD      high-traffic road nearby
  69 DIRT_RD      dir road nearby
  70 DUSTY_RD    Dust from  nearby construction etc
  71 GAR_USE      Park cars in attached  garage?
  72 A_C           air conditioning unit(s) in home
  73 FAN           whole-house or attic fan
  74 S_WIN         Storm windows (0 = none, 0.5 = partial, 1  = all)
  75 C_FUEL       cooking fuel (1 = gas, 2 = electricity)
  76 C_FAN        range hood?
  77 PILOT         Number of pilot lights (0-3)
  78 DRYER        clothes dryer?
  79 DRY_VENT    clothes dryer vented outside?
  80 VAC           vacuum bag type (0 = none, 1 = standard, 2 = HEPA)
  81 AREA          area of house (square footage)                                   feet2
  82 Rooms         Number of rooms
  83 FLRCOVav     Percent of floor covered by carpet                                %
  84 MILDEWavg    mildew noticed by technician
            ::^^                                                              	
                                                             76

-------