United States
               Environmental Protection
               Agency
               Air And Radiation
               (6604J)
EPA402-R-92-011
October 1992
&E PA
National Residential
Radon Survey

Summary Report
                                        Recyctod/Recyctabto
                                        Printed with Soy/Canola Ink on paper that
                                        contain* at toast 50% recycled Iber

-------
                                ACKNOWLEDGEMENTS
       The EPA Project Officer for the National Residential Radon Survey (NRRS) was Frank
Marcinowski of the Radon Division, Office of Radiation and Indoor Air.  The Agency wants to
acknowledge  the efforts of Research Triangle Institute in assisting  the  survey's development,
implementing the field work, compiling the data, and conducting the data analyses. In addition to
this, the Agency greatly benefitted from the comments of its Science Advisory Board on how to
design the survey and how to interpret the results.  In conjunction with this, the radiation staff in
EPA's Regional Offices and the staff of the Radon Division and the Radiation Standards  Division
of the Office of Radiation Programs also provided very useful reviews of the survey results. Their
comments were important to the preparation of the final reports on the NRRS. Steve Page, Director
of the Radon Division, and Kirk Maconaughey,  Chief of the Problem Assessment Branch, also
provided important support and direction throughout the study.  Finally, EPA would like to thank
the participants in the NRRS for their help over the time it took to implement this study.

-------
        SUMMARY REPORT OF THE NATIONAL RESIDENTIAL RADON SURVEY
       The U.S. Environmental Protection Agency (EPA) has completed its National Residential
Radon Survey (NRRS). The project is the first national survey to provide an estimate of radon levels
in the U.S. housing stock.  The results greatly increase EPA's ability to characterize the nature and
extent  of the nation's indoor radon problem. The findings come after over five years of efforts to
develop a survey approach, conduct field work, and carefully evaluate the results.

       This summary report  explains why EPA conducted the NRRS.  It describes what EPA
intended to learn and the survey approach.  The report also summarizes the major conclusions of the
study, outlines the implications of the results for the Radon Program, and discusses potential EPA
initiatives for the  future.  The results section is divided into the following subsections:  findings on
the distribution of radon concentrations in  the  U.S.  housing stock;  housing  and  household
characteristics that are central to EPA's analysis of the radon problem; and an initial assessment of
the importance of certain housing characteristics in estimating residential radon levels.
WHY EPA CONDUCTED THE NRRS

       In October 1986, Congress directed EPA to assess the radon levels in the nation's homes as
part of a larger research effort to address a growing concern over how significant the health threat
was from indoor radon. This mandate was provided in Title IV of the Superfund Amendments and
Reauthorization Act (SARA).

       In 1985, EPA began to recognize that elevated radon levels in homes in several portions of
the country represented a serious public health threat. Homes throughout the Reading Prong area
of New York, New Jersey, and Pennsylvania were registering radon levels that increased residents'
chances of contracting lung cancer to greater than 1 in 10. Similar problems had been identified prior
to 1985 in homes built on mining wastes from uranium and phosphorus in Colorado and Florida.  In
the mid-1980s, however, it had not yet become clear whether or not  this problem was widespread
throughout  the country.   The  best information available at  that  time  was a limited set  of
measurements of residential radon levels compiled and analyzed by scientists at the Lawrence Berkley
Laboratory (LBL) in California.1  Using these measurements, LBL estimated the annual average
radon level for single family homes in the United States and the frequency distribution of radon levels
in the  housing stock.

       Based on these results and other limited information, EPA estimated in 1986 that anywhere
from 5,000 to 20,000 people per year could be dying from lung cancer caused by residential radon
concentrations and that about 7 million homes had radon concentrations above levels at which EPA
would  recommend remedial action.   Subsequently, using radon testing data provided by  national
testing vendors in 1986-1987, EPA analysis indicated that many more homes than originally estimated
might have radon levels above the action level for radon mitigation.

       Given the gravity  of  this  situation,  EPA  recognized  the  importance  of  fulfilling  its
Congressional mandate in SARA to better define the radon problem. In 1987, EPA planned to
conduct a national survey to develop a more rigorous national estimate of the distribution of radon
concentrations in the U.S. housing stock. This effort was intended to focus particularly on assessing
   1 Anthony V. Nero, et al. 1986, "Distribution of Airborne Radon-222 Concentrations to U.S. Homes",
Science.

-------
the differences in radon levels between different regions of the country and types of housing.   EPA
also hoped to use this opportunity to conduct other crucial radon-related research that could provide
a foundation for formulating effective approaches to reducing the problem.
WHAT EPA WANTED TO LEARN

       Based on legislative direction and the program need, EPA established a set of primary and
secondary research objectives for the NRRS that guided the development of the survey. The Agency
also considered carefully the degree of precision necessary to answer critical questions about indoor
radon levels and set objectives for precision levels. The Agency then established precision constraints
(the amount of sampling error that would be acceptable for certain estimates of annual average radon
levels) that would meet  the survey objectives.  The sample size and other aspects of the survey
approach were developed around these precision constraints.  During the survey's planning, EPA
consulted with its Science Advisory Board on the survey's universe (coverage) and on how to measure
annual average residential radon levels, how to select a statistically valid sample of homes, and the
appropriate statistical objectives for the survey.
        The primary research objectives of the survey were to estimate:
               (1)     The  frequency  distribution  of annual  average radon concentrations in
                      occupied residences nationwide (to serve as the basis  for estimating the
                      national residential annual average radon level and the percentage of homes
                      that are above levels where homeowners should take remedial action)3; and

               (2)     The  frequency  distributions of annual average radon concentrations in
                      important subgroups, such as sets of states that  were the responsibility of
                      EPA's ten Regional Offices (to  identify segments of the country and housing
                      stock where radon is more likely to be a problem).4  This data provided the
                      basis for comparison of radon  concentrations between subgroups, such as
                      single family homes versus apartments.

        Secondary research objectives included compiling housing and demographic data that could
 be used for radon program development  and  additional data that could be used  to  assess the
 relationships between indoor radon concentrations and  housing construction and heating, ventilation,
    2 EPA also undertook several other important initiatives. An in-depth assessment of EPA's risk factor
for radon occurred under the auspices of the National Academy of Science and EPA's Science Advisory
Board. Additionally, EPA joined with states to conduct a series of state-wide radon testing surveys that
were designed to collect short-term radon measurements (e.g., screening tests) in homes that EPA could
use to identify areas that have potentially high radon levels. At the end of the summer of 1992, 42 EPA/
State residential radon surveys had been completed.  In addition, six states conducted their own surveys
outside of this program.

    3 The Agency was interested in both the percentage of homes that have radon concentrations above its
action level and the percentage of homes at very high levels where residents faced extremely high risks.
The "action level" is the radon concentration where EPA recommends homeowners act to reduce their
radon risks  through mitigation measures.

    4 EPA administers its program activities through 10 Regional Offices that cover activities in all SO
states and the District of Columbia. Exhibit B-6 in the appendix divides the country by areas covered by
EPA's Regional Offices.

-------
and air conditioning (HVAC) characteristics.  The Agency also wanted to examine how much time
people spend in their homes and where they spend it.  Finally, EPA saw the survey as an opportunity
to see how much radon testing and mitigation the public had done.

       EPA decided to set numerical precision constraints on two key estimates it wanted to obtain
from  the survey.  These constraints also created acceptable error bounds for many other critical
pieces of information as well. EPA set the following survey precision objectives:

       •       The national estimate of the percent of residences with radon concentrations over 10
               picoCuries per liter of air (pCi/L) should have a relative standard error of no more
               than 0.5, if the estimate is approximately, or above 0.5 percent. ' '

       •       The estimate of the percent of residences with radon concentrations over 4 pCi/L for
               an EPA Region should have a relative standard error of no more than  0.5, if the
               estimate is approximately, or above 7.0 percent.
 SURVEY APPROACH

        EPA's primary research objectives and numerical precision constraints guided the survey's
 design, field work, quality assurance efforts, and analysis. The planning of the survey and pretesting
 took place between the beginning of 1986 and the spring of 1989.  Field work  (data collection)
 occurred between the  summer of 1989 and the fall of 1990. Editing of the data, automating the data
 set, and the statistical  analysis occurred from the fall of 1990 through the spring of 1992.

        Coverage and  Sampling Approach

        The target population of the survey included both housing units (for radon measurements)
 and permanent residents of these units (for responding  to the questionnaire).  EPA targeted its
 survey toward the following population: residents whose homes were a principal residence, who had
 no plans to move within twelve months, and who had occupied their homes for at least nine months
 during the radon measurement period (partially to avoid coverage of vacation homes); and housing
 units not located on military installations or thinly populated areas of Alaska. The occupancy criteria
 were important, because the lifestyle of the residents in a home affects  the radon levels present.

        The survey covered all single family detached homes, multi-unit structures (e.g., townhouses,
 apartments, and condominiums)  and mobile homes.  It did not cover group quarters (e.g., college
 dormitories, prisons,  and  other  types  of institutional  housing).   Most  state surveys  are  less
 comprehensive, including only single family homes (both detached and attached, such as townhouses).
    5 A picoCurie per liter of air is a standard unit of measurement of the concentration of radioactivity in
 a gas. A picoCurie corresponds to 0.037 radioactive disintegrations per second in a liter of air.  EPA
 estimates that an individual exposed to 1 pCi/L of radon over her/his entire lifetime would have a 3 in
 1,000 chance of contracting lung cancer, (which is  nearly always fatal).

    6 The relative standard error is the fraction of an estimate that is one standard deviation from the
 estimated value.  There is a two to one probability that the mean value of the distribution will be within 1
 standard deviation above or below the estimated mean value.

    7 Residents spending their lives in a home that has an annual average radon level greater than 10
 pCi/L have greater than a one in twenty-five chance of contracting lung cancer due to radon.

                                             -3-

-------
       The United States was partitioned into 22 strata for sampling in order to identify differences
in radon levels across EPA's ten regions, and to ensure that ample coverage was provided for areas
that were expected to have differing levels of radon.  A multi-stage sampling process was used to
select the sample. First, 125 Primary Sampling Units (PSUs), areas the size of one or more counties,
were selected nationwide. Then, within each PSU, a  random sample of secondary sampling units
(SSUs), which were small land areas such as Census Blocks, were selected. Finally, housing units
within SSUs were randomly selected to be part of the sample. This method was designed to provide
a representative sample of  the  U.S. housing  stock that would enable EPA to meet its survey
objectives at a reasonable cost.  The basis for the stratification of the country by differing radon levels
was the best available information  on radon concentrations throughout the country in 1988 when
some areas  still had done limited  testing.  Some of the assignments were made on the basis of
geologic features. This has the potential for biasing the final results for each  region and should be
considered in evaluating the regional results.

       Data Collection

       ,EPA collected  data  for the survey through personal interviews with  respondents and by
placing radon measurement  devices in their homes for one year. The personal interviews covered
seventy-seven questions related to  characteristics of the  residents' homes, their movements within
their  homes, smoking habits, and steps they may have taken to  test for and mitigate radon levels.
EPA also assisted homeowners in placing alpha track detectors (ATDs) in  their homes.  These
devices were selected because they were the most effective radon  detectors for estimating the annual
average radon levels in homes. Every floor of a home  had at least one radon detector placed in an
appropriate area for the measurement of that floor's annual average radon level. Two detectors were
placed in appropriate  locations  in homes that had only one floor.  Throughout the year, EPA
monitored the  participation of the residents involved  in the study and replaced lost or damaged
detectors.

       Quality Assurance

       Throughout the survey, EPA took numerous steps to ensure that meaningful information  and
measurement data were collected  from the survey participants  and entered  properly into a data
system for statistical analysis.  The survey questionnaire and detector placement procedure  was
pretested with  60 respondents.  A  final questionnaire and detailed field procedures were designed
based on this experience.

       Standard training of interviewers for field work, data editing, and checking of questionnaire
responses were performed.  In addition,  EPA encouraged public participation to  ensure a high
response rate, monitored participation by home residents on a quarterly basis, and addressed field
problems with  detectors as they  arose.   EPA also  documented reasons for dropouts among
participants. Detectors and detector retrieval forms were carefully checked and accounted for. EPA
also ensured that the radon levels from the ATDs were accurately measured while the detectors were
being "read" at  the laboratory.

       The  Agency met the  survey's precision objectives. Through its  audit  of data quality, EPA
determined that:  (1) the measurements collected could be reliably used for estimating radon levels;
and (2) the response rate (after considering the reasons for nonresponse) was sufficient to meet the
                                           -4-

-------
precision objectives it had established.8 EPA needed complete radon measurement information for
approximately 5,000 homes in order to  meet the precision constraints established  in the survey
objectives.   To ensure that this number of reliable measurements would be obtained, the initial
sample of homes drawn for the survey included 11,423 houses. This population was drawn from an
eligible universe of homes (i.e., the target population,) of nearly 72 million households out of the 93
million households in the U.S. There were 7,118 respondents that initially participated in the survey
by completing the survey form and placing the ATDs in their homes for measurements. When the
radon  detectors were retrieved and analyzed, EPA  had exceeded its  goal,  obtaining  radon
measurement data from 5,694 homes and a total of over 15,000 radon detectors that had been placed
in various levels within them.

       The quality of questionnaire responses was also reviewed.  In some cases, questions arose as
to whether respondents were describing their homes properly.  EPA also discovered during its review
of survey responses that many residents were unable to provide accurate  responses to questions
regarding how members of their households spent time on different floors in their homes and how
much time they  spent at  home  in general.   EPA  has  cautiously used  the  data on  housing
characteristics in its analysis. In its estimate of residential radon exposure, EPA based its calculations
on the average radon concentration over all lived-in levels of the homes in the survey.  This estimate
of household radon levels was used in EPA's recent revision of its estimates of the residential risks
of lung cancer to the U.S. population.9  More information on EPA's quality assurance program is
provided in Appendix A.
 RESULTS OF THE SURVEY

        Three types of results were obtained from the NRRS.  First, EPA developed national and
 regional estimates of annual average radon concentrations according to different components of the
 housing stock.  Second, EPA assembled national statistics on housing characteristics, household
 behavior, and  the percentage of households that tested  for radon and reduced their  levels of
 exposure. Third, EPA identified housing/household characteristics that appear to most significantly
 affect radon levels in homes. The major results of EPA's analyses of the survey data are discussed
 below.

        National and Regional Radon Concentrations

        EPA used the measurement results from the NRRS to estimate annual average radon levels
 for homes throughout the United States and homes within  the ten EPA Regions, as well as for
 different components of the housing stock.  For homes covered by the survey, there were three
 approaches EPA considered to estimate annual  averages:   (1) lowest lived-in level; (2) average
 concentration over all lived-in levels; and (3) average concentration over all lived-in levels weighted
 for resident occupancy rates for different home levels.  An individual's risk of getting lung cancer
    Q
     The Agency also carefully considered the response levels throughout the country and weighted the
 data in a manner that was appropriate to draw conclusions for the entire country and the 10 EPA Regions.
 Details on EPA's approach can be found in U.S. EPA, 1992, National Residential Radon Survey Statistical
 Analysis Volume 1: National and Regional Estimates.

    9 The lived-in level is defined as an area of a home that residents spend time on a regular basis (as
 opposed to a storage area).  The average of all lived-in levels is the simple averaging of the results for
 each floor that is a lived-in space. The average concentration over all lived-in levels weighted for  resident
 occupancy rates is the average of all lived-in levels weighted by the fraction of time all residents
 collectively spend in each lived-in level in a home.

                                            -5-

-------
from residential radon exposure is directly proportional to his or her integrated exposure throughout
his/her home on all levels.  Therefore, the third approach represents the ideal way to assess annual
averages in a home to estimate human exposure to radon levels. However, as mentioned earlier,
EPA believes that many respondents were unable to provide answers to the survey questions that
could form the basis for meaningful results.  Therefore, EPA used the average concentration over all
lived-in housing levels to estimate the annual average radon level and  other important statistics
about radon levels in the housing stock.

       From the NRRS, EPA estimates that:

       •      Although most U.S.  homes have relatively low annual average radon levels, a
              significant percentage of homes have levels that are much higher than ambient
              radon concentrations.  The median annual ambient radon level is about 0.4 pCi/L.
              About 64 percent of all homes have annual average levels below 1 pCi/L.  Of the
              remaining homes,  20 percent have annual average levels  between 1 to 2  pCi/L and
              16 percent have radon levels greater than 2 pCi/L.

       •      A disproportionate share of radon exposure is in homes at higher levels.  About 56
              percent of the public's residential exposure to radon occurs in homes above 2 pCi/L.
              Exhibit 1 shows the distribution of homes and radon exposure at various pCi/L levels
              in the U.S. housing stock.  Exhibit 2 presents the risks individuals face at the average
              radon  level at  selected  intervals  within  the radon  distribution  and  EPA's
              recommendations  for dealing with these risks.

       •      The annual average radon concentration in the U.S. housing stock is 1.25 pCi/L.
              EPA believes this estimate may under or overstate the average by +. 9 percent (using
              a  95 percent confidence interval).12  The median value  of the distribution is 0.67
              pCi/L.

       •      About 6 percent of U.S. homes have annual average  radon levels greater than 4
              pCi/L.  This radon concentration is  the action level  at which EPA recommends
              homeowners act to reduce the amount  of radon in their  homes.  Therefore, 5.8
              million homes in the United States in 1990 had radon levels that homeowners should
    10 To estimate the residential risks of lung cancer to the U.S. population, EPA used a range of radon
concentrations around the point estimate of the national annual average radon level that was based on
averaging each home's radon measurement for all lived-in levels. This annual average range was 1.1 to 1.4
pCi/L.  The NRRS estimates of the annual average for the U.S. housing stock for the lowest lived-in level
and the average concentration over all lived-in levels weighted for resident occupancy rates for different
home levels fall .within that range.

    11 Hopper, R.  April 1991.  National Ambient Radon Study, Proceedings of the 1991 International
Symposium on Radon and Radon Reduction Technology. U.S. EPA, Office of Radiation Programs, Las
Vegas Facility.

    12 The most widely accepted estimate of the national average radon level for the housing stock, which
applies only to single family homes, had been 1.5 pCi/L (based on the LBL research mentioned at the
outset of this report).

                                            -6-

-------
                                    EXHIBIT 1
          DISTRIBUTION OF HOMES AND TOTAL EXPOSURE AT SELECTED
                          RADON LEVELS FOR ALL HOMES
    70
    60
    50
    40
    30
    20
     10
     0
0)
^
s.
          64%
            '11 22%
  22%
                    20%
                                  13%
         7%
                                             I/O
                                        4%
                                   Percentage of Homes
                                   Percentage of Predicted
                                   Exposure
                                   rotate will nol sum to 100% du» to rounding oiron.
                                20%
                                                  5%
                                                               11%
                                                            1%
                                                                     0.06
                                                    1%
            0-1
1-2
2-3        3-4        4-8
    PicoCuries Per Liter
8-20
>20
        Based on average over all living levels.

-------
                                    EXHIBIT 2

       RISKS TO RESIDENTS OF HOMES AT DIFFERING RADON LEVELS AND
                   WHAT EPA RECOMMENDS THEY DO ABOUT IT
Radon Level
0-1
1-2
2-3
3-4
4-8
8-20
> 20
If 1,000 people were exposed to this
level over a lifetime...*
About 1 person could get lung cancer
About 4 people could get lung cancer
About 8 people could get lung cancer
About 11 people could get lung
cancer
About 17 people could get lung
cancer
About 35 people could get lung
cancer
About 94 people could get lung
cancer
What EPA Recommends 1
Residents Do :
Difficult to reduce levels
Difficult to reduce levels
Consider fixing home
Consider fixing home
Fix home
Fix home
Fix home
* This is the risk to the general population (including smokers, former smokers and never
smokers). For 1,000 people that smoke, the column entries should be multiplied by 2.33. For
1,000 people that used to smoke, the column entries should be multiplied by 1.03. For 1,000
people that have never smoked, the column entries should be multiplied by 0.212.
                                       -8-

-------
             mitigate.13   This estimate of homes over 4 pCi/L may under or overstate  the
             percentage by +_ 22 percent.

       •      About 0.7 percent of U.S. homes have annual average radon levels greater than 10
             pCi/L.  This estimate of homes over 10 pCi/L may under or overstate the percentage
             by Hh 54 percent.

       •      Apartments   or condominiums  above  the  second  floor  seldom  had  radon
             concentrations  above EPA's  action level.  EPA does not encourage  residents of
             apartments or condominiums (e.g., multi-unit structures) above the second floor to
             test.  The agency recommends residents of single-family homes, multi-unit structures
             below the third floor,  and mobile homes with permanent foundations to test  for
             radon.  Exhibit  3 shows the differences in the average annual radon levels in homes
             that should test and those homes the Agency does not encourage to test. It also
             shows estimates of the percentage of homes at levels higher than EPA's  action level.

       •      Single-family detached homes are four times more likely to require mitigation than
             multi-family homes. The differences in radon levels between these types of housing
             units are shown in Exhibit 3.

       •      Every EPA Region has a significant number  of homes that need radon mitigation,
             although some  Regions have much greater percentages of homes over EPA's action
             level of 4 pCi/L.  Each EPA Region contains anywhere from tens of thousands to
             more than a million homes where residents should reduce their radon levels. In two
             Regions, the percentages of homes over the  action level were well over twice the
             national average of 6 percent.

       More details on these and other findings on radon levels in the housing stock are provided
       in Appendix B.

       In addition to these findings on the distribution and average levels of radon in  the housing
stock, EPA also examined the distributions of radon concentrations within the entire U.S. housing
stock and within major components of it to determine how well they matched common distributional
forms.14  From  this  data, the Agency determined that the shapes of the major radon distributions
   13 Based on 97.1 million single family, multi-family, and mobile homes that were regularly used (i.e.,
occupied year-round and not seasonally used) in the U.S. in 1990 (Census Bureau, 1991). Institutional
housing units (i.e., group quarters), which were not included in the NRRS, are excluded from this
estimate.  They may represent another 3 million housing units at colleges, military reservations and other
facilities using dormitory-style housing arrangements.

   14 The major components examined included single-family detached homes, the radon testing
population, multi-family units, and ground-contact units. For all the components, except ground-contact
units, see  Exhibit 3 for definitions. Ground-contact units include all single-family units, except those that
were 100 percent open underneath (such  as unskirted mobile homes), and residences on the first floor in
multi-family units.

                                            -9-

-------
                                                                        EXHIBIT 3

                        COMPARATIVE STATISTICS FOR RADON LEVELS IN DIFFERENT HOUSING GROUPS
9f*xwrics
Arithmetic Mean (pCi/L)
Median5 (pCi/L)
Geometric Standard
Deviation6 (pCi/L)
Percent > 4 pCi/L
Percent > 10 pCi/L
Percent > 20 pCi/L
A&, &&
HOUSING
fOWHUWHOW
1.25
0.67
3.11
6.0%
0.7%
0.1%
RABO&
TESTING
FOP01ATIOH1
1.30
0.71
3.04
6.3%
0.7%
0.1%
NON-TESTING
POHJL&TtON*
0.27
0.19
4.37
0.0%*
0.0%*
0.0%*
MULTI-FAMILY
0.65
0.36
3.23
1.7%
0.2%
0.0%*
mm&x&mx
DETACHED
M&MIS!4 ;
1.46
0.79
2.97
7.1%
0.8%
0.1%
    EPA recommends that residents of all single-family homes, multi-unit structures (apartments, duplexes, and condominiums) below the third floor, and mobile homes with
    permanent foundations should test them for radon.
2   EPA does not recommend residents of multi-unit structures above the second floor to test.
'   Multi-family homes in this survey are single-family attached homes (townhouses), apartments, duplexes, and condominiums.
4   Single-family detached homes only.
5   The median value should be used as the geometric mean in conjunction with the geometric standard deviation provided below as the best parameters for estimating the lognormal
    distribution of radon levels for each of the survey's populations shown in this exhibit.
6   The best estimate of the GSD using a quantile approach.

* This category contains zero observations.  Although homes of this type have been found with radon levels above the specified concentration, the survey results indicate that the
probability of finding many of these is extremely low.
Note: The standard errors for the arithmetic means and median values can be found in Appendix B.

-------
EPA examined from the NRRS are quite similar to the form of the lognormal distribution.15 This
is the frequency distribution that the LBL scientists believed provided the best fit for their data in
1986. However, the major national distributions resulting from the NRRS do not appear to be truly
lognormal. This suggests that there could be some limitations in estimating the frequency of radon
levels in both "tails" (ends) of these distributions when their parameters are simply derived from the
entire data set available from the NRRS.  EPA did find that using different techniques that adjust
the data and estimate the parameters of the distribution can provide a better "fit" of the lognormal
to the results  for different segments of the major NRRS distributions.16

        EPA also  recognizes  that conclusions can be made only for the  major  NRRS survey
populations it examined.  Other groups and subsets within the NRRS population that EPA did not
examine yet may have a frequency distribution that is lognormal.  In general, EPA expects that at
least a  strong  similarity will exist, according to the present theory regarding natural generation of
radon concentrations  in  homes.   (There is  a combination of many  factors that  lead to radon
concentrations in a home, which have a multiplicative influence on one another.)

        Household Characteristics

        The NRRS  collected  information on  different aspects of the housing stock and habits of
residents that  were relevant to EPA's development of a Radon Program.  Much of these data were
available  from other  sources such as  the Census Bureau,  the Public Health Service, and  the
Department of Energy.  The reasons  for collecting  this data for  the NRRS  was  that different
attributes of the housing stock  and household activities could be examined with respect to radon
levels in the country.  The next section discusses EPA's findings in the initial  relational analyses of
housing/household characteristics and radon levels.

        Respondents to the survey were also asked if they had tested their homes for radon and done
anything about it.  Although  other national EPA surveys have inquired about testing, the NRRS
survey was the first occasion in which this data was collected nationally using personal interviews.

        The highlights  of EPA's findings are provided in Exhibit 4. The exhibit also provides a brief
explanation of why EPA has an interest in each statistic. Further discussion of the relevance of these
findings is  provided below in  an explanation of how EPA is using the NRRS  results  in  the
implementation of its Radon Program.
    15 Data that are distributed lognormally, such as radon concentrations, can potentially take any
 positive values but cannot be zero or negative.  Moreover, most data (that is, radon concentrations for
 most houses) are clustered near the lower end of the distribution around the mean (e.g., 1.25 pCi/L for the
 national residential radon distribution), with a smaller number stretching asymmetrically into a long upper
 "tail" in the positive direction and a short lower "tail" toward zero.  In mathematical terms, a variable is
 said to be distributed lognormally if the natural logarithm of the variable is distributed normally--that is,
 with most data clustered around the mean and  the rest stretching indefinitely into symmetrical "tails" in
 both positive and negative directions.

    16 EPA found that using the median as the geometric mean and using a quantile method for deriving
 the geometric standard deviation  provided the parameters for a lognormal distribution that led to  the best
 fit of a lognormal distribution to  the empirical  data in the NRRS.

                                            -11 -

-------
                                            EXHIBIT 4

        MAJOR FINDINGS FROM THE NATIONAL RESIDENTIAL RADON SURVEY
          QUESTIONNAIRE ON HOUSING AND HOUSEHOLD CHARACTERISTICS
NRRS STATISTICS1
IMPORTANCE TO PROGRAM
HOUSING CHAKACraiUSTICS
71% of Homes Are Single Family Detached
Homes
24% of Homes Are Multi-Family Units
5% of Homes Are Mobile Units
44% of Homes Have Basements
56% of Homes Do Not Have Basements
HOVS1HOLD CHABACTJEHISTICS
51% of Basements Are "Lived-in Space"
49% of Basements Are Not Lived-in Space
2.8 Residents per Home
20% of Residents Were Smokers
80% of Residents Were Former Smokers or
Never Smoked
Important to designing radon testing
protocols, estimating mitigation costs, and
structuring research programs for mitigation
which are influenced by building type
Important for designing mitigation programs
and research which are influenced by
foundation type

Important for evaluating effects of testing
protocols based on placement of devices in
lowest lived-in space as opposed to the lowest
level in a house
Critical to determining risk reductions in
homes that are mitigated
Radon risks are much higher for smokers
than for former smokers and people who
have never smoked
SABOJ* TESTING AND ACTIONS TO REDUCE RADON LEVELS
3% of Homes Had Been Tested by Summer
1989
97% of Homes Had Not Been Tested by
Summer 1989
6 to 8% of Residents in Tested Homes Had
Acted to Reduce Radon Levels by Summer
19892
92 to 94% of Residents in Tested Homes Had
Not Acted to Reduce Radon Levels by Summer
1989
Indicator of radon program progress
Indicator of radon program progress
1 In all cases, except the percentage of residents that have mitigated their homes, the "don't know" responses were
allocated proportionally among the other responses.  For the percentage of residents that have mitigated their homes, the
lower end of the range is the actual percentage of all the respondents answering "yes" and the upper end of the range is
the percentage answering "yes" after a proportional allocation of the "don't know" responses has occurred between the
"yes" and "no" categories.

2 The percentage range given here is the percentage of the 3 percent of homes that had tested by the Summer of 1989.
The multiplying of the 6 to 8 percent of residents who have taken action to reduce radon levels by the 3 percent of
homes that had tested, provides the percentage of homeowners in the entire housing stock that have done something to
reduce their exposure to radon in their homes.
                                               -  12-

-------
       Housing Characteristics Associated with Radon

       EPA wanted to use information collected in the NRRS to determine the relationships
between selected housing characteristics and indoor  radon levels.  Two research objectives were
established:
                                                                                E
       •      Identify housing unit construction, heating, ventilation, and air conditioning (HVAC)
              characteristics associated with radon concentrations in U.S. housing units; and

       •      Ascertain their relative importance in explaining the variation in radon concentrations
              in the housing stock.

       About 50 questions on the questionnaire were aimed at collecting the data for this type of
analysis.  EPA reviewed these data before using them in its relational analyses and found  that some
respondents had problems accurately characterizing their homes.  However, EPA believed the data
could be carefully used to test various hypotheses on what aspects of housing construction and HVAC
systems are important  in determining the level of indoor radon in a home.  EPA examined the
relationship between 50 different home characteristics (independent or explanatory variables) and
indoor  radon  levels specified in different  ways (dependent  variables).  Indoor radon  level
specifications included  the average radon concentrations and the percentage of homes  that were
above 4 pCi/L and  10  pCi/L.  The location of homes in areas that were considered to have high,
medium, and low radon potential was used as an  independent variable in one set of analyses and
excluded from another. It was excluded to allow EPA to better determine what other factors might
be important to the radon levels that were found in a home.

       While EPA has not completed all of the relational analyses it intends to pursue,  the initial
analysis suggests that the most important predictor of radon levels is a home's geographic location.
It further suggests that construction  characteristics  tend to be more  important than  HVAC
characteristics. The independent variables found to have the most predictive value in the analyses
performed are, in rank order:

       •       radon potential of the area;
       •       whether a home was a single family unit;
       •       whether a home had a basement entrance inside the residence;
       •       number of gas appliances used in a residence;
       •       whether a home had a basement;
       •       whether living space was the lowest level;
       •       number of months a home was closed  for heating;
       •       foundation type;
       •       number of home levels;  and
       •       amount of duct work.

       EPA has not completed its examination of all the relationships between household/housing
characteristics and radon levels. Therefore, it would be premature to suggest that other factors are
not also important or to draw conclusions regarding the significance of various  factors.
 HOW EPA IS USING THE NRRS RESULTS

       EPA has used results from the NRRS in a variety of different ways. The survey results have
 allowed EPA to establish a statistically valid national benchmark of radon levels in the U.S. housing
 stock, and thus have improved EP A's risk estimates. The survey's results also have provided valuable

                                           - 13-

-------
corroborative evidence of several operating hypotheses that EPA has used to implement its program,
and have assisted EPA in developing different types of radon testing protocols.

       Establishing a Benchmark for Actions

       One of the most important aspects of the NRRS is that it has established a Radon Program
benchmark for the implementation of EPA's Radon Program. EPA has now refined its assessment
of the severity of the public's exposure to radon, the types of housing units and household habits that
are affected by and influence  the  radon problem, and measures that the public has been willing to
take to address the problem.  Based on the survey, EPA believes that there are 5.8 million homes
that have annual average radon levels above EPA's action level. Residents in approximately 200,000
homes had acted to reduce their radon exposures by the middle of 1989.

       Improving  Risk Estimates

       The most fundamental aspect of assessing  the radon problem has been estimating the public
health risks.  The  NRRS, in  conjunction with  other EPA efforts and the work of other external
scientific organizations, has recently enabled  EPA to revise its estimate of the risks from radon.17
The Agency has used  the NRRS's confidence intervals for its estimate of the annual average radon
concentration and other recent EPA research to improve its quantitative estimation of the range of
uncertainty in EPA's best estimate of the risk.  Given the levels of uncertainty that EPA can quantify
in many of the factors contributing to the risk of radon and based on the NRRS's estimate of the
annual average radon level of 1.25  pCi/L, EPA estimates that indoor radon in homes causes between
7,000 and 30,000 lung cancer deaths per year.18  (EPA's "best estimate" is  14,000 annual lung
cancer deaths are due to residential radon exposure.)

       NRRS data also have  provided EPA with a better estimate of the number of homes in the
United States that are well above its action level of 4 pCi/L. The previous LBL work suggested that
about 1 percent  of U.S. single family homes were above 10 pCi/L. The national survey provides a
fairly comparable estimate that about 0.7 percent of homes are above  10 pCi/L.19

       Providing Corroborative Evidence

       The NRRS has corroborated a number of operating hypotheses that have guided the Radon
Program.  Most importantly, large numbers of homes with annual average radon levels greater than
4 pCi/L exist in all the EPA Regions. This survey supports EPA's national effort to encourage all
homeowners to test for radon and reduce elevated levels.

       In addition, the relational analysis and regional radon estimates from the NRRS have shown
geographic differences in the potential for elevated radon levels.  This supports EPA's strategy to
facilitate the targeting of federal and state resources to areas with the highest radon potential.
    17 EPA has improved its estimate of the risk factor largely based on the contributions of two National
Academy of Science reports entitled Comparative Dosimetry of Radon in Mines and Homes (1991) and
Health Effects of Exposure to Low Levels of Ionizing Radiation:  BEIR IV (1990).

    18 Further information on the risks from residential exposure to radon can be found in Chapter 2 of
the Technical Support Document for the 1992 Citizen's Guide to Radon.  U.S. EPA, 1992.  This estimate is
based on a residential population estimate of 250 million people in 1990.

    19 Using a 95 percent confidence interval, the percentage of homes over 10  pCi/L may be 0.3 to 1.0
percent.

                                           - 14-

-------
       From the outset of the Radon Program, EPA has focused its mitigation research on fixing
single family detached homes.  There has been little  research covering apartments.   The survey
confirms that EPA's emphasis on developing mitigation methods for single family detached homes
was correct, because few homes with elevated radon levels were identified in multi-family homes.
EPA also found that its  use of statistics developed for other purposes by the Census Bureau, the
Public Health Service, and the Department of Energy about housing/household characteristics, such
as occupancy levels and building foundation types, were appropriate for the housing population of
interest.

       Influencing Testing Protocols

       EPA has used NRRS results to critically evaluate different testing protocols for its  1992
Citizen's Guide to Radon and for the upcoming Home Buyers and Sellers Guide to Radon for use
in real estate transactions. In this analysis, EPA considered the reliability of several testing methods,
which include different  types and  combinations of short- and long-term tests that  are used to
determine the necessity of  fixing a home.  This effort led EPA to review and reassess  the advice it
had been giving the public regarding radon testing.  In the past, EPA has urged the public to use
long-term tests in most situations.  The Agency continues  to recommend long-term testing, but has
developed appropriate short-term testing approaches that can be used if test results  are needed
quickly.  EPA also believes the availability of short-term testing approaches will facilitate increased
residential radon testing, because research has shown that more residents are willing to conduct short-
term tests.

       The NRRS data  provided the frequency distribution of the annual average radon levels for
homes for which the Agency recommends testing for radon.20 EPA used the frequency distribution
of radon levels as estimated by of the NRRS and other research on the accuracy of short-term radon
tests to analyze the ability of different  testing protocols to classify homes either above or below the
action level.
 FUTURE RESEARCH QUESTIONS

        EPA is considering several follow-up studies based on the data collected in the NRRS. EPA
 plans to further investigate the significance of certain housing/household characteristics in predicting
 radon levels and whether they also can be considered factors that directly influence existing radon
 levels.  In a related effort, EPA has joined with the U.S. Geological Survey and the Department of
 Energy in a project to develop an approach for  locating  areas of the country with the greatest
 percentage of homes with elevated radon levels.  The Agency is also assessing in more detail the
 uncertainties of short-term  tests  for making mitigation decisions.  As part of that effort, EPA is
 considering the value of future research that would:

        •       Investigate relationships between short and long-term measurements;

        •       Investigate within-year variations in the radon levels in homes;

        •       Evaluate the relationship among annual average radon concentrations and  certain
               weather factors, such as rainfall; and
    20 EPA recommends radon testing by residents of single family homes, multi-unit structures
 (apartments) below the third floor, and mobile homes with permanent foundations.

                                            -15-

-------
              Evaluate and improve, if necessary and possible, the quality of questionnaire data
              regarding selected aspects  of the survey such as occupancy data and actions to
              mitigate radon levels.
WHERE TO GET MORE INFORMATION

       More details on the findings of this summary report and other results from the NRRS are
available in reports EPA has prepared with the results of the analysis. They are:

       •      U.S. EPA, 1992, National Residential Radon Survey Statistical Analysis Volume 1:
              National and Regional Estimates.

       •      U.S. EPA, 1992, National Residential Radon Survey Statistical Analysis Volume 2:
              Summary of the Questionnaire Data.

       •      U.S. EPA, 1992, National Residential Radon Survey Statistical Analysis Volume 3:
              Housing Unit Characteristics Associated with Radon.

       •      U.S. EPA, 1992, National Residential Radon Survey:  Field Operations Report.

       •      U.S. EPA, 1992,  Quality Assurance Support for the National  Residential  Radon
              Survey.

              U.S. EPA, 1989, National Residential Survey Design Report.

       These six reports can be obtained by contacting EPA's Radon Division at (202) 233-9380 or
writing to:

              National Residential Radon Survey Results
              Radon Division (RD-6604J)
              Office of Radiation Programs
              U.S. Environmental Protection Agency
              401 M Street, S.W.
              Washington,  D.C.  20460
                                          - 16-

-------
                                       APPENDIX A

                                  QUALITY ASSURANCE
       EPA prepared a Quality Assurance Project Plan (QAPP) that established standard policies,
procedures, and organizational arrangements to achieve quality assurance goals that were established
for the survey.  The plan had procedures  for data collection, sample custody,  calibration, detector
analysis, and data processing.  It covered both the placement of the alpha track detectors (ATDs)
used to record the radon  levels and the laboratory's process  of analyzing the results from the
detectors.

       The plan also had quality assurance objectives for the survey and the radon measurements.
It specified goals for precision, bias, completeness, representativeness, and comparability. Goals also
were established to ensure the EPA precision objectives mentioned in the text for fractions of homes
> 4 and > 10 pCi/L were met. The key goals were:

       •      At least 5,000 homes would produce usable ATD readings (completeness goal).

       •      ATDs receiving monitored exposures  (spikes) would have  a coefficient of variation
              <, 0.20 (precision goal).

       •      ATDs receiving monitored exposures would have  reported exposures within  15
              percent of the monitored level (bias goal).

       •      Less than 50 percent of unexposed ATDs (blanks) would have  reported exposures
              > 30 pCi-d/L and none would have reported exposure > 60 pCi-d/L.

       •      During data entry, there would be a < 0.5 percent error rate for fields containing
              one- or two-digit/character codes and  < 5.0 percent error rate for fields containing
              longer text (e.g., names and addresses).

       Several audits were  specified in the QAPP to ensure compliance with established procedures
and data quality objectives. These included formal technical system audits, performance evaluation
audits, and an audit of overall data quality.  In addition, less formal activities were  conducted to
ensure data quality. They included careful selection of the type of detectors to be used, development
of detector placement guidelines, and monitoring operations at the laboratory processing the ATDs.

       Based on the audits that occurred during the NRRS, EPA determined that the survey was
administered satisfactorily given its data  quality goals. The NRRS produced usable radon measure-
ments in close to 5,700  homes, substantially exceeding the 5,000 home completeness  goal.  Exhibit
A-l shows the survey's response rates. The error rate for key questionnaire  data was 0.28 percent,
well within the 0.5 percent  objective. The performance of the spiked and duplicate ATDs satisfied
the precision and bias goals specified in  the QAPP. The reported values of the blanks were higher
than EPA had set for its quality assurance goals.  The Agency's evaluation of this result led to the
conclusion that the laboratory processing  the detectors may have underestimated  the background
correction factor for the NRRS detectors and that the effect on annual average concentrations in
most homes was small. Considering the Agency's overall performance in meeting the survey's quality
assurance requirements, EPA concluded that the radon measurements could  be used to achieve the
objectives of the survey.
                                            A-l

-------
       EPA identified certain problems during the survey which have led to careful use of the results
to avoid reaching inappropriate conclusions from data that are identified to have limitations.  For
example, EPA identified potential problems surrounding the fact that certain participants did not
respond to some types of questions accurately. This clearly occurred  in the responses received on
occupancy rates. This problem also arose for some of the questions  on housing characteristics.  For
example, respondents had difficulty in determining whether their homes were "tight" (well-insulated
and constructed to reduce  exfiltration). EPA further recognizes from its  auditing efforts that the
fraction of homes  with low concentrations (< 0.5 pCi/L) may be slightly depressed by inadequate
background correction at the laboratory.  EPA has analyzed the data and reported results with a
sensitivity to the limitations that exist in certain portions of these NRRS data.
                                         Exhibit A-l

                             Summary of Data Collection Results
Result Category
Total Sample Size
Eligibility Status Ascertained
Eligible Cases
Nonrespondents
Respondents
Eligible for Panel Maintenance
Became Ineligible During Monitoring
Period
Eligible at End of Monitoring Period
Final Nonrespondents
Final Respondents
Cumulative Detector Coverage Rate at End of
Study (0.960 x 0.843 x 88.7%)
•:: Number : ,': *•"
11,423
10,961
8,444
1,326
7,118
7,118
699
6,419
725
5,694

•••• ":---,i. Rates "-:.•,':. ?'

96.0%
84.3%

88.7%
71.8%
                                            A-2

-------
                                      APPENDIX B

     ADDITIONAL NRRS RESULTS ON NATIONAL AND REGIONAL RADON LEVELS
                             IN THE U.S. HOUSING STOCK
       The following tables and figures provide more details on the results given in the body of this
summary report and useful supplemental information.  The results that follow are for the estimates
of annual average radon levels for the average of all lived-in levels for year-round occupied housing
units in the United States. The radon measurements were taken for one year during 1989 to 1990.
The results  appear in EPA's National Residential Radon Survey Statistical Analysis Volume 1:
National and Regional Estimates. 1992.

       In reviewing the results showing the differences between EPA Regions, it must be recognized
that there is a potential for a bias that could lead to the understating of the radon levels in several
Regions due to the initial selection of sampling strata for the survey on the basis of limited (although
the best available) data on the variability in radon levels in different areas in each Region (see p.4
of main report).  Also, the regional percentage estimates mask areas within the states in each EPA
Region that may have high residential radon levels by only providing an estimate for an entire region.
Therefore, the regional results should be cautiously used in drawing conclusions about the nature of
the radon problem in various locations in the United States.  More specific regional and state
information on residential radon levels can be obtained from the appropriate EPA Regional Offices
and state radiation programs.
                                            B-l

-------
Percent
70


60


50


40


30


20


10


 0
                           EXHIBIT B-1
             National Distribution of Annual Average
             Radon Levels in the U.S. Housing Stock
                      Average Over All Living Levels
63.6
    0-1     1-2    2-3    3-4    4-5     5-6    6-8    8-10   10-20   >20
                         Radon Concentration (pCi/L)

-------
                                                  EXHIBIT B-2

                  ESTIMATED DISTRIBUTION OF ANNUAL AVERAGE RADON LEVELS BY
                                   EPA REGION OVER ALL LIVING LEVELS
Regi6n\pCi/L
Region 1
Percent
(Std. Err.)
No. of Obs.
Region II
Percent
(Std. Err.)
No. of Obs.
Region III
Percent
(Std. Err.)
No. of Obs.
Region IV
Percent
(Std. Err.)
No. of Obs.
Region V
Percent
(Std. Err.)
No. of Obs.
Region VI
Percent
(Std. Err.)
No. of Obs.
Region VII
Percent
(Std. Err.)
No. of Obs.
Region VIII
Percent
(Std. Err.)
No. of Obs.
Region IX
Percent
(Std. Err.)
No. of Obs.
Region X
Percent
(Std. Err.)
No. of Obs.
United States
Percent
(Std. Err.)
No. of Obs.
0-1
63.7
(6.45)
212
77.4
(6.14)
495
65.0
(5.51)
483
66.4
(3.78)
488
48.0
(2.53)
680
75.6
(2.33)
293
30.4
(9.72)
93
34.0
(13.06)
78
73.3
(4.92)
299
84.9
(6.55)
131
63.6
(1.59)
3,252
1-2
t
19.8
(2.30)
77
11.7
(2.55)
73
18.8
(3.13)
176
20.9
(2.51)
180
24.6
(1.48)
380
16.9
(2.20)
86
31.0
(3.60)
98
23.7
(3.12)
70
19.7
(2.80)
72
11.4
(6.61)
38
19.9
(0.88)
1,250
2-3
8.1
(2.72)
31
4.6
(1.35)
28
5.4
(1.14)
55
5.5
(0.57)
57
12.0
(0.79)
194
3.8
(0.76)
18
12.4
(3.78)
38
14.0
(3.99)
43
4.0
(1.14)
11
1.7
(1.61)
13
6.9
(0.42)
488
3-4
4.3
(1.75)
16
2.3
(1.36)
13
3.4
(0.77)
33
2.4
(0.47)
31
6.2
(0.67)
101
1.0
(0.36)
5
9.4
(2.16)
30
8.3
(2.90)
26
2.4
(2.18)
5
0.6
(0.59)
5
3.7
(0.39)
265
>4
4.1
(1.14)
19
3.9
(1.34)
23
7.4
(2.17)
63
4.8
(1.78)
45
9.2
(1.56)
149
2.7
(1.02)
17
16.8
(6.94)
55
20.1
(6.13)
62
0.7
(0.42)
3
1.4
(1.27)
3
6.0
(0.68)
439
Homes
> 4 pCi/L
212,000
382,000
750,000
867,000
1,644,000
303,000
814,000
604,000
92,000
50,000
5,836,000
Note:  EPA designed the survey to obtain a relative standard error (RSE) (Standard error * values) of no more than 0.5 for its Regional
estimates of the percentage of homes greater than 4 pCi/L, if the estimate is close to 7 percent of homes. For EPA Regions I through VIII
the RSEs are below 0.5. For EPA Regions IX and X the RSEs are .6 and .9, respectively.  This is due to the small number of observatons
found above 4 pCi/L  The relatively large standard errors for these two regions indicate there is greater uncertainty around the estimates
provided in this table relative to the other Regions.  The following exhibit (B-3) shows for the percentage of homes over 4 pCi/L where there
are and are not statistically significant differences between regions. Numbers do not total due to rounding error.

-------
                                        EXHIBIT B-3

                RESULTS OF AN ANALYSIS OF THE STATISTICALLY SIGNIFICANT
                    DIFFERENCES THAT EXIST BETWEEN EPA REGIONS FOR
                     THE PERCENTAGE OF HOMES GREATER THAN 4 pCI/L
EPA Region/
 Percentage
  >4 pCi/L
Key:  Shaded areas show regions that do not have statistically significant differences at the 90 percent
     confidence level. This table is meant to be used in conjunction with the review of Exhibit B-2.

-------
                       EXHIBIT B-4

  ESTIMATES OF SELECTED PARAMETERS OF THE DISTRIBUTION OF
ANNUAL AVERAGE RADON CONCENTRATIONS OF THE AVERAGE OVER
        ALL LIVING LEVELS IN THE U.S. BY RADON LEVELS
Rddon Level
: (pCi/L)
0-1
(Std. Err.)
No. of Obs.
1-2
(Std. Err.)
No. of Obs.
2-3
(Std. Err.)
No. of Obs.
3-4
(Std. Err.)
No. of Obs.
4-5
(Std. Err.)
No. of Obs.
5-6
(Std. Err.)
No. of Obs.
6-8
(Std. Err.)
No. of Obs.
8-10
(Std. Err.)
No. of Obs.
10-20
(Std. Err.)
No. of Obs.
>20
(Std. Err.)
No. of Obs.
Total
(Std. Err.)
No. of Obs.
- --- =^^=^^=^^
Arithmetic Mean
{pCi/L)
0.43
(0.01)
3,252
1.40
(0.01)
1,250
2.44
(0.02)
488
3.43
(0.02)
265
4.45
(0.02)
142
5.47
(0.04)
106
6.90
(0.06)
93
8.95
(0.08)
47
12.89
(0.33)
46
29.25
(4.80)
5
1.25
(0.06)
5,694
==========
Median
(pCi/L)
0.40
3,252
1.36
1,250
2.43
488
3.40
265
4.42
142
5.43
106
6.84
93
9.03
47
11.99
46
25.12
5
0.67
5,694

-------
                             EXHIBIT B-5

   ESTIMATES OF SELECTED PARAMETERS OF THE DISTRIBUTION OF
ANNUAL AVERAGE RADON CONCENTRATIONS OVER ALL LIVING LEVELS
                IN THE UNITED STATES BY EPA REGION
EPA Region
Region 1
Estimate
i (Std. Err.)
No. of Obs.
Region II
Estimate
(Std. Err.)
No. of Obs.
Region III
Estimate
(Std. Err.)
No. of Obs.
Region IV
Estimate
(Std. Err.)
No. of Obs.
Region V
Estimate
(Std. Err.)
No. of Obs.
Region VI
Estimate
(Std. Err.)
No. of Obs.
Arithmetic Mean

-------
   ESTIMATES OF SELECTED PARAMETERS OF THE DISTRIBUTION OF
ANNUAL AVERAGE RADON CONCENTRATIONS OVER ALL LIVING LEVELS
                IN THE UNITED STATES BY EPA REGION
                      (continuation of EXHIBIT B-5)
EPA Region
Region VII
Estimate
(Std. Err.)
No. of Obs.
Region VIII
Estimate
(Std. Err.)
No. of Obs.
Region IX
Estimate
(Std. Err.)
No. of Obs.
Region X
Estimate
(Std. Err.)
No. of Obs.
United States
Estimate
(Std. Err.)
No. of Obs.
Arithmetic Mean
(pCi/L)
2.36
(0.50)
314
2.59
(0.70)
279
0.80
(0.12)
390
0.57
(0.10)
190
1.25
(0.06)
5,694
Median1
(pCi/L)
1.56
314
1.64
279
0.59
390
0.36
190
0.67
5,694
Geometric Standard
Deviation2
2.44
(n.a)
314
2.91
(n.a)
279
2.71
(n.a)
390
2.54
(n.a)
190
3.11
(n.a)
5,694
 1 The median should be used as the geometric mean in conjunction with the geometric standard
 deviation (GSD) in this exhibit as the best parameters for estimating the lognormal distribution of
 radon levels for each area of the country in the exhibit.

 2 The best estimate of the GSD using a quantile approach.

-------
                              EXHIBIT B-6
                   Arithmetic Mean (pCi/L) and Median
           (in parenthesis) of the Distribution of Annual-Average
               Radon Concentrations Over All Living Levels
                             by EPA Region
        o
National Estimate: 1.25
                    (0.67)
                                                                    1.18
                                                                   > (0.69)
                                                                 1.37
                                                                 (0.68)

-------
                                        EXHIBIT B-7

                       Testing Population* Distribution of Annual Average
                Radon Concentrations  in Year-Round Occupied Housing Units
                                      Average Over All Living Levels
    Percent
          0-1      1-2      2-3     3-4     4-5     5-6     6-8    8-10    10-20   >20
                                    Radon Concentration (pCi/L)
1 EPA recommends that residents of all single-family homes, multi-units below the third floor, and mobile homes with permanent foundations should
 test them for radon.

-------
                       EXHIBIT B-8

 ESTIMATES OF SELECTED PARAMETERS FOR INTERVALS IN
THE NATIONAL DISTRIBUTION OF ANNUAL AVERAGE RADON
LEVELS FOR THE AVERAGE OVER ALL LIVING LEVELS IN THE
                 TESTING POPULATION*
ftadon interval (pCi/L)
0-1
(Std. Err.)
No. of Obs.
1-2
(Std. Err.)
No. of Obs.
2-3
(Std. Err.)
No. of Obs.
3-4
(Std. Err.)
No. of Obs.
4-5
(Std. Err.)
No. of Obs.
5-6
(Std. Err.)
No. of Obs.
6-8
(Std. Err.)
No. of Obs.
8-10
(Std. Err.)
No. of Obs.
10-20
(Std. Err.)
No. of Obs.
>20
(Std. Err.)
No. of Obs.
Total
(Std. Err.)
No. of Obs.
Arithmetic Mean (pC(/L)
0.44
(0.01)
3,008
1.40
(0.01)
1,244
2.44
(0.02)
487
3.43
(0.02)
264
4.45
(0.02)
142
5.47
(0.04)
106
6.90
(0.06)
93
8.95
(0.08)
47
12.89
(0.33)
46
29.25
(4.80)
5
1.30
(0.06)
5,442
Median (pCi/L)
0.42
3,008
1.37
1,244
2.42
487
3.40
264
4.42
142
5.43
106
6.84
93
9.03
47
11.99
46
25.12
5
0.71
5,442
* EPA recommends that residents of all single-family homes, multi-unit structures below the third
floor, and mobile homes with permanent foundations should test them for radon.

-------
                            EXHIBIT B-9

          Distribution of Annual Average Radon Levels
                      in Single-Family Units
                        Average Over All Living Levels
Percent
     0-1     1-2    2-3    3-4    4-5     5-6    6-8    8-10    10-20    >20
                         Radon Concentration (pCi/L)         •

-------
                          EXHIBIT B-10
          Distribution of Annual Average Radon Levels
                       in Multi-Family Units
                      Average Over All Living Levels
Percent
100
 80
 60
 40
 20
  0
        84.9
     0-1    1-2     2-3    3-4    4-5    5-6    6-8   8-10  10-20   >20
                          Radon Concentration (pCi/L)

-------
                    EXHIBIT B-11

      ESTIMATES OF SELECTED PARAMETERS OF THE
  DISTRIBUTION OF ANNUAL AVERAGE RADON LEVELS IN THE
UNITED STATES BY FLOOR FOR SINGLE-FAMILY HOUSING UNITS
Floor in Housing Unit
Basement (Total)
(Std. Err.)
No. of Obs.
•• 	 -iii-iii- 	 .•-...••...••—•.•••••••••••••-^•^^ i ^^J^_^,^.^J^^_^^^J^_
Basement (Living Space)
(Std. Err.)
No. of Obs.
Basement (Not Living Space)
(Std. Err.)
No. of Obs.
1st Level Above Ground
(Std. Err.)
No. of Obs.
2nd Level Above Ground
(Std. Err.)
No. of Obs.
3rd and Higher Levels Above Ground
(Std. Err.)
No. of Obs.
Arithmetic
Mean
(pCi/L)
3.30
(0.23)
2,716
3.07
(0.24)
1,398
3.55
(0.26)
1,318
1.31
(0.07)
4,350
1.18
(0.09)
1,561
0.99
(0.16)
174
Median
(pCi/t)
2.06
2,716
—•mill 	 • 	 	 	 la,^.^^^^ , 	 ,,iiii.i,ii,—
1.88
1,398
2.28
1,318
0.77
4,350
0.70
1,561
0.64
174
Geometric
Standard
Deviation
2.90
(n.a)
2,716
• 	 - 	 -•- - •
2.93
(n.a)
1,398
2.81
(n.a)
1,318
2.84
(n.a)
4,350
2.78
(n.a)
1,561
2.61
(n.a)
174

-------
                             EXHIBIT B-12

   COMPARISON OF THE NATIONAL AND GROUND-CONTACT POPULATION
     EMPIRICAL DISTRIBUTIONS WITH THE LOGNORMAL DISTRIBUTIONS
 MODELED USING THE MEDIAN  AND BEST RANGE ESTIMATOR TO ESTIMATE
                    THE LOGNORMAL PARAMETERS
Radon
Category
(pCi/L)
0-1
1-2
2-3
3-4
4-5
5-6
6-8
8-10
10-20
>20

National Population1
Empirical
Estimates
(Percent)
63.61
19.87
6.85
3.66
1.85
1.72
1.15
0.64
0.59
0.06
100.00
Estimate
Percent
63.79
19.44
7.44
3.56
1.94
1.16
1.23
0.58
0.72
0.14
100.00
Difference
0.18
-0.43
0.59
-0.10
0.09
-0.56
0.08
-0.06
0.13
0.08

Ground-Contact Population2
Empircal
Estimates
(Percent)
59.17
22.09
7.75
4.16
2.08
1.97
1.32
0.73
0.68
0.07
100.00
Estimate
Percent
59.65
21.72
8.43
4.01
2.16
1.27
1.32
0.61
0.71
0.12
100.00
Difference
0.48
-0.37
0.68
-0.15
0.08
-0.70
0.00
-0.12
0.03
0.05
•
Key Thresholds
> 4 pCi/L
>10 pCi/L
6.01
0.65
5.77
0.86
-0.24 1 6.85
0.21 | 0.75
6.18
0.83
-0.67
0.08
Assumed Parameters
Geometric mean = median = 0.67 | 0.77
Geometric standard deviation - 3.11 | 2.92
1 All homes included in the survey.
2 All single-family units, except those that were 100 percent open underneath (such as unskirted mobile homes),
and residences on the first floor in multi-family units.

-------
                                     EXHIBIT B-13

                    National Population Empirical  Distribution With
              the Lognormal Distribution Modeled Using the Median and
             Best Range Estimator to Estimate the Lognormal Parameters
 Percent
 70
 60
 50
 40
 30
 20
 10
        0-1
1-2
3-4
6-8
          4-5      5-6
      Radon Concentration (pCi/L)

Empirical Estimate • Model-Based Estimate
8-10
                                          10-20
                                                                                     >20
Note: Actual values are presented in Exhibit B-12.

-------
Percent
70
                                      EXHIBITS-14

                    Ground-Contact Population Empirical Distribution
               With the Lognormal Distribution Modeled Using the Median
            and Best Range Estimator to Estimate the Lognormal Parameters
60
50
40
30
20
10
       0-1
1-2
3-4
    4-5      5-6
Radon Concentration (pCi/L)
6-8
8-10
10-20
                                                    >20
                               Empirical Estimate  H Model-Based Estimate
Note: Actual values are presented in Exhibit B-12.

-------