October 30, 2007

  SEQ CHAPTER \h \r 1 

FINAL REPORT

Statistical Analyses to Compare Average Levels of Lead in Paint 

Between Housing Age Categories, for Household Surfaces Classified as
Containing Lead-Based Paint in Residences, 

Using Data from the National Survey of Lead and Allergens in Housing

Prepared by

BATTELLE

505 King Avenue

Columbus, Ohio 43201-2693

Ron Morony, Work Assignment Manager

Sineta Wooten, Project Officer

Work Assignment 3-08

EPA Contract No. EP-W-04-021

Office of Pollution Prevention and Toxics

U.S. Environmental Protection Agency

Washington, DC 20460 

  SEQ CHAPTER \h \r 1 Battelle Disclaimer

This report is a work prepared for the United States Government by
Battelle. In no event shall either the United States Government or
Battelle have any responsibility or liability for any consequences of
any use, misuse, inability to use, or reliance upon the information
contained herein, nor does either warrant or otherwise represent in any
way the accuracy, adequacy, efficacy, or applicability of the contents
hereof.

EPA Disclaimer

The material in this document has not been subject to Agency technical
and policy review.  Views expressed by the authors are their own and do
not necessarily reflect those of the U.S. Environmental Protection
Agency.  Mention of trade names, products, or services does not convey,
and should not be interpreted as conveying, official EPA approval,
endorsement, or recommendation.  Do not quote or cite this document.

Statistical Analyses to Compare Average Levels of Lead in Paint 

Between Housing Age Categories, for Household Surfaces Classified as
Containing Lead-Based Paint

Introduction and Objectives

	The U.S. Environmental Protection Agency (EPA)’s Office of Pollution
Prevention and Toxics (OPPT) is currently working to finalize the
Renovation, Repair, and Painting (RR&P) Rule, which will establish work
practice standards and training and certification requirements for
persons engaged in renovation activities for compensation within certain
housing units built prior to 1978 and found to contain lead-based paint.
 EPA’s proposed rule, published on January 10, 2006, as supplemented
on June 5, 2007 contained requirements that are aimed to reduce exposure
to lead hazards in such housing that may be created by performing
renovation, repair, and painting activities that disturb lead-based
paint. The RR&P Rule supports the attainment of the Federal government's
goal of eliminating childhood lead poisoning in the U.S. by 2010. 

	An important component to the economic analysis supporting the RR&P
Rule, which assesses the costs and benefits associated with the proposed
requirements, is a characterization of the levels of lead in paint on
building components within pre-1978 housing units, conditional on the
paint being characterized as lead-based (i.e., equal to or exceeding 1.0
mg of lead per square centimeter of painted surface area). This
characterization, with a focus on how lead levels in paint may differ
for housing units of different age ranges that were built while
lead-based paint was still sold for residential use, will aid in the use
of dust study data in the economic analysis.  The National Survey of
Lead and Allergens in Housing (NSLAH) is the most recent and complete
source of lead levels on painted surfaces in the nation’s housing. 
The NSLAH, jointly sponsored by the U.S. Department of Housing and Urban
Development (HUD) and the National Institute of Environmental Health
Sciences (NIEHS), was conducted in 1998-1999 to assess children's
potential exposure to lead and allergens in the home.  The target
population in this survey consisted of all permanently-occupied,
noninstitutional housing units in the U.S. housing stock, including
single- and multi-family buildings and manufactured housing units, in
which young children were permitted to live.  (Approximately 96 million
housing units existed within the target population at the time of the
survey.)  The NSLAH measured lead levels in various media, including
dust, soil, and paint, at each selected housing unit.  In particular,
within each unit, x-ray fluorescence (XRF) technology was used to
measure the concentration of lead on each painted surface in each of
four rooms, with each selected room representing one of the following
four room types: kitchen, common living areas, bedrooms (preferably
those containing children), and other rooms. 

	The primary objectives of the statistical analyses presented in this
report were 1) to use XRF data collected within the NSLAH to
characterize lead levels in lead-based paint found in pre-1978 housing
according to housing stock vintage and specific building component
types, and 2) to identify those building components where statistically
significant differences exist in average XRF lead levels between
different housing age categories.  

Conclusions

A statistical analysis was applied to XRF measurements within the NSLAH
that equaled or exceeded 1.0 mg/cm2 lead (i.e., representing lead levels
in lead-based paint) to determine the presence of significant
differences in average XRF measurements between specified housing age
categories (as defined by year built).  Data for all housing sampled in
the NSLAH and built prior to 1978 were included in this analysis. 
Separate data analyses were performed for different types of interior
and exterior painted building components.  While EPA selected one set of
housing age categories for this analysis (pre-1930, 1930-1949,
1950-1959, 1960-1977), an alternative set was also considered for
illustrative purposes, and its results are presented in Appendix A.

Significant differences among pairs of housing age categories (at an
overall 0.05 level) in average XRF measurements (where these
measurements equaled or exceeded 1.0 mg/cm2 lead) occurred for exterior
building components but not interior components.  When exterior surfaces
were considered as a single group, housing built prior to 1930 had
significantly different average XRF measurements for this group compared
to housing in each of the other three age categories (i.e., 1930-1949,
1950-1959, 1960-1977), with higher average XRF measurements occurring in
the pre-1930 category.  When building components were further
categorized by their specific type (i.e., walls/floors/ceilings,
windows, doors, and trim), significant differences in average XRF
measurements continued to exist between housing age categories for each
exterior building component type, but not for interior components.  For
each of the exterior component types, the significant differences
consisted of the following age categories being significantly different
from the pre-1930 age category:

1930-1949 homes, for exterior walls and windows.

1950-1959 homes, for exterior walls, windows, doors, and trim.

1960-1977 homes, for exterior windows, doors, and trim.

Methods

	The NSLAH featured a complex probabilistic survey design, where
eligible units within the target population differed in their
probability of selection to the survey.  As a result, survey weights
were assigned to each of the 831 housing units that were selected to
participate in the survey.  Two sets of weights have been calculated: 
one set representing all stages of sampling to the housing unit level,
and another set representing all stages to the room/component level.  By
incorporating these weights properly within data summaries, it is
possible to use data collected at the housing unit level to obtain
representative, unbiased estimates of unknown parameters, such as
average XRF lead levels, within the nation’s housing stock.  The
summaries and analyses presented in this report account for the survey
weights.

	Housing Age Categories.  Within the NSLAH database, each surveyed
housing unit is identified according to the year in which it was built,
with this value occasionally having been derived (as noted in the
survey’s data file documentation) if the respondent did not specify a
year within the questionnaires.  Thus, the housing units could be
categorized within any pre-defined set of housing age categories. 
Published NSLAH reports focus on the following categorization:  pre-1940
housing, 1940-1959 housing (broken down into 1940-1945 and 1946-1959
groups within the NSLAH database), 1960-1977 housing, and 1978-1998
housing.  For the analyses presented in this report, data for housing
units built in 1978 or later (when the sale of lead-based paint was
banned for residential use in the United States) were excluded.  Upon
reviewing several different sets of housing age categories, EPA selected
the following set of housing age categories within which average XRF
lead levels would be summarized and the results used in further
evaluations:

Selected set 

Units built prior to 1930 

Units built between 1930 and 1949

Units built between 1950 and 1959

Units built between 1960 and 1977.

This report summarizes XRF levels in lead-based paint for specific
building component types according to this housing age categorization. 
In addition, for illustrative purposes, the report presents a second set
of summaries (within Appendix A) for the following alternative set of
housing age categories:  

Alternative set 

Units built prior to 1920 

Units built between 1920 and 1939

Units built between 1940 and 1959

Units built between 1960 and 1977.

Other sets of housing age categories which EPA had also considered
previously include the five sets listed in Table 1.  Outcomes of the
statistical analyses associated with these other sets of categories were
similar in nature to the findings summarized in this report.

Table 1.	Other Housing Age Categories which EPA Considered When
Summarizing Average XRF Measurements from Lead-Based Painted Surfaces
for Use in the RRP Rulemaking Process

Pre-1950 

1950 - 1959

1960 - 1978	Pre-1950 

1950 - 1978	Pre-1930 

1930 - 1949

1950 - 1978	Pre-1940 

1940 - 1959

1960 - 1978	Pre-1930 

1930 - 1939

1940 - 1949

1950 - 1959

1960 - 1969

1970 - 1977



Building Component Categories.  For each NSLAH housing unit, database
records containing XRF measurements were categorized according to the
type of building component that they represented (as specified within a
database variable named COMPCAT).  The following eight building
component categories were defined:

Interior walls, ceilings, or floors (COMPCAT=1)

Exterior walls (COMPCAT=51)

Interior windows (COMPCAT=2)

Exterior windows (COMPCAT=52)

Interior doors (COMPCAT=3) 

Exterior doors (COMPCAT=53)

Interior trim (COMPCAT=4)

Exterior trim (COMPCAT=54).

XRF measurements for other building components that were not among these
eight categories were not included in the analysis.  This report
presents XRF data summaries for each of these eight categories, as well
as for two more general categories of interior and exterior building
components, where the four interior building component categories were
collapsed into a single “interior” category, and the four exterior
component categories were collapsed into an “exterior” category.  

	For a given building component category, each housing unit was
classified as to whether or not at least one tested component within the
category resulted in an XRF measurement that equaled or exceeded 1.0
mg/cm2 of lead (meaning   SEQ CHAPTER \h \r 1 lead-based paint was found
to be present in at least one instance).  For a given housing age
category, this classification was used to determine the number of
sampled units with lead-based paint somewhere on that particular
component type, as well as to generate an estimate for the total number
of units in the housing stock with this characteristic (using weights
that account for sampling at the housing unit level).

	Calculation of Descriptive Statistics.  Only NSLAH database records
corresponding to building components with XRF measurements exceeding 1.0
mg/cm2 of lead were considered when calculating descriptive statistics
of XRF measurements.  These statistics were calculated across housing
units, rooms, and tested components, with separate sets of statistics
calculated for each housing age category and building component type. 
To make the statistics representative of building components containing
lead-based paint within the housing stock population targeted in the
NSLAH, they were calculated utilizing sampling weights that account for
sampling down to the building component level.  The DESCRIPT procedure
in Version 9.0.1 of SUDAAN® was used to calculate these descriptive
statistics, which included the arithmetic and geometric means, standard
error of the mean, and the 25th, 50th (i.e., median), and 75th
percentiles.  In addition to the sampling weights, this procedure used a
set of jackknife replicate weights provided in the NSLAH database when
calculating the standard error of the mean.

	Statistical Analysis.  The statistical analysis, which compared average
XRF measurements (for components found to contain lead-based paint)
between housing age categories for a given building component type,
utilized a one-way analysis of variance (ANOVA) approach.  This analysis
was performed using the REGRESS procedure in Version 9.0.1 of SUDAAN®
and utilized the sampling weights (and corresponding jackknife replicate
weights) that account for sampling down to the building component level.
 The ANOVA model expressed XRF measurements as an overall average that
is allowed to vary by housing age category, plus random variation. 
Within the ANOVA, statistical tests were performed to make comparisons
between each pair of housing age categories in order to determine which
pairs have arithmetic means that differ significantly from each other. 
Because a set of housing age categories consisted of four categories, a
total of six pairs of categories existed.  To keep the overall Type I
error probability among all six pairwise comparisons (i.e., making at
least one incorrect conclusion of significant difference) to no higher
than 0.05, each comparison was tested at a 0.05/6=0.0083 significance
level.  These comparisons were performed using a chi-square test
statistic and Satterthwaite’s adjusted degrees of freedom.

	

Results

For the set of housing age categories that EPA selected for
consideration in the RRP rulemaking process, Tables 2 and 3 present the
values of the descriptive statistics that were calculated from XRF
measurements in the NSLAH database that equaled or exceeded 1.0 mg/cm2,
calculated for each housing age category and building component type. 
Table 2 divides building components among two categories (interior and
exterior), while Table 3 considers eight different building component
categories.  

The arithmetic means and standard errors listed in Table 2 are expressed
graphically in Figure 1a, and those listed in Table 3 are presented in
Figure 2a (for interior components) and Figure 2b (for exterior
components).  Within each of these figures, the bars represent the
arithmetic means, while plus and minus one standard error of the mean
emanate as lines from the top of each bar.  By observing the heights of
the bars and the size of the standard errors, these charts allow for
significant differences in the arithmetic means among housing age
categories to be more easily noticed.  Utilizing the same bar chart
format, Figure 1b presents the geometric means from Table 2, while
Figure 2c (for interior components) and Figure 2d (for exterior
components) present geometric means from Table 3.  (Note that standard
errors could be calculated for arithmetic means only, and therefore, the
bar charts of geometric means do not have standard error bars included.)

When classifying building components as only interior vs. exterior
(Table 2), both arithmetic and geometric means are similar between these
two types of components for a given housing age category.  The pre-1930
category has arithmetic and geometric means that are nearly twice the
size of those observed for 1930-1949 housing, and over twice the size of
the means for 1950-1959 and 1960-1977 housing.  

Regardless of whether building components were classified into two or
eight categories, significant differences (at the 0.0083 level) in the
arithmetic means between pairs of age categories occurred only among
exterior building components, and not for interior components.  When all
exterior building components were grouped together (Table 2), each of
the three categories covering 1930-1977 housing were significantly
different from pre-1930 housing for exterior surfaces.  When data for
the four exterior building component types were analyzed separately
(Table 3), the following housing age categories were found to be
significantly different from pre-1930 homes for the specified exterior
building components:

1930-1949 homes, for exterior walls and windows.

1950-1959 homes, for exterior walls, windows, doors, and trim.

1960-1977 homes, for exterior windows, doors, and trim.

Thus, the analysis in Table 3 concluded that all three age categories
representing housing built in 1930 or later had significantly different
average XRF levels on exterior windows compared to pre-1930 housing,
while for exterior doors and trim, only the two age categories
representing housing built in 1950 or later differed significantly from
pre-1930 housing.  None of the building component types had significant
differences present among the three age categories representing housing
built in 1930 or later.  When significant differences with pre-1930
housing were present among exterior building components, average XRF
measurements were higher in pre-1930 housing.  

	For the alternative set of housing age categories, Appendix A contains
the same set of descriptive statistics, tables, and bar charts for XRF
measurements within these categories.

Further investigation of the XRF data demonstrated that there is
considerable skewness in the distribution of XRF measurements within
each age category for both interior and exterior building components, as
would be expected with such measures, with most of the measurements
clustered between 1.0 and 5.0 mg/cm2 and occasional large measurements
leading to extended right tails in the distribution.  This finding
suggests that the geometric means in Figure 1b and Figure 2b are more
representative of the central tendency of the distribution than are the
arithmetic means, due to the heavier influence that large XRF measures
have on the arithmetic mean compared to the geometric mean.  When the
statistical analyses in Table 2 were repeated on log-transformed XRF
measures (to better account for the skewed distribution), significant
differences among age categories in their average log-transformed XRF
measures were noted for exterior components, but not for interior
components.  For exterior components, those pairs of housing age
categories that the analyses of the log-transformed XRF measures found
to be significantly different at an overall 0.05 level included the
pre-1930 category paired with each of the other three categories (which
Table 2 also notes as being significantly different when untransformed
data were analyzed), as well as the pair consisting of 1930-1949 housing
and 1950-1959 housing. 



Table 2.	Descriptive Statistics on XRF Measures from Pre-1978 Housing
Units Participating in the National Survey of Lead and Allergens in
Housing, from Building Components Found to Contain Lead-Based Paint, and
Estimated Numbers of Housing Units with Lead-Based Paint, for the
Selected Set of Housing Age Categories and Interior vs. Exterior
Building Components

House Age Categories	# Houses Sampled 	# Sampled

Housing Units with LBP	Estimated #

U.S. Housing Units with LBP (1,000)	Summary of XRF Measures (mg/cm2)
Exceeding 1.0 mg/cm2





Arithmetic

Mean	Standard Error of the Mean	Geometric Mean	25th

Percentile	Median	75th

Percentile

Interior Components

Pre-1930	140	108	12,233	9.92	4.89	6.30	2.59	6.43	14.86

1930-1949	74	48	5,558	5.34	2.40	3.34	1.36	2.36	7.95

1950-1959	130	46	4,940	4.73	4.01	3.14	1.41	2.71	5.43

1960-1977	267	40	4,298	5.99	9.68	3.66	1.90	3.06	5.06

Exterior Components

Pre-1930	134	93	10,643	10.68	1.03	6.07	2.39	5.93	15.86

1930-1949	72	51	6,145	5.49(1)	0.79	3.77	1.86	3.43	7.58

1950-1959	125	65	6,843	3.58(1)	0.45	2.53	1.46	1.95	3.53

1960-1977	251	34	3,323	4.80(1)	0.70	3.35	1.73	2.98	5.95

(1)  Significantly different from the pre-1930 housing group at a
0.05/6=0.0083 level.

Table 3.	Descriptive Statistics on XRF Measures from Pre-1978 Housing
Units Participating in the National Survey of Lead and Allergens in
Housing, from Building Components Found to Contain Lead-Based Paint, and
Estimated Numbers of Housing Units with Lead-Based Paint, for the
Selected Set of Housing Age Categories and Eight Different Building
Components

Age Categories	# Houses Sampled 	# Sampled

Housing Units with LBP	Estimated #

U.S. Housing Units with LBP (1,000)	Summary of XRF Measures (mg/cm2)
Exceeding 1.0 mg/cm2





Arithmetic

Mean	Standard Error of the Mean	Geometric Mean	25th

Percentile	Median	75th

Percentile

Interior Walls, Ceilings, or Floors

Pre-1930	139	54	6,241	8.88	3.04	5.17	1.95	4.46	14.00

1930-1949	73	24	2,224	4.73	0.85	3.25	1.62	2.45	5.92

1950-1959	130	20	2,140	7.33	2.76	4.98	2.20	4.50	12.94

1960-1977	267	15	1,514	13.03	7.41	6.76	1.75	6.80	24.45

Exterior Walls

Pre-1930	100	53	6,167	7.12	1.01	4.29	1.96	3.61	8.02

1930-1949	58	20	2,415	4.04(1)	0.90	3.01	1.82	2.35	3.87

1950-1959	96	26	2,618	2.84(1)	0.42	2.29	1.45	1.88	3.02

1960-1977	180	18	1,962	5.57	1.21	3.71	1.97	3.20	6.61

Interior Windows

Pre-1930	135	90	9,830	10.51	3.18	6.77	2.98	6.74	15.95

1930-1949	71	30	3,452	6.16	0.99	4.06	1.78	3.32	9.09

1950-1959	120	25	2,753	4.13	4.62	2.73	1.24	2.06	5.00

1960-1977	236	20	2,092	3.33	2.74	2.80	1.97	2.19	3.85



Table 3.  (cont.)

Age Categories	# Houses Sampled 	# Sampled

Housing Units with LBP	Estimated #

U.S. Housing Units with LBP (1,000)	Summary of XRF Measures (mg/cm2)
Exceeding 1.0 mg/cm2





Arithmetic

Mean	Standard Error of the Mean	Geometric Mean	25th

Percentile	Median	75th

Percentile

Exterior Windows

Pre-1930	96	58	6,630	13.92	2.32	8.57	3.72	7.68	20.88

1930-1949	47	28	3,271	5.57(1)	0.80	3.92	1.96	4.04	6.61

1950-1959	84	39	3,903	4.34(1)	0.84	3.00	1.65	2.31	4.39

1960-1977	120	17	1,798	4.15(1)	1.05	2.96	1.56	3.08	3.43

Interior Doors

Pre-1930	138	66	7,889	9.51	7.81	6.27	2.58	6.98	14.51

1930-1949	74	31	3,599	3.59	10.59	2.10	1.15	1.35	2.38

1950-1959	124	27	2,936	4.09	1.30	3.01	1.48	3.52	4.30

1960-1977	262	19	1,541	2.85	1.36	2.32	1.09	2.78	4.18

Exterior Doors

Pre-1930	125	64	6,907	11.07	1.90	6.31	2.47	4.47	17.32

1930-1949	70	35	4,113	6.72	1.32	4.41	2.05	3.66	9.53

1950-1959	119	36	4,003	3.89(1)	0.84	2.58	1.40	1.98	3.49

1960-1977	233	16	1,717	4.09(1)	1.02	2.99	1.48	2.58	4.38

Interior Trim

Pre-1930	130	50	6,330	8.55	1.18	5.43	2.06	6.36	13.32

1930-1949	64	13	1,110	7.34	2.01	5.15	2.22	4.59	9.39

1950-1959	105	12	1,222	4.16	1.14	2.81	1.35	2.87	3.08

1960-1977	219	9	1,075	8.85	3.09	7.20	4.44	8.90	---

Exterior Trim

Pre-1930	62	20	2,302	15.94	3.45	9.01	2.56	15.91	20.63

1930-1949	31	2	382	6.59	4.50	4.99	---	---	---

1950-1959	64	14	1,737	2.74(1)	0.91	1.98	---	1.51	2.26

1960-1977	128	14	1,429	4.47(1)	0.82	3.53	1.78	3.00	6.09

(1)  Significantly different from the pre-1930 housing group at a
0.05/6=0.0083 level.

Figure 1a.	Arithmetic Means (± standard error bars) for XRF
Measurements Exceeding 1.0 mg/cm2 on Painted Surfaces in U.S. Housing,
for Each of the Selected Set of Housing Age Categories and Interior vs.
Exterior Building Components 

Figure 1b.	Geometric Means for XRF Measurements Exceeding 1.0 mg/cm2 on
Painted Surfaces in U.S. Housing, for Each of the Selected Set of
Housing Age Categories and Interior vs. Exterior Building Components 



Figure 2a.	Arithmetic Means (± standard error bars) for XRF
Measurements Exceeding 1.0 mg/cm2 on Painted Surfaces in U.S. Housing,
for Each of the Selected Set of Housing Age Categories and Eight
Different Interior Building Components 

Figure 2b.	Arithmetic Means (± standard error bars) for XRF
Measurements Exceeding 1.0 mg/cm2 on Painted Surfaces in U.S. Housing,
for Each of the Selected Set of Housing Age Categories and Eight
Different Exterior Building Components 

Figure 2c.	Geometric Means for XRF Measurements Exceeding 1.0 mg/cm2 on
Painted Surfaces in U.S. Housing, for Each of the Selected Set of
Housing Age Categories and Eight Different Interior Building Components 

Figure 2d.	Geometric Means for XRF Measurements Exceeding 1.0 mg/cm2 on
Painted Surfaces in U.S. Housing, for Each of the Selected Set of
Housing Age Categories and Eight Different Exterior Building Components 



Appendix A

Statistical Summaries for an Alternative Set 

of Housing Age Categories

For the alternative set of housing age categories (pre-1920, 1920-1939,
1940-1959, 1960-1977), Tables A-1 and A-2 present the values of the
descriptive statistics that were calculated from XRF measurements in the
NSLAH database that equaled or exceeded 1.0 mg/cm2, calculated for each
housing age category and building component type.  Table A-1 divides
building components among two categories (interior and exterior), while
Table A-2 considers eight different building component categories.  

The arithmetic means and standard errors listed in Table A-1 are
expressed graphically in Figure A-1a, and those listed in Table A-2 are
presented in Figure A-2a (for interior components) and Figure A-2b (for
exterior components).  Figure A-1b presents the geometric means from
Table A-1, while Figure A-2c (for interior components) and Figure A-2d
(for exterior components) present geometric means from Table A-2.  

Table A-1 indicates that for exterior building components, each of the
two most recent categories (covering housing built after 1939) had
average XRF measures (when the XRF measures represented lead-based
paint) that were significantly different from each of the two earliest
categories (covering pre-1940 housing).   Significant differences were
not observed for interior building components, due to considerably
larger variability estimates within each age category (as seen by the
sizes of the standard errors of the mean of the untransformed XRF
measures) compared to exterior components.

As noted in Table A-2, all but one instance of significant differences
among arithmetic means occurred with exterior building components. 
Among interior building components, significant differences were seen
only for interior windows, where 1960-1977 housing was significantly
different from pre-1920 housing.  Among exterior components, significant
differences in arithmetic means from pre-1920 homes were observed in
only one instance:  on exterior walls in 1940-1959 homes.  The most
frequent occurrences of significant differences within exterior
components were with 1920-1939 homes, which were seen in the following
instances:    

1940-1959 homes, for exterior walls, windows, doors, and trim.  

1960-1977 homes, for exterior windows and doors.  

All cases of significant differences among pairs of building age
categories found higher averages for the older housing category within a
pair.



Table A-1.	Descriptive Statistics on XRF Measures from Pre-1978 Housing
Units Participating in the National Survey of Lead and Allergens in
Housing, from Building Components Found to Contain Lead-Based Paint, and
Estimated Numbers of Housing Units with Lead-Based Paint, for the
Alternative Set of Housing Age Categories and Interior vs. Exterior
Building Components

House Age Categories	# Houses Sampled 	# Sampled

Housing Units with LBP	Estimated #

U.S. Housing Units with LBP (1,000)	Summary of XRF Measures (mg/cm2)
Exceeding 1.0 mg/cm2





Arithmetic

Mean	Standard Error of the Mean	Geometric Mean	25th

Percentile	Median	75th

Percentile

Interior Components

Pre-1920	30	28	3,625	11.67	3.29	7.03	2.60	7.38	21.23

1920-1939	128	94	10,040	8.60	5.84	5.58	2.35	5.80	13.03

1940-1959	186	80	9,066	4.70	3.11	3.04	1.31	2.24	5.92

1960-1977	267	40	4,298	5.99	9.68	3.66	1.90	3.06	5.06

Exterior Components

Pre-1920	29	23	2,948	13.40	2.56	7.46	2.59	6.92	19.85

1920-1939	123	85	9,255	8.83	0.72	5.29	2.31	4.11	12.77

1940-1959	179	101	11,428	4.09(1) (2)	0.41	2.85	1.50	2.27	4.44

1960-1977	251	34	3,323	4.80(1) (2)	0.70	3.35	1.73	2.98	5.95

(1)  Significantly different from the pre-1920 housing group at a
0.05/6=0.0083 level

(2)  Significantly different from the 1920-1939 housing group at a
0.05/6=0.0083 level.

Table A-2.	Descriptive Statistics on XRF Measures from Pre-1978 Housing
Units Participating in the National Survey of Lead and Allergens in
Housing, from Building Components Found to Contain Lead-Based Paint, and
Estimated Numbers of Housing Units with Lead-Based Paint, for the
Alternative Set of Housing Age Categories and Eight Different Building
Components

Age Categories	# Houses Sampled 	# Sampled

Housing Units with LBP	Estimated #

U.S. Housing Units with LBP (1,000)	Summary of XRF Measures (mg/cm2)
Exceeding 1.0 mg/cm2





Arithmetic

Mean	Standard Error of the Mean	Geometric Mean	25th

Percentile	Median	75th

Percentile

Interior Walls, Ceilings, or Floors

Pre-1920	29	17	2,031	11.25	6.09	6.43	2.11	4.78	22.01

1920-1939	128	44	4,939	6.39	0.82	4.04	1.63	3.27	9.47

1940-1959	185	37	3,634	6.26	1.56	4.20	1.92	4.18	7.47

1960-1977	267	15	1,514	13.03	7.41	6.76	1.75	6.80	24.45

Exterior Walls

Pre-1920	24	16	2,221	8.83	1.95	5.11	1.58	5.83	8.78

1920-1939	92	42	4,524	5.95	0.97	3.89	2.13	2.68	7.64

1940-1959	138	41	4,456	3.08(1)(2)	0.43	2.40	1.48	1.90	3.29

1960-1977	180	18	1,962	5.57	1.21	3.71	1.97	3.20	6.61

Interior Windows

Pre-1920	29	23	2,853	12.19	2.12	7.49	2.94	8.16	21.19

1920-1939	124	79	8,254	9.17	5.34	6.05	2.71	5.98	13.24

1940-1959	173	43	4,928	4.98	2.52	3.18	1.31	2.22	7.59

1960-1977	236	20	2,092	3.33(1)	2.74	2.80	1.97	2.19	3.85

Table A-2.  (cont.)

Age Categories	# Houses Sampled 	# Sampled

Housing Units with LBP	Estimated #

U.S. Housing Units with LBP (1,000)	Summary of XRF Measures (mg/cm2)
Exceeding 1.0 mg/cm2





Arithmetic

Mean	Standard Error of the Mean	Geometric Mean	25th

Percentile	Median	75th

Percentile

Exterior Windows

Pre-1920	21	16	2,079	16.41	5.93	9.73	5.10	7.15	28.40

1920-1939	85	50	5,373	11.08	1.63	6.93	2.53	7.48	16.01

1940-1959	121	59	6,353	4.71(2)	0.59	3.30	1.64	2.81	5.96

1960-1977	120	17	1,798	4.15(2)	1.05	2.96	1.56	3.08	3.43

Interior Doors

Pre-1920	29	17	2,096	10.34	7.80	6.51	2.68	8.66	14.69

1920-1939	127	60	6,937	8.62	5.43	5.45	2.20	6.00	13.76

1940-1959	180	47	5,391	3.22	3.89	2.28	1.23	1.75	3.73

1960-1977	262	19	1,541	2.85	1.36	2.32	1.09	2.78	4.18

Exterior Doors

Pre-1920	28	15	1,852	15.87	5.26	9.23	2.48	9.62	25.91

1920-1939	115	59	5,982	9.46	1.36	5.65	2.43	3.80	16.99

1940-1959	171	61	7,189	4.65(2)	0.74	3.09	1.51	2.54	4.78

1960-1977	233	16	1,717	4.09(2)	1.02	2.99	1.48	2.58	4.38

Interior Trim

Pre-1920	28	13	1,839	11.70	3.66	6.61	1.96	10.55	19.96

1920-1939	118	43	5,129	7.58	1.01	5.24	2.23	6.05	10.00

1940-1959	153	19	1,695	4.30	1.09	2.89	1.38	2.71	3.09

1960-1977	219	9	1,075	8.85	3.09	7.20	4.44	8.90	---

Exterior Trim

Pre-1920	18	8	818	20.01	8.07	11.89	3.10	16.22	20.67

1920-1939	51	12	1,484	13.59	3.42	7.67	1.91	9.97	19.12

1940-1959	88	16	2,119	3.44(2)	1.09	2.34	---	1.68	2.93

1960-1977	128	14	1,429	4.47	0.82	3.53	1.78	3.00	6.09

(1)  Significantly different from the pre-1920 housing group at a
0.05/6=0.0083 level.

(2)  Significantly different from the 1920-1939 housing group at a
0.05/6=0.0083 level.



Figure A-1a.	Arithmetic Means (± standard error bars) for XRF
Measurements Exceeding 1.0 mg/cm2 on Painted Surfaces in U.S. Housing,
for Each of the Alternative Set of Housing Age Categories and Interior
vs. Exterior Building Components 

Figure A-1b.	Geometric Means for XRF Measurements Exceeding 1.0 mg/cm2
on Painted Surfaces in U.S. Housing, for Each of the Alternative Set of
Housing Age Categories and Interior vs. Exterior Building Components 



Figure A-2a.	Arithmetic Means (± standard error bars) for XRF
Measurements Exceeding 1.0 mg/cm2 on Painted Surfaces in U.S. Housing,
for Each of the Alternative Set of Housing Age Categories and Eight
Different Interior Building Components 

Figure A-2b.	Arithmetic Means (± standard error bars) for XRF
Measurements Exceeding 1.0 mg/cm2 on Painted Surfaces in U.S. Housing,
for Each of the Alternative Set of Housing Age Categories and Eight
Different Exterior Building Components 



Figure A-2c.	Geometric Means for XRF Measurements Exceeding 1.0 mg/cm2
on Painted Surfaces in U.S. Housing, for Each of the Alternative Set of
Housing Age Categories and Eight Different Interior Building Components 

Figure A-2d.	Geometric Means for XRF Measurements Exceeding 1.0 mg/cm2
on Painted Surfaces in U.S. Housing, for Each of the Alternative Set of
Housing Age Categories and Eight Different Exterior Building Components 

 The Lead-Based Paint Poisoning Prevention Act of 1971, as amended by
the Housing and Community Development Act of 1987, established 1.0
mg/cm2 as the Federal threshold for lead on painted architectural
components in public and Indian housing, where exceeding this threshold
(along with paint condition standards) can trigger abatement of the
paint due to the lead hazard.  Title X of the Housing and Community
Development Act of 1992 provides the Federal definition of lead-based
paint, applicable to all housing, as paint containing 1.0 mg/cm2 of lead
or 0.5% lead by weight.

 See footnote on previous page.

Battelle	  PAGE  8 	June 25, 2007

WA 3-08, Contract No. EP-W-04-021

Battelle	  PAGE  1 	October 30, 2007

WA 3-08, Contract No. EP-W-04-021

