Interest in tracking the employment status of persons with disabilities has grown since the
passage of the Americans with Disabilities Act of 1990 (ADA). The primary national source of
data on labor force status (the Current Population Survey (CPS)) has been used by some analysts,
even though that survey does not actually attempt to identify the appropriate universe -- persons
with a physical or mental impairment that substantially limits one or more of the major life
activities.
The disability questions that are asked in the monthly CPS and in the annual March supplement
have specific purposes that make them inappropriate for identifying such persons. The disability
question that is asked in the monthly core is intended to identify persons who are out of the labor
force because they are "prevented from working". The March supplement question on whether
there are persons in the household who are "prevented from working" or who are "limited in the
kind or amount of work they can do" is asked as a screener to identify those households in which
there is some likelihood that one or more persons may be receiving disability-related income.
On the other hand, the disability supplements that have been asked in the Survey of Income and
Program Participation (SIPP) were designed to be consistent with the ADA definition of
disability. The supplements obtain information on
. the ability to perform specific functional activities (seeing, hearing, having one's speech
understood, lifting and carrying, climbing stairs, and walking),
. certain ADL's or activities of daily living (getting around inside the home, getting in and
out of a bed or chair, bathing, dressing, eating, and toileting), and
. certain IADL's or instrumental activities of daily living (going outside the home, keeping
track of money and bills, preparing meals, doing housework, and using the telephone).
The survey also collects information on the use of special aids such as wheelchairs and canes, the
presence of certain conditions related to mental functioning, and the ability to work at a job or
business.
We have now compiled three SIPP files containing data on disability status and a number of
other characteristics including employment status. These files make use of the overlapping panel
design of SIPP by combining data from two panels when the same disability questions were
asked at the same time on both panels.
Reference months
File for employment status
90 panel:Wave 6 / 91 panel:Wave 3 September - December 1991
92 panel:Wave 6 / 93 panel:Wave 3 September - December 1993
92 panel:Wave 9 / 93 panel:Wave 6 September - December 1994
Whenever comparisons are made over time, the data sets must be as comparable as possible. The
SIPP data sets described above are comparable in terms of the questions that were asked in the
surveys, but there are other issues of comparability that should be addressed. It has been
established that, when household surveys make use of a panel design (periodic visits to the same
households over time), certain measures are affected by the problems of time-in-sample bias and
attrition. In the CPS, for example, the reported rate of unemployment is affected by the number
of times a household has been included in the CPS sample. The CPS has solved the problem by
adopting a survey design that rotates households into and out of the sample, so that the average
time-in-sample is the same for each monthly CPS sample. The SIPP does not have such a
design, and for some measures there is a concern that the effects of time-in-sample bias or
attrition bias may be significant. (Attrition bias may occur if the characteristics of households
that drop out of the sample over time differ from the characteristics of households that remain.)
The potential for these problems to affect SIPP estimates has been examined with some care, and
analysts have generally concluded that these problems are of concern for some measures (e.g.,
poverty status), but statistically significant effects have been found only when comparisons
involve data sets that do not have the same proportion of wave 1 interviews. The SIPP data sets
used in this analysis do not make use of any interviews collected in waves earlier than wave 3, so
it seems unlikely that any comparisons are strongly affected by time-in-sample or attrition bias.
Another measurement issue concerns the criteria used to classify someone of working-age "with
a severe disability". Because working at a job or business is clearly a major life activity, SIPP
disability data treat persons who report that they are "unable to work at a job" as "with a severe
disability". But if we are tracking the employment status of persons with a severe disability, is it
proper to include in that category persons who report that they are "unable to work"? Part of this
issue concerns the meaning of "unable to work". When a person reports that he or she is "unable
to work" that person is making a judgement about workplace accommodations and the
availability of transportation. Changes in the workplace and transportation environments could
result in far fewer persons describing themselves as "unable to work".
The figures below show the number of persons 21 to 64 with a severe disability based on three
alternative definitions and shows the corresponding employment rates. Definition 1 is the
definition used in the publication "Americans With Disabilities: 1991-92" (P70-33), Definition 2
excludes from the severe category all persons self-identified as "unable to work", and Definition
3 excludes work disability as a criterion for determining whether a person has a severe disability.
To understand the difference between the last two definitions, consider a person who uses a
wheelchair and reports that he or she is unable to work; the person would be excluded from
Definition 2, but would be included in Definition 3.
1991 1994
(September-December) (September-December)
Number Percent Number Percent
Persons With Severe Disability (thousands) employed (thousands) employed
Definition 1 (basic SIPP
definition) 12,494 23.3 14,219 26.1
Definition 2 (excludes persons
"unable to work") 4,311 60.3 4,337 72.7
Definition 3 (does not use
ability to work as a
criterion) 8,624 30.1 9,506 33.9
The 1991 to 1994 changes in the percent employed are statistically significant at the 90-percent
level for all three definitions.
The attached tables show employment rates by disability status and show which 1991 to 1994
changes are statistically significant. The ability to measure change is determined, in large part,
by sample size, and the desire to look at samples as large as possible is the reason the data
presented here are based on combined panels. The approximate size of the first file was 34,000
households, and the approximate size of the second and third files was 40,000 households. The
next full SIPP disability supplements are scheduled to be asked in the fifth and eleventh waves of
the 1996 panel (June 1997 - Sept. 1997 and June 1999 - Sept. 1999 respectively). The 1996
panel will have a sample size of 37,000 households and will include 12 waves (households will
be visited 12 times at four-month intervals).
Contact the HHES Information area at 301-763-3242 or visit
ask.census.gov for further information on Disability Statistics.
Source: U.S. Census Bureau, Housing and Household Economic Statistics Division
