May 2002 Article

The lack of a disability measure in today's Current Population Survey

Thomas W. Hale

A common practice is to use data based on the Current Population Survey (CPS) to demonstrate that the employment-population ratios for people with disabilities deteriorated over the 1990s. This finding is counter-intuitive given the employment growth in the general population over this period, so various researchers have attributed the decline to the Americans with Disabilities Act (ADA) requirement for reasonable accommodations for persons with disabilities and the failure of the U.S. Equal Employment Opportunity Commission to adequately enforce the ADA.

There are two flaws with the analyses. The first defect, and the focus of this article, is that there are no questions in the Current Population Survey that identify persons with disabilities.¹ The second defect is attributing the decline to a specific statute or Federal agency when there is no data from the CPS that would provide empirical evidence linking the decline to a specific agency or statute. Therefore, conclusions by researchers about the employment rate trend for persons with disabilities and the underlying causes are not valid.

The questions

There are two sources in the CPS that have been used to measure employment trends among persons with disabilities. The first is from the basic CPS. (The basic CPS contains the questions that are asked every month to determine the employment status of household members.) The second is from the CPS income supplement that is appended once a year to the March basic CPS. (The income supplement provides information about work experience, earnings, and income in the previous calendar year.) Typically, researchers use data derived from both sources of the CPS to estimate the employment-population ratio for the population with disabilities.

The basic CPS questions are organized to first determine if the respondent is employed or unemployed (that is, actively looking and available for work). If the respondent does not meet the criteria for either category, he or she then is classified as "not in the labor force." When individual respondents or their proxies are asked if they did any work for pay or profit (a "yes" or "no" question), some instead respond that they have a disability, which is recorded. (No list of possible alternative responses is read to the respondent.) Following is the question on disability from the basic CPS:

Q20. LAST WEEK, did you do ANY work for (either) pay (or profit)?

                    o Yes             o Retired              o Don't know
                    o No              o Disabled             o Refused
                                        o Unable to work

Respondents who answer "yes" to this question are asked to give details about their employment, such as how many hours they worked, and in which industry and occupation they were employed. If the respondents answer "no," or give an alternative answer such as "disabled," the interviewer probes further to make sure they have not forgotten about some kind of work they did, or whether they had a job from which they were absent because of vacation, illness, and so forth. If it is determined that the respondents did no work for pay or profit, the interviewer then moves to another battery of questions to try to determine if respondents meet the criteria for being classified as unemployed. Failing that, respondents are classified by default as "not in the labor force."

If the respondent meets the criteria for either employed or unemployed, the response of "disabled" is erased from the file because it is inconsistent with the labor force categories of employed and unemployed. Because of this, researchers have no information on the disability status of employed and unemployed persons from the basic CPS.

Moreover, the alternative answers to the question are not mutually exclusive. The fact that a person said he or she was retired did not mean they did not have a disability. Hence, not only is the basic CPS unable to identify employed and unemployed persons with disabilities, it cannot fully measure the extent of disability among the "not in labor force" group. Perhaps most important is that most retired people and people with disabilities simply say "no" to the work question because the interviewer had not asked about their retirement or disability status.

Most researchers who use the CPS to develop data on the employment status of persons with disabilities realize that the basic CPS does not provide them with the information they need. Thus, they turn to the second source-income supplement data from the March CPS. As the name suggests, a primary purpose of the income supplement is to determine sources and amounts of income. The income supplement is the source of Federal data on poverty and the working poor.

From the income supplement, researchers typically use the questions on work limitation and income to inquire about the receipt of Supplemental Security Income (SSI) or Social Security Disability Income (SSDI) payments to define the population with disabilities.

Following is the work limitation question from the March 1999 Current Population Survey income supplement:

Q59a. (Do you/Does anyone in this household) have a health problem or disability which prevents (you/them) from working or which limits the kind or amount of work (you/they) can do?

o Yes o No

If individuals respond "yes" to the work limitation question, they are assumed to have a disability. The questioning then moves to identify sources of income related to the disability, such as workers' compensation, disability insurance, State or local government disability, and State temporary sickness. Persons collecting such benefits might be assumed to have a disability. (As noted earlier, they would have been asked this question only if they responded positively to the work limitation question.) Researchers then cross tabulate the responses from the question on work limitation with the responses from questions on labor force status in the basic (March) CPS to determine the employment and unemployment rates of those who responded positively to the work limitation and disability income questions.

At first glance, these questions seem to provide a reasonable means of identifying the population with disabilities. However, a closer examination leads to the conclusion that the questions lack validity as identifiers of persons with disabilities. In the case of disability, a valid question would be written in a way such that the people who were envisioned to have a disability would identify themselves when asked the question. Understanding the validity characteristics of survey questions is key to understanding what the data mean. For example, suppose one defines a person with a disability as someone with an impairment or condition that limits him or her in a major life activity. This is the first prong of the ADA definition of disability. Question 59a in the March CPS supplement certainly identifies some such people, but also may identify people with the flu, colds, broken legs, and other temporary illnesses or conditions. In this case, the survey question is not valid. Because a "yes" response to the work limitation question does not differentiate between a person with an impairment that limits the ability to work, and a person with a temporary health problem, this question could not be used to positively identify the population with disabilities (as defined in the ADA in our example).

The procedure for designing valid disability questions starts with the definition of disability. Then questions should be constructed in a way that will evoke positive responses from those who fit that definition. Tests, such as cognitive tests and field tests, should be undertaken to determine if a question does the job for which it was designed.

The current CPS questions on work limitation and disability income did not begin with any definition of disability. In fact, the purpose of the work limitation question is to serve as a screen for the income questions. It was specifically designed to direct respondents to questions on sources of income. Neither the work limitation nor the income questions were designed to identify the population with disabilities, nor were they tested to determine if they do so.

In the March CPS income supplement, disability is defined as a health condition or other disability that makes a person unable to work or limits the ability to work. This is an extremely narrow (and circular) definition of disability. In the ADA definition (which BLS has adopted in its efforts to design proper disability questions for the CPS), there is no reason to believe a person with a disability is limited or unable to work. Indeed, the purpose of the ADA is to enhance the employment prospects of people with disabilities and, therefore, the default assumption of the ADA is that such individuals can work. The work limitation and income questions in the March income supplement might identify a subset of the disability population (an untested empirical question), but they are not likely to capture the larger population with disabilities.

Click here for Part Two

This article is from the U.S. Department of Labor
Monthly Labor Review June 2001