Data Points

Journal of Legal Economics, Mar 2008 by Hale, Thomas W

Special Editors Thomas Hale and Kurt V. Krueger

This article is a first in a series of Journal of Legal Economics articles under the Data Points Associate Editorship category as edited by Tom Hale and Kurt Krueger. Tom Hale of the Social Security Administration leads off this new section of the JLE with an article analyzing the disability questions contained within the Current Population Survey. We are soliciting papers from JLE readers for the Data Points section that describe data sources, computer software, and sites on the Internet related to the computation of damages in litigation or business valuation. We encourage all forms and lengths of submissions from brief announcements to detailed research. We also encourage anyone with suggestions of relevant topics they would like seen presented in Data Points to send an email to us at route216@verizon.net or Krueger@JohnWardEconomics.com.

Development of the New Disability Questions for the CPS and What They Can Tell Us About the Employment Status of People With Disabilities*

The CPS is the government's official source of employment and unemployment data. All demographic groups protected by civil rights statutes are separately identified in the CPS, except people with disabilities. The Bureau of Labor Statistics (BLS), the sponsor of the CPS, began its work to develop a disability measure in 1996. The effort was given a higher priority when President Clinton signed Executive Order 13078 in March 1997, mandating that the BLS, in conjunction with other agencies, develop an accurate and reliable measure of the employment rate of adults with disabilities. While the effort took some interesting turns over the years, a different set of questions, originally developed for the American Communities Survey (ACS), have been placed in the February 2008 CPS and will be administered to all survey households in their first month in sample (months 1 and 5). If things go well, the data will be available early in 2009, and these questions these could be a permanent fixture in the C

This Data Point will discuss the methodology used to develop BLS' questions, demonstrating how difficult it is to measure something as complex as disability. It will also briefly discuss the uses and limitations of the disability data.

Early History

The Executive Order that mandated development of a disability measure required disability to be defined as in the first prong of the Americans with Disabilities Act. That is, a person with a disability is one who, because of a condition or impairment, is limited in a major life activity. The key to this definition is the interaction of a person with a condition and impairment and their environment.

Several questions already appeared in the basic CPS and in the Annual Social and Economic Supplement, administered in March, which used the word "disability." Unfortunately, the questions were developed for a different purpose than to identify the disability population. In the case of the basic CPS, the objective of the question about why one is not working is to identify those who are not in the labor force, so as not to burden them with further questions about their labor force activity. The response categories refer to the reason or reasons a person did not work. Only one response is recorded, although several may apply. Individuals are not prompted to a list of responses as to why they are not working and the list the respondent never sees includes things like child care problems, transportation problems, retired, and in one response category, "sick or disabled." When the questions were developed, BLS did not particularly care why one was not in the labor force, they just wanted to skip individuals out of the remainder of the survey if work activity questions did not apply.

The March questions were developed to be screener questions with the objective of determining who should get various questions about their sources of income - SSI, SSDI, VA, Workmen's Compensation, etc. By design, screener questions cast a wide net to catch more people than would a classification question, because allowing false positives is considered the "better" mistake than allowing false negatives. Here, false negatives refer to those who said they were not limited in the type or amount of work they can do, but they did have a disability income. In short, the existing questions do not accurately identify the disability population, however defined, nor were they intended to. The Census web site says of the basic CPS question, that it represents a count of those who said "yes." The same could be said of the March question. (Census Bureau, 2004)

In other words, we do not know the characteristics of those who endorse the questions. This point has been made in numerous venues. (Hale, 2001; Silverstein et al., 2005)

The Executive Order governing BLS's disability research required them to develop accurate and reliable questions out of recognition that the existing questions did not have these properties. There are several types of accuracy and reliability in this context. In this case, accuracy refers to the ability of a question to elicit responses from the individuals envisioned to fit the definition of disability.