SIDEBAR: Measuring Disabilities for Persons in the Labor Force
As noted in chapter 1, there is no consensus on the definition of disabilities. This means that in examining statistics related to disabilities, it is necessary to understand the definition used in compiling the statistics.
The decennial census has two relevant questions on work-related disabilities. Individuals are considered to have a disability if they answered "yes" to the question, "Does [the person under discussion] have a physical, mental, or other health condition that has lasted for 6 or more months and which limits the kind or amount of work [the person] can do at a job?" or "yes" to a similar question indicating that the disability made the person unable to work. This definition is not adequate for current purposes for two reasons. First, individuals with what are usually regarded as significant disabilities may respond that they do not have a work disability if they regard their work as being consistent with their education and other skills. This is especially important in understanding the representation of those with disabilities in science and engineering fields, because the work is primarily intellectual. With appropriate accommodation, individuals with significant disabilities that impair their sensory functions or mobility can be highly productive and may not regard themselves as having a disability that affects their ability to work. Second, the measure does not distinguish among types of disabilities. Some disabilities (e.g., disabilities that significantly impair mental functioning) would preclude individuals from attaining the necessary skills for science and engineering employment. It is important, though not always easy, to distinguish between those with disabilities that cannot be accommodated within the science and engineering labor force and those with disabilities that can be accommodated.
To address the problems with the Census Bureau's definition of disabilities, NSF's surveys use a functional definition of disability patterned after one developed for a planned survey of individuals with disabilities developed by the Census Bureau. This measure is based on asking individuals, "What is the USUAL degree of difficulty you have with [specific tasks involving seeing, hearing, walking, and lifting]?  Respondents are given five choices for each response, ranging from "none" to "unable to do." Unless elsewhere noted, having a disability is defined for this survey as having at least moderate difficulty in performing one or more of these tasks. Although this definition was designed to provide a relatively objective measure of disability, it is important to note that not all disabilities are captured by this measure. For example, learning disabilities and behavioral disorders are not included. 
The 1991-92 Survey of Income and Program Participation (SIPP) used questions for measuring disability that are quite similar to those in the Survey of Doctorate Recipients (McNeil 1993). This provides an opportunity to make some approximate comparisons between the science and engineering doctoral population and the larger population.
 The full wording of these alternatives in the survey forms is "SEEING words or letters in ordinary newsprint (with glasses/contact lenses if you usually wear them)," "HEARING what is normally said in conversation with another person (with hearing aid, if you usually wear one)," "WALKING without assistance (human or mechanical) or using stairs," "LIFTING or carrying something as heavy as 10 pounds, such as a bag of groceries."
 Additional measures of types of disability were omitted from the surveys due to practical limitations. The disability questions included in the questionnaires were considered burdensome and intrusive by many respondents. The survey designers were concerned that additional questions in this area would have a serious negative impact on the overall response rate and the validity of the surveys. This would be especially true if the surveys requested information on highly sensitive disabilities.