Factors associated with workforce participation among SSDI beneficiaries.

The Social Security Disability Insurance (SSDI) program provides cash benefits and health insurance to approximately 6.8 million disabled workers and their families, at a total annual cost of about $66 billion (Social Security Administration, 2004). The program has grown rapidly, a trend attributed to various factors, including general population growth (Yeas, 1995), enrollment of workers at younger ages (Rupp & Stapleton, 1995), greater longevity of workers (Riley, Lubitz, & Zhang, 2003), SSA policy changes (Burkhauser, Butler, & Weathers, 2001), and dislocations and transformations of the u.s. labor market (Catalano & Kennedy, 1998; Khan, Gerdtham, & Jansson, 2004; Yelin & Katz, 1994). Once participants enroll in the SSDI program, they tend to stay in the program until they reach age 65 and receive regular retirement benefits. The U.S. General Accounting Office reports that only one-half of one percent of beneficiaries leave the program rolls and return to work (GAO, 2001). This persistently low rate of disenrollment is inconsistent with a growing political consensus, bolstered by survey data, that people with disabilities want to work and, with sufficient support services and adequate opportunities, can work (Kaye, 1997; National Council on Disability, 1997).

In 1999, Congress passed the Ticket-to-Work and Work Incentives Improvement Act (TWWIIA, PL 106-170), which mandates a series of program changes to encourage employment among both SSDI beneficiaries and Supplemental Security Income (SSI) recipients. These changes for SSDI beneficiaries included vouchers for rehabilitation services, maintenance of Medicare coverage for working beneficiaries, and a streamlined re-enrollment process (SSA, 2001). The economic ramifications of such efforts are significant: doubling the small proportion of SSDI beneficiaries who return to work (e.g., from 0.5% to 1%) would return billions to the Social Security trust fund (GAO, 1999).

Kennedy, Olney and Schiro-Geist (2004) using data from the 1994 and 1995 National Health Interview Survey (NHIS) disability supplement, found that about 9.2% of SSDI beneficiaries reported being employed, and another 1.3% reported looking for work. This paper uses data from the 2001 and 2002 NHIS to investigate the intensity (hours per week) and duration of employment among SSDI recipients, and identify factors associated with employment.

While the SSA has regularly monitored work among SSI beneficiaries (SSA, 2000), little comparable data has been available for working SSDI beneficiaries. This study roughly mirrors the published SSA reports on working SSI beneficiaries. Prior research has identified a core set of person-level factors associated with workforce and program participation (e.g., Bound, Cullen, Nichols & Schimdt, 2001; Favreault, 2002; GAO, 1998; Gruber & Kubik, 2002; Kruse, 1997; Mitchell & Phillips, 2002; NASI, 1996; RTI, 2000; Schechter, 1999; Stern, 1989), including age, gender, race, education, income, marital status, health status, chronic condition, and severity of disability. In this study, working and nonworking beneficiaries are compared on each of these factors, and significant factors are included in a multivariate model predicting employment among SSDI beneficiaries.

Methods

This study is a secondary analysis of the 2001 and 2002 National Health Interview Surveys (NHIS). The NHIS is an ongoing household survey, conducted by the Centers for Disease Control's National Center for Health Statistics (Adams & Benson, 1991; Massey, Moore, Parsons, & Tadros, 1989). Such national surveys offer a timely and cost-effective approach for studying program participation (Apfel, 2000), and have played a crucial role in informing SSA policy.

All population estimates were weighted using U.S. decennial census data. Because of the relatively low rates of workforce participation among SSDI beneficiaries, two panels of the NHIS (2001 and 2002) were merged, and the weights were adjusted accordingly. To address concerns regarding sampling error, the SUDMN CROSSTABS procedure was used to generate standard errors for all prevalence estimates (RTI, 1998). Following a protocol established by the NCHS, estimates with relative standard errors over 30% were considered statistically unreliable and collapsed or flagged in summary tables. Simple group comparisons were tested with a Wald chi-square, using denominator degrees of freedom equal to the number of primary sampling units (PSUs) minus the number of strata. SUDMN LOGISTIC was used for model specification and analysis.

Within the household sample of 117,781 working age adults (age 18-64), 2,619 respondents reported receiving SSDI because of their own disability. A separate survey item asked about current employment status and, not surprisingly, the great majority of SSDI beneficiaries reported that they are neither working nor looking for work (see table 1). However, 174 respondents (6.8%) who said they received SSDI also said they had engaged in paid work during the preceding week. An additional 39 respondents claimed to have a job, but for various reasons didn't work in the preceding week, 26 said they were looking for work, and 7 said they were working without pay.

For both practical (small cell size) and conceptual (ambiguous interpretation) reasons, respondents who claimed to be employed but not working, seeking work, or doing volunteer work were omitted from subsequent comparisons. Bivariate and multivariate comparisons focused on SSDI beneficiaries who reported working in the past week and those who did not work or seek work in the past week (N=2,546).

Results

An estimated 260,000 beneficiaries reported working for pay in the past week, and most respondents also specified the amount of hours they worked. As seen in Table 2, nearly two thirds of respondents worked 25 hours or less. However, 21.4% reported working over 35 hours per week.

Table 3 shows a reduced logistic regression model that predicts employment among SSDI beneficiaries. There are a number of significant sociodemographic differences in rates of work. Men are slightly more likely than women to work ([beta]= 0.4, t=1.9, p=0.06), and Hispanics ([beta]=-1.1, t=2.8, p=0.02) are less likely to work than non-Hispanic whites; while African Americans and other racial or ethnic groups did not differ from whites. The likelihood of working declines with age: beneficiaries under 35 ([beta]=1.1, t=3.3, p=0.001), age 35 to 44 ([beta]=0.9, t=3.6, p=0.001), and those age 45-54 ([beta]=0.8, t=3.3, p=0.001) all had higher rates of employment than beneficiaries age 55 to 64. Beneficiaries who were never married ([beta]=0.8, t=3.4, p=0.001) or were divorced ([beta]=0.8, t=3.3, p=0.001) were more likely to work than married beneficiaries. Beneficiaries in the Midwestern ([beta] =0.8, t=3.1, p=0.002) or Northeastern U.S. ([beta] 0.6, t=2.2, p=0.023) were more likely to work than beneficiaries in the South.

Higher socioeconomic status was generally associated with higher rates of employment. College graduates were over twice as likely to work as those who did not graduate from high school ([beta]=1.3, t=4.3, p<0.001). Recipients with incomes over the federal poverty level were more likely to work than those below the poverty level ([beta]=0.9, t=3.5, p=0.001). Recipients without health insurance were more likely to work than those with health insurance ([beta]=1.0, t=2.8, p=0.005). Health and disability factors were most closely associated with work status. Recipients who were in good, very good or excellent health were much more likely to work than those in fair or poor health (13=0.9, t=4.7, p<0.001). Those without mobility limitations ([beta]=1.0, t=4.0, p<0.001) and those with developmental disabilities ([beta]=1.0, t=3.4, p=0.001) were also more likely to work.

Discussion

The analysis confirms that a small, but economically and politically important portion of SSDI recipients appear to be actively engaged in paid work, despite strict program eligibility criteria premised on "the inability to engage in any substantial gainful activity (SGA) because of a medically determinable physical or mental impairment(s)" (SSA, 2005). Survey responses suggest that most of this work is part time, due either to the limited work capacity of beneficiaries or the desire to stay below income limits for SSDI eligibility. However, the fact that nearly 7% of current beneficiaries are demonstrably willing and able to work suggests that recent program changes and new initiatives could encourage these beneficiaries to leave the program and return to the workforce.

Many of the factors associated with workforce participation among SSDI beneficiaries are consistent with employment patterns observed among other adults with disabilities. For example, workers tend to be younger and male. There are marked regional differences in employment, with SSDI beneficiaries in the Midwestern U.S. working at over twice the rate of Southern beneficiaries. Single adults (except for those who are widowed) work at higher rates, and this association remains statistically significant even after controlling for associated factors like age, suggesting that spousal income and/or social support factors into employment decisions for this population.

The relatively low rate of employment among Hispanic beneficiaries, compared to other racial and ethnic groups, is provocative. However, low cell size prevents more fine-grained analysis of this disparity. Preliminary comparisons (not shown) found no " significant difference in the average age or health status of SSDI beneficiaries in each of these groups. Additional research should assess various cultural and economic factors associated with race and ethnicity.

The relationship between employment and socioeconomic status is complex. Beneficiaries with higher levels of education were more likely to work, presumably due to the greater earning capacity of this group. Beneficiaries with family incomes below the federal poverty level were less likely to work, though it is unclear if this reflects the propensity of beneficiaries with higher incomes to enter the workforce, or the effect of having the SSDI benefit as the sole source of income.

Lack of health insurance is typically associated with lower socioeconomic status in the general population, so in this sense the higher rates of workforce participation are inconsistent with the relationships found between income, education, and employment. However, among SSDI beneficiaries, the lack of health insurance is a crude marker for recent program eligibility (i.e., SSDI recipients are eligible for Medicare, but only after a 24 month waiting period). The higher rate of paid work (nearly twice 4 that of insured beneficiaries) could represent a recency effect: newly awarded beneficiaries are more likely to maintain ties with former employers or the general labor market. This could strengthen the case for early intervention programming of the type advocated by Berkowitz (2002) and other analysts.

Poor health and mobility limitation, not surprisingly, are associated with lower rates of workforce participation among SSDI beneficiaries. Appropriate medical and rehabilitative care, including assistive technology, could maintain or improve health and reduce limitations, improving quality of life and increasing the likelihood of return to work.

A notable finding was the strong positive association between employment and developmental conditions such as mental retardation. vocational services, like supported employment, are relatively common for adults with developmental disabilities, and this apparently translates into higher rates of paid work for SSDI beneficiaries with developmental conditions. Improving access to vocational services for beneficiaries with other conditions may translate to higher levels of workforce participation.

Policy Implications

There has been a paradigm shift in American disability policy that recognizes the need for full community participation by persons with disabilities, beginning with active participation in the labor market. A number of policy changes have been instituted in an attempt to increase employment among SSDI beneficiaries, including a trial work period, continuation of medical coverage, vouchers for vocational services, and extended period of eligibility. The data presented here, along with previously cited disenrollment data (GAO, 2001) and recent employment rates among disabled workers (Maher, 2005), suggest that these attempts have so far met with limited success. Yet the estimated 260,000 working beneficiaries also provide concrete evidence that employment and eventual disenrollment is a viable SSDI program goal. Moreover, greater scrutiny of this subpopulation can inform early intervention programming which diverts work-capable applicants from enrolling in the first place.

The finding that enrollees who engage in paid work tend to be younger and healthier should surprise no one, and it follows that program initiatives to encourage employment will tend be more attractive to this subset of beneficiaries. Consequently, the success of new program initiatives (greater rates of disenrollment for younger and healthier beneficiaries) will likely raise questions about program equity and about the rigor of initial eligibility determination procedures.

The 1998 Amendments to the Rehabilitation Act clearly make a commitment to serve those with the most significant disabilities first. Eligibility requirements for SSDI assure that beneficiaries have work disabilities, typically making them eligible for vocational rehabilitation services through the state-federal program.

However, those with less significant disabilities might not actually receive services. The "creaming" of less severely disabled SSDI recipients into work programs (Etzoni, 1994) could reduce total SSA program costs (or at least growth of program costs) and offer all beneficiaries greater choice in the type and duration of program participation. Programs such as TWIIAA and related work incentive programs offered through the Social Security Administration could help to create a bifurcated system through which individuals with both significant and less significant disabilities receive vocational rehabilitation services that lead to employment.

Part-time employment is common among workers with disabilities in the u.s. (Hotchkiss, 2004). Particularly for SSDI beneficiaries, this may be a rational response on to current program rules--if an individual earns more that $830 per month (in 2005) for 12 months (nine months of Trial Work Period plus a three month grace period), cash benefits cease (Social Security, 2005). However, the decision to enter SSDI and the decision to remain in the program also reflect social and economic barriers to employment that are particularly daunting for adults with significant disabilities (Krause, 2003). For example, experiences of work discrimination are reported among roughly 10% of working age adults with disabilities, but discrimination experiences among SSDI recipients were nearly three times (27%) as common (Kennedy & Olney, 2001).

Lack of services is also a problem for this population. In particular, vocational rehabilitation has proven to have a positive impact on return to work, but only the minority SSDI beneficiaries have received these services (Rogers, Crystal & Bishop, 2005). Access to health insurance can also be a problem. For example, many states have established "Medicaid Buy-In" programs that allow beneficiaries to return to work and maintain Medicaid eligibility (Williams, Claypool, & Crowley, 2005), but the integration of program benefits for dual eligibles is a serious challenge for administrators and program participants (Nemore, 2004). Implementation of the new Medicare prescription drug benefit may prove particularly problematic, because Medicaid drug benefits are an important and heavily used part of these programs (Liu, Ireys, White, & Black, 2005).

Study Limitations

Population based surveys like the NHIS are a useful tool for assessing population prevalence and subpopulation variance, but all rely on voluntary self-reporting of complex personal information, and are therefore prone to a variety of general threats to internal validity such as recall bias and social desirability (Babbie, 1995). This study should be viewed as a preliminary effort to investigate the scope of paid work among SSDI- beneficiaries and the characteristics of workers.

The selection criteria used for identifying SSDI beneficiaries and the work status of these beneficiaries are of particular concern, given the widely recognized problems with self-report of program participation (Maag & Wittenburg, 2003) and work disability (Kate, 2003). The phrasing of these questions was straightforward, but there was no way to systematically verify program participation or employment status (e.g. examination of recent check stubs or tax forms). For example, despite explicitly linking the SSDI program to receipt of Social Security benefits, it is hypothetically possible that some respondents could have confused the SSDI program with the SSI program, which has a similar acronym but very different set of program policies and procedures regarding employment and health insurance coverage.

Given the explicit limits on substantial gainful employment in the SSDI program, beneficiaries may also have been reluctant to report paid work in the previous week. If some respondents chose not to tell a government-funded interviewer about pay received (particularly "under the table" cash payments), then this study would underestimate the prevalence of paid work. Analysis of earnings data (e.g. through tax records) would improve the internal validity of earnings overall, but would not address the issue of unreported earnings.

Finally, statistical power (i.e., sample size) was an issue in this study. Despite the merging of two panels of data, the sample size of working beneficiaries was quite low (N=174), and this precluded extensive follow-up analysis of intriguing results like the racial and ethnic disparities in work rates. SSA program record reviews and beneficiary interviews should conducted to independently verify and extend these findings.

Conclusion

The SSDI program is, and will continue to be, a vital source of income and health insurance for American workers with disabilities. However, the fact that so few program beneficiaries leave the program rolls and return to the workplace is inconsistent with the broader policy goals of social and economic participation for persons with disabilities. The fundamental challenge is to identify and encourage adults with disabilities who are willing and able to return to the workforce, given the proper supports and incentives (Kaye, 2003). The SSDI program has attempted to limit program eligibility to those who cannot work, but this study suggests that nearly 7% percent of beneficiaries are engaged in paid work. This suggests that the eligibility criteria do not match the capacity of some recipients: at least 260 thousand beneficiaries are demonstrably capable of working.

From the broader perspective of increasing participation opportunities for Americans with disabilities through responsive and personalized programming, this group offers hope to reformers. New program incentives from the TWIIAA could increase the proportion of working beneficiaries, at least in the short term, though this should eventually translate into greater rates of disenrollment. Finally, early intervention programming could offer an alternative to workers with disabilities who need support, but not necessarily the entire package of program benefits and associated restrictions on earnings.

Acknowledgements

This study was funded by an award from the 2003 Upjohn Institute grant program. The analysis developed from earlier work on a subcontract awarded by the US Social Security Administration to the Disability Research Institute at the University of Illinois. The authors wish to acknowledge the contributions of Mark Newsom, Chrisann Schiro-Geist, and Theresa Richer in our preliminary work on SSDI beneficiaries using the National Health Interview Surveys. The analysis and interpretation of this survey data is my own, and does not represent the policies or positions of the Upjohn Institute, The Disability Research Institute, The National Center for Health Statistics, or the Social Security Administration.

References

Adams, P. & Benson, V. (1991). Current estimates from the National Health Interview Survey. Vital Health Stat, 10 (181).

Apfel, K. S. (2000). Mastering the challenge. Washington, D.C.: Social Security Administration. Babbie E. (1995). The practice of social research, 7th edition. Belmont: Wadsworth Publishing.

Berkowitz, E. (1997). The historical development of Social Security in the United States. In E. Kingson and E. Shulz (eds.) Social Security in the 21st century. New York: Oxford University Press.

Berkowitz, M. (2002, January). The selection of Social Security Disability applicants for an early intervention program: Identifying probable beneficiaries who are likely to return to work. Urbana: DRI.

Bound, J., J. Cullen, A. Nichols, & L. Schimdt (2001, May). The welfare implications of increasing DI benefit generosity. Washington DC: Annual Conference of the Retirement Research Consortium.

Burkhauser, R., Haverman, R., & Wolfe, B. (1993). How people with disabilities fare when public policies change. Journal of Policy Analysis and Management, 12(Spring), 252-269.

Burkhauser, R. V., Butler, J. S., & Weathers, R. R., 2nd. (2001). How policy variables influence the timing of applications for Social Security Disability Insurance. Social Security Bulletin 64(1), 52-83.

Catalano, R. & Kennedy, J. (1998). The effect of unemployment on disability caseloads in California. Journal of Community and Applied Social Psychology 8, 137-144.

Dale, S. & Verdier, J. (2003). Elimination of Medicare's waiting period for seriously disabled adults: impact on coverage and costs. Commonwealth Fund Issue Brief July, NYC: Task Force on the Future of Health Insurance.

Etzioni, A. (1994, July). Incorrigible. Atlantic Monthly, 14-15.

Favreault, M. (2002, December). Forecasting incidence of work limitations, disability insurance receipt, and mortality in dynamic simulation models using Social Security administrative records: a research note. Center for Retirement Research Working Paper 2002-09, Boston College.

Gruber, J. & Kubik, J. (2002, September). Health insurance coverage and the disability insurance application decision. Center for Retirement Research Working Paper 2002(04), Boston College.

Hotchkiss, J. L. (2004). Growing part-time employment among workers with disabilities: Marginalization or opportunity. Economic Review, 89(3), 25-41.

Khan, J., Gerdtham, U. G., & Jansson, B. (2004). Effects of macroeconomic trends on social security spending due to sickness and disability. American Journal of Public Health 94(11), 2004-2009.

Kaye, S. (1997). Disability watch: The status of people with disabilities in the United States. Volcano, CA: Volcano Press.

Kaye, S. (2003). Improved employment opportunities for people with disabilities. Disability Statistics Report, 17, UC San Francisco: Disability Statistics Center.

Kennedy, J., M. Olney & C. Schiro-Geist (2004). A national profile of SSDI applicants: implications for early intervention. Journal of Disability Policy Studies. 15{3), 178-185.

Kennedy, J., & Olney, M. (2001). Job Discrimination in the Post-ADA Era: Estimates from the 1994 and 1995 National Health Interview Surveys. Rehabilitation Counseling Bulletin 45(1), 24-30.

Kruse, D. (1997, September). Employment and disability: Characteristics of employed and non-employed people with disabilities. New Brunswick: Rutgers University.

Lando, M., Coate, M., & Krause, R. (1979). Disability benefit applications and the economy. Social Security Bulletin, 42(10), 3-10.

Lewin-VHI, Inc. (1995, October 31). Longer term factors affecting disability program applications and awards. Report prepared by D. Stapleton, K. Dietrich, J. Furman, & G. Lo. Fairfax, VA: Lewin-VHI.

Liu, S., Ireys, H., White, J., & Black, W. (2005). Enrollment patterns and medical expenditures for Medicaid Buy-in participants in five states. Washington DC: Mathematica.

Maag, E. & Wittenburg (2003, May). Real trends or measurement problems? Disability and employment trends from the Survey of Income and Program Participation. Washington DC: Urban Institute.

Marini, I. & Reid, C. (2001). A survey of rehabilitation professionals as alternate provider contractors with Social Security: problems and solutions. Journal of Rehabilitation. 67(2), 36-44.

Massey, J., Moore, T., Parsons, V., & Tadros, W. (1989). Design and Estimation for the National Health Survey, 1985-94. Vital and Health Statistics, 2(110).

Mitchell, O. & Phillips, J. (2002, May 18). Eligibility for Social Security Disability Insurance. Washington, DC: Joint Conference for the Retirement Research Consortium.

National Academy of Social Insurance (1996). Balancing security and opportunity: The challenge of disability income policy. Mashaw & V. Reno (eds). Washington, DC: NASI.

National Council on Disability. (1997). Return to work for people with disabilities: Action proposals for the 105th Congress. Washington, DC.

Nemore, P. (2004). Dual eligibles. Issue Brief; Center for Medicare Education, 5(2), 1-6.

O'Day, B. (1999). Policy barriers for people with disabilities who want to work. American Rehabilitation 25(1),8-19.

Research Triangle Institute. (December 2000). Analysis of the National Health Interview Survey: The vocational rehabilitation perspective. Research Triangle Park, NC: Author.

Research Triangle Institute. (1998). Software for the Statistical Analysis of Correlated Data (SUDAAN); Version 7.5) [Computer Software]. Research Triangle Park, NC: Author.

Riley, G. F., Lubitz, J. D., & Zhang, N. (2003). Patterns of health care and disability for Medicare beneficiaries under 65. Inquiry, 40(1), 71-83.

Rogers, J. B., Crystal, R. M., & Bishop, M. (2005). Predicting rehabilitation outcome for Supplemental Security Income and Social Security Disability Income recipients: Implications for consideration with the Ticket to Work program. Journal of Rehabilitation, 71(3), 5-10.

Rupp, K., & Stapleton, D. (1995). Determinants of the growth in the Social Security Administration's disability programs--an overview. Social Security Bulletin, 58(4), 43-70:

Schechter, E. (1999). Industry, occupation, and disability insurance beneficiary work return. Social Security Bulletin, 62(1 ), 10-22.

Sim, J. (1999) Improving return-to-work strategies in U.S. Disability programs, with analysis of program practices in Germany and Sweden. Social Security Bulletin, 59(3), 41-49.

Social Security Administration (SSA) (2000). SSI Disabled Recipients Who Work--December 2000. Washington DC: SSA.

Social Security Administration. (2004). Annual Statistical Supplement, 2003. Washington DC: SSA.

Social Security Administration (SSA) (2001). Report on Supplemental Security Income and resource exclusions and disability insurance earnings related provisions. Washington DC: SSA.

Social Security Administration (SSA) (2005). Red Book: A summary guide to employment support for individuals with disabilities under the Social Security Disability Insurance and Supplemental Security Income programs (SSA Pub. No. 64-030/ICN 436900). Baltimore, MD: Project ABLE.

Stern, S. (1989). Measuring the effect of disability on labor force participation. Journal of Human Resources, 24(3), 361-395.

U.S. General Accounting Office (GAO) (1998). Social Security Disability Insurance: Factors affecting beneficiaries' return to work. Washington, D.C.: Government Printing Office.

U.S. General Accounting Office (GAO) (1999). Multiple factors affect return to work. Washington, D.C.: Government Printing Office.

U.S. General Accounting Office (GAO) (2001). Other programs may provide lessons for return to work efforts. Washington, D.C.: Government Printing Office.

Williams, B., Claypool, H., & Crowley, J. S. (2005). Keeping Medicare and Medicaid when you work." A resource guide for people with disabilities, their families, and their advocates. Menlo Park: Kaiser Family Foundation.

Ycas, M. A. (1995). The issue unresolved: innovating and adapting disability programs for the third era of Social Security. Social Security Bulletin, 58(1), 48-56.

Yelin, E. (1989). Displaced concern: the social context of the work-disability problem. Milbank Quarterly, 67(2), 114-165.

Yelin, E. (1992). Disability and the displaced worker. New Brunswick, NJ: Rutgers University Press. Yelin, E. H., & Katz, P. P. (1994). Making work more central to work disability policy. Milbank Quarterly, 72(4), 593-619.

Jae Kennedy

Washington State University

Marjorie F. Olney

San Diego State University

Jae Kennedy, Ph.D., Associate Professor, Department of Health Policy Administration, Washington State University.

Table 1. Self-reported employment status of SSDI beneficiaries

                              sample n   est. N (a)   col %

Current employment status       2,619     3,795.5     100.0%
of SSDI recipients

  Working for pay at a            174       259.7       6.8%
  job or business

  Looking for work                 39        52.0       1.4%

  With a job or business           26        42.0       1.1%
  but not at work

  Working, but not for pay,         7        14.0       0.4%
  at a job or business

  Not working and not           2,372     3,425.4      90.2%
  looking for work

Source: Merged data from the core 2001 and 2002 National
Health Interview Surveys (NCHS, 2003)

(a) population estimates in 1000's

(+) estimate unstable - relative standard of error
(std. error/estimate) > 30%

Table 2. Hours of work reported by DI beneficiaries

                                           est. N (a)   col %

Number of hours worked in the past week      251.8      100.0%
  1-15                                        88.1       35.0%
  16-25                                       86.2       34.2%
  26-35                                       23.6        9.4%
  over 35                                     53.9       21.4%

Source: Merged data from the core 2001 and 2002 National
Health Interview Surveys (NCHS, 2003)

(a) population estimates in 1000's

Table 3. Logistic regression model predicting employment
among DI beneficiaries

Independent variables              Total DI        Working DI
                                   Beneficiaries   beneficiaries

                                   est. N (a)      est. N (a)   row %

Model intercept

Age
  18-34                              340            59          17.2%
  35-44                              661            73          11.1%
  45-54                            1,117            82           7.4%
  55-64                            1,567            45           2.9%

Sex
  female                           1,735           113           6.5%
  male                             1,950           146           7.5%

Race/ethnicity
  white                            2,626           193           7.3%
  Hispanic                           287             7           2.6%
  black                              673            52           7.7%
  other                               99             8           8.2%

Education
  did not graduate from            1,138            53           4.6%
    high school
  high school graduate or GED      1,284            98           7.7%
  some college                       655            42           6.4%
  college graduate                   531            62          11.6%

Income at or below poverty level
  no                               2,933           227           7.7%
  yes                                752            33           4.4%

Has health insurance
  no                               143.7           19.3         13.4%
  yes                              3,528           240           6.8%

Marital status
  married                          1,662            69           4.2%
  divorced or separated              899            65           7.2%
  never married                      829           115          13.9%
  widowed                            277            10           3.4%

Health status
  fair-poor                        2,355            86           3.7%
  excellent-good                   1,320           172          13.0%

Needs help w/one or more ADLs
  no                               3,136           235           7.5%
  yes                                548            24           4.4%

Diagnosed with developmental
    condition (b)
  no                               3,397           201           5.9%
  yes                                288            59          20.5%

Region
  Southern U.S.                    1,600            76           4.7%
  Northeastern U.S.                  694            63           9.1%
  Midwestern U.S.                    816            86          10.5%
  Western U.S.                       575            35           6.1%

Total model fit
-2 normalized log likelihood
  (full model)
Wald F (full model - intercept)

Independent variables              Model Coefficients

                                   Beta    SE     t-test   p

Model intercept                    -5.04   0.48   -10.56   0.000

Age
  18-34                             1.17   0.31   3.74     0.000
  35-44                             1.00   0.26   3.90     0.000
  45-54                             0.85   0.24   3.58     0.000
  55-64                             0.00   0.00.

Sex
  female                            0.00   0.00.
  male                              0.32   0.19   1.73     0.084

Race/ethnicity
  white                             0.00   0.00.
  Hispanic                         -1.10   0.40   -2.77    0.006
  black                             0.17   0.23   0.74     0.461
  other                             0.23   0.64   0.36     0.716

Education
  did not graduate from             0.00   0.00.
    high school
  high school graduate or GED       0.43   0.25   1.71     0.088
  some college                      0.31   0.29   1.08     0.281
  college graduate                  1.10   0.29   3.74     0.000

Income at or below poverty level
  no                                0.00   0.00.
  yes                               0.87   0.25   3.47     0.001

Has health insurance
  no                                1.02   0.36   2.82     0.005
  yes                               0.00   0.00.

Marital status
  married                           0.00   0.00.
  divorced or separated             0.72   0.23   3.20     0.002
  never married                     0.85   0.24   3.55     0.001
  widowed                           0.46   0.50   0.92     0.357

Health status
  fair-poor                         0.00   0.00.
  excellent-good                   -0.99   0.19   -5.12    0.000

Needs help w/one or more ADLs
  no                                0.00   0.00.
  yes                              -0.75   0.32   -2.32    0.021

Diagnosed with developmental
    condition (b)
  no                                0.00   0.00.
  yes                               0.96   0.28   3.40     0.001

Region
  Southern U.S.                     0.00   0.00.
  Northeastern U.S.                 0.59   0.27   2.18     0.030
  Midwestern U.S.                   0.79   0.26   3.01     0.003
  Western U.S.                      0.34   0.29   1.17     0.242

Total model fit
-2 normalized log likelihood                      1037.6   0.000
  (full model)
Wald F (full model - intercept)                   10.6     0.000

Source: Merged data from the core 2001 and 2002 National Health
Interview Surveys (NCH5, 2003)

(a) population estimates in 1000s

(b) conditions included are mental retardation, birth defects,
or other developmental conditions causing limitation