Abstract

Public benefits are widely used by people with intellectual and development disabilities (IDD) as crucial financial supports. Using Rehabilitation Service Administration 911 and Annual Review Report datasets to account for individual and state vocational rehabilitation (VR) agency variables, a sample of 21,869 people with IDD were analyzed using hierarchical linear modeling to model the effects of public benefits on hours worked per week. Findings point to associations that indicate that public benefits not only limit access to employment participation, they also have a restricting effect on growth of weekly hours that typically come with higher wage positions, compared those that do not access benefits. The article also lays out important implications and recommendations to increase the inclusion of people with IDD in the workplace.

There is growing attention on the importance of increasing the employment participation of individuals with intellectual and developmental disabilities (IDD). Numerous federal and state policies and programs have been designed to specifically target the promotion of employment; and though, relative to other types of day and employment services, there is significantly lower financial investment in individual employment services in the community (Butterworth et al., 2014). Despite this focus, people with IDD continue to be underrepresented in the general workforce (American Association on Intellectual and Developmental Disabilities & The Arc of the United States, 2008; Migliore, Mank, Grossi, & Rogan, 2007). At the same time, it is known that most adults with IDD seek economic self-sufficiency (Gray, McDermott, & Butkus, 2000) and define community employment as a primary goal (Migliore, et al., 2007). Many factors influence this divide that people with IDD face, including personal, systemic, social, and policy barriers (Nord, Luecking, Mank, Kiernan, & Wray, 2013). The purpose of this article is to explore the influence benefit programs have on the employment experiences of people with IDD.

Public Cash Benefit Programs and Work

Two federal cash benefit programs are widely used and play a critical economic role in lives of many individuals with IDD. Supplemental Security Income (SSI) acts as the federal cash assistance program for the poor and Social Security Disability Insurance (SSDI) acts as the major income replacement program. To qualify for SSI and SSDI programs, both of which are subject to income limits, an individual must have a medically proven disability that prevents the person from achieving work activities and earnings that are substantial and gainful, also known as Substantial Gainful Activity (SGA). In 2014, the SGA threshold was $1,070 per month (Social Security Administration, 2014a). There are nearly 13 million people aged 18–64 years receiving these benefits on the basis of disability (Social Security Administration, 2013).

Supplemental Security Income is a means-tested program providing cash assistance to eligible individuals who are blind, aged, or disabled, including children under the age of 18 (U.S. House of Representatives Committee on Ways and Means, 2011). In 2012, there were a total of 4.9 million working-age people with disabilities that received SSI, resulting in $4.6 billion in SSI cash payments, with an average monthly federal payment of $528.25 (Social Security Administration, 2013). Many states also supplement the federal benefit to increase the cash payment using state tax revenues. Because the SSI cash benefit is designed to supplement income rather than replace income, SSI alone will not raise an individual out of poverty (Iceland, 2012; Kingdon & Shultz, 1997); thus, coupling SSI benefit usage with employment can be an effective poverty-reduction intervention in that SSI cash payments gradually taper to zero as a recipient receives more earned income (Social Security Administration, 2012, 2014b).

The SSDI entitlement program is part of the federal Social Security program under the Old Age, Survivors, and Disability Insurance (OASDI). Eligibility for SSDI is not means tested, rather it is based on an individual worker's or parental contribution into the Social Security system. In 2012, Social Security provided SSDI benefits to 7.9 million recipients that qualified due to a disabling condition; and another 1.4 million recipients also received SSI due to having a limited income (Social Security Administration, 2013).

Individuals with IDD regularly access cash benefits, but how these benefits influence an individual in working and accessing employment services is not well understood. The complex web of benefits available to individuals with IDD, on one hand, provide a much needed financial support but can inadvertently impact participation in employment and, thus, their integration into their workplaces, and ultimately their communities. It has been reported that only 3% to 8% of low-income individuals with disabilities receiving cash benefits were participating in employment programs (Butterworth, et al., 2014; Riley & Rupp, 2012).

Dean, Pepper, Schmidt, and Stern (2014) explored vocational rehabilitation (VR) service usage in an attempt to understand the relationship between employment support services and SSI and SSDI usage in Virginia. Findings revealed that SSI and SSDI recipients were more likely to be unemployed when they entered VR and monthly income at application tended to be below SGA. It was also found that the receipt of VR services were negatively associated with SSI and SSDI uptake overall. However, this result did not hold for those with VR-defined cognitive disabilities, which includes people with IDD. The opposite was observed: SSI and SSDI usage increased faster for this population. For those VR recipients who did receive SSI and SSDI, it was found that a higher level of VR service was required in order for this population to obtain employment.

As we investigate community access and integration with regards to work, a useful measure to consider is the amount of time a person is employed in an individual job. Butterworth et al. (2014) recently found people with IDD in the VR system worked an average of 23.6 hr per week at closure. This is compared to all VR recipients that worked an average 30.3 hr per week.

Research Question

This study sought to expand what is known about the relationship between public benefit usage and employment outcomes of people with IDD by filling a gap related to community access and integration, particularly for those accessing VR services throughout the United States. Studying any outcome within the VR system requires caution. As demonstrated by past research, individual employment outcomes within the VR system differ greatly depending on the state in which a person resides and receive support services (Berry & Caplan, 2010; Yamamota & Alverson, 2013); therefore, these important dependencies must be accounted for to ensure valid results. In response, this study sought to answer the following research questions:

  1. How much variance in individual hours worked is attributable to the state in which people with IDD receive VR services and individual characteristics?

  2. What effect does public cash benefit usage have on work hours, after accounting for individual characteristics, hourly wages, and state-level VR service characteristics?

  3. How does hourly wage moderate the effects of benefit programs on weekly work hours?

Data Sources, Measures, and Definitions

Individual subjects were drawn from Rehabilitation Services Administration case closure data by way of the RSA-911 public access database from fiscal year 2011. The RSA-911 dataset includes individual-level data of VR recipients that experienced a case closure during the fiscal year. Data included in this dataset includes demographic information, benefit usage across various public programs at program entry and closure, VR service provision and funding, employment-related outcomes, and reason for case closure.

The focus of the study required a restricted sample; thus, the inclusion criteria limited subject participation on a number factors. Due to focus of this study on people with IDD specifically, only those with primary or secondary IDD disability cause were included. These included those with an autism, cerebral palsy intellectual disability, epilepsy, and intellectual disability. It was then decided to exclude subjects that received services from VR programs specifically targeting people who are blind or visually impaired, thus retaining only subjects receiving services from the general or combined VR agency. Next, only those with IDD that achieved successful case closure in integrated employment were included, as defined in the dataset as case closure reason and primary or secondary disability source, respectively. This criterion isolated a total of 69,614 subjects with IDD. From this group, those experiencing a case closure before receiving services were removed from the sample because program participation did not occur, resulting in the removal of 46,836 subjects. Subjects receiving services from the United States territories were also removed, resulting in the exclusion of an additional 352 subjects. Last, 558 subjects were removed due to unavailable or extreme outlier data, resulting in a final sample of 21,869 subjects.

Due to the multilevel nature of this study, available state data were also used. These data were obtained from the RSA Annual Review Report (RSA-ARR) that provided state VR system results for fiscal year 2011. The RSA-ARR is produced every fiscal year and provides a state-by-state summary of key data about all services and service users from VR programs. The datasets represented in the RSA-ARR include Annual Vocational Rehabilitation Program/Cost Report (RSA-2) used to report program expenditures and Case Service Report (RSA-911) used to collect information about people exiting the VR program.

The outcome of interest was the number of weekly hours worked at the time of VR case closure. This variable was chosen due to its noted relationship with benefit participation and its importance as a key indicator of employment quality. Additionally, the number of weekly hours serves as a proxy for community integration, where those working more hours in community settings are, by their very nature, integrated at higher levels. Independent variables were chosen to extend previous studies investigating VR service usage (Burgess & Cimera, 2014; Wehman, Chan, Ditchman, & Kang, 2014) and cash benefit programs (Houtenville & Brucker, 2014; Riley & Rupp, 2012). On average, subjects in this study worked 24.14 7(sd = 10.97) hr per week at closure.

Individual Measures: Level-1

Six level-1 main effects were selected for this model. Level-1 descriptive statistics are summarized in Table 1. The six variables and their resulting coding structure can be found on Table 2.

Table 1

Level-1 Characteristics and Benefit Usage (N = 21,869)

Level-1 Characteristics and Benefit Usage (N = 21,869)
Level-1 Characteristics and Benefit Usage (N = 21,869)
Table 2

Main Effects Variables, Definitions, and Coding Scheme

Main Effects Variables, Definitions, and Coding Scheme
Main Effects Variables, Definitions, and Coding Scheme

Finally, an individual's hourly wage was thought to moderate the effects of benefits on weekly hours. In other words, those using benefit programs that earn higher hourly wages were expected to work fewer hours than those with lower wages. This was expected due to existing SGA eligibility criterion and payment relationships that exist between cash benefit programs and monthly earnings. In response, a benefit-usage-by-hourly-wage interaction effect was thought to exist; and a variable was constructed and entered through dummy coding, using the “Neither SSI nor SSDI”-group-by-continuous-hourly-wage interaction as the reference group.

State Vocational Rehabilitation Measures: Level-2

Two state VR agency characteristics are included in the model to control for unique state service make up of people on different benefit programs. These level-2 variables included the percent of total VR service users that received SSI (x-state = 18.8%, sd = 5.8%) and the percent of total users that received SSDI (x-state = 19.3%, sd = 6.2%). Both variables were grand mean centered.

Method and Strategy for Analysis

It has been shown that some outcomes of VR recipients tend to be more similar to others obtaining services in their own state system than would randomly be expected (Berry & Caplan, 2010; Yamamota & Alverson, 2013). Though the extent of this dependency varies by outcome assessed, it is critical to evaluate for it and, if present, account for it or risk increasing Type I error and miss-specifying the effects of important and potentially influential variables. In response, this research employed hierarchical linear modeling (HLM), a structured regression approach that accounted for the natural nesting of individual VR within a state VR system.

This study used a four-step modeling process, each building off the previous model. These included an unconditional model to assess the variance explained by state nesting and appropriateness of HLM, a demographic model to explain variance through individual characteristics, a benefits model to explain variation due to benefit usage, and a benefit-bywage-interaction model to understand the moderating effects of wages on benefits. Normality and homoscedasticity at levels 1 and 2 were evaluated and met. Related variables were entered into the model in distinct blocks, allowing for the assessment of variance explained by these variable groupings. The first model included an unconditional model to evaluate the appropriateness of HLM by assessing the existence of state dependency on the outcome. The remaining included a demographic, wage–and-benefit-usage, and benefit-usage-by-wage interaction model. Predictors were found to be significant at p < .05. The level-1 and level-2 full conditional model equation included the following:

 
formula

Results

Research Question 1: How much variance in individual hours worked is attributable to the state in which people with IDD receive VR services and individual characteristics?

The HLM results can be found in Table 3. As demonstrated by the significant intercept of the unconditional model, the number of hours an individual is employed does depend on the state where they receive VR services. Of the total variance in weekly hours in these data (X2 = 124.22), 12% can be partitioned out and attributed to state nesting, denoted by the intra-class correlation (). Not only does this model confirm that individual outcomes depend on the state they receive services, it also confirms the appropriateness of HLM in understanding and predicting weekly hours of VR service recipients.

Table 3

Unconditional and Demographic Models of Predictors of Weekly Hours (Nwithin state = 21,868 & Nbetween state = 51)

Unconditional and Demographic Models of Predictors of Weekly Hours (Nwithin state = 21,868 & Nbetween state = 51)
Unconditional and Demographic Models of Predictors of Weekly Hours (Nwithin state = 21,868 & Nbetween state = 51)

Next, the demographic model shows all demographic predictors were found to be significant (see Table 3). Education greater than the grand mean had a positive effect hours, whereas age over the grand mean and the presence of a significant disability had a negative effect on hours (p < .001). Those determined to have cognitive impairments as their primary impairment, on average, worked significantly fewer hours than people with other impairments. This model explained a total of 2% of level-1 variance.

Research Question 2: What effect does public cash benefit usage have on work hours, after accounting for individual characteristics, hourly wages, and state-level VR service characteristics?

The third model introduces individual benefit usage and wages, as well as VR agency benefit variables into the model (see Table 4). The effects of age and having a psychosocial or other mental impairment and physical impairment dissipated, with the inclusion of the new variables. Additionally, hourly wages higher than the grand mean were found to have a significantly positive effect on weekly hours (p < .001). The benefit usage variable indicated that those on SSI, SSDI, or both SSI and SSDI worked significantly lower hours when compared to the reference group, those that did not receive these benefits (all sig. at p < .001). Holding all else constant, those receiving SSI averaged nearly 11 fewer hours than the reference group, whereas those on SSDI or both SSI and SSDI averaged nearly 8 fewer hours per week. At level-2, individuals receiving services from VR agencies serving a greater proportion of SSDI recipients, on average, experienced lower weekly hours. This model accounted for 24% of all variance, a considerable amount. Specifically, 20% of the level-1 variance and 50% of the level- 2 variance was explained.

Table 4

Benefits and Wage, and Interaction Models of Predictors of Weekly Hours (Nwithin state = 21,868 & Nbetween state = 51)

Benefits and Wage, and Interaction Models of Predictors of Weekly Hours (Nwithin state = 21,868 & Nbetween state = 51)
Benefits and Wage, and Interaction Models of Predictors of Weekly Hours (Nwithin state = 21,868 & Nbetween state = 51)

Research Question 3: How does hourly wage moderate the effects of benefit programs on weekly work hours?

The final model retains previously entered variable and incorporates the benefits by hourly wage interaction effect. The effects of education, having a significant disability, and primary impairment remained significant in this model (all sig. at p < .001). With the inclusion of the interaction term, benefit usage becomes a conditional effect and its interpretation changes. Assessment of the interaction variable found a moderating effect does exist. Specifically, the weekly hours of those receiving SSDI and both SSI and SSDI were found to be significantly influenced by hourly wage. The lack of a significant interaction of the SSI-only group may be a result of limited power, due to a small sample of SSI-only recipients. This model explains 25% of all variance in weekly hours. Level-1 variables account for 19% of level-1 variance, and the level-2 variables explain 51% of available level-2 variance.

Using the interaction model, predicted weekly hours of VR recipients with IDD across different hourly wages were graphed. As shown in Figure 1, after allowing hourly wages to vary while holding all else constant, those accessing benefit programs, on average, worked considerably fewer hours than those not receiving benefits, regardless of the hourly wage they were paid. Those earning the federal minimum wage, $7.25, and receiving SSDI, worked an average of 18.7 hr per week; and those on SSDI and SSI worked 18.3 hr per week. The outcomes of people on benefit programs were far lower than those that received no benefits, who averaged 25.8 hr of work per week. As hourly wages increased, so too did the weekly hours of reference group. Those earning $12 per hour worked an average of 31.6 hr per week, an increase of nearly 6 hr per week. Contrasted with benefit users earning $12 per hour, the SSDI users averaged 20.9 hr, denoting a nearly 2-hr increase, and SSI and SSDI users averaged 18.3 hr per week, denoting a 0.4 hr increase.

Figure 1. 

Predicted weekly hours of vocational rehabilitation (VR) recipients by benefit program, moderated by hourly wage. Note. SSI = Supplemental Security Income; SSDI = Social Security Disability Insurance. Predicted values are based on the following held constant: Education = grand mean, age = grand mean, significant disability = 1, primary impairment = cognitive, hourly wage = grand mean.

Figure 1. 

Predicted weekly hours of vocational rehabilitation (VR) recipients by benefit program, moderated by hourly wage. Note. SSI = Supplemental Security Income; SSDI = Social Security Disability Insurance. Predicted values are based on the following held constant: Education = grand mean, age = grand mean, significant disability = 1, primary impairment = cognitive, hourly wage = grand mean.

Discussion

Limitations

This research presents a number of limitations. The reliance on the RSA administrative datasets creates unique challenges. As a panel dataset, capturing all people that experienced case closure and all state agencies providing services during a specific fiscal year, it is not possible to determine causal relationships. Instead, this study highlights correlational findings. Additionally, because its primary purpose is to capture programmatically relevant data, some important variables may not be available to the researcher or defined in a manner that is ideal for comparisons. For example, important services and supports that people with IDD often rely on outside of the non-RSA services system are not identified or accounted for in this study. Another example is the disability-specific variables. Another example is the RSA-911 dataset does not articulate the significance of a subject's disability, limiting the utility of cross-disability comparisons. National administrative datasets may also encounter differences in data entry procedures across states. The extent of such states differences are unknown and, as a result, an unknown amount of error is likely entered into the model.

Finally, the outcome variable selected, the typical hours worked per week at closure, does not extend past the 90 days postemployment placement nor does it assess integration outside of the work environment. As a result, this study is unable to evaluate the relationship of the outcome and benefit usage beyond this narrow time period. It is also unable to evaluate community integration in its entirety.

Multilevel Nature of RSA Outcomes

The RSA-911 dataset is popular among researchers in the field of employment and disability services. However, despite the major differences in implementation of VR services across states, the majority of research using this dataset ignores these critical state differences. This study adds to the small but growing evidence base showing that RSA outcomes depend on the state that an individual receives services and supports (Berry & Caplan, 2010; Yamamota & Alverson, 2013). The picture beginning to emerge from this multilevel research is that state matters. As shown in this study by way of the intraclass correlation, 12% of the total variability of individual weekly hours was partitioned to the state; and over half of this level-2 variance was explained by state-level VR variables. As researchers continue to make sense of the outcomes of VR services through regression. it is imperative to account for the state systems by employing multilevel modeling or risk model misspecification.

Public Benefits Impede Full Community Access and Integration

Employment in community settings can be a primary outlet for community integration and inclusion for people with IDD (Hall, 2009). Unfortunately, many people with IDD continue to experience minimal work hours in individual community jobs (Butterworth et al., 2014; Kas & McKimmie, 2014), resulting in large spans of time in segregated day programs or isolated at home alone (Kas and McKimmie, 2014). This study adds to this line of research by demonstrating that benefit usage plays a role in restricting community integration and inclusion that employment can provide by affecting the amount of hours individuals work in the community. Thus, to improve inclusion of people with IDD working, as new support and benefit interventions are considered, strategies that expand work hours are necessary.

Modernizing Benefit Programs

As this study showed, people with IDD in the VR system experience disparate outcomes depending on their use of public benefits. A notable result was the large percent of people with IDD receiving no benefits (52%) and the high level of work hours of this no-benefits group, regardless of disability. These findings challenge the misconceptions that people with IDD are unable to work at levels comparable to the general workforce (U.S. Census Bureau, 2014). Yet, as also shown by this study, there is a suppressing effect of benefit programs on employment outcomes for people with IDD. Benefit use had a negative effect on hours worked. Additionally, it was found that work hours were predicted to increase in jobs where an individual was paid a higher hourly wage. However, these results were not equitable across the different benefit user groups; specifically, those receiving benefits experienced near zero growth in weekly hours as wages increased, compared to a marked increase in weekly hours for those receiving no benefits.

In considering areas ripe for future research and potential reform, researchers need to include the work incentives available to encourage federal benefit program recipients. Work incentives can provide enhanced opportunities for public benefit recipients to work and earn more. Though this research did not address the role of work incentives, it is an important area for future research in relation to hours worked. Existing literature indicates that 5,683 individuals with IDD accessed work incentives in 2012 by way of a Plan for Achieving Self-Support, Impairment-Related Work Expense, or Blind Work Expense (Butterworth, et al., 2014). Based on this dismal uptake, changes are clearly needed to expand the usage of benefit incentives to facilitate greater workforce participation in the United States. The scarcity of research on incentives interventions has led to a void in developing useful policy alternatives. A recent randomized experiment in Norway targeted disability insurance policies to encourage higher levels of employment. Such policies, coupled with financial work incentives to reduce the negative impact of work on benefits, can have notable positive effects on labor force participation and can decrease program costs for those furthest from retirement age (18–49 years). Though these findings should be applied with caution, it is important to consider encouragement and financial-incentive models as areas requiring attention by the policy and research communities (Kostl & Mogstad, 2014).

Conclusion

Public benefits play an important role in the lives of many people with IDD. To qualify individuals must undergo an extensive medical review to demonstrate permanent disability and meet specific thresholds related to income or work experience. Once accepted into the benefit programs, the perceived complexities, coupled with the personal financial consequences of potentially losing one's cash benefit, appear to dissuade many people from maximizing their employment potential. In current form, as this study demonstrates, decisions about accessing these benefits have potentially enormous personal, financial, and social costs over a lifetime. As the field grapples with a comprehensive approach to improve employment outcomes, there is a need for an honest assessment of benefit programs and the negative and positive effects they have on the employment of people with IDD. Ignoring the effects of these programs will stunt progress in achieving better outcomes in the long run.

Acknowledgments

This publication is supported by Cooperative Agreement #H133B080005 from the National Institute on Disability Independent Living and Rehabilitation Research (NIDILRR), U.S. Department of Health and Human Services.

References

References
American Association on Intellectual and Developmental Disabilities & The Arc of the United States
. (
2008
).
Employment: Joint position statement
.
Berry
,
H. G.
,
&
Caplan
,
L. J.
(
2010
).
Employment and earnings growth among transition-age Supplemental Security Income Program participants
.
Journal of Disability Policy Studies
,
21
(
3
),
151
159
.
Burgess
,
S.
,
&
Cimera
,
R.
(
2014
).
Employment outcomes of transition-aged adults with Autism Spectrum Disorders: A state of the states report
.
American Journal on Intellectual and Developmental Disabilities
,
119
(
1
),
64
83
.
Butterworth
,
J.
,
Smith
,
F. A.
,
Hall
,
A.C.
,
Migliore
,
A.
,
Winsor
,
J.
,
&
Domin
,
D.
(
2014
).
StateData: The national report on employment services and outcomes
.
Boston, MA
:
University of Massachusetts Boston, Institute for Community Inclusion.
Dean
,
D.
,
Pepper
,
J. V
,
Schmidt
,
R. M.
,
&
Stern
,
S.
(
2014
).
State vocational rehabilitation programs and federal disability insurance: An analysis of Virginia's vocational rehabilitation program
.
Journal of Labor Policy
,
3
(
7
),
1
19
.
Gray
,
B.
,
McDermott
,
S.
,
&
Butkus
,
S.
(
2000
).
Effect of job coaches on employment likelihood for individuals with mental retardation
.
Journal of Vocational Rehabilitation
,
14
,
5
11
.
Hall
,
S. A.
(
2009
).
The social inclusion of people with disabilities: A qualitative meta- analysis
.
Journal of Ethnographic and Qualitative Research
,
3
,
162
173
.
Houtenville
,
A. J.
,
&
Brucker
,
D. L.
(
2014
).
Participation in safety-net programs and the utilization of employment services among working-age persons with disabilities
.
Journal of Disability Policy Studies
,
25
(
2
),
91
105
.
Iceland
,
J.
(
2013
).
Poverty in America: A handbook
.
Berkeley and Los Angeles, CA
:
University of California Press
.
Kas
,
S.
,
&
McKimmie
,
H.
(
2014
).
Hours that count: Making employment supports work for Washingtonians with developmental disabilities
.
Kingdon
,
E.
,
&
Schultz
,
J.
(
1997
).
Social Security in the 21st century
.
New York, NY
:
Oxford
.
Kostl
,
A.
,
&
Mogstad
,
M.
(
2014
).
How financial incentives induce disability insurance recipients to return to work
.
American Economic Review
,
104
(
2
),
pp
.
624
655
.
Migliore
,
A.
,
Mank
,
D.
,
Grossi
,
T.
,
&
Rogan
,
P.
(
2007
).
Integrated employment or sheltered workshops: Preferences of adults with intellectual disabilities, their families, and staff
.
Journal of Vocational Rehabilitation
,
26
,
5
19
.
Nord
,
D.
,
Luecking
,
R.
,
Mank
,
D.
,
Kiernan
,
W.
,
&
Wray
,
C.
(
2013
).
The State of the science of employment and economic self-sufficiency for people with intellectual and developmental disabilities
.
Intellectual and Developmental Disabilities, 51
,
5
,
376
384
.
Riley
,
G. F.
,
&
Rupp
,
K.
(
2012
).
Expenditure patterns under the four major public cash benefit and health insurance programs for working-age adults with disabilities
.
Journal of Disability Policy Studies
,
25
(
2
),
71
80
.
Social Security Administration
. (
2012
).
Social Security handbook: Your guide to the Social Security Programs
.
Washington, DC
:
U.S. Government Printing Office
.
Social Security Administration
. (
2013
).
Annual statistical report on the Social Security Disability Insurance Program, 2012 (SSA Publication No.13-1182)
.
Washington, DC
:
U.S. Government Printing Office
.
Social Security Administration
. (
2014a
).
2014 Redbook: A summary guide to employment supports for persons with disabilities under the Social Security Disability Insurance and Supplemental Security Income programs (SSA Publication No. 64-030)
.
Washington, DC
:
U.S. Government Printing Office
.
Social Security Administration
. (
2014b
).
Annual statistical supplement to the Social Security Bulletin, 2013 (SSA Publication No. 13-11700)
.
Washington, DC
:
U.S. Government Printing Office
.
U.S. Census Bureau
. (
2014
),
Work status in the past 12 months, 2009 −2013 5 year American Community Survey
.
Retrieved from http://www.census.gov
U.S. House of Representatives Committee on Ways and Means
. (
2011
).
Green book
.
Washington, DC
:
U.S. Government Printing Office
.
Wehman
,
P.
,
Chan
,
F.
,
Ditchman
,
N.
,
&
Kang
,
H.-J.
(
2014
).
Effect of supported employment on vocational rehabilitation outcomes of transition-age youth with intellectual and developmental disabilities: A case control study
.
Intellectual and Developmental Disabilities
,
52
(
4
),
296
310
.
Yamamoto
,
S. H.
,
&
Alverson
,
C. Y.
(
2013
).
Successful vocational outcomes: A multi-level analysis of self-employment through U.S. Vocational Rehabilitation agencies
.
Journal of Vocational Rehabilitation
,
38
(
1
),
15
27
.

Author notes

Derek Nord and Kelly Nye-Lengerman, Research and Training Center on Community Living, University of Minnesota, Minneapolis.