Propensity Score Analysis of RSA 911 Supported and Customized Employment Outcomes

Abstract

Background

Supported Employment (SE) and Customized Employment (CE) are vocational rehabilitation services designed to support individuals with significant disabilities to achieve competitive integrated employment. Both services target individuals for whom employment has historically been difficult to obtain. but they differ in implementation. These differences may lead to varying employment outcomes, such as occupation types, wages, and hours worked.

Objective

This study analyzes FY17-FY21 RSA-911 data to compare the outcomes at exit for individuals receiving SE and CE vocational rehabilitation services.

Methods

Propensity score matching was used to create comparable SE and CE samples to (a) examine the prevalence of CE and SE services within vocational rehabilitation programs, (b) describe exit patterns among individuals receiving these services, and (c) identify differences in employment outcomes.

Results

A smaller proportion of individuals receiving CE exited into competitive integrated employment compared to those receiving SE. However, CE participants were more likely to enter self-employment and a broader range of occupations, indicating that CE may provide more personalized and flexible employment options.

Conclusion

This analysis highlights disparities in the use and outcomes of CE and SE services within state vocational rehabilitation programs.

Keywords

supported employment customized employment RSA-911 outcomes propensity score analysis

Introduction

Supported Employment (SE) was formally established as a vocational rehabilitation service through the 1986 amendments to the Rehabilitation Act. These amendments introduced formula grant funding to help state vocational rehabilitation agencies (SVRAs) collaborate with public and private entities to deliver SE services. The purpose of SE is to provide ongoing support to job seekers with significant disabilities, enabling them to obtain and maintain competitive integrated employment (CIE). The effectiveness of SE programs for individuals with disabilities is a well-documented vocational rehabilitation service (Hughes & Carter, 2001; Wehman et al., 2007, 2018) to help individuals secure meaningful employment, which in turn promotes greater independence and social inclusion. However, despite SE's successes, there are ongoing concerns about how SE engages those with higher-level support needs (Mank et al., 1998; Wehman et al., 2018; Winsor et al., 2023). Specifically, individuals with the most significant disabilities often encounter barriers to participation, resulting in lower engagement rates in SE compared to those with less intensive support needs. This disparity raises important questions about why individuals with the most significant disabilities are not achieving outcomes on par with their peers who require less support.

Customized Employment (CE) emerged as a strategy to help individuals with significant disabilities find and sustain jobs tailored to their unique strengths in competitive, integrated settings. The concept of CE gained prominence when the U.S. Department of Labor's Office of Disability Employment Policy (ODEP) introduced competitive grants aimed at improving employment outcomes for people with disabilities through strategic planning and implementation at One-Stop Career Centers (Federal Register, 2002). Subsequent ODEP demonstration and systems change projects showed that CE was effective in achieving meaningful employment outcomes for individuals with significant disabilities (Citron et al., 2008; Elinson et al., 2008; Fesko et al., 2008). In response to the success of these early initiatives and to address the ongoing disparities in employment opportunities for people with significant disabilities, the Rehabilitation Act was amended under the Workforce Innovation and Opportunity Act (WIOA, 2014) to include CE as a vocational rehabilitation service. CE was then incorporated into the definition of supported employment, described as “competitive integrated employment for an individual with a significant disability, based on an individualized determination of the strengths, needs, and interests of the individual, and designed to meet the specific abilities of the individual and the business needs of the employer” (29 U.S.C § 705(7), p. 1634).

Both SE and CE are designed to help individuals with significant disabilities obtain CIE. While both vocational rehabilitation services aim to assist those who have historically faced barriers to employment, there are key differences in their implementation. CE specifically focuses on individuals whose past employment experiences have been limited to segregated settings, such as facility-based programs, sheltered workshops, subminimum wage positions, or those who have not had opportunities to work due to low expectations (VRTAC-QE, 2022).

The primary differences between SE and CE emerge during the career planning and job development stages. In SE, traditional vocational assessments and evaluations are used to identify a job seeker's strengths and interests. This might include informal observations, job shadowing, work experiences, work assessments, or situational assessments. In contrast, CE employs a qualitative discovery process to uncover the job seeker's strengths, interests, and needs. This process involves interviews, observations, document reviews, and direct interactions with the job seeker (WINTAC, 2017). The discovery process is a “no-fail” approach that seeks to fully understand the individual's employment potential (VRTAC-QE, 2022).

SE and CE differ fundamentally in how jobs are identified and developed for job seekers. SE typically focuses on adapting preexisting job descriptions or tasks to match the interests of the job seeker. The process relies on conventional strategies to identify employers with current vacancies. These vacancies are based on existing roles within the business, outlined in standard job descriptions. Job seekers are either matched to these preexisting positions or, in some cases, job tasks are “carved out” from the broader job description to better suit the individual. In contrast, CE adopts a person-centered, customized approach that moves beyond conventional hiring practices. Instead of starting with existing job openings, businesses are identified that align with the job seeker's interests, regardless of whether they have current vacancies or standard job descriptions. The CE process begins with informational interviews and exploratory conversations to understand the business's operations, workplace culture, and unmet needs. Based on this information, a job proposal is created, detailing how the job seeker's specific skills and contributions can address the employer's needs or add value to their operations. Ultimately, the customized job is a unique position, negotiated and develop specifically for the job seeker. These roles are not adapted from existing vacancies—they are entirely new positions, created through collaboration between the employer and the employment specialist.

Given the differences in service delivery approaches between SE and CE, each service may lead to different employment outcomes in terms of occupation types, wages, and hours worked for individuals. To explore these differences, Riesen, Juhasz, et al. (2023) analyzed Rehabilitation Service Administration (RSA-911) data from fiscal years 2017–2020 to assess the outcomes for individuals receiving SE and CE services upon exit. Specifically, the study examined (a) the occupations, according to the Standard Occupational Classification (SOC) system, do individuals obtain after exiting state vocational rehabilitation SE and CE services, (b) the differences in occupational types at exit between individuals receiving SE and CE services and the differences in hourly wages and hours worked at exit between individuals receiving SE and CE services? The authors found that most individuals receiving SE and CE services exit into three broad SOC code categories: food preparation and serving-related occupations, office and administrative support, and building and grounds cleaning and maintenance. Proportional differences suggested that individuals receiving CE services explore a wider range of occupations compared to those receiving SE services. There was a significant difference in wages between SE and CE service recipients, but no difference in hours worked. Riesen, Juhasz, et al. (2023) noted specific limitations in their analysis and recommended that future RSA-911 analyses use propensity score matching (PSM) to equalize group sizes while adjusting for selection bias by matching cases on specific covariates.

Objective

The purpose of this study is to examine FY17-FY21 RSA-911 data to determine the outcomes at exit for individuals receiving supported and customized employment vocational rehabilitation services. We used PSM to create a more evenly match SE and CE sample and the following research aims guiding our analysis:

Describe the prevalence of use of CE and SE services within Vocational Rehabilitation programs.

Describe exit patterns among people who receive CE and SE services.

Determine the differences in employment outcomes between those who receive CE services versus those who receive SE services after accounting for initial differences between the two groups.

We hypothesize that SE and CE services will produce different employment outcomes at exit in terms of occupation types, wages, and hours worked for individuals receiving employment services.

Methods

Data for this study were extracted from the Rehabilitation Services Administration, Case Service report (RSA-911) database for FY17-FY21. SVRAs are required to collect and provide data on the performance of the vocational rehabilitation and supported employment program through the case service report, RSA-911 data system. RSA-911 is a rich data source that includes information captured at time of referral, during service provision delivery, and employment outcomes at exit for unique individuals receiving state vocational rehabilitation services. These individuals are assigned an identifier that allows RSA to provide an unduplicated count of the array of vocational rehabilitation services provided and case closure information such as occupation and wage outcomes for each state. Files are organized by cases closed during each fiscal year. Deidentified data is publicly available for secondary analysis.

Variables

We examined several variables for this study including demographic, employment status, primary occupation at exit, hourly wage, and hours worked per week at exit. Detailed descriptions of all available categories for each variable are publicly available on the U.S. rehabilitation services administration website (https://rsa.ed.gov/performance/rsa-911-policy-directive). Demographic information included gender (male, female), race (American Indian, Asian, Black, Hawaiian, White, Hispanic), primarily disability type, significance of disability, and whether a participant was a student or a veteran. RSA-911 datasets indicate which services each client receives while participating in the vocational rehabilitation program. Codes are provided within the RSA-911 datasets to indicate the exit type and employment outcome following the allowable codes accepted within the RSA-911 data collection system. For those who exit into employment, an occupational classification using the 2018 SOC code, is provided. The SOC system is a federal statistical standard used to classify workers into occupational categories. Occupations are combined to form 459 broad occupations, 98 minor groups, and 23 major groups. As described in the analysis section, these categories were further collapsed to meet the assumptions of specific analyses.

The final variable reviewed was hourly wage and hours worked per week at exit. These two variables were multiplied to create weekly total earnings, which provides a more wholistic view of take-home income for each individual.

Sample Selection

For research questions one and two, the cases for the analyses were selected based on whether they received SE or CE services regardless of whether the service was provided by VR agency staff or through VR agency purchase. This included 3,335 cases who received CE services, and 109,609 who received SE services. To prepare for analyses related to research question three to compare employment outcomes, additional selection criteria were applied before identifying matched cases between the two groups (i.e., those who had received CE services, and those who had received SE services). First, only cases that exited “after a signed IPE in competitive and integrated employment or supported employment” were retained. Next only exit employment outcomes in the following categories were retained: (a) CIE, (b) self-employment, (c) SE in CIE, and (d) supported employment on a short-term basis. Finally, to maintain dichotomous groups who had received only CE or SE services, 1,206 (0.10%) cases that received both SE and CE services were excluded from analyses. The final sample included 1,000 CE only cases and 75,771 SE only cases. The large difference in sample size between the groups prompted the application of PSM to balance the two groups and control for potential differences that may have initially influenced the imbalance. PSM is a quasi-experimental approach that has been commonly used as an alternative method to estimate causal effects in observational studies where potential selection biases need to be corrected (Rajeev & Sadek, 2002; Rosenbaum & Rubin, 1983). A one-to-one nearest neighbor approach was applied using MatchIt package in R (Ho et al., 2011). Participants from the SE group were matched to the 1,000 CE participants based on gender, race, and age at exit of the VR program. These matching covariates were selected based on research indicating that gender (Boeltzig et al. 2009; Kang et al., 2019) and race/ethnicity (Austin et al., 2019) are likely to impact wages among people with disabilities, while age is positively related to both wages and disability. Fit statistics indicated excellent balance between the two groups in the final sample of 2,000 cases.

Design and Analysis

We used a combination of Microsoft Excel and the Statistical Package for Social Scientists (SPSS) version 28, and RStudio for data analyses. A descriptive analysis and frequency tests were conducted to address research questions one and two. To address aim three, a chi-square test for goodness of fit was conducted to determine whether there were statistically significant group differences in the distribution of major SOC codes for each group (CE and SE) across all years (2017–2020). Preliminary calculations of expected cell values revealed that more than 20% of cells would have fewer than 5 cases. This violates an assumption of the chi-square test that is often attributed to Cochran (Bland, 2015). This is a common issue when there are a large number of categories, and can be remedied by collapsing categories (Bewick et al., 2004; McHugh, 2013). Starting with the smallest groups, SOC categories were combined into a single “All Others” group until at least 80% of cells had 5 or more cases expected. The resulting “All Others” group represents 13 SOC categories listed in the note of Table 5.

To account for unequal group variances, an independent samples Welch's t-test for exit hourly and weekly wages and exit hours worked was conducted to inspect whether there was a significant difference between participants who received CE and those who received SE. Finally, multiple regression was applied to test whether significant group differences in weekly earnings were present after controlling for demographic variables (gender, race, and age). Given the large sample size, and in alignment with other publications using RSA-911 data with similar types of analyses (see Alsaman & Lee, 2017; Ethridge et al., 2020) the alpha level a priori was set at p < .001. Power analyses conducted with G*Power (Faul et al., 2009) indicated that all analyses were sufficiently powered given sample sizes, selected alpha level, with 0.95 power (β) the analyses conducted could detect small effect sizes: t-test analyses d = .193, chi-square goodness-of-fit d = .027, regression analyses d = .001. Given the large amount of data available, and the small amount of missing data, listwise deletion was selected as the preferable method to handle missing data.

Results

Prevalence of CE services within vocational rehabilitation programs. There was an overall downward trend in the number of cases. In 2017 there were a total of 503,238 cases closed that received any type of service. In 2018, that number rose by 3,980 for a total of 507,218. This was followed by a decline of 22,578 cases (484,640 available for analyses) in 2019, and a sharper decline of 136,355 in 2020. The number of cases in 2021 increased slightly (367,810 total cases) but was still more than 116,000 fewer than in 2019.

Inspection of the type of service provided revealed that across all five years 3,335 people (0.15%) received customized employment services either through VR Agency Purchase or by VR agency staff (in-house). Comparatively, 109,609 (2.96%) clients received SE services. Inspection of use of CE services by state revealed that the majority of cases were clustered in a small number of states and territories (See Table 1). Between 2017 and 2021 only nine states provided CE services to more than 100 cases. Conversely, only four states and territories had fewer than 100 SE cases. When the raw number of CE cases was compared to the total number of cases for each state, there were only 2 states that provided CE services to more than 1% of their clients. Conversely, the average percentage of SE cases across all states and territories was 4.45% with only 6 states and territories that provided SE services to less than 1% of all clients served.

Table 1.

Number and Proportion of People Receiving CE and SE Services 2017–2021 by State and Territory.

State	Number of CE Cases	CE % of total Cases	Number of SE Cases	SE % of total Cases
Alabama	35	0.06%	2,541	4.69%
American Samoa	4	19.05%	0	0.00%
Alaska	0	0.00%	123	1.76%
Arizona	13	0.04%	365	1.04%
Arkansas	1	0.00%	282	0.90%
California	15	0.01%	4,444	2.83%
Colorado	8	0.03%	1,277	4.35%
Connecticut	1	0.01%	95	0.59%
Delaware	0	0.00%	642	6.33%
District of Columbia	4	0.04%	317	2.84%
Florida	797	0.72%	4,209	3.80%
Georgia	34	0.07%	2,255	4.62%
Guam	0	0.00%	1	0.24%
Hawaii	61	1.54%	77	1.94%
Idaho	22	0.11%	533	2.56%
Indiana	0	0.00%	2,325	6.89%
Iowa	63	0.19%	1,478	4.52%
Kansas	0	0.00%	1,119	5.53%
Kentucky	16	0.04%	2,061	4.68%
Louisiana	5	0.02%	997	3.53%
Maine	0	0.00%	214	1.36%
Maryland	0	0.00%	1,890	5.98%
Massachusetts	379	0.73%	1,926	3.70%
Michigan	264	0.31%	721	0.85%
Minnesota	200	0.55%	95	0.26%
Mississippi	16	0.05%	388	1.21%
Missouri	19	0.03%	11,618	17.24%
Montana	22	0.18%	256	2.07%
Nebraska	115	0.63%	2,049	11.22%
Nevada	34	0.28%	310	2.59%
New Hampshire	0	0.00%	162	1.66%
New Jersey	17	0.03%	1,517	3.01%
New Mexico	0	0.00%	187	1.22%
New York	29	0.02%	9,718	5.73%
North Carolina	312	0.35%	9,161	10.39%
North Dakota	4	0.04%	337	3.74%
Northern Marianas	0	0.00%	6	1.44%
Ohio	12	0.01%	7,233	8.17%
Oklahoma	52	0.21%	1,133	4.60%
Oregon	7	0.02%	2,816	8.68%
Pennsylvania	19	0.02%	3,788	4.11%
Puerto Rico	0	0.00%	459	1.47%
Rhode Island	5	0.08%	718	10.94%
South Carolina	6	0.01%	4,175	5.88%
South Dakota	17	0.16%	547	5.19%
Tennessee	13	0.05%	3,288	11.59%
Texas	16	0.01%	3,432	2.28%
Utah	44	0.13%	748	2.25%
Vermont	103	0.76%	361	2.68%
Virginia	31	0.06%	8,816	18.21%
Virgin Islands	0	0.00%	7	2.12%
Washington	365	1.04%	2,330	6.66%
West Virginia	6	0.04%	219	1.37%
Wisconsin	113	0.20%	3,409	5.94%
Wyoming	36	0.43%	434	5.19%

Among the 3,335 clients who received CE services from 2017–20211, slightly more than half (58.6%) self-identified as male. The average age at application was 28.63 years (median = 23, mode = 18, SD = 13.19). Race categories were not mutually exclusive and “Participant did not self-identify” is an option for all race categories: 76.6% self-identified as White, 16.9% Black, 12.4% Hispanic, 3.9% Asian, 2.0% American Indian, 1.5% Hawaiian. Very few (1.1%) were veterans, and most (72.5%) were not students. The majority (84.0%) were also most significantly disabled. Cognitive disabilities (e.g., disabilities involving learning, thinking, processing information and concentration) was the most common primary disability, accounting for nearly half (48.4%) of participants, followed by psychosocial disabilities (e.g., interpersonal and behavioral disabilities, difficulty coping) which accounted for 28.6% of clients. The next largest primary disability was “Other Mental Disabilities” which accounted for 5.1% of the sample. All other primary disability types accounted for less than 4% of the total sample.

For the 109,609 clients who received SE services from 2017–20211, slightly more than half (61.5%) self-identified as male. The average age at application was 30.39 years (median = 26, mode = 18, SD = 13.04). Race categories were not mutually exclusive and “Participant did not self-identify” is an option for all race categories: 70.3% self-identified as White, 26.8% Black, 7.8% Hispanic, 2.4% Asian, 1.7% American Indian, 0.5% Hawaiian. Very few (1.6%) were veterans, and most (85.2%) were not students. The majority (95.8%) were also most significantly disabled. Cognitive disabilities (e.g., disabilities involving learning, thinking, processing information and concentration) was the most common primary disability, accounting for over half (51.2%) of participants, followed by psychosocial disabilities (e.g., interpersonal and behavioral disabilities, difficulty coping) which accounted for 31.5% of clients. The next largest primary disability was “Other Mental Disabilities” which accounted for 6.1% of the sample. All other primary disability types accounted for less than 3% of the total sample.

Exit patterns among people who receive CE services. Examination of exit type revealed that the most common exit type was “after a signed IPE in competitive and integrated employment or supported employment” (CE = 59.9%, SE = 70.2%). However, this leaves a large proportion of individuals who exited without an employment outcome (see Table 2). Similarly, the top reason for exit was achieving CIE, with the second and third most common exit reasons “the person was no longer interested in receiving services or further services” (CE = 19.2%, SE = 14.3%), and “unable to locate or contact” (CE = 9.6%, SE = 7.3%).

Table 2.

Type of Exit by CE and SE.

	CE		SE
Exit type	Frequency	Percent	Frequency	Percent
Individual exited after a signed IPE in competitive and integrated employment or supported employment	1,998	59.9%	76.891	70.2%
Individual exited after a signed IPE without an employment outcome	1,276	38.3%	32,367	29.5%
Individual exited during or after a trial work experience	23	0.7%	3	0.0%
Individual exited after eligibility, but prior to a signed IPE	17	0.5%	3	0.0%
Individual exited as an applicant, prior to eligibility determination or trial work	3	0.1%	1	0.0%
Individual exited after a signed IPE in noncompetitive and/or nonintegrated employment	1	0.0%	15	0.0%
Missing	17	0.5%	329	0.3%
Total	3,335	100.0%	109,609	100.0%

Table 3 provides information about exit outcomes of CE and SE service recipients.

Table 3.

Employment Outcomes by CE and SE.

	CE		SE
	Frequency	Percent	Frequency	Percent
Supported employment in competitive integrated employment and	1,011	50.7%	52,774	68.6%
competitive integrated employment	957	47.9%	23,802	31.0%
Self-employment	22	1.1%	131	0.2%
Other*	6	0.4%	178	0.2%
Total	1,996	100.0%	76,885	100.0%

*Other category includes Homemaker, Supported Employment on Short-term Basis, State Agency Managed BEP, and Randolph-Sheppard BEP.

Exploration of employment outcomes for only those coded as “individuals exited after a signed IPE in competitive and integrated employment or supported employment,” revealed that the vast majority exited into either Supported Employment in (CIE = 50.7%, SE = 68.6%) or (CIE = 47.9%, SE = 31.0%). It is also notable that a larger proportion of individuals receiving CE exited into self-employment compared to those receiving SE (CE = 1.1%, SE =0.2%).

Employment outcomes for CE and SE combined groups. The remaining analyses reported were conducted using the sample of cases that had been matched through the PSM procedure described in the analysis section. Descriptive analyses of occupations revealed that the most common occupations for combined CE and SE groups was Food Preparation and Serving Related Occupations (17.3%). Table 4 presents the 23 Major SOC groups for both groups combined.

Table 4.

Major SOC Groups for Both Groups Combined Largest to Smallest.

Major SOC Group	Percent
Food preparation and serving related occupations	17.3%
Building and grounds cleaning and maintenance occupations	16.8%
Office and administrative support occupations	15.7%
Sales and related occupations	12.4%
Transportation and material moving occupations	9.3%
Production occupations	8.3%
Personal care and service occupations	5.6%
Healthcare support occupations	2.4%
Educational instruction and library occupations	2.3%
Installation, maintenance, and repair occupations	2.0%
Community and social service occupations	1.6%
Protective service occupations	1.2%
Computer and mathematical occupations	1.0%
Management occupations	0.9%
Arts, design, entertainment, sports, and media occupations	0.9%
Healthcare practitioners and technical occupations	0.8%
Business and financial operations occupations	0.6%
Construction and extraction occupations	0.6%
Architecture and engineering occupations	0.3%
Life, physical, and social science occupations	0.2%
Farming, fishing, and forestry occupations	0.2%
Legal occupations	0.1%
Military specific occupations	0.1%

Differences in employment outcomes between CE and SE service. Using the sample of cases matched using propensity scores, analyses were conducted to describe the types of jobs obtained. A chi-square test comparing major occupations at exit revealed there were statistically significant differences between those who had received SE services only and those who had received CE services only (χ² (9, N = 2,000) = 107.526, p < .001), and a Cramer's V test of strength of association indicated the relationship was weak (Cramer's V = .232). Although the effect was weak, inspection of the differences was telling (see Table 5). At exit, compared to those who received SE services, a smaller proportion of participants who received CE services were employed in Food Preparation and Serving Related Occupations, Building and Grounds Cleaning and Maintenance Occupations, Office and Administrative Support Occupations, and Production Occupations. There were virtually no differences in the proportion of SE and CE cases employed in Sales and Related Occupations. However, there were noticeably a greater number of CE cases employed in Transportation and Material Moving Occupations, Educational Instruction and Library Occupations, Healthcare Support Occupations, and Other Occupations.

Table 5.

Crosstabs SE and CE Groups by Major Occupation.

Major Occupation Codes	SE	CE	Total
Food preparation and serving related occupations	210	136	346
Building and grounds cleaning and maintenance occupations	196	139	335
Office and administrative support occupations	171	142	313
Sales and related occupations	123	124	247
Other occupations*	53	149	202
Transportation and material moving occupations	78	108	186
Production occupations	97	69	166
Personal care and service occupations	48	64	112
Healthcare support occupations	13	35	48
Educational instruction and library occupations	11	34	45

* Includes the following occupation categories: 1) Installation, Maintenance, and Repair, 2) Community and Social Service, 3) Protective Service, Computer and Mathematical, 4) Management, 5) Arts, Design, Entertainment, Sports, and Media, 6) Healthcare Practitioners and Technical, 7) Business and Financial Operations, 8) Construction and Extraction, 9) Architecture and Engineering, 10) Life, Physical, and Social Science, 11) Farming, Fishing, and Forestry, 12) Legal, 13) Military Specific.

Difference in weekly or hourly wages and hours worked at exit for SE and CE. Results of the Welch's t-test indicated that there were significant differences in hourly wages (M_SE= $10.77, SD_SE= 3.67; M_CE= $13.12, SD_CE= 5.66) t(1712.763) = 11.00, p < .001, d = .492 nearly Cohen's (1988) medium effect convention (d = .50); and weekly wages (M_SE= $233.48, SD_SE= 155.78; M_CE= $299.58, SD_CE= 248.76) t(1678.083) = 7.122, p < .001, d = .318 greater than Cohen's (1988) small effect convention (d = .20); but not in hours worked (M_SE= 20.95, SD_SE= 10.74; M_CE= 21.58, SD_CE= 11.69), t(1975.297) = 1.241, p = .107. In summary, participants who had received CE services exited into CIE earning more per hour and worked a similar number of hours.

Race and sex differences in wages at exit for SE and CE. A multiple regression analysis was conducted to test whether between group differences in weekly wages persisted after controlling for gender, race, age, and disability significance. Dichotomous variables were entered into the model indicating whether the participant had received CE or SE services (1 = CE, 0 = SE), sex (1 = Male, 2 = Female), race (0 = non-white, 1 = white), and disability significance (0 = Individual has no significant disability, 1 = Individual has a significant disability, 2 = Individual is most significantly disabled). Regression analyses yielded a statistically significant equation, F (5, 1,994) = 37.515, p < .001, with the predictors accounting for 8.6% of the variance in weekly wages. The effect size (r = .293) was more than Cohen's (1988) convention for small effects (r = .2) indicating that, while statistically significant, gender, race, age, disability significance and whether the participant received CE or SE services explained a small amount of the variability in weekly wages at program exit. After controlling for all other variables, only disability significance and whether the client received CE or SE services significantly predicted weekly wages. Clients who had received SE services made an average of $37.86 less per week as compared to those who had received CE services (See Table 6).

Table 6.

Hierarchical Regression Predicting Weekly Wages at Program Exit.

Variable	Unstandardized B			95% Confidence Interval
Constant	517.23***	(30.19)	Standardized B	[458.03,	576.44]
Sex	−10.38	(7.56)	−.029	[−25.22,	4.45]
White	−2.33	(4.61)	−.011	[−11.37,	6.71]
Age at exit	0.16	(0.34)	.010	[−0.50,	0.82]
Disability significance	−139.86***	(12.25)	−.254	[−163.88,	−115.84]
CE SE group	37.86***	(9.33)	.090	[19.57,	56.16]
R²	.086
F	37.52***

Discussion

The purpose of this study was to examine FY17-FY21 RSA-911 data to describe who received SE and CE services in state VR programs and the outcomes for those individuals in each of the services. We employed a quasi-experimental design, by applying PSM to balance uneven sample sizes, which allowed for a more balanced comparison between the two groups. Our approach helped address potential selection biases and provided stronger evidence related to the differences in outcomes. Our findings provide insights into the employment outcomes of individuals receiving SE and CE services.

In response to questions regarding the prevalence and utilization CE and SE services, our analysis uncovered notable disparities across SVRA's. Despite more than a decade of legislative support for CE, the widespread adoption of CE remains limited. A relatively small percentage of VR consumers receive CE services compared to SE services. This finding aligns with previous research, which indicated that a handful of states had high frequency of CE utilization (Kim et al., 2023). Given the promising outcomes highlighted in our findings, there is a pressing need to increase awareness of CE and to provide more extensive training on implementing discovery and customized job development to fidelity. Enhancing the dissemination of information about CE is crucial, as existing research has identified that one of the main challenges in CE implementation is the incomplete operationalization of key components of the intervention (Riesen, Hall, et al., 2023; Riesen, Snyder, et al., 2023; Wehman, 2023).

Regarding our question about exit patterns, there were notable differences in exit patterns between the two services. Only 59.9% of CE cases, and 70.2% of SE cases exited with an employment outcome, leaving more than a third of each group without employment when they exited the program. Comparatively, a larger proportion of CE cases exited without an employment outcome than SE cases. The fact that a substantial number of individuals in CE did not obtain CEI raises concerns about broad performance or design challenges and resource allocation. The higher rate of CE cases exiting without employment suggests that participants may not be receiving the level of individualized support they need. SVRAs could improve the process by requiring highly trained job coaches who can tailor CE supports to participants’ unique needs can improve their chances of securing and maintaining employment. To ensure continuous improvement, SVRAs should establish clear metrics for success and regularly monitor CE outcomes, including the use of validated fidelity scales (Riesen, Hall, et al., 2023). By using process data obtained from fidelity measures, SVRAs can examine CE job seeker progress and exit patterns in real time and can make data informed decisions to quickly identify and address areas in the CE process that need to be improved. The disparity between CE and SE outcomes may also be tied to how resources are distributed across these services. SVRAs could review the allocation of funding to support CE service delivery as supporting a person with more significant disability who has limited employment experience may require more time to engage in discovery and customized job development. Ensuring that CE participants have access to flexible funding resources could help improve employment outcomes. An examination of employment outcomes also reveals that individuals receiving CE services demonstrated a higher likelihood of exiting into self-employment compared to those receiving SE services. CE focus on customizing job roles to fit the needs and abilities of the job seeker aligns well with the principles of self-employment as it offers the highest level of flexibility for job seekers who are seeking autonomy or the ability to work outside of traditional job structures.

The analysis also revealed that individuals receiving CE services were more likely to exit into a wider range of occupations compared to those receiving SE services. The differences in occupations may be attributed to implementation differences between the two services. CE involves the discovery process designed to uncover a job seeker's strengths, interests, and needs. This process typically includes interviews, observations, review of documentation, and direct interactions with the job seeker (WINTAC, 2017). Unlike SE, which focuses on matching job seekers with existing job opening, CE uses a customized approach to job development that includes conducting informational interviews to learn more about employers, their working conditions, and identifying other potential employers engaged in similar work. Based on this information, customized jobs are negotiated through an employment proposal that reflects the job seeker's unique skills and interests. The job developer then creates a detailed job site analysis and plan. These findings strengthen those reported previously ([name deleted to maintain the integrity of the review process], 2023). Compared to previous findings, this study, which applied a quasi-experimental design and an additional year of data, achieved a larger effect size than previously reported. Given the small proportion of cases that exit into some occupational types, these methodological additions significantly strengthen these findings suggesting the CE services do indeed diversify employment opportunities for individuals with significant disabilities.

However, the results about the occupational differences must also be tempered by the relatively large proportion of both SE and CE clients who continue to achieve work in three job clusters: (1) Food Preparation and Serving Related, (2) Building and Grounds Cleaning and Maintenance, and (3) Office and Administrative Support. The lack of occupational diversity is problematic as it suggests SE and CE job seekers are not provided opportunities to explore the full range of occupations and opportunities available in their local labor market. This restriction may indicate the presence of stereotypes about the types of work individuals with significant disabilities can engage in, and thus can limit the potential for individuals to find roles that truly match their skills and interests. The significant differences in occupational distribution between CE and SE cases highlight the potential of CE to diversify employment opportunities for individuals with significant disabilities. By focusing on the unique strengths and interests of job seekers, CE can help break down these stereotypes and expand the range of occupations available to them. This approach not only benefits the individuals but also enriches the workforce by bringing diverse talents and perspectives into various industries.

Finally, our analysis revealed statistically significant differences between wages with individuals receiving CE services, on average, earning $2.35 more per hour than those receiving SE services. This finding underscores the potential for CE services to secure better-paying jobs for individuals with significant disabilities. Over the course of one year of full-time work, this equates to $1,969 more per year for those receiving CE services. Furthermore, regression analysis revealed that those gains in earnings were still significant after controlling for gender, race, age, and significance of disability. It is important to note, however, that significance of disability had a much larger impact on wages than whether the participant received CE or SE services. As noted in the regression analyses, as the significance of an individual's disability increased, wages decreased. While exploring these differences is outside of the scope of the current investigation, this finding should be investigated more in-depth in future research. A promising finding is that the average wages for SE and CE cases are higher than those previously reported (Riesen, Juhasz, et al., 2023). The wage increase is likely due to the addition of data from 2021, which may suggest that as wages increase nationally, SE and CE wages are also increasing. However, there are still wide hourly wage disparities between employees with and without disabilities. For instance, the U.S. Bureau of Labor Statistics (2023) reports national average hourly wages of $17.03 per hour Food Preparation and Service-Related Occupations. Neither SE ($10.77) nor CE ($13.12) average hourly wages reached these averages. The fact that SE and CE recipients are not compensated at rates comparable to national averages raises concerns about their financial sufficiency and ability to meet regional cost of living standards.

Conclusions

The findings from this study underscore some disparities in the utilization and outcomes of CE and SE services within state vocational rehabilitation programs. Despite legislative support for CE, its adoption remains limited, with only a small percentage of job seekers with significant disabilities receiving these services compared to SE. This is particularly evident in the concentration of CE services in a few states, highlighting a need for broader awareness and training to ensure the effective implementation of CE nationwide. The differences in exit patterns and employment outcomes between CE and SE are telling. Individuals receiving CE services are less likely to exit to CEI compared to individuals receiving SE. Individuals receiving CE services are more likely to enter self-employment and a wider range of occupations, suggesting that CE may offer more personalized and flexible pathways to employment. However, the concentration of both SE and CE clients in a narrow range of job clusters, such as Food Preparation and Building Maintenance, raises concerns about the limited occupational diversity afforded to these individuals. The wage disparities between CE and SE recipients further highlight the potential advantages of CE in securing better-paying jobs, though both groups still earn below national averages. These findings suggest that while CE shows promise in enhancing employment outcomes for individuals with significant disabilities, there is a critical need for continued efforts to expand its reach, improve occupational diversity, and address wage gaps to ensure financial stability and meaningful inclusion in the workforce.

While this study provides important insights, it is not without limitations. The reliance on RSA-911 data, which is self-reported by SVRAs, may introduce reporting errors. For example, a small number of cases were excluded from some analyses because these cases were coded as receiving both SE and CE services, which is inconsistent with practice recommendations. Similarly, we report on all exit outcome codes including “supported employment in CIE” and “competitive integrated employment” alone even though it is unclear in the RSA-911 instructions how these are distinguished in practice. Furthermore, there is no information provided by the RSA-911 datasets related to the fidelity and quality of the implementation of SE and CE services. Additional information is needed to understand the mechanisms related to SE and CE service delivery to understand how these approaches positively or negatively affect SE or CE employment outcomes which may shed light on the finding here that compared to those who received SE services, a larger proportion of CE participants did not exit into CIE. This is especially true in light of findings reported here that much of CE service delivery is clustered within a small number of states. State policy that either supports, or hinders implementation of CE services may dramatically impact service delivery and should be accounted for, and further explored in future research. Furthermore, qualitative studies could provide deeper insights into the experiences of individuals receiving CE and SE services, shedding light on the factors that contribute to successful employment outcomes.

Footnotes

Acknowledgements

The authors declare no acknowledgements.

ORCID iDs

Audrey Juhasz

Tim Riesen

Ethics Statement

This study used publicly available data. Institutional Review Board (IRB) approval was therefore not necessary.

Informed Consent

Not applicable.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: One of the authors, Tim Riesen, is a member of the Journal of Vocational Rehabilitation editorial board.

References

Alsaman

M. A.

Lee

C. L.

(2017). Employment outcomes of youth with disabilities in vocational rehabilitation: A multilevel analysis of RSA-911 data. Rehabilitation Counseling Bulletin, 60(2), 98–107. https://doi.org/10.1177/0034355216632363

Austin

B. S.

Lee

C.-L.

Fleming

(2019). Vocational rehabilitation outcomes for individuals with intellectual disabilities and co-occurring psychiatric disorders. Journal of Rehabilitation, 85(4), 14–23.

Bewick

Cheek

Ball

(2004). Statistics review 8: Qualitative data - tests of association. Critical Care, 8(1), 46–53. https://doi.org/10.1186/cc2428

Bland

(2015). An introduction to medical statistics. Oxford university press.

Boeltzig

Timmons

J. C.

Butterworth

(2009). Gender differences in employment outcomes of individuals with developmental disabilities. Journal of Vocational Rehabilitation, 31(1), 29–38. https://doi.org/10.3233/JVR-2009-0469

Citron

Brooks-Lane

Crandell

Brady

Cooper

Revell

(2008). A revolution in the employment process of individuals with disabilities: Customized employment as the catalyst for system change. Journal of Vocational Rehabilitation, 28(3), 169–179. https://doi.org/10.3233/JVR-2008-00417

Cohen

(1988). Statistical power analysis for the behavioral science (2nd ed.). Lawrence Erlbaum Associates.

Elinson

Frey

Palan

Horne

M. A.

(2008). Evaluation of customized employment in building the capacity of the workforce development system. Journal of Vocational Rehabilitation, 28(3), 141–158. https://doi.org/10.3233/JVR-2008-00415

Ethridge

Dowden

A. R.

Brooks

Kwan

Harley

(2020). Employment and earnings among ex-offenders with disabilities: A multivariate analysis of RSA-911 data. Journal of Vocational Rehabilitation, 52(3), 279–289. https://doi.org/10.3233/JVR-201077

10.

Faul

Erdfelder

Buchner

Lang

A.-G.

(2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41(4), 1149–1160. https://doi.org/10.3758/BRM.41.4.1149

11.

Federal Register, June 26, 2002, Vol. 67, No. 123 pp. 43154–43149. https://www.govinfo.gov/content/pkg/FR-2002-06-26/pdf/02-16098.pdf

12.

Fesko

Varney

DiBiase

Hippenstiel

(2008). Effective partnerships: Collaborative efforts that support customized employment. Journal of Vocational Rehabilitation, 28(3), 159–168. https://doi.org/10.3233/JVR-2008-00416

13.

Imai

King

Stuart

(2011). Matchit: Nonparametric preprocessing for parametric causal inference. Journal of Statistical Software, 42(8), 1–28. https://doi.org/10.18637/jss.v042.i08

14.

Hughes

Carter

(2001). Supporting transition from school to work. In Wehman

(Ed.), Supported employment in business: Expanding the capacity of workers with disabilities (pp. 239–250). TRN.

15.

Kang

Nord

D. K.

Nye-Lengerman

K. M.

(2019). Weekly wage exploration of vocational rehabilitation service recipients: A quantile regression approach. Journal of Rehabilitation, 85(1), 4–14.

16.

Kim

Riesen

Inge

Keeton

Weathers

Tansey

(2023). Customized employment as a pathway to competitive integrated employment: An analysis of RSA 911 data of state vocational rehabilitation agencies with the highest use of this intervention. Journal of Vocational Rehabilitation, 58(1), 89–98. https://doi.org/10.3233/JVR-221227

17.

Mank

Cioffi

Yovanoff

(1998). Employment opportunities for people with severe disabilities: Opportunities for improvement. Mental Retardation, 36(3), 205–216. https://doi.org/10.1352/0047-6765(1998)036<0205:EOFPWS>2.0.CO;2

18.

McHugh

M. L.

(2013). The Chi-square test of independence. Biochemia Medica, 23(2), 143–149. https://doi.org/10.11613/issn.1846-7482

19.

Rajeev

H. D.

Sadek

(2002). Propensity score-matching methods for nonexperimental causal studies. Review of Economics and Statistics, 84(1), 151–161. https://doi.org/10.1162/003465302317331982

20.

Riesen

Hall

Keeton

Snyder

(2023). Internal consistency of the customized employment discovery fidelity scale: A preliminary study. Rehabilitation Counseling Bulletin, 66(3), 195–202. https://doi.org/10.1177/00343552211043259

21.

Riesen

Juhasz

A. C.

Remund

(2023). An analysis of the rehabilitation service administration 911 supported and customized employment outcome data for fiscal years 2017–2020. Research and Practice for Persons with Severe Disabilities, 48(3), 115–126. https://doi.org/10.1177/15407969231181901

22.

Riesen

Snyder

Byers

Keeton

Inge

(2023). An updated review of the customized employment literature. Journal of Vocational Rehabilitation, 58(1), 27–38. https://doi.org/10.3233/JVR-221222

23.

Rosenbaum

P. R.

Rubin

D. B.

(1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41–55. https://doi.org/10.1093/biomet/70.1.41

24.

U.S. Bureau of Labor Statistics. (2023). Occupational employment and wage statistics: May 2021 occupation profiles. https://www.bls.gov/oes/2023/may/oes359099.htm

25.

VRTAC-QE . (2022, November 15). Supported and customized employment side by side referral decision guide . Vocational Rehabilitation Technical Assistance Center for Quality Employment (VRTAC-QE). Retrieved November, 2024, from https://tacqe.com/supported-and-customized-employment-referral-decision-guide/

26.

Wehman

(2023). Supported employment and customized employment: How effective are these interventions and what has been their impact on the field? Journal of Vocational Rehabilitation, 58(3), 237–247. https://doi.org/10.3233/JVR-230022

27.

Wehman

Inge

K. J.

Revell

G. W.

(2007). Supported employment and workplace supports: Overview and Background. In Wehman

Inge

K. J.

Revell

G. W.

Brook

V. A.

(Eds.), Real work for real pay: Inclusive employment for people with disabilities (pp. 117–138). Brookes.

28.

Wehman

Taylor

Brooke

Avellone

Whittenburg

Ham

Brooke

A. M.

Carr

(2018). Toward competitive employment for persons with intellectual and developmental disabilities: What progress have we made and where do we need to go. Research and Practice for Persons with Severe Disabilities, 43(3), 131–144. https://doi.org/10.1177/1540796918777730

29.

Winsor

Butterworth

Migliore

Domin

Zalewska

Shepard

Kamau

Wedeking

Edelstein

(2023). StateData: The national report on employment services and outcomes through 2021. University of Massachusetts Boston, Institute for Community Inclusion.

30.

WINTAC. (2017). The essential elements of customized employment for universal application. Workforce Innovation Technical Assistance Center. www.wintac.org/topic-areas/resources-and-strategies-competitive-integrated-employment/resources

31.

Workforce Innovation and Opportunity Act (WIOA). 29 U.S.C § 705 et seq. (2014).