Predictors of Employment Following Postsecondary Education for Vocational Rehabilitation Participants With Traumatic Brain Injury

Abstract

The purpose of this study was to identify vocational rehabilitation (VR) services associated with employment outcomes of individuals with traumatic brain injury (TBI) who received college training and increased their highest level of postsecondary education completed. The participants were 1,221 individuals with TBI who increased their highest level of postsecondary education while receiving VR services. A logistic regression model containing five demographic and six VR service variables correctly classified 75.4% of cases as competitively employed or not competitively employed. The model explained approximately 10.0% of the variance in competitive employment. Significant predictors associated with competitive employment were the demographic characteristics of ethnicity and age at application. VR service variables that were significant predictors of competitive employment were (a) vocational rehabilitation counseling and guidance, (b) job readiness training, (c) job placement assistance, and (d) personal assistance services. While demographic characteristics and VR services explained a relatively small proportion of the variance in employment outcomes, more robust measurements of the variables have the potential to enhance prediction of outcomes. Provision of job readiness training and personal assistance services should be investigated further to determine the extent to which they serve as indicators of the presence of additional barriers to competitive employment.

Keywords

traumatic brain injury postsecondary education vocational rehabilitation

Traumatic brain injury (TBI) presents a significant public and social welfare concern in the United States. From 2002 to 2006, the Centers for Disease Control and Prevention (CDC; 2010) reported approximately 1.7 million TBIs occurred each year in the United States, with this number increasing to approximately 2.5 million in 2010 (CDC, 2015). The CDC (2016) estimated that approximately 5.3 million Americans are living with a disability as a consequence of TBI. TBI increasingly enters the public consciousness due to the high number of veterans injured in Iraq and Afghanistan (Degeneffe, Tucker, & Griffin, 2015) as well as growing awareness of the consequences of repeated sports injuries such as chronic traumatic encephalopathy (Jordan, 2013).

The Context of TBI

Many persons with TBIs are challenged by the difficulty of integrating their pre- and postinjury lives. There is often a pervasive impulse to compare what the person did prior to their injuries in such contexts as their family and work lives (Prigatano, 1995). In attempting to come to terms with the realities of their lives, some turn to drugs and alcohol as a means of coping (Taylor, Kreutzer, Demm, & Meade, 2003), and some injured persons consider suicide (Baker, Tandy, & Dixon, 2002). Persons with TBIs and their families are also faced with a system of long-term support that is often insufficient (Degeneffe & Tucker, 2014) and fragmented (Degeneffe et al., 2008). Families are therefore often tasked with extended caregiving responsibilities (Degeneffe & Lee, 2015), with many facing elevated rates of depression and anxiety (Degeneffe, 2001).

In spite of the difficulties presented by this disability, it is possible for persons with TBIs to attain a satisfying perception of the quality of their lives. One of the key factors associated with positive quality of life (Berger, Leven, Pirente, Bouillon, & Neugebauer, 1999) and life satisfaction (Corrigan, Bogner, Mysiw, Clinchot, & Fugate, 2001) is the ability of the person with TBI to engage in productive activities. The importance of persons with TBI returning to work and pursuing career goals cannot be overstated, as work provides individuals with senses of affiliation, financial well-being, and identity (Brown, 2011). Vocational participation also can serve as a sign of recovery from TBI for injured persons (Levack, McPherson, & McNaughton, 2004).

It is difficult to determine precisely how many individuals with TBIs return to work as differences occur due to methodological inconsistencies. It is generally accepted however that a sizable proportion of persons with TBIs do not engage in employment postinjury, likely due to the many functional impairments experienced after injury. According to the National Data and Statistical Center, Traumatic Brain Injury Model Systems (2015), among persons receiving inpatient rehabilitation care, the employment rate of persons with TBIs at the time of injury was 61%, which dropped to 28% at a 1-year follow-up, 31% at 2 years, and 35% 5 years postinjury. When served by the state/federal vocational rehabilitation (VR) system however, persons with TBIs are often able to reach successful employment outcomes. Wehman, Targett, West, and Kregel’s (2005) review of biennial RSA-911 data indicated successful closure rates of 57% (1993) to 60% (2001). Of those successfully rehabilitated during this period, competitive employment outcomes improved with an 81% rate in 1993 to 93.9% in 2001. A total of approximately 54,000 persons with TBIs were served during this 8-year period. In addition, Catalano, Pereira, Wu, Ho, and Chan’s (2006) review of Fiscal Year 2004 RSA-911 data found a 50% integrated employment rate among the 7,366 persons with TBIs closed as either employed or not employed.

The Value of Postsecondary Participation

In the general population, it is generally accepted that postsecondary participation leads to positive economic outcomes. The U.S. Census Bureau (2012) estimated that for persons aged 25 to 64 working full-time over a 40-year period, lifetime earnings of someone with a bachelor’s degree were US$2,422,000 compared with US$1,371,000 for a high school graduate. A similar assumption about the value of college training also applies to VR. A number of studies that have examined level of education as a predictor of rehabilitation outcomes can be found in the rehabilitation literature (Gamble & Moore, 2003; James, DeVivo, & Richards, 1993; Roessler, Rumrill, & Fitzgerald, 2004; Rogers, Crystal, & Bishop, 2005; Waghorn & Chant, 2002). Higher levels of education when applying for rehabilitation services have generally been found to predict more favorable VR outcomes. Such studies are part of a body of research focused upon identifying variables associated with successful rehabilitation outcomes and identifying variables that may help identify individuals who are less likely to be successful; individuals who might benefit from additional attention and supports throughout the VR process (Bolton, Bellini, & Brookings, 2000; Capella-McDonnall, 2005).

Given the known association between postsecondary education and employment, studies focusing upon VR participants who receive college or university training would appear to be worthwhile, yet they appear to be published relatively infrequently (Tucker, 2007). A small number of studies have been published which examine college or university training provided as a VR service to specific disability groups (Boutin & Accordino, 2009; Boutin & Wilson, 2009; Nakaji, 2014).

Specific to TBI, there appears to be no consistent pattern of predicting employment status based on preinjury education (Catalano et al., 2006). However, consistent with other disability populations, providing college training in VR service plans for individuals with TBIs demonstrates positive associations with integrated employment outcomes (Catalano et al., 2006; Gamble & Moore, 2003). Using binary logistic regression analysis and factorial ANOVA, Gamble and Moore (2003) examined RSA-911 data pertaining to 1,073 individuals with TBI in a southeastern state whose cases were closed between October 1992 and September 2000 in a southeastern state. They examined the results of employment status and earnings based on receiving assessment, college training, counseling/guidance, job placement, restoration, and work adjustment. While job placement was the most important variable related to competitive employment (with a 20.77 odds ratio [OR]), consumers with TBIs were also 5.21 times more likely to attain competitive employment when receiving college training. Also, for consumers with severe levels of TBI, those who received college training had higher weekly earnings compared with those without college training.

Using chi-square automatic interaction detection (CHAID) analyses, Catalano and associates’ (2006) examination of Fiscal Year 2004 RSA-911 data of 7,366 closed cases for consumers with TBIs found similar results. As Gamble and Moore (2003) concluded, job placement was the most important service related to integrated competitive employment. Catalano et al. also found that for consumers with TBIs, the following conditions in combination were positively related to integrated employment outcomes: (a) receiving job placement services, (b) not receiving work disincentives (e.g., Supplemental Security Income [SSI]), and (c) high expenditure services (i.e., more than US$4,824).

Several authors state that persons with TBIs can encounter a range of challenges in attempting postsecondary education due to the possible cognitive, physical, and psychological implications of this disability. For example, Kennedy, Krause, and Turkstra (2008) administered an online survey to 35 persons with TBIs in Minnesota, Kansas, and Wisconsin regarding the challenges they experienced while attending college. Kennedy and associates found separate significant positive correlations for post-TBI (a) cognitive effects and (b) psychosocial effects with academic challenges. A limited body of research also attempts to identify which factors predict postsecondary participation among persons with TBIs. Todis, Glang, Bullis, Ettel, and Hood (2011) assessed postsecondary participation among 66 persons with TBIs in Oregon and Washington over a 6-year period, with the first assessment occurring on average 1.3 years after leaving high school. Higher socioeconomic status was related to a shorter time to enrollment in postsecondary studies, and females were more likely to attend postsecondary education than males. Those injured earlier in life were less likely to participate in postsecondary education than those who incurred TBIs at a later point. Injury severity did not predict postsecondary participation. In contrast to this finding, Willmott, Ponsford, Downing, and Carty (2014) tracked outcomes among 295 persons with TBIs for 10 years following inpatient care. All participants had participated in either secondary or tertiary education at the time of injury. Those with longer posttraumatic amnesia were less likely to return to some form of education.

While the extant research on college training in VR provides evidence of the value of this resource, an incomplete understanding of this service for individuals with TBIs remains. That is to say, prior research does not identify the specific value of advancing in educational attainment during VR service provision. This is an important gap in the literature, given the demonstrated value of educational attainment found in the general population. Furthermore, while VR research preliminarily demonstrates the value of college training as a binary variable (i.e., yes or no), outcomes related to how college training is used in VR are not understood. To the best of our knowledge, no studies have been published that examine correlates of successful employment for individuals with TBIs who completed postsecondary certificates or degrees while receiving VR services.

During the 5 federal fiscal years between 2009 and 2013, only 46,806 of the 2,959,994 cases closed by VR agencies were individuals whose primary cause of disability was TBI despite more than 5 million people in the United States living with long-term disability following TBI (CDC, 2016). These figures suggest that persons with TBIs are likely underserved in the VR system. While not all individuals with TBIs are eligible for VR services and some may seek rehabilitation services through other providers (e.g., Department of Veterans Affairs), state-federal VR agencies represent one of the few public resources available to many individuals with TBIs. There is considerable variability between states and communities with respect to public services for individuals with TBIs (Degeneffe et al., 2008; Degeneffe & Tucker, 2014). Most individuals with TBIs do not benefit from services provided by the Department of Veterans Affairs or other disability service systems with more robust public supports, such as services for people with intellectual disabilities.

As suggested by Gamble and Moore (2003) and Catalano et al. (2006), VR service plans for persons with TBIs involve the provision of multiple services. Therefore, the present investigation examined the value of educational attainment in the context of commonly provided VR services for consumers with TBIs. An assumption here is that college attainment in isolation is not as likely to result in successful integrated employment outcomes as when paired with other supports designed to address the many physical, cognitive, and psychosocial concerns commonly associated with TBI.

Given the high cost (e.g., Catalano et al., 2006) of VR services involving college training, the central goal of our study was to identify a parsimonious set of VR services most strongly correlated with integrated competitive employment of persons with TBIs when consumers were able to use college training to advance educationally. Accordingly, our exploratory study focused upon the following research questions.

Research Questions

Research Question 1: After controlling for the contributions of demographic variables, which combination of VR service variables best predicts competitive employment outcomes of VR participants with TBIs who advanced their postsecondary education?

Research Question 2: Within the prediction model, which specific VR services and demographic characteristics have the strongest associations with competitive employment for VR participants with TBIs who advanced their postsecondary education?

Method

Data Source

The source of data for the study was Rehabilitation Services Administration administrative data from federal Fiscal Years 2009 through 2013, commonly known as RSA-911. RSA-911 data, obtained from the U.S. Department of Education, consist of all VR cases closed during a given federal fiscal year. Cases selected were those where the individual had a primary cause of impairment of TBI, the individual had received college or university training while his or her case was open, the individual’s level of education increased on an ordinal education level scale between application for services and case closure, and the individual’s level of education at case closure was “associate degree or vocational/technical certificate,” “bachelor’s degree,” or “master’s degree or higher.”

All the participants had a VR case that was closed during the 5 federal fiscal years between 2009 and 2013. Federal Fiscal Year 2013 was the most recent year available at the time the study was initiated. The RSA-911 data set for this 5-year period contained 2,959,994 case records; a much smaller subset, 46,806 of these cases (approximately 1.6%) had a primary cause of impairment identified as TBI. Of the 46,806 cases that had a primary cause of impairment identified as TBI, 4,128 (approximately 8.8%) were provided with college or university training while their VR case was open. From these 4,128 cases, the study population was selected by identifying those cases where the individual’s level of education increased between application and case closure on an ordinal scale measuring highest education level completed to one of three scale points, indicating completion of postsecondary education: “associate degree or vocational/technical certificate,” “bachelor’s degree,” or “master’s degree or higher.” This selection process yielded 1,225 individuals. Four of these individuals were missing data essential to the logistic regression analysis and were thus excluded from the study, yielding a study population of 1,221 cases.

Table 1 summarizes selected demographic characteristics of the participants. Six hundred seventy-five of the participants (55.3%) were male, and 546 (44.7%) were female. Participants were largely White, with 1,087 of the 1,221 participants (89.0%) identified as White (it should be noted that more than one race could be recorded for each individual, and some may have been of mixed race or of Hispanic ethnicity). Thirty-six participants (2.9%) were identified as veterans. Case records indicated that 690 cases (56.5%) had a secondary disability. In all, 201 participants (16.5%) received SSI when the case was closed, whereas 275 participants (22.5%) received Social Security Disability Insurance (SSDI) at case closure. Four hundred sixty-nine (38.4%) of the participants completed an associate’s degree or vocational technical certificate; 626 individuals (51.3%) completed a bachelor’s degree; and 126 individuals (10.3%) completed a master’s degree or higher at case closure.

Table 1.

Demographic Characteristic of Participants.

Characteristic	n	%
Gender
Male	675	55.3
Female	546	44.7
Race/ethnicity^a
American Indian or Alaska Native	18	1.5
Asian	22	1.8
Black	100	8.2
Hawaiian or Other Pacific Islander	8	0.7
Hispanic or Latino	92	7.5
White	1,087	89.0
Veteran status
Yes	36	2.9
No	1,138	93.2
Missing	47	3.8
Secondary disability
Yes	690	56.5
No	531	43.5
SSI recipient at case closure
Yes	201	16.5
No	1,001	82.0
Missing	19	1.6
SSDI recipient at case closure
Yes	275	22.5
No	932	76.3
Missing	14	1.1
Highest level of education completed at case closure
Associate’s degree or vocational/technical certificate	469	38.4
Bachelor’s degree	626	51.3
Master’s degree	126	10.3

Note. SSI = Supplemental Security Income; SSDI = Social Security Disability Income.

More than one race or ethnicity could be recorded for each individual.

In addition to the cause of the disability, the case record contains a variable that indicates the individual’s area of impairment. Table 2 summarizes the areas of impairment recorded for the 1,221 individuals with a primary cause of disability of TBI who received college training and increased their postsecondary education level. The most commonly reported area of impairment was cognitive impairments, indicated in 53.2% of the cases. The second most commonly reported area of impairment was other physical impairments, which was indicated in 14.2% of the cases; the third most commonly reported area was mobility/manipulation/dexterity and orthopedic/neurological impairment, indicated in 9.5% of cases. It should be noted that the case record permits reporting only one area of impairment associated with the primary cause of disability; as a result, the record may contain incomplete information about functional limitations if an individual experiences impairment in two or more areas of function.

Table 2.

Area of Impairment.

Impairment code	n	%
Blindness	12	1.0
Other visual impairments	22	1.8
Deafness, primary communication visual	3	0.2
Hearing loss, primary communication auditory	8	0.7
Communication impairments (expressive/receptive)	12	1.0
Mobility orthopedic/neurological impairments	53	4.3
Manipulation/dexterity/orthopedic/neurological impairments	31	2.5
Mobility/manipulation/dexterity and orthopedic/neurological impairments	116	9.5
Other orthopedic impairments	17	1.4
General physical debilitation	18	1.5
Other physical impairments	173	14.2
Cognitive impairments	650	53.2
Psychosocial impairments	50	4.1
Other mental impairments	56	4.6

Variables

The criterion variable for both research questions was competitive employment at case closure. Competitive employment is defined by the Rehabilitation Services Administration (2013) as “Employment in an integrated setting, self-employment, or state-managed Business Enterprise Program that is performed on a full-time or part-time basis for which an individual is compensated at or above the minimum wage” (p. 33). Competitive employment may be considered one indicator of successful or unsuccessful VR outcome and is recorded in the RSA-911 data set in a dichotomous manner (competitively employed or not competitively employed).

The demographic predictor variables addressed gender, race, ethnicity, age at application, and reported income at application. A number of studies have identified gender, race, age, and socioeconomic status as associated with postsecondary enrollment or success (e.g., Cooper, 2003; Kim, 2012; Tekleselassie, Mallery, & Choi, 2013). Gender was recorded as a dichotomous variable indicating male or female. Due to the relatively homogeneous race background of the study population and the comparatively small number of non-White participants, the race variable used in this analysis was a dichotomous variable with one level, indicating that the individual was White only, no other race and the other level indicating non-White race or mixed race. Ethnicity was recorded dichotomously; the variable indicated whether an individual was of Hispanic ethnicity. Age at application was a continuous variable recorded in 1-year increments. Income at application was a continuous variable which summed all reported forms of income at the time of application, including monthly earnings at application, SSI income at application, SSDI income at application, Temporary Assistance for Needy Families (TANF) income at application, and other income at application.

The VR service predictor variables were drawn from a collection of 22 rehabilitation service variables that were contained in the RSA-911 data set. One service variable, college or university training, was a constant in this study (i.e., every case included in the study had received the service). Table 3 summarizes the number of VR cases in the study that received or did not receive each of the rehabilitation services that were utilized as predictor variables in the study.

Table 3.

Rehabilitation Services Provided to Participants.

Rehabilitation service	n	%
Assessment
Provided	890	72.9
Not provided	331	27.1
Diagnosis and treatment of impairments
Provided	533	42.7
Not provided	688	56.3
Vocational rehabilitation counseling and guidance
Provided	844	69.1
Not provided	377	30.9
Occupational or vocational training
Provided	171	14.0
Not provided	1,050	86.0
On-the-job training
Provided	27	2.2
Not provided	1,194	97.8
Basic academic remedial or literacy training
Provided	32	2.6
Not provided	1,189	97.4
Job readiness training
Provided	136	11.1
Not provided	1,085	88.9
Disability-related augmentative skills training
Provided	42	3.4
Not provided	1,179	96.6
Miscellaneous training
Provided	191	15.6
Not provided	1,030	84.4
Job search assistance
Provided	325	26.6
Not provided	896	73.4
Job placement assistance
Provided	415	34.0
Not provided	806	66.0
On-the-job supports
Provided	93	7.6
Not provided	1,128	92.4
Transportation services
Provided	558	45.7
Not provided	663	54.3
Maintenance
Provided	349	28.6
Not provided	872	71.4
Rehabilitation technology
Provided	237	19.4
Not provided	984	80.6
Reader
Provided	7	0.6
Not provided	1,214	99.4
Interpreter
Provided	3	0.2
Not provided	1,218	99.8
Personal attendant
Provided	15	1.2
Not provided	1,206	98.8
Technical assistance
Provided	29	2.4
Not provided	1,192	97.6
Information and referral services
Provided	280	22.9
Not provided	941	77.1
Other services
Provided	529	44.3
Not provided	692	56.7

Procedure

To obtain additional contextual information on the sample, prior to conducting the main analysis designed to answer the primary research question, the authors employed ANOVA to examine differences in weekly earnings at case closure and differences in case service costs. Mean weekly earnings at closure were compared for three groups who completed degrees or certificates while their rehabilitation cases were open: those who advanced to an associate’s degree or vocational/technical certificate, those who advanced to a bachelor’s degree, and those who advanced to a master’s degree or higher. It was expected that those with master’s degrees or higher would have the highest mean weekly earnings, followed by those with bachelor’s degrees. Case service costs were then compared for the same three groups. It was expected that those with master’s degrees or higher would have the highest case costs, followed by those with bachelor’s degrees.

To address the primary research questions, a hierarchical logistic regression procedure was employed to identify(a) a prediction model consisting of demographic and case service variables, and (b) demographic and VR case services significantly associated with competitive employment. Logistic regression facilitates prediction of a categorical outcome (e.g., whether an individual was competitively employed when his or her VR case was closed) from a set of predictor variables that may be categorical or a combination of categorical and continuous variables (Mertler & Vannatta, 2005). The first block of the regression procedure consisted of entering demographic variables into the regression model using standard logistic regression. When standard logistic regression procedures are employed, variables are entered into the regression equation and remain in the model (i.e., they are not removed at any subsequent step in the analysis). The second block of the regression procedure consisted of entering the VR service variable into the regression model using backward stepwise logistic regression. A stepwise approach was utilized with the second block of variables (the VR services) as the investigation was exploratory in nature and not theory driven, given the limited amount of research addressing the association between postsecondary education completion and competitively employment outcomes for state-federal VR participants with TBIs. When conducting exploratory work, stepwise approach can be used to examine many predictors and identify covariates with significant associations with the outcome under investigation (Hosmer & Lemeshow, 2000; Tabachnick & Fidell, 2001).

When stepwise logistic regression procedures are employed, statistical criteria are used to determine which predictors are included or excluded from the logistic regression equation (Tabachnick & Fidell, 2001). Backward stepwise logistic regression begins with all specified predictors in the equation; statistical criteria are then employed over a number of steps to remove variables from the equation (although variables removed at one step may reenter the equation at a later step) to arrive at a parsimonious model that predicts the criterion variable (Hosmer & Lemeshow, 2000).

VR participant data were imported into SPSS (Statistical Package for the Social Sciences) Version 22, a quantitative data analysis program. Each of the 21 case service variables was coded dichotomously (service provided or service not provided). The outcome variable, competitive employment, was also coded dichotomously (competitively employed or not competitively employed). The ratio of cases to predictor variables exceeded the minimum ratio of 10:1 described by Hosmer and Lemeshow (1989).

Backward stepwise logistic regression was employed using a criterion for inclusion in the model of p = .15 and a criterion for exclusion of p = .20. Hosmer and Lemeshow (1989) indicated that the customary significance level of .05 is too restrictive for stepwise regression and recommend using a criterion for inclusion between .15 and .20. The criterion for exclusion must be greater than the criterion for inclusion in the model to avoid including and excluding the same variable in consecutive steps; for that reason, a criterion for exclusion of p = .20 was established. The maximum number of stepwise iterations was limited to 20, which is the default maximum value in SPSS.

Results

Proportion Competitively Employed

Of the 1,221 VR participants with a primary cause of impairment of TBI who received college training and advanced their postsecondary education level, 922 (75.5%) individuals were competitively employed at case closure. The remaining 299 (24.5%) individuals were not competitively employed when their case was closed.

Average Weekly Earnings at Closure for Individuals Completing Degrees or Certificates

Of the 943 cases with values recorded for weekly earnings at closure, the mean weekly earnings at closure were US$384.14 (SD = US$242.97) for individuals with an associate’s degree or vocational/technical certificate, US$498.93 (SD = US$300.37) for individuals with bachelor’s degrees, and US$638.41 (SD = US$335.24) for individuals with a master’s degree or higher. A one-way ANOVA robust to violations of the homogeneity of variance assumption (Welch ANOVA) indicated that there were statistically significant differences between groups (p < .001). Post hoc tests (Games-Howell) indicated that each of the group means was significantly different from the others. The average number of hours worked per week at closure was 31.58 (SD = 11.55) for individuals with an associate’s degree or vocational/technical certificate, 33.80 (SD = 10.30) for individuals with bachelor’s degrees, and 33.06 (SD = 10.35) for individuals with a master’s degree or higher.

Average Cost of Purchased Services

The average cost of purchased services (inclusive of all rehabilitation services purchased for individuals whether costs were related to postsecondary education) was US$10,748.40 (SD = US$15,768.31) for individuals with an associate’s degree or vocational/technical certificate, US$18,634.62 (SD = US$20,128.87) for individuals with bachelor’s degrees, and US$27,881.56 (SD = US$23,835.56) for individuals with a master’s degree or higher. A one-way ANOVA robust to violations of the homogeneity of variance assumption (Welch ANOVA) indicated that there were statistically significant differences between group means (p < .001). Post hoc tests (Games-Howell) indicated that each of the group means was significantly different from the others.

Predictors of Competitive Employment

With respect to Research Question 1 (After controlling for the contributions of demographic variables, which combination of VR services variables best predicts competitive employment outcomes of VR participants with TBI who advanced their postsecondary education?), the results of the logistic regression analysis appear in Table 4. The stepwise logistic regression procedure terminated after 16 stepwise iterations in a model with all five demographic variables and six service variables. The model predicted a modest but statistically significant proportion of the variance (approximately 10.0%) in the competitive employment outcome variable (p < .001). The final prediction model included the following VR services: vocational rehabilitation counseling and guidance, job readiness training, job search assistance, job placement assistance, personal assistance services, and interpreter services. The following services were excluded from the final prediction model: Assessment, Diagnosis and Treatment of Impairments, Occupational or Vocational Training, On-the-Job Training, Basic Academic Remedial or Literacy Training, Disability-Related Augmentative Skills Training, Miscellaneous Training, On-the-Job Supports, Transportation Services, Maintenance, Rehabilitation Technology, Reader Services, Technical Assistance Services, Information and Referral, and Other Services. Additional items of note in Table 4 are the regression coefficient, standard error, and OR associated with Interpreter Services, which are considerably different than the other values in the table. It is important to note that Interpreter Services was clearly not a statistically significant predictor of competitive employment by itself, and the values obtained are likely attributable to both the limited number of cases that were provided with interpreter services and the fact that every individual who was provided with interpreter services was competitively employed at case closure.

Table 4.

Summary of Logistic Regression Procedure.

Predictor	β	SE	Wald	p	Exp(β)
Gender	−.031	0.140	0.050	.822	0.969
Race	.289	0.207	1.944	.163	1.335
Hispanic ethnicity	−.550	0.240	5.266	.022	0.577
Age	−.045	0.007	44.027	.000	0.956
Income at application	.000	0.000	0.017	.897	1.000
Counseling and guidance	.318	0.148	4.611	.032	1.375
Job readiness training	−.650	0.227	8.236	.004	0.522
Job search assistance	.240	0.178	1.816	.178	1.271
Job placement assistance	.470	0.163	8.302	.004	1.600
Personal attendant services	−1.382	0.555	6.189	.013	0.251
Interpreter services	20.291	23,150.587	0.000	.999	649,185,279.6

One of the functions of logistic regression is to predict group membership, in this case to predict whether cases would be competitively employed at case closure based upon patterns of VR services received or not received by individuals. Overall, the prediction model correctly classified 75.4% of the cases as competitively employed or not competitively employed. The correct classification rate for cases that were competitively employed was 97.2% (896 of 922 cases); however, the correct classification rate for cases that were not competitively employed was much lower, at 8.4% (25 of 274 cases). It should be noted that in this case, the prediction model’s correct classification rate was not an improvement upon simply predicting that all participants would be competitively employed at case closure, as 922 (75.5%) of the 1,221 participants were competitively employed when their cases were closed.

The second research question was, within the prediction model, which specific VR services and demographic characteristics have the strongest association with competitive employment for VR participants with TBIs who advanced their postsecondary education? In the regression model, the contributions of the VR service variables were assessed after the demographic variables had already been entered into the model. Of the demographic variables in the final prediction model, two of these variables were statistically significant predictors of competitive employment using a criterion of p ≤ .05: ethnicity and age at application. The ORs associated with these demographic characteristics (which appear under the column titled Exp(β) in Table 4) indicate that an individual of Hispanic ethnicity was only 0.58 times as likely to be competitively employed (i.e., less likely to be competitively employed) at case closure as an individual not of Hispanic ethnicity. The OR associated with age at application was 0.956, which indicated that the likelihood of being competitively employed at case closure decreased with each 1-unit (i.e., 1 year) increase in age at the time of application.

Of the case service variables in the final prediction model, the following four VR services were statistically significant predictors of competitive employment using a criterion of p ≤ .05: vocational rehabilitation counseling and guidance, job readiness training, job placement assistance, and personal assistance services. The ORs associated with these VR services indicate that an individual who received vocational rehabilitation counseling and guidance was approximately 1.4 times more likely to be competitively employed when his or her case was closed than an individual who did not receive the service. An individual who received job readiness training was less likely to be competitively employed when his or her case was closed than an individual who did not receive the service (OR = 0.52). Individuals who received job placement assistance were 1.6 times more likely to be competitively employed at case closure than individuals who did not receive the service. And finally, an individual who received personal assistance services was much less likely to be competitively employed when his or her case was closed than an individual who did not receive the service (OR = 0.25).

Discussion

Analyses of differences in costs of purchased services and weekly earnings at case closure by level of education attained indicated significant differences between groups on both measures. Case costs increased significantly as the level of education completed increased; in addition, weekly earnings at closure increased as the level of education completed increased, suggesting that the investment of resources may be warranted.

Of those who completed a degree or certificate that increased their highest level of postsecondary education completed, 75.5% were competitively employed at case closure. This figure is considerably higher than the rate of competitive employment among all individuals with an Individualized Plan for Employment who had a primary cause of impairment of TBI during the same time period, which was 45.6%. The competitive employment figure of 75.5% is higher than the successful closure rates of 57% to 60% reported by Wehman, Targett, and Kregel (2005) and higher than the integrated employment rate of 50% reported by Catalano et al. (2006). It is important to note that the population investigated in the present study consisted entirely of individuals who received college or university training and advanced their highest level of postsecondary education completed while their case was open; a consideration which distinguishes the study population from many of the individuals in the earlier studies of employment outcomes of individuals with TBIs.

The demographic characteristics and VR services investigated in this study predicted a statistically significant, but relatively small proportion of the variance in employment outcomes of the participants. It should be noted that while the VR service data permitted us to determine whether a specific service was provided, no information was available indicating the intensity, duration, quality, or timing of the specific services. We speculate that more robust measures of the service variables would facilitate the development of prediction models that explain a greater proportion of the variance in employment outcomes among those who advanced in their postsecondary education attainment. Nonetheless, a key implication of our findings is that college training alone may not be sufficient to facilitate a successful employment outcome for persons with TBIs. These individuals likely benefit from additional VR supports.

VR counseling and guidance demonstrated a significant association with competitive employment for persons with TBIs who increased their postsecondary education levels. It is perhaps not surprising that provision of this service is associated with favorable employment outcomes for VR consumers with TBIs. Individuals with TBIs often face a difficult process of psychosocial adjustment to disability. Preparing for, obtaining, and retaining employment may intensify the barriers to psychosocial adjustment as individuals seek to return to their predisability lives, work identities, and career plans. Due to the nature and severity of the injury, some individuals may not fully understand the ramifications of their brain injury. This lack of insight presents additional challenges to successful adjustment. Furthermore, elevated levels of depression and risk of suicide are present in the population of individuals with TBIs. This constellation of challenges underscores the importance of helping individuals to manage the transition to employment, and the results of this inquiry highlight the critical role that counseling and guidance may play in that process.

Job placement assistance has been identified in a number of prior studies as a significant predictor of VR outcomes for individuals with a variety of disabilities (Boutin & Wilson, 2009; Chan et al., 2011; Gamble & Moore, 2003; Rogers et al., 2005; Tucker, 2007). That it surfaces regularly as a predictor of rehabilitation outcomes is likely due to the utility of the service itself and the conditions under which it is offered to VR consumers. Job placement assistance is usually provided after a working relationship has been established with one or more VR practitioners, a plan for employment has been developed, and services deemed necessary to the realization of the plan delivered. The participant has also participated actively in his or her rehabilitation plan, and the counselor and consumer have determined that the consumer is “job ready.” Beyond the provision of the service itself, it is likely that there are additional characteristics or attributes (e.g., motivation, interpersonal skills, organizational skills, access to additional supports and resources) that are associated with provision of job placement assistance and might themselves be understood as significant predictors of employment outcomes if they were to be measured and assessed for their predictive utility. Effective job placement support helps direct consumers with TBIs to attain jobs that match their postsecondary attainment, transferable skills, postinjury function, and knowledge. It is likely that without this support, many individuals with TBIs would not be able to utilize the training and knowledge acquired through VR-supported college training.

The provision of two VR services in the model was significantly associated with a decreased likelihood of obtaining competitive employment for individuals with TBIs: job readiness training and personal assistance services. Job readiness training is defined as “training to prepare an individual for the world of work (e.g., appropriate work behaviors, getting to work on time, appropriate dress and grooming, increasing productivity)” (Rehabilitation Services Administration, 2012, p. 25). Personal assistance services are services designed to help individuals with disabilities to manage activities of daily living at work and elsewhere. It is possible that persons with TBIs represented in this study who required these services had high levels of post-TBI impairment which could have presented substantial barriers to searching for jobs, getting hired, and maintaining employment regardless of their postsecondary accomplishments.

With respect to the demographic variables in the prediction model, when considered individually gender, race (White only or non-White/mixed race), and income at application did not predict a statistically significant proportion of the variance in competitive employment. Ethnicity (which indicated whether an individual was or was not of Hispanic ethnicity) and age at application did predict significant proportions of the variance in competitive employment. Being of Hispanic ethnicity and being older at the time of application were associated with decreased likelihoods of competitive employment at case closure. These attributes cannot be influenced or changed by the client or the counselor, and should not guide counselors’ decisions about whether to provide college training services. Rather, when these attributes are present, they may serve as indicators that additional supports and services could be necessary to increase the likelihood of competitive employment for these individuals.

Limitations

Conclusions drawn from this inquiry should be done so with the understanding that the ex-post-facto nature of the research methods can be used to develop an understanding of associations or relationships between variables but should not be used to imply cause-and-effect relationships. Furthermore, the nature of the stepwise logistic regression model, which relies upon the associations between many variables, can present difficulties with replication. Data analyzed in this study represent VR cases closed during one 5-year period, and data from the same state/federal VR system drawn from different years may differ considerably with respect to VR service provision and employment outcomes.

It is important to note that when backward stepwise logistic regression models are employed, it is essential to avoid misinterpreting the exclusion of a predictor variable from a prediction model as an indication that the predictor is not meaningfully associated with the outcome variable. In fact, in many instances predictors that have strong associations with the outcome variable may be excluded from the final prediction model by other predictors or combinations of predictor variables that explain much of the same variance in the outcome variable.

An additional limitation pertains to the statistical power that is associated with data sets as large as the Rehabilitation Services Administration case records examined in this study. Wheaton and Wilson (1996) stated that when working with large samples, it can be easy to achieve high levels of statistical power. In response to this concern, interpretation of the results of this study focused upon both the significance and the practical implications of the findings.

The way that participants were identified for inclusion into the study presents some limitations. The ordinal scale used to identify the highest level of education completed at application and at case closure makes it impossible to identify individuals who complete another equivalent degree (e.g., an individual who has a bachelor’s degree when he or she applies for VR services who then proceeds to complete a second bachelor’s degree), or those who go from one level of graduate degree to another (e.g., an individual who advances from a master’s degree at application to a doctoral degree at case closure). As a result of this limitation, these individuals, although likely few in number, were excluded from this study.

The influence that severity of disability may have had on the study is unknown. The RSA-911 data set does not contain reliable indicators that can be used effectively to assess severity of disability. It is possible that those who completed postsecondary degrees and certificates represent individuals served by state-federal VR programs who have milder forms of impairment with respect to the functions that are closely associated with successfully navigating the college experience. Severe impairments in other areas of function, however, may not inhibit postsecondary success to the same extent. Incorporating one or more valid indicators of the severity of an individual’s disability or disabilities into the case service records would permit more robust examination of the influence of severity of disability and its associations with rehabilitation services and outcomes.

Finally, the analysis did not assess differences in utilization of support services provided by the postsecondary institutions the participants attended. It is possible that campus-based supports such as academic accommodations might be related to the participants’ ability to successfully complete postsecondary education and subsequently find employment.

Conclusion

In this study, the significance of the VR counseling function is supported, as the service had a significant association with favorable vocational outcomes for consumers with TBIs in the VR system. The provision of job placement assistance, which has emerged as a significant predictor of employment in studies of other VR populations, has likewise been found to be associated with desirable employment outcomes for the participants in the present study. VR cases which include the provision of college or university training represent significant investments of resources (including time, effort, and money) on the part of the individual and the VR service providers. The resource investments are often even greater when an individual’s educational endeavors rise to the level of advancing their postsecondary education attainment. Given the significance of the investments made by all parties, it is essential that individuals be offered services known to be associated with favorable employment outcomes.

An incidental finding of note pertains to the number of individuals with a primary cause of impairment of TBI among the cases closed during the 2009–2013 federal fiscal years. Of the 2,959,994 cases closed, only 46,806 (1.6%) had a primary cause of impairment of TBI. The magnitude of the incidence of TBI, when compared with the number of closed cases in the RSA-911 data, suggests opportunities to serve a much larger proportion of individuals living with the effects of TBI. The implications of failing to meet the support needs of individuals with TBIs are substantial as the consequences of this disability are often severe, and in many cases limitations are lasting; they do not diminish over time. For this reason, the involvement of rehabilitation professionals, who can help individuals, families, and employers to address limitations and maximize the independence and vocational functioning of a person with brain injury, is essential to maximizing human potential.

Recommendations for Future Research

We recommend replicating this study with other causes of impairment to determine the extent to which services identified as significant predictors for individuals with TBIs are similar or different for VR consumers with other types of disabilities who advance their postsecondary education. When possible, we suggest that studies build upon this work by incorporating more robust measures of the VR services provided to VR participants, as the relatively superficial way that VR services were recorded in the data set utilized for this study likely leads to research findings that underestimate the true impact of the services. Likewise, more complete information on the VR participants (such as information pertaining to severity of disability and functional limitations) would permit examination of the interactions between personal and VR service variables.

We also suggest further examination of job readiness training and personal assistance services as predictors of employment outcomes. Exploring the association between the need for these services and other barriers known to moderate employment outcomes may help researchers and practitioners to better understand the extent to which these variables serve as indicators of additional rehabilitation needs. Finally, we recommend conducting longitudinal studies to examine changing expectations about postsecondary education for individuals with TBIs and the changes in services or supports that are available to assist postsecondary education students with disabilities.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Baker

K. A.

Tandy

C. C.

Dixon

D. R.

(2002). Traumatic brain injury: A social worker primer with implications for practice [Electronic version]. Journal of Social Work in Disability & Rehabilitation, 1(4), 25–42.

Berger

Leven

Pirente

Bouillon

Neugebauer

(1999). Quality of life after traumatic brain injury: A systematic review of the literature [Electronic version]. Restorative Neurology and Neuroscience, 14, 93–102.

Bolton

B. F.

Bellini

J. L.

Brookings

J. B.

(2000). Predicting consumer employment outcomes from personal history, functional limitations, and rehabilitation services. Rehabilitation Counseling Bulletin, 44, 10–21.

Boutin

D. L.

Accordino

M. P.

(2009). Exploring postsecondary education and competitive employment for people with mental illness. Journal of Applied Rehabilitation Counseling, 40(2), 13–21.

Boutin

D. L.

Wilson

(2009). An analysis of vocational rehabilitation services for consumers with hearing impairments who received college or university training. Rehabilitation Counseling Bulletin, 52, 156–166.

Brown

(2011). Career information, career counseling, and career development (10th ed.). Boston, MA: Allyn & Bacon.

Capella-McDonnall

M. E.

(2005). Predictors of competitive employment for blind and visually impaired consumers of vocational rehabilitation services. Journal of Visual Impairment and Blindness, 99, 303–315.

Catalano

Pereira

A. P.

Chan

(2006). Service patterns related to successful employment outcomes of persons with traumatic brain injury in vocational rehabilitation. NeuroRehabilitation, 21, 279–293.

Centers for Disease Control and Prevention. (2010). Traumatic brain injury in the United States: Emergency department visits, hospitalizations and deaths 2002-2006. Retrieved from http://www.cdc.gov/traumaticbraininjury/pdf/blue_book.pdf

10.

Centers for Disease Control and Prevention. (2015). Traumatic brain injury in the United States: Fact sheet. Retrieved from http://www.cdc.gov/traumaticbraininjury/get_the_facts.html

11.

Centers for Disease Control and Prevention. (2016). Severe TBI. Retrieved from http://www.cdc.gov/traumaticbraininjury/severe.html

12.

Chan

J. Y.

Keegan

J. P.

Ditchman

Gonzalez

Zheng

L. X.

Chan

(2011). Stigmatizing attributions and vocational rehabilitation outcomes of people with disabilities. Rehabilitation Research, Policy, and Education, 25, 135-148.

13.

Cooper

(2003). Impact of trends in primary, secondary, and postsecondary education on applications to medical school. II: Considerations of race, ethnicity, and income. Academic Medicine, 78, 864–876.

14.

Corrigan

J. D.

Bogner

J. A.

Mysiw

W. J.

Clinchot

Fugate

(2001). Life satisfaction after traumatic brain injury [Electronic Version]. The Journal of Head Trauma Rehabilitation, 16, 543–555.

15.

Degeneffe

C.E.

(2001). Family caregiving and traumatic brain injury. Health & Social Work, 26, 257–268.

16.

Degeneffe

C. E.

Boot

Kuehne

Kuraishi

Maristela

Noyes

… Will

(2008). Community-based interventions for persons with traumatic brain injury: A primer for rehabilitation counselors. Journal of Applied Rehabilitation Counseling, 39(1), 42–52.

17.

Degeneffe

C. E.

Lee

(2015). Brain injury and the family. A guide for rehabilitation counselors. In Millington

Marini

(Eds.), Families in rehabilitation counseling: A community-based model (pp. 153–170). New York, NY: Springer.

18.

Degeneffe

C. E.

Tucker

(2014). Community-based support and unmet needs among families of persons with brain injuries: A mixed methods study with the Brain Injury Association of America state affiliates. In Wadsworth

S. M.

Riggs

D. S.

(Eds.), Military deployment and its consequences for families (pp. 293–313). New York, NY: Springer.

19.

Degeneffe

C. E.

Tucker

Griffin

(2015). Families of veterans with polytrauma injuries in the United States: A guide for rehabilitation counseling practice. In Millington

M. J.

Marini

(Eds.), Families in rehabilitation counseling: A community-based rehabilitation approach (pp. 213–230). New York, NY: Springer.

20.

Gamble

Moore

C. L.

(2003). The relation between VR services and employment outcomes of individuals with traumatic brain injury. Journal of Rehabilitation, 69(3), 31–38.

21.

Hosmer

D. W.

Lemeshow

(1989). Applied logistic regression. New York, NY: Wiley.

22.

Hosmer

D. W.

Lemeshow

(2000). Applied logistic regression (2nd ed.). New York, NY: Wiley.

23.

James

DeVivo

M. J.

Richards

J. S.

(1993). Postinjury employment outcomes among African-American and White persons with spinal cord injury. Rehabilitation Psychology, 38, 151–164.

24.

Jordan

B. D.

(2013). The clinical spectrum of sport-related traumatic brain injury. Nature Reviews Neurology, 9, 222–230.

25.

Kennedy

M. R.

Krause

M. O.

Turkstra

L. S.

(2008). An electronic survey about college experiences after traumatic brain injury. NeuroRehabilitation, 23, 511–520.

26.

Kim

(2012). Exploring the relationship between state financial aid policy and postsecondary enrollment choices: A focus on income and race differences. Research in Higher Education, 53, 123–151.

27.

Levack

McPherson

McNaughton

(2004). Success in the workplace following traumatic brain injury: Are we evaluating what is most important? Disability and Rehabilitation, 26, 290–298.

28.

Mertler

C. A.

Vannatta

R. A.

(2005). Advanced and multivariate statistical methods (3rd ed.). Glendale, CA: Pryczak.

29.

Nakaji

(2014). Vocational rehabilitation outcomes for deaf and hard of hearing consumers who were supported by the Rehabilitation Services Administration to pursue postsecondary education. Available from ProQuest Dissertations and Theses database. (3632592)

30.

National Data and Statistical Center, Traumatic Brain Injury Model Systems. (2015). National database: 2015 profile of people within the Traumatic Brain Injury Model Systems. Retrieved from https://www.tbindsc.org/StaticFiles/Documents/2015%20TBIMS%20National%20Database%20Update.pdf

31.

Prigatano

G. P.

(1995). The problem of lost normality after brain injury. The Journal of Head Trauma Rehabilitation, 10(3), 87–95.

32.

Rehabilitation Services Administration. (2012). Reporting manual for the case service report (RSA-911). Washington, DC: Rehabilitation Services Administration.

33.

Rehabilitation Services Administration. (2013). Case service record report (RSA-911) (PD-13–05). Washington, DC: Rehabilitation Services Administration.

34.

Roessler

R. T.

Rumrill

P. D.

Fitzgerald

S. M.

(2004). Predictors of employment status for people with multiple sclerosis. Rehabilitation Counseling Bulletin, 47, 96–103.

35.

Rogers

J. B.

Crystal

R. M.

Bishop

(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.

36.

Tabachnick

B. G.

Fidell

L. S.

(2001). Using multivariate statistics (4th ed.). Needham Heights, MA: Allyn & Bacon.

37.

Taylor

L. A.

Kreutzer

J. S.

Demm

S. R.

Meade

M. A.

(2003). Traumatic brain injury and substance abuse: A review and analysis of the literature. Neuropsychological Rehabilitation, 13, 165–188.

38.

Tekleselassie

Mallery

Choi

(2013). Unpacking the gender gap in postsecondary participation among African Americans and Caucasians using hierarchical generalized linear modeling. The Journal of Negro Education, 82, 139–156.

39.

Todis

Glang

Bullis

Ettel

Hood

(2011). Longitudinal investigation of the post-high school transition experiences of adolescents with traumatic brain injury. The Journal of Head Trauma Rehabilitation, 26, 138–149.

40.

Tucker

(2007). Predictors of competitive employment for vocational rehabilitation participants who complete undergraduate or graduate degrees (Doctoral dissertation). Available from ProQuest Dissertations and Theses database. (3280275)

41.

U.S. Census Bureau. (2012). Work-life earnings by field of degree and occupation for people with a Bachelor’s degree: 2011. Retrieved from https://www.census.gov/prod/2012pubs/acsbr11-04.pdf

42.

Waghorn

Chant

(2002). Clinical and non-clinical predictors of vocational recovery for Australians with psychotic disorders. Journal of Rehabilitation, 68(4), 40–51.

43.

Wehman

Targett

West

Kregel

(2005). Productive work and employment for persons with traumatic brain injury: What have we learned after 20 years? The Journal of Head Trauma Rehabilitation, 20, 115–127.

44.

Wheaton

J. E.

Wilson

K. B.

(1996). The relationship between vocational rehabilitation services and the consumer’s sex, race, and closure status. Rehabilitation Counseling Bulletin, 40, 116–133.

45.

Willmott

Ponsford

Downing

Carty

(2014). Frequency and quality of return to study following traumatic brain injury. The Journal of Head Trauma Rehabilitation, 29, 248–256.