Acceptance rates of African-American versus white consumers of vocational rehabilitation services: a meta-analysis.
African-Americans comprise about 12% (34-35 million) of the total United States population and represent the second largest minority group in the nation (US Census. 2000). Although African-Americans have experienced much progress in improving opportunities in the past 20 years, they still lag behind other segments of the population in many respects. In fact, the most recent US Census Bureau data (2000) indicates that that African-Americans are subject to some of the most severe unemployment of any racial or ethnic group in American society with the possible exception of American Indians. More specifically, 22.9% of individuals reporting as African Americans in the last U.S. Census were living below poverty levels in contrast with 11.6% for Whites. In addition, individuals reporting as African Americans in the last U.S. Census were much more likely to have disabilities than other racial groups, again with exception to American Indians: 24.3% of African Americans and American Indians reported having disabilities. This is in contrast with 18.5% of Whites despite the fact that the median age of Whites was significantly older than the other groups. These recent census findings support other researchers' contentions that ethnicity is related to the overall disability rates in the United States (Allen, 1976: Bowe, 1984: Hayes-Bautista, 1992; U.S. Department of Education, 1902: Walker, Adbury, Maholmes, & Rackely, 1992).
The demographic characteristics within the U. S. reflect an increasingly diverse population. Recent census data indicates that African Americans. Hispanic Americans, and Asian Americans presently comprise approximately 33% of the U. S. population, and the U.S. Census Bureau has announced that by the year 2010, Whites seem certain to be a distinct numerical minority (U.S. Department of Labor, 2000). Thus, the workplace as we now know it will become increasingly diverse with composition changing from mainly Whites to mainly women and individuals of other racial and ethnic groups.
Since African Americans comprise a largo, significant, and growing minority group with a high prevalence of disability, it is anticipated that African Americans will increasingly come into contact with rehabilitation agencies. Thus, it is important to study the manner in which they are accepted into rehabilitation services. If African American clients are actually denied rehabilitation services in disproportion to White counterparts, then the rehabilitation needs of African Americans may not be getting met, their potential may be limited, and opportunities may be denied them that would have led to success in education, job training, and employment (Rosenthal & Berven, 1999). Certainly many variables other than race may influence eligibility determination (Bolton & Cooper, 1980; Wheaten: 1995; Wilson, 2000); however, legislative mandates such as the 1992 and 1998 amendments to the Rehabilitation Act of 1973 as well as continued attention to the potential for other inequitable service patterns for minority consumers suggest that racial discrepancies in MR acceptance remains an issue in many states (Wilson, 2000).
The first study to compare VR outcomes among African Americans and Whites in the U. S. was conducted by Atkins and Wright (1980). These researchers found that African Americans were accepted less frequently for VR services than were Whites in most Rehabilitation Services Administration (RSA) regions. Interestingly, after reanalyzing the data from the Atkins and Wright study, Bolton and Cooper (1980) questioned whether the results (a 5.5% difference in acceptance rates) were a true indication of differential acceptance rates for African Americans. Bolton and Cooper also questioned if VR acceptance is influenced more by variables associated with a lower socioeconomic status (e.g., education or vocation), than by ethnic group membership. Although the findings of the Atkins and Wright study were challenged as not being conclusive, the reasons for eligibility determination discrepancies between White and African Americans remain a highly debated issue.
Herbert and Martinez (1992) investigated whether ethnicity (Native American/Alaskan Native, Asian/Pacific Islander, African American, or White) had an effect oil acceptance rates. The authors found that African Americans and Hispanics were more likely to be found ineligible liar VR services than any other underserved and underrepresented group under investigation, and that Whites were more likely to be accepted for VR services than were their African American counterparts.
In 1993, Dziekan and Okocha explored the accessibility of VR services for African Americans, Hispanics, Native Americans, Asian Americans, and Whites. Their investigation supported previous findings (Atkins & Wright, 1980: Herbert & Martinez, 1992) indicating that African Americans were less likely to be found eligible for VR services than their White counterparts. Shortly thereafter, Feist-Price (1995) reported results indicating comparable patterns of inequitable acceptance rates for African Americans. Thus, with exception to the questions raised by Bolton and Cooper (1980), all initial studies investigating VR acceptance and ethnicity (1980 to 1995) suggested differential acceptance rates existed for African Americans in contrast to White counterparts.
Contrary to the findings of initial studies, Wheaten (1995) concluded "the proportions of Whites and African Americans found eligible for VR services are not significantly different statistically" (p. 228). Wheaten questioned results of previous studies due to research design and statistical procedures. Differing from the analyses used in the past studies, Wheaten employed a symmetrical hypothesis and controlled for homogeneity of proportions within sampling procedures and found that although Whites had higher acceptance rates than African Americans, there was no significant difference in the proportion of acceptance rates for rehabilitation services of White clients versus African American. Wilson's study (1999) supported Wheaton's findings. reporting no significant differences between African Americans and Whites in VR acceptance rates. The contradictory findings revealed questions whether African Americans are accepted less for VR services than Whites, and several of the authors stated a need for further investigation of VR eligibility and acceptance rates.
The studies conducted in the past several years have not brought consensus to the issue of acceptance rates. Wilson (2000) included three other predictor variables (education, work status, and source of support at application) to determine which variables would predict VR acceptance. Using a stepwise regression analysis, Wilson found "primary source of support at referral" and "ethnicity" to be statistically significant. Although Wilson employed a multivariate analysis while previous researchers used univariate analyses (e.g., Atkins & Wright, 1980; Wheaten, 1995), his results supported the earlier contentions that African Americans were less likely than Whites to be accepted for VR services (Atkins & Wright: Dziekan & Okocha, 1993: Feist-Price, 1995: Herbert & Martinez, 1992). Supporting Wilson's (2000) findings, Wilson, Harley, and Alston (2001) replicated the Wilson (1999) study and found Whites are more likely to be accepted for VR services than African Americans. However, a subsequent, national investigation by Wilson, Alston, Harley, and Mitchell (2002) found African Americans to be over twice as likely to be accepted for services than Whites.
Lastly, a recent examination of RSA 911 data from fiscal year 2001 (Chan, Wong, Rosenthal, Kundu & Dutta (in press) revealed an approximate six percent difference between acceptance rates of African Americans and White consumers, favoring acceptance of White consumers, in the national data-base. This figure supports other research indicating similar differences across groups (e. g. Capella, 2002).
Does a significant difference exist in the VR acceptance between African Americans and Whites within the aggregate data from studies that met inclusion for the meta-analysis? Our primary hypothesis was that VR acceptance rates are dependent on ethnicity within the aggregate data from the included studies. More specifically, it was anticipated that the acceptance rates of Whites would be significantly greater than those of African Americans.
Twelve published articles based on RSA-911 data were identified that pertained to VR acceptance rates. The two following variables had to be present in the studies to warrant inclusion: Racial/Ethnic Membership--This predictor variable was categorical, with two levels (African American and White. Race/ethnicity was defined as the race/ethnic group reported by customers on their application for VR services (RSA, 1995). Acceptance for VR Services--This dependent variable also was a categorical variable. Closure statuses from the state or national RSA-911 data grouped the categories by the RSA definitions of acceptance and nonacceptance for VR services (RSA 1995). It is important to note that we did not have access to RSA 911 data for all fiscal years included in the study. Thus. the nine fiscal years used in the meta-analysis are derived from the acceptance literature instead of directly from the RSA 911 data.
The assembly of the present meta-analysis model was based on the work of Fleiss (1994) and Shadish and Haddock (1994). Articles for the current meta-analysis were evaluated for inclusion based on the following specific criteria (see Table 1). Articles had to report data extracted from the RAS-911 database for one or more fiscal year(s) on race and acceptance into the state VR system. Articles needed to contain race (either White or African American client) as the predictor variable and acceptance (either accepted or not accepted) as the criterion variable. Articles evaluated for inclusion into the meta-analysis required statistical tests that could be converted into proportions and then into an odds ratio. Ten studies met the inclusion criteria.
However, not all ten studies were included in the meta-analysis because of duplicated RSA-911 data for certain fiscal years. Duplicated data was present in four articles for fiscal year 1997-1998: a) Wilson, Harley, and Alston, (2001), Wilson (2002), Capella (2002), and Wilson, Alston, Harley, & Mitchell (2002); and present in two articles for fiscal year 1995-1996: a) Wilson (2000) and b) Wilson (1999). Duplicate articles (see Table 1) were excluded to reduce the threat to statistical independence (Matt & Cook, 1994). Failure to recognize dependence can result to inaccurate estimates of the aggregated effect size and the standard error (Hedges, 1994) and can inflate the overall outcome in a particular direction. It was determined that, in cases of duplication, the studies used for the meta-analysis would be chosen randomly. This is of particular importance when different studies from the same timeframes, seem to indicate different patterns of acceptance (e.g., Capella, 2002, and Wilson, Alston, Harley, & Mitchell, 2002). In each case, the studies were randomized within years they represented and one was chosen for inclusion in the analysis. Thus, one article for each fiscal year period was included in the meta-analysis, Wilson et al. (2002) for 1997-1998 and Wilson (2000) for 1996. Four studies did not have any duplication and covered: a) fiscal year 1992 (Wheaton, 1995), b) fiscal years from 1985 to fiscal year 1989 (Dziekan et al., 1993), c) fiscal year 1991 (Feist-Price, 1995), and fiscal year 1976 (Atkins & Wright, 1980). The first three articles were included in the analysis, but the Atkins & Wright (1980) was not because of data conversion problems. Several individuals knowledgeable in meta-analysis techniques unsuccessfully attempted to convert the Atkins and Wright data into proportions/odds ratio. This article therefore was excluded from the analysis. Thus, a total of five articles covering nine fiscal years were used in the meta-analysis.
There is evidence that five articles is a sufficient number for a meta-analysis. Synthesizing research with as few as three articles is legitimate (Hunt, 1997). According to Hunt if meta-analysis had been applied to the treatment of myocardial infarctions, it would have shown that after three studies, streptokinase (a thrombolytic agent, i.e. it dissolves vascular thrombi) was an ineffective treatment, saving hundreds of thousands of lives. This supports the authors' contention that the five articles utilized in the present study is a sufficient number.
Statistical Hypothesis Testing
This meta-analysis employed a fixed-effects model. A fixed effects model refers to an investigator fixing the independent variable at a particular value. Therefore the estimate of any effect size will differ due to sampling error. A fixed effect model was chosen for this study because it was assumed that the investigators in each research article fixed the predictor variable as race, either White or African American client while the criterion variable was acceptances.
The data for each predictor and criterion variable is in the form of count (the number within each racial group that was either accepted or not accepted). Count converts to proportion or ratio (a ratio scale without intervals) and into odds ratio. Both the predictor and the criterion variables are categorical so measures of effect size must fit categorical data (Fleiss, 1994) instead of effect size of mean differences (parametric measures). According to Fleiss measures of effect size for categorical data include the difference between two probabilities, the ratio of two probabilities, the Phi coefficient, and the odds ratio. Effect size for categorical data is distinguished from effect size for parametric measures. Parametric measures effect size fall into one of two families; either the family of Pearson product moment correlation or the family of indices of differences (Hedges's g, Glass's [DELTA] and Cohen's d; Rosenthal, 1994).
Odds ratio, variance, standard error, and confidence intervals were calculated by hand for each data set and for the aggregate effect size over all fiscal years. The nine individual effect sizes (per year) are presented along with their respective geographic region (see Table 2). The alpha (a) in the present study was set at .05.
The data for each fiscal year was converted into a proportion and then into an odds ratio. Odds ratios were chosen as a measure of relationship because of their interpretive value. The odds ratio becomes the effect size measure for both the individual fiscal years as well as for the aggregated effect size. It is useful for describing the results of studies that employ categorical outcomes. The odds ratio is a measure of association, but unlike other measures of association, "1.0" means that there is no relationship between the variables. The size of any relationship is measured by the difference (in either direction) from 1.0. An odds ratio less than 1.0 indicates an inverse or negative relation; an odds ratio greater than 1.0 indicates a direct or positive relation. In odds ratios the mill hypothesis uses 1.0 instead of the traditional zero in testing the null to determine if there is no association or no difference. Thus, when interpreting confidence intervals the reader must examine whether 1 is included instead of zero.
Finally, as suggested by Shadish and Haddock (1994), we conducted a homogeneity test (Q) of assumption to ascertain if the studies share a common population effect size. Once derived, the Q value is compared to a chi-square at k - 1 degree1 of freedom. If Q exceeds the critical value of chi-square, then the observed variance in study effect sizes is significantly greater than what could be expected by chance if all studies shared a common population effect size. Therefore, the rejection of the null hypothesis and acceptance of the alternative hypothesis indicates heterogeneity of effect sizes. Heterogeneity can occur in a meta-analysis in spite of a significant confidence interval over all studies.
Table 2 illustrates the results for the statistical hypothesis testing for each fiscal year. These results are presented first because they indicate the make-up for each geographic area/fiscal year and serve the basis in calculating the overall or aggregated odds ratio. Finally, results are presented from the test of homogeneity.
Statistical Hypothesis Testing for Each Fiscal Year
According to Table 2 the odds ratio reported for fiscal years 1998 through 1985 ranged between 0.46 up to 2.04. The 95% confidence intervals indicate that these odds ratios are all within significance levels, except for fiscal year 1996 where the confidence interval contains 1. The overall odds favor White over African American clients being accepted for VR services. However, an interesting result surfaced for the USA data. The odds ratio for this geographic area during fiscal year 1998 is between 1 and zero (.46). This value indicates that for the US data, the odds favor African American clients being accepted for VR services over Whites.
Statistical Hypothesis Testing for the Aggregated Effect Size
The aggregated odds ratio is 1.54 over all fiscal years (variance = .0004 and standard error = .02). This odds ratio resides within a confidence interval of 1.60 (upper limit) to 1.48 (lower limit), which indicates rejection of the null hypothesis [alpha (a) = .05 or 1.96]. This result indicates a 1.54 likelihood of acceptance for VR services for clients who are White. It is noted that generalization of this result might be limited to the geographical regions reported in this meta-analysis.
The test of homogeneity, however produced a value large enough to reject the null hypothesis that a common population effect size underlies the 9 data sets [Q = 182.91 (a = .05), ?2 (8) is 15.51]. The alternative hypothesis that the studies are heterogeneous must be accepted. Therefore, although the aggregated odds ratio for all studies was sufficiently large to reject me null hypothesis mat the population effect is zero.
Statistical Hypothesis Testing of the Impact of the Cultural Diversity Initiative
The Rehabilitation Cultural Diversity Initiative in the 1992 Rehabilitation Act Amendment signaled the federal government's intent to improve the quality of access into VR services for persons with disabilities from minority backgrounds. More specifically, Goal V of the Rehabilitation Cultural Diversity Initiative in the 1992 Rehabilitation Act Amendment was "To implement a comprehensive program to reach and serve traditionally unserved and underserved persons with disabilities." Thus, we asked the question whether the Cultural Diversity Initiative impacted the VR acceptance rates for those studies that used data collected after the passage of the Amendment. We addressed this question by examining data for two groups. We calculated both the aggregated effect size (odds ratio) for fiscal year 1992 through 1998 and the aggregated effect size (odds ratio) for fiscal years between 1984 through 1990. We hypothesized that, if the Cultural Diversity Initiative improved the quality of access into VR services for African American consumers, the effect size (odds ratio) of the fiscal years 1992-1998 would be nor demonstrate an association and be significantly closer to 1. We also hypothesized that the Q chi-square statistic would be non-significant indicating a common population effect size underling the 9 data sets; and that the 1984-1990 data, the effect size, or odds ratio, will be significantly greater than 1 and the homogeneity Q chi-square statistic will be statistically significant, meaning that the effect sizes are not equal.
The results of the subsequent data analyses indicate that the fiscal 1992-1998 data analysis resulted in a significant effect size (odds ratio = 1.15, 95% confidence interval 1.29, 1.01) but a non-significant Q [[chi square] (2, N = 20,211) = 4.98, p = .083]. The results of the fiscal 1984-1990 data analysis also resulted in a significant effect size (odds ratio = 1.63, 95% confidence interval 1.67, 1.59) and a significant Q [[chi square] (5, N = 143,645) - 372.47, p = < .0001].
The results of this study revealed statistically significant differences between VR acceptance rates for Whites versus African Americans. Whites were found to be more likely to be accepted for VR services than African Americans. There is evidence however, that prior to the 1992 amendments, acceptance rates may have been affected by other variables such as geographic region. While some of the past investigations have revealed that ethnicity accounts for a negligible amount of variance in VR acceptance (Wheaton, 1995; Wilson, 1999), this research indicates a more robust effect size. The aggregate findings are conclusive even with the inclusion of the Wilson, Alston, Harley, and Mitchell (2002) study that revealed a converse pattern in which African American clients were more likely to be accepted for services with a ratio of 2.17 chances of being accepted for VR services for African American clients over White applicants. In addition, the significant Q statistic revealed in the aggregate data indicates that there may be error as a consequence of the presence of random effects such as geographic region.
The results of the present study stand in contrast to what several earlier studies found regarding ethnicity and VR acceptance (e.g., Wheaton, 1995; Wilson, 1999) in that ethnicity and VR acceptance are independent. In contrast to the meta-analysis results, the studies conducted by Wheaton, and Wilson used state RSA- 911 databases, and the variances explained by ethnicity reported by Wheaton and Wilson were 3% and less than 1%, respectively.
Importantly, although results of the post 1992 amendment analysis are statistically significant, it is possible that the initiatives introduced by the 1992 amendments may have had a positive impact in reducing the discrepant acceptance rates between African Americans and Whites as demonstrated by an odds ratio much closer to 1. In addition, the non-significant Q statistic revealed homogeneity within the post 1992 amendment data, in contrast with the heterogeneity indicated in the overall analysis as well as the pre 1992 amendment analysis.
Although there are many variables that may influence the findings that African Americans are less likely to be accepted for VR services than are Whites, one potential influence that must be considered is the possibility discrimination against African Americans (Herbert & Martinez, 1992; Wilson, 2000; and Wilson, Harley, McCormick, Jolivette, & Jackson, 2001). Many studies within the social sciences, particularly within social cognition literature, have found that stereotypes of African Americans tend to be extremely negative (Devine & Elliot, 1995). Atkins (1986) argued that, although American society has made progress in recognition of minority issues, social and cultural pluralism is still not prevalent. Subgroups exist within American society, which are isolated in varying degrees from the dominant group. Rehabilitation counselors must give careful consideration to the notions that (a) racism is prevalent in our society; (b) minorities are not treated equally relative to their majority peers; and (c) rehabilitation counselors can be part of both the problem and the solution. Boski (1988) reported that when African Americans present themselves in ways that are consistent with negative stereotypes held by Whites, they tend to trigger or exacerbate negative evaluations. As postulated by Dziekan and Okocha (1993), a VR counselor's negative perception of a potential customer's capacity for success (or failure) may produce an inaccurate determination of the customer's ability to benefit from VR services, resulting in disproportionate numbers of underrepresented customers found ineligible for VR services.
Findings from investigations conducted by Rosenthal and Berven (1999), and Rosenthal (2004) indicate that White rehabilitation counselors may jump to conclusions about African American clients in early stages of the rehabilitation processes. Rosenthal (2004), Rosenthal and Berven (1999), and Rosenthal and Kosciulek (1996) attribute counselor biases due to client race as a major factor for inequitable treatment of people with disabilities from minority backgrounds. Such conclusions are often based on stereotypes of African Americans and are found to be resistant to change, even in the face of contradictory information. These researchers contended that counselor biases based on selected client characteristics could contribute negatively to influence diagnostic impressions and decisions about eligibility determination, plan development, and service provision for their clients. Judgments regarding client potential may determine the educational and career opportunities that clients ultimately pursue, and dramatically impact their future direction and quality of life. Such findings (Rosenthal, 2002; Rosenthal & Berven, 1999) support the contentions of authors (Dziekan & Okocha, 1993: Boski, 1988; Wilson, et al., 2001) that VR counselor bias may be a source of differential determinations that discriminate against African American VR applicants. Given that the current U.S. VR demographics indicate that about 93% of VR counselors and 92% of VR administrators classify themselves as White (Whitney-Thomas, Timmons, Gilmore, & Thomas, 1999; cited in Wilson et al., 2002), it is important to consider the potential effects of counselor bias and race.
Two true-experimental studies were conducted to examine the effects of client race on clinical judgment of White graduate students in rehabilitation counseling (Rosenthal & Berven, 1999), and practicing VR counselors (Rosenthal, 2004). In both studies, two groups of Whites were asked to review case materials for an identical client with the exception of race. For one group, the hypothetical client was reported to be White and for the other, African American. In the African American condition, the client was judged to have less potential for education and employment. Given that all case information was identical with exception to race, the group differences were attributed to racial bias (Rosenthal, 2004; Rosenthal & Berven, 1999).
The results of the Rosenthal studies are consistent with Strohmer and Leierer's (2000) review of the counselor bias literature. Counselors have been found to be prone to being susceptible to systematic biases associated with specific client variables such as gender, age, sexual preference, social class, and disability type. Phenomena such as diagnostic overshadowing may lead counselors to give undue weight on one salient variable, while disregarding or missing other important information (Spengler, Strohmer, & Prom, 1990). Once counselors formulate negative hypotheses regarding clients, they may demonstrate confirmatory bias, seeking confirmatory information while paying less attention to disconfirmatory information, even in the face of contradictory evidence (Strohmer & Shivy, 1994; Strohmer, Shivy, & Chiodo, 1990). McGinn, Flowers, and Rubin (1994) further suggested that cultural biases were, at least in part, responsible for the inequitable patterns of rehabilitation counseling service delivery for African American consumers. It is important to note that such biases may not be overt, intentional or even within consciousness of the practitioner. Research regarding the implicit nature of stereotypes indicates that even when stereotypes are not explicitly recognized or noted, implicit stereotypes can have significant influence on perceptions (Fazio & Olson, 2003), thus, influencing clinical decisions (Rosenthal & Berven, 1999).
This study is the first meta-analysis investigating VR acceptance rates between African Americans and Whites. Although the findings indicate that race does have a robust effect on VR acceptance, results should be interpreted cautiously. Most of the studies use samples of customers drawn from Midwestern state VR agencies. As indicated previously in the discussion, it is quite possible that region may serve as a variable that moderates the effects between race of the customer and VR acceptance rates. Given the fact that most of the studies are drawn from one region of the country limits the generalizabilty of the meta-analysis. The rejection of homogeneity in the aggregate analysis indicates that other variables such as regional effects are contributing to the significant effects. It is important to note that although the significant results indicated a 1.54 likelihood of being accepted for VR services for clients who are White, Wilson, Alston. Harley, & Mitchell's (2002), was a national study and revealed a converse pattern in which African American clients were more likely to be accepted for services with a ratio of 2.17 chances of being accepted for VR services for African American clients over White applicants.
It is also important to recognize that RSA-911 data has been criticized in terms of both reliability and validity. As previously noted by Wilson et al. (2002), it is not possible to control for the RSA- 911 race/ethnicity classification features. For purposes of this recta-analysis, we applied the term race using the federal government terminologies and definitions. The federal government still uses the term race when referring to African Americans, Whites, Native Americans/Alaskan Natives, and Asians/ Pacific Islanders (Rehabilitation Services Administration, 1995).
Another limitation of this investigation is the lack of uniformity in delineating closure statuses. The definitions of closure status are coded on the national RSA-911 reporting form. In the first national study (Atkins & Wright, 1980), it is not clear how the variables indicating closure status were coded. Thus, this study was not included in the present recta-analysis. Most of the studies that used state data used the explanatory and criterion variables of Status 08 (not accepted for VR services) and Status 10 (accepted for VR services) (e.g., Wheaton, 1995; Wilson, 1999). This differs from the Wilson et al. (2002) national investigation, in which the researchers collapsed closure statuses. In the Wilson et al. (2002) study, VR applicants were labeled (categorized) 1 through 6, and collapsed into categories as defined by RSA (Rehabilitation Services Administration. 1995). Thus, in the Wilson et al. (2002) study, the criterion variable of VR acceptance included two levels: Status 08 from 02 and Status 08 from 06 were coded as 0 (not accepted for VR services). Statuses 38 from 04, 28, and 30 were coded as 1 (accepted for VR services). Wilson et al. (2002) noted that "As a result, not only were customers who were accepted for VR services included in Category 2 but also customers who had their cases closed before and after the IEP. The authors conceded that the results of their study might have been different if the national RSA-911 database included these two separate closure statuses in the RSA-911 data reporting form. Thus, inconsistencies from state to national coding, as well as the potential for different interpretations of closure status coding is noted as a potential limitation in many of the acceptance studies.
In addition, statistical analyses may be problematic in some of the previous research regarding race and VR acceptance. Supporting contentions that the analyses using RSA-911 data may be highly suspect in the studies regarding race and VR acceptance rates, Wheaton contended that
As sample size increases so does statistical power, and very large sample will, therefore produce statistically significant results even when the effect size is so tiny as to be trivial. This is because many of the findings regarding race and acceptance rise the RSA-911, which has literally hundreds of thousands of cases. The RSA-911 uncritically reports statistical significance as thought had substantive significance. I believe that this is the fundamental error of this research" (J. Wheaton, personal communication, Marcia 14. 2002, as cited by Thomas & Wienrach, 2002).
In reference to Wheaton's discussion, it is important to note that the effect size of the aggregated data used in this meta-analysis does not indicate that these are tiny or trivial effects. In fact, one could consider the odds ratio of a 1.54 likelihood of Whites (versus African Americans) being accepted for VR services to be quite robust. However, the indications of heterogeneity across the studies points to the possibility that other variables may also be influencing the results.
Lastly, as noted by Wilson et al., (2002) other factors may be hidden from research results pertaining to factors influencing VR acceptance. Wilson and his associates speculated that some customer may be perceived by VR counselors as more "marketable" than others, and that, a severe disability might predispose a counselor to reject a customer for VR services given the pressure on VR counselors to increase the number of successful rehabilitation closures (Status 26). Wise (1988) asserted that disability severity might predispose some customers to acceptance or rejection in the VR system despite the order of selection mandates to serve people with the most severe disabilities. More research is needed to determine other possible relationships between VR eligibility and race in interaction with severity of disability.
Although the results reported here indicate discrepancies in acceptance rates between the two ethnic groups (African Americans and Whites) seeking VR services, this study revealed significant heterogeneity across the sample studies. This provides insight that there are potentially other variables also affecting acceptance rates.
Given that within the fixed effects model used for the present study the test for homogeneity of effect size was rejected, subsequent studies are warranted to identify other variables that may influence acceptance rates. Many researchers have speculated that variables other than ethnicity may influence VR acceptance (Bolton & Cooper, 1980: Wheaton, 1995: Wilson, 2000; Wilson, Alston, & Harley). Researchers using the RSA-911 national database or other national databases should investigate other variables within multivariate analyses to determine which variables might explain variance in VR acceptance. For example, using regression analyses, Wilson (2000) found source of support at application as well as ethnicity to be statistically significant. Wilson's application of logistic regression to predict VR acceptance from a host of explanatory variables may provide a guide for researchers studying VR acceptance and ethnicity.
A recent study conducted by Allen, Parnell, Crawford, and Beardall (2000) demonstrated the importance of geographic region and service location when examining equitable treatment in the rehabilitation process. The authors asserted "Although significant differences between Whites and African Americans in terms of expenditure and cases closed as rehabilitation were found on a state-wide basis, statistically significant differences did not exist in all districts" (p. 16). Future research might investigate RSA by regions to determine which ones are most likely to reveal discrepancies in VR acceptance (Wilson, 2002).
Lastly, this study was limited to African Americans and Whites, excluding the experiences of individuals within important groups such as Native Americans/Alaskan Natives, and Asians/Pacific Islanders who sought VR services in the United States. Continued investigation of the experiences of all minority consumers requires further examination.
Causation cannot be discerned within the data accumulated regarding VR acceptance and race. However, one can safely determine that race and VR acceptance are not independent phenomenon, and that it appears, over the time period spanning between 1984 and 1998 there was a discernable and robust pattern of acceptance with Whites more likely to be accepted for services than African Americans. Within the 1992 amendments to the Rehabilitation Act it was stated that:
Patterns of inequitable treatment of minorities have been documented in all major junctures of the vocational rehabilitation process. As compared to Whites, a larger percentage of African American applicants to the vocational rehabilitation system is denied acceptance. Of applicants for service, a larger percentage of African-American cases are closed without being rehabilitated. Minorities are provided less training than their White counterparts. Consistently, less money is spent on minorities than their White counterparts (p. 4364).
The assertions within the Rehabilitation Act Amendments (1992) were based on archival research studies of state vocational rehabilitation (VR) agency data suggesting the existence of inequitable patterns of VR service delivery for African American consumers. Given the results of this meta-analysis, it appears these concerns remain as important today as when written.
Allen, T. B., Parnell, L., Crawford, R., & Beardall, R. L. (2000). Equitable treatment in the rehabilitation process: Implication for future investigations related to ethnicity. Journal of Rehabilitation, 66(2), 14-18.
Atkins, B. J., & Wright, G. N. (1980). Three views vocational rehabilitation of African Americans the statement. Journal of Rehabilitation, 46(2), 40, 42-46.
Bolton, B., & Cooper, P. G. (1980). Three views: Vocational rehabilitation of African Americans: The comment. Journal of Rehabilitation, 46(2), 41-49.
Boski, P. (1988). Cross-cultural studies of person-perception. Effects of ingroup/outgroup membership and ethnic schemata. Journal of Cross-Cultural Psychology, 19, 287-328.
Bowe, F. (1984) .U. S. census and disabled adults. Hot Springs: University of Arkansas, Arkansas Rehabilitation Services, Arkansas Rehabilitation Research and Training Center.
Capella, M. E. (2002). Inequities in the VR system: Do they still exist? Rehabilitation Counseling Bulletin, 45, 143-153.
Chan, F., Wong, D., Rosenthal, D., Kundu, M., & Dutta, A. (in press). Eligibility Rates of Traditionally Underserved Individuals with Disabilities Revisited: A Data Mining Approach. Journal of Applied Rehabilitation Counseling.
Devine, P. G., & Elliot, A., J. (1995). Are racial stereotypes really fading? The Princeton trilogy revisited. Personality and Social Psychology Bulletin, 21, 1139-1150.
* Dziekan, K. I., & Okocha, A. A. G. (1993). Accessibility of rehabilitation services: Comparison by racial-ethnic status. Rehabilitation Counseling Bulletin, 36, 183-189.
* Feist-Price, S. (1995). African Americans with disabilities and equity in vocational rehabilitation services: One State's review. Rehabilitation Counseling Bulletin. 39, 119-129.
Fleiss, J. L. (1994). Measures of effect size for categorical data. In H. Cooper & L. V. Hedges (Eds.), The handbook of research synthesis (pp. 245-260). New York: Russell Sage Foundation.
Fraenkel, J. R., & Wallen, N. E. (1993). How to design and evaluate research in education. New York: McGraw-Hill.
Hayes-Bautista, D. E. (1992). Latino health indicators and the underclass model: From paradox to new policy models. In A. Furino (Ed.), Health policy and the Hispanic (pp. 32-47). Boulder, CO: Westview.
Hedges, L. V. (1994). Statistical considerations. In H. Cooper & L. V. H edges (Eds.), The handbook of research synthesis (pp. 29-38). New York: Russell Sage Foundation.
Herbert, J. T., & Martinez, M. Y. (1992). Client ethnicity and vocational rehabilitation case service outcome. Journal of Job Placement, 8(1), 10-16.
Matt, G. E., & Cook, T. D. (1994). Threats to the validity of research syntheses. In H. Cooper & L. V. Hedges (Eds.), The handbook of research synthesis (pp. 503-520). New York: Russell Sage Foundation.
McGinn, F., Flowers, C. R., & Rubin, S. E., (1994). In quest of an explicit multicultural emphasis in ethical standards for rehabilitation counselors Rehabilitation Education, 7, 261-268.
Rehabilitation Act Amendments of 1992, Pub. L. No. 102-569, 106 Stat. 4344-4488 (1992).
Rehabilitation Act Amendments of 1998, Pub. g. 105-220, 112 Stat. 1092-1247 (1998).
Rehabilitation Services Administration. (1995). Reporting manual for the case service report (RSA-911) (RSA-PD-95-04). Washington, DC: Author.
Rosenthal, D.A., (2004). Effects of client race on clinical judgment of practicing European American vocational rehabilitation counselors. Rehabilitation Counseling Bulletin, 47, 131-141.
Rosenthal, D. A., & Berven, N. L. (1999). Effects of client race on clinical judgment. Rehabilitation Counseling Bulletin, 42, 243-264.
Rosenthal, D. A., & Kosciulek, J. F. (1996). Clinical judgment and bias due to client race or ethnicity: An overview with implications for rehabilitation counselors. Journal of Applied Rehabilitation Counseling, 27, 30-36.
Shadish, W. R., & Haddock, C. K. (1994). Combining estimates of effect size. In H. Cooper & L. V. Hedges (Eds.), The handbook of research synthesis (pp. 261-281). New York: Russell Sage Foundation.
Spengler, P. M., Strohmer, D. C., & Prout, H. T., (1990). Testing the robustness of the diagnostic overshadowing bias. American Journal on Mental Retardation, 95, 204-214.
Strohmer, D. C., & Chiodo, A. L. (1984). Counselor hypothesis testing strategies: The role of initial impressions and self-schema. Journal of Counseling Psychology, 31, 510-519.
Strohmer, D. C., & Leierer, S. J. (2000). Modeling rehabilitation counselor clinical judgment. Rehabilitation Counseling Bulletin, 44(1), 3-11.
Strohmer, D. C., & Newman, L. J. (1983). Counselor hypothesis-testing strategies. Journal of Counseling Psychology, 30, 557-565.
Strohmer, D. C., & Shivy, V. A. (1994). Bias in counselor hypothesis testing: Testing the robustness of counselor confirmatory bias. Journal of Counseling and Development, 73(2), 191-198.
Strohmer, D. C., Shivy, V. A., & Chiodo, A. L. (1990). Information processing strategies in counselor hypothesis testing: The role of selective memory and expectancy. Journal of Counseling Psychology, 37, 465-472.
U.S. Census Bureau. (2000). Population projections of the United States by age, sex, race, and Hispanic origin: Retrieved July 12, 2003, from the World Wide Web: http:www.census.gov/ main/www.cen2000.html
U.S. Department of Education, National Institute on Disability and Rehabilitation Research. (1992). Disability statistics abstract number 4. Washington, DC: Author.
Walker, S., Adbury, C., Maholmes, V., & Rackley, C. (1992). Prevalence, distribution and impact of disability among ethnic minorities. In S. Walker, F. Z. Belgrave, R. W. Nicholls, & K. A. Turner (Eds.), Future frontiers in the employment of minority persons with disabilities (pp. 10-24). Washington, DC: Howard University, Research and Training Center.
* Wheaton, J. E. (1995). Vocational rehabilitation acceptance rates for Whites and African Americans: Another look. Rehabilitation Counseling Bulletin, 38, 224-231.
Whitney-Thomas, J., Timmons, J. C., Gilmore, D. S., & Thomas, D. M. (1999). Expanding access: Changes in vocational rehabilitation practice since the 1992 Rehabilitation Act Amendments. Rehabilitation Counseling Bulletin, 43, 30-40. Wilson, K. B. (2002). Exploration of VR acceptance and ethnicity: A national investigation. Rehabilitation Counseling Bulletin, 45, 168-176.
* Wilson, K. B. (2000). Predicting vocational rehabilitation acceptance based on race, education, work status, and source of support at application. Rehabilitation Counseling Bulletin, 43, 97-105.
Wilson, K. B. (1999). Vocational rehabilitation acceptance: A tale of two races in a large Midwestern state. Journal of Applied Rehabilitation Counseling, 30(4), 25-31.
* Wilson, K. B., Alston, R. J., Harley, D. A., & Mitchell, N. A. (2002). Predicting VR acceptance based on race, gender, education, work status at application, and primary source of support at application. Rehabilitation Counseling Bulletin, 45, 132-142.
Wilson, K. B., Harley, D. A., & Alston, R. (2001). Race as a correlate of vocational rehabilitation acceptance: Revisited. The Journal of Rehabilitation, 67(3), 35-41.
Wilson, K. B., Harley, D. A., McCormick, K., Jolivette, K., & Jackson, R. L. II (2001). A literature review of vocational rehabilitation acceptance and explaining bias in the rehabilitation process. Journal of Applied Rehabilitation Counseling, 32(I), 24-35.
Wise, S. A. (1988). Service equity and program effectiveness in the rehabilitation process. Journal of Rehabilitation, 54(4), 68-72.
Note: * indicates studies used in the meta-analysis
David A. Rosenthal
University of Wisconsin-Madison
James Micheal Ferrin
University of South Carolina
Pennsylvania State University
Florida Atlantic University
David A. Rosenthal, Department of Rehabilitation Psychology and Special Education, University of Wisconsin-Madison, Room 413,432 North Murray Street, Madison, WI 53706-1496. E-mail: email@example.com
|Printer friendly Cite/link Email Feedback|
|Publication:||The Journal of Rehabilitation|
|Date:||Jul 1, 2005|
|Previous Article:||An investigation of the job tasks and functions of providers of job placement activities.|
|Next Article:||Contemporary Mental Health Issues Among African Americans.|