Abstract: Using Hierarchical Logistic Regression to Examine the Impact of Disability-Risk and Student-Teacher Racial Similarity on Special Education Service Identification (Society for Social Work and Research 26th Annual Conference - Social Work Science for Racial, Social, and Political Justice)

Using Hierarchical Logistic Regression to Examine the Impact of Disability-Risk and Student-Teacher Racial Similarity on Special Education Service Identification

Friday, January 14, 2022
Liberty Ballroom O, ML 4 (Marriott Marquis Washington, DC)
* noted as presenting author
Melissa Jenkins, MSW, Doctoral Student, University of North Carolina at Chapel Hill, Chapel Hill, NC
Sarah Eisensmith, PhD, Academic Consultant; Researcher, Duke University
Background and Purpose: One of the greatest challenges of special education is the decades-long issue of racial disproportionality in disability categorization. Some scholars hypothesize that implicit bias and a cultural mismatch between students and teachers influences Latinx and/or Black youths’ overrepresentation in special education, particularly in the identification and subsequent diagnosis of emotional and/or behavioral disorders. Still, other researchers suggest that these youth are underrepresented in special education and attribute this disproportionality to disparities in socioeconomic status (SES). While lower SES is a commonly researched risk factor for poor educational outcomes, familial “fragility” defined as the impact of unmarried parents on child wellbeing is understudied. The current study asks the following research questions: Does race moderate the relationship between disability identification risk and likelihood of “fragile” students having an individualized education program (IEP)? In addition to the above predictors, does student–teacher racial similarity moderate the relationship between teacher-reported problem behavior and likelihood of having an IEP?

Methods: Data were obtained from the fifth wave of the Fragile Families and Child Wellbeing Study when youth were approximately 9-years-old. Students were dichotomized as being high-risk for a disability if they scored one standard deviation below average on one or both measures of receptive vocabulary (Peabody Picture Vocabulary Test) and working memory (Wechsler Intelligence Scale for Children, Digit Span subtest). Due to the overrepresentation of White (75%) and Black (17%) teachers surveyed, observations were included for children who self-identified as White or Black at the sixth wave, resulting in an analysis sample of N=1,267. Stata was used to perform hierarchical logistic regression to probe the interactions between a) student race and disability risk, and b) student-teacher racial similarity and problem behavior as measured via the Social Skills Rating System – Problem Behavior subscale (SSRS-PB).

Results: The log odds of having an IEP was not statistically significantly different for White or Black children at low- or high-risk for a disability (β=-.40, ns). However, the difference in the predicted probability of service identification between high-risk White students (47%) and high-risk Black students (24%) when controlling for problem behavior and student-teacher racial similarity is notable. Additionally, the odds of a student having an IEP was approximately 3.9 times lower if their racial identity matched with their teacher (p<.001). When examining predicted probabilities, students with SSRS-PB scores at approximately twice the average score (and thus more problem behaviors) had an IEP when their teachers matched (11%) and did not match (17%) their racial identity.

Conclusions and Implications: While race was not a statistically significant moderator, Black students at high-risk for disability identification were less likely to have an IEP even when controlling for student-teacher racial similarity and teacher-reported problem behavior. This has practical implications for schools seeking to target areas of racial disproportionality, especially when considering how disability risk is identified (e.g., standardized measures validated among racially diverse samples). The underrepresentation of non-White teachers in the sample warrants the need to understand how the cultural mismatch hypothesis can be empirically tested among a more racially diverse sample.