Poster #11 - Specific Learning and Speech and Language Disability Identification: Disproportionality and the Ever-English Learner Framework

• In Event: Poster Session 16

Abstract

There is conflicting evidence regarding the over- or underrepresentation of English Learners (ELs) in special education nationwide: current-ELs were overrepresented in special education at the secondary level, while students who have been EL (ever-ELs) were underrepresented (Umansky et al., 2017). Our study focuses on two common disability categories: specific learning disabilities (SLD) and speech and language impairments (SLI). Under-identification of ELs suggests that we fail to provide essential support services to those ELs with special education needs. Meanwhile, over-identification of ELs is also harmful, limiting their inclusion in general education classrooms (Samson & Lesaux, 2009). No work has yet been conducted using the ever-EL framework and nationally representative data to examine the disproportionate identification for SLI and SLD.

We analyzed restricted student-level National Assessment of Educational Progress. We examined fourth and eighth grade reading and mathematics surveys from 2009 to 2019 to provide the latest national evidence of special education identification. Our analyses incorporated weighting to account for complex sample design and nonresponse in achieving nationally representative estimates. We conducted logistic regression with three models: Model 1 includes EL status without any covariate adjustments; Model 2 added all the covariates; and Model 3 added school-fixed effect. We report results based on 2019 and 2017 Reading assessment datasets. The observed patterns were consistent when we conducted same analyses using Math assessment datasets.

Fourth Grade (see Table 1)
SLD. We found that current EL students change from more likely (OR=1.472019, 1.652017, p<0.001) to less likely (OR=0.552019, 0.522017, p<0.001) identified as SLD than never EL students after controlling covariates and school fixed effects. Meanwhile, former EL students were less likely (OR=0.41, p<0.001) to be identified as SLD in 2017 while not in 2019 after adjustment. SLI. In 2019, there were no significant disparities among SLI identification rate between former EL and never EL. Current EL students were less likely to be identified as SLI after adjustment (OR=0.79, p<0.001). In 2017, current EL and former EL students were less likely to be identified as SLI than never EL students after adjustment, respectively (OR=0.84current, p<0.01; OR=0.69former, p<0.05).

Eighth Grade (see Table 2)
SLD. We found that current-EL students change from more likely (OR=2.342019, 2.402017, p<0.001) to less likely (OR=0.732019, 0.732017, p<0.001) be identified as SLD than never-EL students after controlling covariates and school fixed effect. Meanwhile, former-EL students were less likely (OR=0.78, p<0.05) to be identified as SLD in 2019 while not in 2017 after adjustment. SLI. In 2019, there was no significant disparities among SLI identification rate between current-EL and never-EL, while former-EL students were less likely to be identified as SLI after adjustment (OR=0.48, p<0.05). In 2017, on the contrary, current-EL students were less likely to be identified as SLI after adjustment (OR=0.78, p<0.05), while there were no significant identification disparities between former-EL and never-EL students.

Our preliminary results indicate that ELs are disproportionately identified as SLD and SLI during the past decade. Moreover, such disproportion identification differed by students’ EL status, grade, and across years.

Authors

• Eric Hengyu Hu, Pennsylvania State University, University Park
  Presenting Author

• Cecelia Gloski

• Jessica Liu, Pennsylvania State University, University Park

• Paul L. Morgan, Pennsylvania State University, University Park