About this poster

---

Panel information
Panel 3. Schooling and Education

Abstract

---

Background: Extending previous work on linguistically-diverse students’ special education (SPED) representation in the U.S., we longitudinally examine SPED representation by language status in Tennessee, a state that has experienced significant growth in the number of school-age linguistically diverse students. We ask: Across a 10-year span (2009-2019), how are students represented in SPED services by language status between kindergarten and eight grade? We aim to make a unique contribution to the SPED disproportionality literature by: (1) focusing on the heterogeneity of language status and (2) taking a longitudinal approach using large-scale state-level historical data. Examining longitudinal SPED representation patterns by specific language status has the potential to enhance research and policy-level understanding at the intersection of English learner (EL) status and SPED status. Further to note, we conceptualize language status into four main categories: Current ELs, Former ELs, Never ELs, and native English speakers (NES).

Methods: The person-period dataset was prepared by merging deidentified student characteristics and enrollment datasets. We conducted discrete-time hazard models (Singer & Willett, 2003) to explore the relation between SPED placement and language status over time (from kindergarten to eighth grade), controlling for female, Hispanic background, lower-income status, and chronic absenteeism. This analytical approach is commonly used in research examining SPED placement likelihood of cultural and linguistic minority learners (also referred to as event history analysis or discrete-time survival analysis; e.g., Cruz & Firestone, 2022; Farkas et al., 2020; Umansky et al., 2017). To estimate SPED placement likelihood by language status, we fit two models: (1) the unadjusted discrete-time hazard model (i.e., language status as the only predictor) and (2) the adjusted discrete-time hazard model (i.e., with student-level covariates in addition to language status). Given that this exploratory study aimed to examine the relation between students’ language status and SPED placement outcomes, we included school fixed effects to adjust for systematic differences between students across schools, to focus on the association between student-level language status and SPED placement.

Findings: The analysis revealed that at any given timepoint between kindergarten and eighth grade, Never ELs and Former ELs were less likely than NES students to have SPED status. In contrast, Current ELs were 120% more likely than NES students to be represented in SPED. Our findings suggest that English proficiency seems to be associated with likelihood of SPED placement even after controlling student-level characteristics. Additionally, regardless of language status, all students (NES, Never EL, Former EL, and Current EL) were most likely to be in SPED services by third grade (i.e., the start of upper elementary grades) and the likelihood decreased thereafter. Finally, the findings revealed magnitude differences in the cumulative probability of being in SPED emerged by language status, such that Current ELs had the highest probability of being in SPED across all grade levels while Former ELs had the lowest probability across all grade levels.

Practically, states may consider: (1) how funding is allocated to support the unique needs of ELs with disabilities, (2) whether the teaching force is equipped to support the needs of this student population, and (3) review collaboration efforts between their SPED and EL teams to support student’s language and academic needs.

Author information

Author	Role
Bhabika Joshi, Vanderbilt University	Presenting author
Jiaxin Jessie Wang, Vanderbilt University, United States	Non-presenting author
Jeannette Mancilla-Martinez, Vanderbilt University, United States	Non-presenting author
Min Oh, University of Virginia, United States	Non-presenting author

⇦ Back to session