Association for Public Policy Analysis & Management

Students in California schools without proficient English are assigned to English Learner (EL) support services. This paper uses student-level data from two school districts to ask “Are ELs being reclassified at the right time?” Prior research has not conclusively established whether removing language support causes better outcomes for ELs or simply more efficiently selects students who do well without continued support. We ask if students are reclassified at the right time and what reclassification policy is associated with better academic outcomes. Only a few studies (Robinson 2011, Robinson-Cimpian and Thompson 2015) have been able to acquire the student-level data required to establish causality. Ideally, students are reclassified at the moment when EL support no longer benefits them and they are prepared to undertake an English-only instructional program. Capitalizing on our ability to precisely identify the reclassification policies used to decide when EL support should be removed, we will estimate the causal relationship between continued EL instructional support and academic outcomes for students who are performing near reclassification cutoffs under four different reclassification policies. We use a Regression Discontinuity (RD) to examine outcomes for students just above and just below the cutpoints for EL reclassification. Because the two districts have used four different reclassification policies, we will be able not only to identify any causal relationship between duration of English language support and student outcomes, but also identify which of the reclassification standards comes closest to reclassifying a student at the appropriate time. Data for the research are from the two largest school districts in California: Los Angeles Unified and San Diego Unified School Districts. Together, they serve 14% of the state’s EL students, and 4% of EL students nationwide. We have over a decade of individual student level data that allows us to follow students as they receive EL instruction, are reclassified (or not), and track their academic outcomes, including post-secondary enrollment for a few cohorts. Results from our analysis illuminate the causal effect of reclassification policies and help evaluate which reclassification policies are the most beneficial to EL students. We examine the possibility that reclassification policies could have different effects by grade level and consider how moderating factors of the student, school, and community relate to EL student outcomes. We also ask i) whether students are reclassified in accordance with district guidelines and ii) whether there is fidelity of implementation to district policy regarding assignment to English Language Development (ELD) courses (i.e. ELs not yet reclassified as fluent receive the prescribed coursework). We find that students are largely reclassified in accordance with district guidelines, and that the ELD assignments vary in how closely they follow the district guidelines by year and grade. Results from this paper could help design reclassification policies to optimize EL treatment. California and other states are working to understand how new assessments measure EL students’ English proficiency and academic English and how to use those assessments to standardize their reclassification policies. Getting these policies right could improve outcomes for a large and vulnerable student population.