The Effect of English Learner Reclassification on Student Achievement and Noncognitive Outcomes

Abstract

English learners’ (ELs) day-to-day experiences in school change when reclassified as fully English proficient. Prior research, however, is mixed on how reclassification influences outcomes. Many studies also do not or cannot explore key long-term outcomes or identify impacts over time. In this study, I leverage longitudinal student data in a regression discontinuity and find that reclassification after third grade affects ELs’ achievement in the short and longer term. Reclassified ELs score considerably higher on mathematics and reading standardized tests in fifth and eighth grade. I also provide the first causal evidence for the impact of reclassification on several theoretically affected noncognitive outcomes. I find that reclassification substantially lowers the level of challenge for work assigned by teachers and increases ELs’ out-of-school engagement in the short term. However, effects on noncognitive outcomes attenuate or reverse direction in the longer term. Together, these findings highlight the need for evaluations to consider multiple measures and to identify impacts over time when possible, especially when data on long-term outcomes such as high school graduation, college persistence, or labor market success are unavailable.

Keywords:

• English learner
• regression discontinuity design
• policy evaluation
• achievement
• noncognitive outcomes

Acknowledgments

This work would not have been possible without Glenda Harrell, Matthew Lenard, and Bradley McMillen. I thank Desmond Ang, David Deming, Thomas Kane, Martin West, and seminar participants at Harvard, the Wake County Public School System, and AERA for valuable feedback. The opinions expressed are those of the author and do not represent the views of the Institute or the US Department of Education. All errors are my own.

Notes

1 As I demonstrate below, compliance to this reclassification policy is high in WCPSS.

2 Because students’ ACCESS scores are discrete, in all models controlling for these scores I follow Lee and Card’s recommendation (2008) and cluster standard errors at the score level.

3 I also use a triangular kernel to estimate models using rdrobust, also the default, but present results without any weighting (i.e. a uniform kernel).

4 Because nearly all ELs who scored just below the ACCESS reclassification thresholds in third grade eventually achieved reclassification status by eighth grade, I focus on the years reclassified instrumented variable.

5 As I describe in my conclusion, though these estimates may be limited in their generalizability, policymakers will find them useful in choosing reforms that influence the educational experiences of ELs.

Additional information

Funding

The research reported here was supported in part by the Institute of Education Sciences, US Department of Education, through [grant R305B150010] for the Partnering in Education Research Fellowship in collaboration with the Center for Education Policy Research at Harvard University.