The Big Picture: A Meta-Analysis of Program Effectiveness Research on English Language Learners

17 studies

English language learners in the United States

The authors state explicitly that they follow the Glass strategy of INCLUDING AS MANY STUDIES AS POSSIBLE rather than excluding any on a priori study-quality criteria. That is a defensible position in meta-analytic methodology and it is fatal for this archive's purposes: a pooled estimate that deliberately declines to weight by design quality inherits every confound in the underlying literature, which here is mostly non-randomised programme comparison. Graded D on that basis, not on the topic.

Bilingual education is reported as consistently superior to all-English approaches, and developmental bilingual superior to transitional bilingual. Positive effect for bilingual education of +0.23 SD on English-language outcomes, and +0.86 SD on outcomes measured in the NATIVE LANGUAGE. The +0.86 figure is regularly quoted without the qualifier; it is the effect of teaching children in a language on tests administered in that language.

High. Family and district selection into programme type is uncontrolled across most included studies and the meta-analysis makes no attempt to weight it out.

Directionally consistent with Slavin & Cheung on English outcomes but at a much lower evidential standard, and its headline +0.86 is not comparable to the English-outcome estimates it is usually quoted alongside.