Matsudaira.

In recent years, the role of bilingual education as the dominant pedagogy for teaching immigrant students has been challenged by federal and state policy reforms. Critics of bilingual education contend that bilingual programs hinder achievement by reducing incentives to learn English and by trapping students in classrooms with low-performing peers. Proponents argue that learning occurs most rapidly when students master concepts in their native language first, and therefore that bilingual education programs promote achievement. To distinguish between these competing views, Matsudaira exploits quasi-random assignment of students to bilingual and mainstream (English-immersion) classes generated by discontinuous program eligibility rules in a large urban school district. In the District, eligibility for bilingual programs is determined by a test of English proficiency: students scoring below a preset threshold level are eligible for bilingual classes; students scoring above this threshold are not eligible. Using information on achievement and program participation from a large administrative dataset, Matsudaira compares students scoring just below and just above this threshold for several years following the proficiency test. Compared to students scoring just above the threshold, students scoring just below it are nearly 90 percent more likely to participate in a bilingual program and are surrounded by peers with significantly lower average achievement scores. Despite these striking differences in classroom environments, however, he finds negligible differences in achievement in both reading and math (measured by tests given in English) across these two groups of students. These results speak directly to current policy debates over the most effective method of educating immigrant children. In addition, the results inform a broader debate regarding the importance of educational peer effects, because bilingual programs involve learning in classes surrounded by a group of peers who have markedly lower average achievement scores and are more similar in ethnic background than those in mainstream English-immersion classes.