“Comparing Achievement between K-8 & Middle Schools: A Large Scale Empirical Study”

This paper reports on a natural experiment in the Philadelphia School District. It compares Middle Schools to established K-8 schools, as well as to newly formed K-8 schools that are part of the district’s K-8 conversion policy, in order to determine if the different school structures have an effect on student academic achievement. The outcome is students’ 8 grade mathematics achievement and our sample includes 40,883 8 grade students taken from 95 schools across 5 cohorts from the 1999-2000 to the 2003-04 school years. The analysis uses multi-level modeling to account for differences between student, cohort, and school level variation, and includes a large set of statistical controls that include student demographics, teacher characteristics, school transition, and several cohort and school level factors including average school size. The results find that older K-8 schools do perform significantly better than Middle Schools, and that this advantage is adequately explained by the two school type’s differing student and teacher populations, differences in average grade size, and the extra school transition that Middle School students must make from elementary to the middle grades. Newer K-8 schools created as part of the district’s reform efforts outperformed Middle Schools but not by as much or as significantly as did older K-8 schools, despite having smaller grade sizes and lower rates of school transition. We found that this was likely due to their student populations, which like those of Middle Schools, consisted primarily of minority students from high-poverty backgrounds. We conclude that while K-8 schools do perform better in terms of student achievement, the advantage exists for several reasons and may not be easily replicated or represent a solution to the problem of low achieving schools and students in large urban public school districts that serve highminority and low-poverty student populations.