Search

Recent studies have used increasingly complex methodologies to estimate the effect of peer characteristics--race, poverty, and ability--on student achievement. A paper by Hanushek, Kain, and Rivkin using Texas state testing data has received particularly wide attention because it found a large negative effect of school percent black on black math achievement. This paper replicates the HKR models using state testing data from North and South Carolina and national testing data from the Early Childhood Longitudinal Study. The replications fail to support the Texas results. In most models tested, black peer effects are small and not statistically significant, and in the few instances where effects are significant, they are much weaker than those found in Texas. Moreover, it appears that computational problems in the HKR study led to incorrect estimates for black peer effects. An appendix is included. (Contains 8 tables, 5 figures and 15 footnotes.)