Differential Test Performance and Peer Effects

TL;DR

This paper develops a method to identify peer effects in student test scores using differences in related test scores, avoiding the need for randomized group assignment, and applies it to early education data.

Contribution

It introduces a novel identification strategy using differences in related test scores to measure peer effects without relying on random group assignment.

Findings

Significant peer effects found in early education classes.

Peer effects are at the lower end of existing literature estimates.

Methodology allows for identification of peer effects using test score differences.

Abstract

We use variation of test scores measuring closely related skills to isolate peer effects. The intuition for our identification strategy is that the difference in closely related scores eliminates factors common to the performance in either test while retaining idiosyncratic test specific variation. Common factors include unobserved teacher and group effects as well as test invariant ability and factors relevant for peer group formation. Peer effects work through idiosyncratic shocks which have the interpretation of individual and test specific ability or effort. We use education production functions as well as restrictions on the information content of unobserved test taking ability to formalize our approach. An important implication of our identifying assumptions is that we do not need to rely on randomized group assignment. We show that our model restrictions are sufficient for the formulation of linear and quadratic moment conditions that identify the peer effects parameter of interest. We use Project STAR data to empirically measure peer effects in kindergarten through third grade classes. We find evidence of highly significant peer effects with magnitudes that are at the lower end of the range of estimates found in the literature.