- Students placed with higher achieving peers, and in peer groups with more dispersed ability levels, tend to perform better on standardized tests. The impact of peer characteristics appear to be persistent through time.
- Peer characteristics are significantly associated with outcomes even before actual exposure to the peer group occurs.
- Peer effects estimated here and elsewhere in existing literature do not reflect true casual relationship.
- Racial or ethnic achievement gaps are most pronounced in schools serving small minority populations.
- Students of all races now appear to receive lower test scores when the share Black or Hispanic in their classroom is high relative to other classrooms in the same school.
- Classroom racial composition has a significant instantaneous effect on achievement, but in most cases this effect decays rapidly over time, to the extent that it appears irrelevant for most students by the end of their 8th grade year.
- The observed correlation in peer and individual achievement can be attributed primarily to sorting and not any causal impact.
- According to this research findings could be because peer effects may indeed exist, but in observational data they are swamped by omitted variable bias generated by selection into peer groups.