2004 · 18 citations · 6 references
EducationEducational MeasurementHigher EducationCausal InferenceEconometricsStatisticsAbility GroupingPeer Group DynamicsSelection BiasSpillover EffectsMonte Carlo ResamplingPeer EffectsPeer GroupingSpillover Effect Estimation
We develop a new strategy for estimating peer efiects when there are multiple observations per person and the peer group varies across observations. This technique allows us to explicitly account for student flxed efiects and uses these student flxed efiects to formulate the ability level of the peer group. Monte Carlo evidence shows that our algorithm performs well even when with a short panel. While it is generally thought that peer efiect estimates are biased upward, we flnd strong evidence to the contrary for both our technique and the standard selection on observables approach. The bias due to measurement error in peer ability is downward and stronger than the upward bias associated with selection into the peer group. We demonstrate the technique using transcript data from University of Maryland undergraduates. We flnd statistically signiflcant peer efiects, particularly in courses of a collaborative nature.
6
Teachers, Schools, and Academic Achievement
Steven G. Rivkin, Eric A. Hanushek, John F. Kain · Econometrica · 2005 · 4.4K citations
Does peer ability affect student achievement?
Eric A. Hanushek, John F. Kain, Jacob M. Markman et al. · Journal of Applied Econometrics · 2003 · 1.2K citations