Estimating the impact of peer group effects on socioeconomic outcomes: Does the distribution of peer group characteristics matter?

For some socioeconomic outcomes of interest to government policy-makers, the influence of ''peer groups'' may be important. For example, students' performance in school may be affected by characteristics of their classmates. If peer group effects exist, governments may be able to manipulate them to better achieve policy objectives. An example of this is the choice between ''mixing'' and ''streaming'' students of different ''abilities'' in public schools. Several recent studies have examined the influence of peer group effects. This paper argues that past attempts to estimate their impact may have used insufficiently flexible techniques. In particular, they have emphasized the mean of peer characteristics without taking into account their overall distribution. The paper shows how estimation can he done that takes the shape of the distribution into account, and how failure to do so can yield seriously misleading results. Copyright (C) 1997 Elsevier Science Ltd