This paper investigates community effects in the determination of desired s
chooling in a sample of more than 300 school children and their parents in
three Mexican cities. Community residence is found to be a significant pred
ictor of desired schooling of parents and children, even with comprehensive
controls for child and family traits. Measurement error and omitted variab
le bias are considered, but rejected, as principal causes of this result. A
comparison of recent and long-term residents of a community reveals that t
he predictive power of residence is much stronger for long-term residents.
This result is interpreted as evidence of community effects, since the alte
rnative hypothesis of Tiebout behavior predicts a stronger common effect fo
r recent migrants. potential sources of the community effects are then inve
stigated with neighborhood-level data from the 1990 Mexican Census. (C) 199
9 Elsevier Science Ltd. All rights reserved.