One of the most important theories of ethnic conflict focuses on the c
ompetition hypothesis: conflict occurs between ethnic groups when they
interact more rather than when they are kept separated. If true, this
theory implies that attempts at desegregation will necessarily be ass
ociated with ethnic conflict and violence. The author presents a forma
l model that suggests that the apparent competition effect uncovered i
n the empirical literature may, in fact, be spurious. Ethnic conflict
can appear to increase with competition, even in the case where indivi
duals are less Likely to attack individuals of other ethnicities than
members of their own ethnicity. A consideration of some alterations in
the basic model suggests that the result is robust. An exploratory ex
ercise shows that after controlling for the selection bias problems in
volved in studying ethnic conflict, the predicted competition effects
disappear.