Previous studies of group conflict theory look at the relationship between racial balance in a particular area (precinct, county, etc.) and racial attitudes or political behaviour within that area. While of value, there are significant methodological disadvantages to this approach. Here, I address those problems using public opinion experiments in which I ask respondents whether they would allocate political goods proportionally given different (randomly assigned) hypothetical racial environments. The experiment yields confirmatory results, with non-blacks more likely to support proportionality in less black than heavily black environments. In a second experiment, I find that the same relationship is much weaker when the variations in population are non-racial. Finally, I show that the relationship also holds for blacks and argue that this is theoretically consistent.