Using a sample of 628 white, black, and Hispanic voters in a large urban school district, we test a series of hypotheses about voting in a school bond election. We find that there is a core of similar results across racial/ethnic groups. All three groups show strong, directly measured, self-interest effects. We also find some distinct group differences. Symbolic values played a limited role for white voters, but a stronger role for minorities. In addition, for white voters we find a substantial drop in support for the bond across age cohorts, but no such drop among black and Hispanic voters.