High school students participated in a field experiment that tested the effects of exposure to early election returns in a nonpartisan referendum. On a pretest of attitudes, students stated their preferences on the issue of the school's grading policy. One week later, just before voting, subjects in the experimental condition received information announcing the probable outcome of the election. There was a significant relationship between class achievement level and (a) likelihood of changing vote, and (b) the direction of change. Students in the brighter classes changed their votes less frequently, but when they changed their preferences they showed a greater bandwagon effect. Students in the classes of lower academic achievement changed their votes more frequently, but when they changed their preferences they showed a greater underdog effect. Differences between the more typical political election in which bandwagon and underdog effects are seldom reported and the experimental election were discussed.