In this work we analyze the impact of the voting rule on individual behavior. We use a sample of individuals naturally motivated by real candidates. Then, in our methodology we do not induce preferences. Moreover, up to our knowledge, this is the first work in which both individual behavior and aggregate results are studied with a relatively large sample. We implemented an online simulation during the presidential campaigns for 2010 in Colombia. Voters were asked to submit experimental ballots under four different voting rules: plurality rule, approval voting, Borda rule and majority rule with runoff elections. We compared the observed individual behavior with two benchmarks.The fist one considers sincere voting and the second one instrumen...