Holding an unpopular position on an issue important to voters can endanger a candidate’s electoral success. What is the candidate’s best agenda-setting strategy? To focus on other issue positions congruent with the same ideological stereotype, shoring up support among like-minded voters? Or to “go maverick” by discussing some issues that signal liberal positions and some that signal conservative positions? Existing voting models suggest the answer depends on voter preferences, since going maverick should have symmetric effects—support among voters who agree with the candidate’s positions will decrease, proportionally, as support increases among voters who disagree. We argue, however, that stereotype incongruence prompts these voters to process information differently, yielding asymmetric effects. We test our expectations experimentally, using a fictional candidate webpage to show how the benefits of going maverick can outweigh the costs.