In addition to mobilizing supporters to vote, partisan campaigns use get-out-the-vote tactics as a means to boost support for their candidate. Although observational studies have attempted to estimate the effects of grassroots campaigning on political attitudes, they are unable to establish causality convincingly. Because campaigns strategically target potential supporters, comparing the attitudes of those whom campaigns contact to those they do not may only reveal spurious and biased relationships. In this paper I use a randomized field experiment to isolate the influence of personally delivered campaign messages on candidate support and attitude formation. I find that both door-to-door canvassing and commercial phone bank calls can have strong effects on voting preferences, but these tactics appear to have only weak effects on the actual beliefs that subjects possess about candidates and the degree to which those beliefs are weighted in their candidate preference. Although previous field experiments show that phone calls are less cost-effective at boosting turnout than door-to-door canvassing, they may be equally effective at increasing candidate support.