We conduct a field experiment with a sample of retail investors. We first elicit beliefs about the returns to active investing. We then generate exogenous variation in beliefs by providing treated respondents with information about index funds historically outperforming active funds. Treated respondents are 17.4 percentage points more likely to think that index funds will outperform active funds in the future. Four months after the experiment, we collect data on the survey respondents' portfolio allocations. Our main result is that treated respondents increase the index fund share of their portfolio by 4.6 percentage points (39 percent) relative to control group respondents.