In many African states, voters and civil society typically play only a minor role in policymaking. Information about constituent preferences reaches politicians infrequently and unsystematically. This could mean that elected politicians are underinformed and/or under-incentivized to act in the interest of constituents. This project uses an experiment to estimate the impact of new information about voter’s preferences on their representatives’ stated preferences and legislative behavior. In June 2014, the Civil Society Coalition to Stop Maternal, Newborn, and Child Mortality in Uganda polled voters around the country on the extent to which they supported budget increases in health, and the extent to which they were likely to hold their elected representatives to account for failure to deliver adequate health services. The poll revealed that health was the most salient issue for voters and had grown in importance relative to other issues over the past years. Unsurprisingly, a majority of Ugandans felt that their representatives were doing lessthan they should to address systematic failures in service delivery, and reported that this fact was likely to drive their future voting decisions. This information was then distributed to some randomly selected Members of Parliament (MPs) and not others ahead of the final vote on the health budget. Using a petition, we are able to show that learning about citizen priorities pushed MPs, when contacted, to publicly support budget increases for health service provision at a higher rate than their colleagues without this information. Interestingly, treated (informed) MPs were less likely to respond to the petition at all. When it came to the budget, however, it passed unamended and without increases in spending. We interpret this result as illustrative of the extent to which politicians feel their actual actions in Parliament are hidden.