Political influence through microtargeting

Political actors routinely target custom audiences on social media in order to influence elections. We model this process, focusing on the way in which it induces voters to learn about their own preferences. This differs from the past literature, which has focused on party platforms and the effects of bias. We find that the optimal strategy based on some empirically estimated parameters is to target groups favoring one’s opponents, providing a rational explanation for negative campaigning. More generally, log-concave cost of voting distributions can give rise to a non-convex set being targeted—weak supporters of the politician and strong sup porters of their opponent. We make use of this setup to provide a novel analysis of the effects of micro-targeting on turnout, and find a sense in which lower costs of voting encourage negative campaigning.