Biased learning from elections

A foundational premise in democratic theory is that political competition encourages parties to be responsive to voters. Parties have incentives to respond to the median voter's preferences in order to win elections, and should learn from the results of elections where their platforms diverge from what the electorate wants. However, parties may be subject to motivated reasoning, wanting to believe that the electorate favors their own policy preferences. We develop a repeated model of elections with motivated beliefs to explore how this bias affects how parties compete with one another for popular support. Motivated beliefs lead to excessive platform divergence, and allow parties to infer from poor electoral outcomes that elections are unfair rather than that their platforms are unpopular. Disagreement about the fairness of the electoral system increases over time, even if platform divergence decreases. Our analysis reveals how motivated beliefs inhibit parties' ability to learn what voters want while encouraging partisans to distrust the electoral process itself.