Legislature integration and bipartisanship : a natural experiment in Iceland

Nearly all legislatures segregate politicians by party. We use seating lotteries in the Icelandic Parliament to estimate the effects of seating integration on bipartisanship. When two politicians from different parties are randomly assigned to sit together, they are roughly 1 percentage point more likely to vote alike. Despite this effect, other-party neighbors do not affect general bipartisan voting, as measured by the likelihood that a politician deviates from their party leader's vote. Furthermore, the pair-level similarity effect is temporary, disappearing the following year. The pattern of results support cue-taking and social pressure as mechanisms for the effects of proximity.