Interest group information in elections

Do interest groups provide information in ways that systematically bias and polarize candidates? I consider a Downsian model where candidates are uncertain about the median voter's preference. Up to two interest groups, completely biased, observe voter preferences in the extremes and send costless messages to candidates. Starting with one interest group, I show all informative equilibria are asymmetric: the interest group “plays favorites” by revealing coarse information to one candidate, causing policy divergence. Informative equilibria exist only if the interest group has sufficiently broad information. With opposing interest groups, this requirement is relaxed. I describe equilibria where each candidate is more sensitive to voter shocks in one tail, leading to policy convergence in the center and divergence in the tails. The presence of interest groups reframes the policy space, demarcating the consensus “moderate” regions from the “extremes”.