Reconciling Candidate Extremism And Spatial Voting

We propose a modified spatial-voting rule to explain why congressional candidates adopt more extreme ideological positions than their constituents' preferences. Our model accepts the standard spatial-voting model with one critical exception: voters in the same party as a candidate tolerate extremism without imposing an electoral penalty. This, in turn, creates "leeway" for candidates to adopt extreme positions as they increasingly depend on voters from their own party. Electoral simulations demonstrate that a key election-level implication of this model is that it explains candidate polarization without relying on institutional factors like primary elections. Finally, we show that asymmetry in perceptual bias is one possible mechanism and that real-world patterns of ideological representation are consistent with our simulation results.