The proximity paradox: the legislative agenda and?the?electoral success of ideological extremists

This paper presents a new approach to spatial models of legislative elections in which voters have preferences over the bundles of roll call votes implied by candidate locations rather than over the locations themselves. With such preferences, voters with single-peaked, symmetric preferences and perfect information can sincerely prefer a distant candidate to a more proximate candidate. Moreover, negative agenda control in Congress makes such preference orderings inevitable, so party agenda control can allow majority party extremists to defeat more centrist minority party candidates. The model has implications for theories of parties in Congress, and spatial modeling more broadly.