I would like to thank Kevin Arceneaux and Ryan Vander Wielen for their review of Incremental Polarization. I appreciate the positive nature of their assessment, as well as the acuity of their critiques.

I will begin with a brief correction. Arceneaux and Vander Wielen state that the “proximity paradox” explains how voters can prefer legislators with “unrevealed” extreme ideal points to more moderate challengers. In fact, even legislative agendas that reveal extremism can permit extremists to defeat centrists when there are “asymmetric differentials” in the legislative agenda, which occur when some roll call votes are more substantively important than others.

Primarily, though, I will focus my response on two elements that I did not think I addressed satisfactorily in Incremental Polarization, and which Arceneaux and Vander Wielen noted. First, they point out one of the bigger weaknesses in the argument—my inability to specify much about the party leaders themselves. They note, for example, the difficulty of enforcing discipline without accurate information, and in the face of competing incentives at the leadership level, which my model did not specify. These are all valid points, which I recognized while writing, and simply did not have a way to capture in the model. Every model makes sacrifices.

The second observation is the extent to which my model conflicts with Cartel Theory. Cartel Theory asserts that party control of legislative procedure secures a partywide electoral benefit, but my model argues that party discipline produces policy gains at an electoral cost. I remain, at this point, conflicted about this tension, as I see much of value in Cartel Theory. I see the models as capturing different processes. When party pressure induces noncentrist roll call votes, as it sometimes does, the result is to reduce the expected vote shares of pressured legislators. My argument, though, is that partywide electoral benefits are unnecessary for explaining why legislators submit to such pressure. Sincere voting is a collective-action problem, to be solved by parties when legislators want to achieve noncentrist policy objectives. I do not argue, and do not believe that parties cannot produce partywide electoral benefits, but I do not see such benefits as necessary for explaining the otherwise electorally suboptimal behavior that we observe by individual incumbent members of Congress. Fundamentally, procedural models are aimed at explaining the manner in which parties manage the floor, whereas my own argument focuses on how parties manage “substantive” roll call votes via party discipline. I see no intrinsic conflict if the electoral effects of these tactics work in opposing directions since the actions themselves are separable.

One of the points I hope to make is the importance of complementary and alternative models. I see the model at the core of Incremental Polarization not as competing with the Cartel model, or even conventional spatial models, but as a complementary one. While one of my objectives was to unify spatial models of elections and spatial models of roll call voting, I did not intend for the product to replace existing models, but rather to show the value of alternative structures and unusual assumptions because no single model can explain any political institution.