Why are There so Many Divided Senate Delegations?

The last three decades have witnessed a sharp increase in the number of states with spilt Senate delegations, featuring two senators of different parties. In addition, there is evidence that senators of different parties do not cluster in the middle: they are genuinely polarized. We propose a model which explains this phenomenon. Our argument builds upon the fact that when a Senate election is held, there is already a sitting senator. If the voters care about the policy position of their state delegation in each election, they may favor the candidate of the party which is not holding the other seat. We show that, in general: (1) a candidate benefits if the non-running senator is of the opposing parry; (2) the more extreme the position of the non-running senator, the more extreme may be the position of the opposing party candidate. Our 'opposite party advantage' hypothesis is tested on a sample including every Senate race from 1946 to 1986. After controlling for other important factors, such as incumbency advantage, coattails end economic conditions, we find reasonably strong evidence of the 'opposite party advantage.'