ABSTRACT
This chapter explores the logic underpinning the divisive primary hypothesis, discussing its causal mechanisms and providing an example for each. It reviews the literature which has investigated the divisive primary effect at each electoral level. The chapter presents an empirical test of the divisive primary effect in Senate and gubernatorial elections which controls for the type of endogeneity observed in House elections. It details strong evidence of a divisiveness effect in Senate elections; in gubernatorial elections the evidence is much weaker. The chapter shows that the primary difference between races where divisive primaries hurt and the ones where they don't is voter attention. If voters aren't paying enough attention during the primary to know what happened during that stage of the election, it's not possible for them to factor primary election results into their general election decision-making.
