Abstract
The verdict on the divisive primary hypothesis has fluctuated as investigators have pursued a variety of theoretical and methodological debates. Although most recent studies find that divisive primaries harm general election outcomes, some claim that this effect is spurious, an artifact of uncontrolled electoral prospects phenomena. I argue that this claim is debatable because it rests on questionable conceptual and model constructs and evidence inconsistent with an investigation that controls for the phenomena central to the spurious effect claim. Instead, I show that null and alternative hypothesis findings turn on an unfeatured design characteristic, pooling election years. After controlling for phenomena central to the spurious effect claim and pooling election years, I extend the divisive primary hypothesis to the arena currently void of evidentiary support: elections to the House of Representatives. In addition, I find that the divisive primary penalty is asymmetric, applying in incumbent but not open-seat races.