If citizens are to hold politicians accountable for their performance, they probably must have some sense of the relevant facts, such as whether the economy is growing. In surveys, Democrats and Republicans often claim to hold different beliefs about these facts, which raises normative concerns. However, it is not clear that their divergent survey responses reflect actual divergence of beliefs. In this review, we conclude that partisan divergence in survey responses is often not due to sincere, considered differences of belief that fall along party lines-but determining what it is due to is difficult. We review the evidence for possible explanations, especially insincere responding and congenial inference. Research in this area is still nascent, and much more will be required before we can speak with precision about the causes of partisan divergence in responses to factual questions.