The second problem is that social science theories are, by their nature, reflexive. They try to explain human behavior, but humans can, in turn, read about these theories and adapt to them. We know from reporting on how both parties are attempting to better use data to reach voters that Republicans have paid close attention to political science theories about campaigns. If GOP decision-makers read these analyses about Trump, they might have concluded that they did not need to do anything to stop him, because he would inevitably fail anyway.

That conclusion would have been wrong, as voters in 12 states are poised to demonstrate Tuesday. Most of the political analyses concluding that Trump would not win were based on variations of “The Party Decides.” But that book argues that the reason insurgents are rarely successful is that party elites coalesce around an establishment front-runner, giving that person a commanding edge in resources, news coverage and endorsements over any alternatives before the voting starts and before most of the public is really paying attention.

In other words, party leaders actually have to do something to stop an insurgent. The whole reason smart analysts believed that Trump had no chance was because they thought GOP leaders would eventually strike against him. But that didn’t happen.