Political science racks up some big wins

At this point, I’d like to be more specific and say that the political science profession had a blind spot about the transformation of the G.O.P. into an extremist party. That’s why Norman Ornstein at The Atlantic, who did face up to that reality, was so much closer to the mark than most of his colleagues.

But while political science had a big miss in the fall, it has done very well since the primary season got underway, at least as compared with standard political punditry.

Think of it this way: There have been two narratives of the campaign. One is full of ups and downs, momentum and stunning reversals. Donald Trump is doomed! He’s inevitable! Bernie Sanders has won seven in a row! Can his candidacy ever recover after New York?.

The other narrative sees a fairly stable race, with state-by-state results mostly reflecting demographic differences. In this view, momentum is just a bad metaphor; it involves treating what is basically noise, a string of very white states with open primaries, which favor Mr. Sanders, for instance, as a signal. As the political scientist Alan Abramowitz has noted, the Democratic race in particular is quite well explained by a model in which just three factors determine the vote share for Hillary Clinton: whether the primary is being held in the South, the percentage of African-Americans and the share of Democrats (as opposed to independents) in the voting. (See what Abramowitz’s analysis looks like here:

Thus, Mrs. Clinton’s big win in New York recently wasn’t a shocking reversal of Mr. Sanders’s momentum; it was what you’d expect in a state whose demographics look much more like the Democratic Party as a whole than the states Mr. Sanders had won in the preceding weeks. (Notice that Mrs. Clinton’s overall lead in the popular vote, 15%, is almost the same as her margin in New York.)

The point is that statistical analysis has consistently trumped horse race punditry all along. Quantitative political science is looking pretty good.