[...] I still have a couple of points of caution.

One is that it isn’t uncommon for candidates to run up the score in primaries that appear to be noncompetitive. In fact, this happens all the time. Polls usually call the winners right in primaries, but they often lowball their margins of victory. It isn’t very motivating to turn out for a guy who’s going to lose a state by 30 percentage points, even if doing so might win him an extra delegate under some obscure provision of the delegate rules. In Indiana, a genuinely competitive, winner-take-all race where Cruz is (theoretically) the clear alternative to Trump, that won’t be such a problem.

The other caveat is that the Republican race has not only defied “momentum” but often contradicted it. Whenever Trump seemed to be on a glide path to the nomination (such as after Super Tuesday or March 15), he’s had a setback. When he’s seemed to be vulnerable (such as after losing Iowa or Wisconsin), he’s rebounded.

This maybe — or probably — is just our reading too much into noisy data. But it’s possible there’s some sort of thermostatic effect at work. When Trump seems to be on the verge of becoming the presumptive nominee, there’s more focus on his awful general election numbers; meanwhile, the media’s incentives for covering him change, with the possibility of Trump imploding at a contested convention becoming a more attractive story than the man-bites-dog narrative of Trump winning the nomination in the first place. When Trump seems to be in trouble, conversely, Republicans are forced to contemplate the problems of a contested convention and the inadequacies of Cruz and Kasich, and the media becomes more eager to tell a Trump comeback or pivot story.

Indiana is important not only because of its delegates, but also because it will give us an indication as to whether the apparent change in Republican attitudes is temporary or permanent. If Trump wins Indiana despite its middling-to-fair (from his standpoint) demographics, he won’t quite be the presumptive nominee because he’ll still need to follow through with a decent performance in California. But he’ll at least be in the liminal zone that Hillary Clinton spent a lot of time in, with the race not quite wrapped up mathematically but close enough that something (a gaffe, a scandal) would have to intervene to deny him the nomination. Incidentally, Trump’s potential support from the uncommitted delegates in Pennsylvania will give him more margin of error in that situation.

If Trump loses Indiana, however, that will suggest the race is still fairly volatile week-to-week, that he’s very likely to lose states such as Nebraska that vote later in May, and that the geographic and demographic divergences in the GOP haven’t reversed themselves so much as they’ve become more exaggerated. It will improve the morale of anti-Trump voters and change the tone of press coverage. And mathematically, it will make it hard (although not quite impossible) for Trump to win 1,237 delegates outright; he’d be back to fighting tooth-and-nail for every uncommitted delegate.

I don’t know what’s going to happen in Indiana. But Trump’s strong results over the past two weeks have changed the Hoosier State from potentially being “must-win” for Trump to probably being “must-win” for his opponents.