We’re missing quite a bit of data here, but we still have exit polls for states with over half the total delegates, and they’re reasonably representative of the sample as a whole. Notice that these late primary states are overall substantially more moderate, and significantly less evangelical than the earlier states.

This could prove critical in a drawn-out race. The reason is simple, and yet not well-known. The RNC has provided that states holding primaries before April 1 must allocate delegates proportionately. But after that date, states may opt for winner-take-all primaries, and many of these states have done so. In other words, we could have a situation where a conservative candidate (or a pair of conservative candidates) does well in the first three months, but has to give some delegates to the more moderate candidate. This is similar to what happened to Clinton, who won crucial primary battles late in the game, but couldn’t make much headway in the delegate count because of how these delegates were allocated. So despite winning the majority of primaries, the conservative candidate could end up with only a small lead in delegates over the more moderate candidate. If the moderate candidate then performs well in April or afterward, he could quickly rack up enough delegates to break away and claim the nomination.

Of course, the moderate candidate in this scenario would have to win some primaries in February and March if he hoped to stay in it. But there are probably enough big, moderate states like Illinois, Michigan, Florida and New Jersey that a moderate candidate could stay in it. In other words, an extended race probably favors a candidate like Romney.