Tomorrow, Tuesday March 15, like many Tuesdays over the past couple of months, is THE DAY THAT IS GOING TO SETTLE THE PRIMARY ONCE AND FOR ALL!!1!!!!!111!
Which is to say, there are a fair number of delegates at stake, 691 out of the 4051 pledged delegates, over a quarter of the 2724 remaining pledged delegates.
[Aside: I’ll be drawing delegate counts from The Green Papers, confirming with 538’s counts, noting any discrepancies. Also, I’ll be ignoring superdelegates, because their votes are not pledged, and I honestly don’t believe that they are not going to swing for whichever candidate has more pledged delegates at the convention.]
At the moment, the pledged delegate counts are Clinton 775, Sanders 552. So, how many would Sanders need to win in order to be on track to capture 50% of the pledged delegates before the convention?
Sanders needs 1474 of the remaining 2724 delegates, or 54%. So, the simplest calculation would say that he would need to win at least 374 of tomorrow’s delegates.
But that calculation ignores the fact that the states vary in systematic ways. If Sanders had won 51% of the vote in Vermont, for instance, that would not be an indication that we was on track to win 51% of the pledged delegates nationally.
To address this issue, 538 put together their delegate tracker, which attempts to adjust for demographic variation among the states. Their model estimates the number of delegates each candidate would be expected to win in a given state in order to wind up with 50% total. For example, their model projects that Sanders should perform substantially better in Nebraska than in other states. Sanders won 15 delegates there to Clinton’s 10, which is exactly the split given by their model.
That is, to the extent to which their model is accurate, the results from Nebraska point to a very close primary race nationally. (Personally, I’m not sure about their model, as it seems to rely more on conventional wisdom and media narratives than on data — sort of a microcosm of the decline in quality of 538 overall. But, it is probably a decent first-order correction to a simple delegate-count horse race.)
According to the 538 model, Clinton’s target number is 365 delegates, while Sanders’s is 326. The difference comes from Florida and Illinois (and, to a lesser extent, North Carolina), where they expect Clinton to overperform relative to her national standing.
So, for example, if Sanders were to win 330 delegates, Clinton would extend her lead over him, but it would suggest that, if he keeps performing at that level, he will win more than 50% of the remaining delegates.
The problem (and the reason I wrote this) is that the number 326 does not account for the fact that Clinton already has a lead of more than 200 delegates. At this point for Sanders, winning 50% of the remaining delegates means losing the nomination.
So, I’m combining the two calculations — the demographic corrections from 538 and the current delegate totals — to come up with a number that I think represents a reasonable target for tomorrow’s primaries.
There are a couple of different ways to do this. One sets Sanders’s target at 353, and the other at 349. So, something in that vicinity, let’s call it 351. And Clinton’s corresponding target would be 340 delegates.
Of course, regardless of the specific outcome tomorrow, both campaigns will continue on, notwithstanding whatever predictable and idiotic statements come from the media. And once those results are in, I’ll update this calculation for next week’s primaries in Arizona, Idaho, and Utah.