How much trouble will Hillary Clinton be in if she loses in Nevada, where Democrats will caucus on Saturday? How close does Bernie Sanders need to come in South Carolina, which votes a week later? And which states are really “must-wins” for Sanders in March, April and beyond?
We can try to answer all of those questions with the help of the gigantic chart you’ll see below. On the left-hand side of the chart, you’ll find a projection for how each state might go if recent national polls are right, with Clinton ahead of Sanders by about 12 percentage points nationally. The right-hand side is more crucial: It shows how the states might line up if the vote were split 50-50 nationally. Since the Democrats’ delegate allocation is highly proportional to the vote in each state, that means Sanders will be on track to win the nomination if he consistently beats these 50-50 benchmarks. Conversely, Clinton will very probably win the nomination if Sanders fails to do so, especially since superdelegates would likely tip a nearly tied race toward Clinton.
The starting point for these estimates is state-by-state polling from Morning Consult, a non-partisan polling and media firm that has surveyed about 8,000 Democrats online since Jan. 1. That’s a lot of responses, although not enough to provide an adequate sample size for all 50 states; while there are about 800 respondents from California in the sample, for instance, there are only a dozen or so from Montana.
The solution is to blend the polling results with other data. In particular, I used exit polls to determine the nonwhite share of the Democratic electorate in each state and how each state lines up on a liberal-conservative scale (Sanders does better in white and liberal states). I also included the amount of money raised by Clinton and Sanders in each state in individual, itemized contributions, and their ratio of Facebook likes. In states like California where there’s an adequate sample size from the Morning Consult polling, the polling gets a fair amount of weight, but in the smaller states the other factors predominante. (For a more technical explanation of how this is accomplished, check out the footnotes.exit polls from the 2008 general election. (Why use exit polls from 2008 instead of 2012, or from the primaries instead of the general? Because the exit poll was conducted in all 50 states that November, which was not true in the primaries or in 2012.)
The regression result is treated as equivalent to 500 Morning Consult interviews and then blended with the Morning Consult polls.">1)
Don’t get too attached to these: The state-by-state estimates are pretty rough. But they’re calibrated in such a wayProbit transformation.">2 so as to provide a reasonable benchmark of what a 50-50 race would look like. Maybe Michigan is less favorable to Sanders than this estimate holds, for example. That’s fine, but it means he’ll need to make up ground in another state.
What about the states that have already voted? We estimate that in a 50-50 national race, Sanders would win Iowa by about 6 percentage points, and New Hampshire by 26 points. He didn’t quite hit those targets in either state, although he came close — several percentage points better than you’d expect from his current national polling. As we’ve said, however, the real challenges for Sanders lie ahead.
A quick look at the calendar
Nevada. Nevada has a fairly high nonwhite population, but it isn’t especially liberal. Clinton was also well ahead of Sanders in the (relatively small sample) of interviews Morning Consult conducted there earlier this year. It’s possible that Clinton will be hurt because the state holds a caucus, although we don’t have a lot of evidence yet about which Democrat that benefits. In other words — and as much as her campaign might try to avoid admitting it — it’s a state that Clinton “should” win. Conversely, a Sanders win would be a sign he has staying power.
South Carolina. Clinton is the overwhelming favorite in South Carolina, but her margin of victory could be a useful benchmark for where the race stands nationally. Suppose, for instance, that Clinton winds up winning by 17 percentage points in South Carolina (a bit closer than most polls have it). Would that be a good result for her or a bad one? Our chart projects that Clinton would win South Carolina by 11 points in a 50-50 race, so she’d be doing a little bit better than that benchmark. But not a lot better: Such a result would still suggest that the national race had tightened.
Super Tuesday (March 1). Clinton is likely to compile lots of delegates from the seven Southern states that vote on Super Tuesday, although Oklahoma — which is quasi-Midwestern and relatively white — might be Sanders’s best shot at an upset. Sanders should win Vermont by a huge margin, meanwhile. That leaves the Minnesota caucus, Colorado caucus and Massachusetts primary as the races to watch; they’re the sorts of states Sanders absolutely needs to win to have a shot at taking the nomination.
Big-state primaries on March 8 and March 15. This is probably the most important eight-day stretch on the Democratic calendar. Michigan votes on March 8 (as does Mississippi), followed by Florida, Ohio, Illinois, North Carolina and Missouri on March 15. Together, these states will put 857 pledged delegates at stake, or more than 20 percent of the Democrats’ pledged total. (Pledged delegates, chosen by voters, are distinct from superdelegates.) Based on current polling, most of these states favor Clinton either narrowly or substantially, so Sanders will have to make up ground, perhaps enough to win a couple of them outright.
Favorable terrain for Sanders in late March. A series of Western states vote between March 22 and April 9, as does Wisconsin. Almost all of them figure to be favorable to Sanders — including Wisconsin, where he was already almost tied with Clinton in the polls before his New Hampshire win. A possible exception is Arizona, where Clinton beat Barack Obama in 2008 and where the electorate can be tricky to predict.
New York, California and a big blue finale. With some exceptions — Kentucky, West Virginia and Indiana will be interesting to watch — the last quarter of the Democratic calendar mostly resides on the coasts. And there are some big prizes: New York, Pennsylvania and California foremost among them. All three offer advantages and disadvantages to each candidate. For instance, will California’s left-wing politics, which help Sanders, prevail over its racially diverse population, which helps Clinton? Sanders probably needs at least two of the three states, and maybe all of them given Clinton’s lead in superdelegates. A win in California on June 7 would also carry symbolic power, as it’s the last state to vote,3 possibly allowing the winner to claim a mandate from the Democratic electorate.
But we’re getting ahead of ourselves. If Sanders can hang tight with Clinton in Nevada on Saturday, his chance of eventually notching a win in California and securing the nomination will look a lot better.
And check out the “perfect” Democratic stump speech.
Listen to the latest episode of the FiveThirtyEight politics podcast.