Leave a comment, and send us questions @FiveThirtyEight.
We’re calling it a night. (Well, sort of: Harry will have more about Bernie Sanders’s shocking win in Michigan in a separate article later on.) In the meantime, here are my initial thoughts on why the polls in Michigan, which had Clinton ahead by 21 percentage points, got things so wrong.
I’ll also be posting some extended thoughts about Marco Rubio, who obviously had a terrible evening. In fact, Rubio may not net any delegates from any of Michigan, Mississippi and Idaho, having failed to hit delegate thresholds in all three states. It appears that some of this reflects tactical voting — there were a lot of late-deciders for Kasich — rather than a total collapse of Rubio’s image. So perhaps, just perhaps, he can hold out hope of that tactical voting working in his favor in his home state, Florida. But it’s an awfully long parlay — he’ll have to overcome very skeptical media coverage this week, then somehow win Florida despite already having been behind in polls, and then overcome a huge delegate deficit (or win at a contested convention) even if he wins his home state. Anyway, more on what’s going wrong for Rubio later.
For the other three Republicans, things were reasonably in line with expectations (or at least, FiveThirtyEight’s expectations). But note that things being in line with expectations is basically good news for Trump since he’s currently leading the nomination race.
Trump wound up with 37 percent of the vote in Michigan, a good sign with three big Midwestern states, Ohio, Illinois and Missouri, set to vote March 15. He ran especially strongly in southeastern Michigan, including in Macomb County, which was one famous for its “Reagan Democrats,” as well as in rural northern Michigan. There were still a few trouble signs — the exit poll showed Trump tying Cruz among women and losing to him in a hypothetical one-on-one matchup and (once again) faring mediocrely among late-deciders. But the results tonight take the notion that Trump was in some sort of free-fall largely off the table. He’s not invincible, but he won’t be easy to beat.
Trump also won Mississippi, getting 47 percent of the vote to 36 percent for Cruz. What’s the difference between Mississippi and Louisiana, which was closer? The difference may be that Mississippi was an open primary while Louisiana was a closed one, a factor to keep in mind going forward.
Cruz’s evening was reasonably good also, however, with two second-place finishes (very narrowly in Michigan ahead of Kasich) along with what looks like a fairly emphatic win in Idaho. He won’t lose many delegates to Trump — he’s down about 10 as I write this, with a chance to gain some back in Hawaii early this morning. Cruz had a reputation for being a regional candidate, but he now has won states in all four regions of the country: the Northeast (Maine), the Midwest (Iowa and Kansas), the South (Texas and Oklahoma) and the West (Idaho and Alaska). His chances look pretty good of emerging as the main challenger to Trump, much to the GOP establishment’s chagrin.
Kasich’s performance, on the surface, was somewhere between a par and a bogey. His final results in Michigan were in line with polling averages, although expectations were probably inflated in light of one poll that, in contradiction to the polling average, had Kasich winning Michigan. However, Kasich won’t pick up any delegates in Mississippi or Idaho. More to the point, he doesn’t really seem to have a plan to win the nomination without a contested convention and has admitted as much. Still, Kasich could possibly benefit from the fact that Rubio had an even worse night, especially given that Kasich seems more likely to win Ohio than Rubio is to win Florida.
There are a few possible explanations for the huge polling miss in the Michigan Democratic primary: For instance, if a significant number of Clinton supporters stayed home out of complacency or crossed over to the Republican primary to oppose Trump, that may have contributed to Sanders’s shocking win. We might get more explanations from pollsters soon — I’ve emailed several who showed Clinton ahead by 10 percentage points or more to ask why they think the polls were so far from the voting results.
Here’s another possible explanation: The most recent Michigan polls in our database stopped contacting voters Sunday, the night of the last debate, held in Flint, Michigan. Although many thought Clinton performed better than Sanders in the debate, perhaps voters felt differently. After many pollsters missed Ted Cruz’s Iowa win and suggested that Clinton would win Iowa easily — she won so narrowly that some call it a tie — several pollsters told us a lesson they learned: “Keep on contacting voters as late as possible.” But it’s up to poll sponsors to pay to contact voters until the final days of the race, and none did so in Michigan.
Just when you thought it couldn’t get any worse for Marco Rubio, it has tonight. Buried in the drama is the fact that Rubio has yet to win a single delegate in tonight’s primaries. Not only did he fall woefully short of hitting Mississippi’s 15 percent delegate threshold and 6 percentage points shy of hitting Michigan’s 15 percent threshold, his current 18 percent in Idaho is just barely below the Gem State’s 20 percent threshold. Luckily for Rubio, Hawaii’s lack of a delegate threshold is likely to save him from a humiliating shutout. But Rubio’s mainland fortunes couldn’t have fallen any further.
Our colleagues at ABC News have called Idaho for Cruz.
Cruz now leads Trump by about 11 percentage points in Idaho, according to both The New York Times and Decision Desk HQ. The one caveat is that there are relatively few returns in from Ada County (Boise), where Trump is running a little better than his statewide figures. Still, it’s not a bad night for Cruz, who’s also pulled narrowly ahead of Kasich for second place in Michigan.
Michigan is a large and diverse state, and this makes summarizing its recent economic picture kind of difficult. The Bureau of Labor Statistics reported Michigan unemployment in December 2015 as 5.1 percent, about the same as the U.S. overall.
But Michigan has been at the center of symbolic discussions over the 2008 economic meltdown and the impact of inequality and environmental racism. After the big crash, the question about whether to bail out the auto industry symbolized what had gone wrong with the economy and forced Republicans and Democrats to articulate positions on government interference in the economy. The concept of “bailouts” spurred the tea party movement. In 2012, Mitt Romney’s “let Detroit go bankrupt” statement came back to haunt him. Michigan, where Romney’s father served as governor, handed Obama a nearly 10-point victory. Last weekend, Hillary Clinton tried to depict Sanders as having been on the wrong side of the issue, pointing out that he voted against the auto bailout when it was packaged with the financial industry bailout.
This year, Michigan — specifically Flint — has been at the center of the debate about the great disparity in health and safety across economic and racial lines. Democrats have worked to use the Flint water crisis to cast a shadow on Michigan’s Republican governor, Rick Snyder, and Republican governing philosophy in general. But it also highlights the fact that at the end of Obama’s presidency, all is not well with many of the people Democrats claim to represent.
Clinton’s accusation doesn’t appear to have stuck, however. As Nate, Clare and Harry have pointed out tonight, Sanders’s victory — just called as I write this — is a big surprise. And we’ll have to wait for more data to know exactly which voters drove the outcome. Symbolically, though, Michigan represents the efforts and limitations of the Obama administration. As Clinton runs to carry on the Obama legacy, this is a significant upset.
Q: Not sure if 538 is reading this message board — but I would really be interested in an article that addresses the question of whether Sanders or Clinton is “more electable” or actually (at this point in time) can be said to have a “better chance” against Trump/Cruz/Kasich (whomever wins). — commenter Josh Medly
A: Electability is tough to pin down. I think we can say with confidence how the potential candidates compare to candidates who’ve won (and lost) in the past. This is probably what makes Democrats, especially older ones who remember the 1980s, nervous — Sanders reminds them of liberal Democrats who got trounced in the general election. What we don’t know for sure is how these characteristics — ideology, or even personal characteristics such as age, religion or home state — will matter in 2016. The electorate evolves over time, both in terms of its ideological preferences and in terms of the kinds of qualities it looks for in a president.
Nate, you noted one reason Sanders pulled out his victory in Michigan: He’s losing to Clinton among black voters in the state by much less than he lost to her among black voters in previous states. That may be a sign that he gets more support from black voters outside the South, which if it persists past Michigan could help him stay competitive in the Democratic race. Most of the previous primary states with enough black voters to measure their presidential preferences in exit polls were in the Southeast and the Southwest. In those regions, Clinton led Sanders among black Democratic voters by 73 percent to 19 percent, or 54 percentage points, in an aggregation of all polls so far this year by Reuters. Everywhere else, her lead narrowed to 35 percentage points: 64 percent to 29 percent. Clinton’s lead in Michigan among black voters is exactly that: 35 percent.
In what might be one of the greatest shockers in presidential primary history, The Associated Press has called Michigan for Sanders. Most of the polls were not close, and any thought that Sanders would exit this race in the foreseeable future has been put to rest by a stunning victory.
With vote-counting in Mississippi almost complete, Trump leads Cruz 47.4 percent to 36.5. A very good performance, although it appears Trump will come in just shy of 50 percent. That extends a streak we’ve been tracking: So far, no Republican has achieved an outright majority of the vote in any of the 21 states to have voted to date (I’m counting Michigan and Mississippi among those 21 but not yet Idaho and Hawaii). Rubio did achieve a majority in a territory, Puerto Rico.
I am skeptical of the math working out for Clinton in Michigan given that Kent County (home to Grand Rapids) just reported a ton of votes on its election department page, and it has Sanders expanding his lead there.
One reminder about the Democratic calendar: Although there are a lot of reasonably good Clinton states on March 15 — according to both polls and demographics — we then have a stretch of states that look quite strong for Sanders. These include Idaho, Utah, Arizona, Alaska, Washington, Hawaii, Wisconsin and Wyoming, all of which vote from March 22 to April 9.
So if Clinton underperforms her polls on March 15 in states such as Ohio as she did tonight in Michigan — and pollsters probably ought to be checking their turnout models carefully after tonight — she could have a really long few weeks ahead, even though Sanders’s delegate math remains highly challenging.
We won’t know exactly how many delegates Sanders or Clinton will take out of Michigan until the results are final, but based on the numbers so far, The Green Papers estimates that Sanders would get 72 delegates and Clinton 58. That would put Sanders five delegates above his FiveThirtyEight target for Michigan and Clinton five delegates below.
Another chunk of Wayne just came in and Clinton is now down just 17,000 votes or so, but that’s still not a small advantage for Sanders. Still votes left in Wayne, Grand Rapids and Flint. We’ll see if it’s enough to make up the difference. I’m somewhat skeptical.
We’re getting some initial results in from Idaho, although they contradict each other. The New York Times had it Cruz 39, Rubio 28, Trump 22 with about 4,400 votes reporting. Conversely Decision Desk HQ, reporting different initial precincts, had it Trump 36, Cruz 33, Rubio 18 after about 4,300 votes.
Sanders’s lead just dropped to about 24,000 votes with some more of Wayne County reporting. Wayne (at 75 percent) is still behind the overall state count (81 percent) in terms of precincts reporting, but 24,000 is still a healthy lead. We’ll see what happens next.
Donald Trump has won Michigan, his first Midwestern victory.
One of the questions going forward, as the Republican “establishment” casts about for a path to stop Trump, is what will happen in the race for second place. John Kasich was the main hope for beating Trump in Michigan, but Ted Cruz is the clear second-place finisher in the GOP nomination race overall. Rubio’s strong second-place hopes seem far behind us.
From where we stand now, the South Carolina primary seems like a turning point. Trump showed that he could beat Cruz in the South, and a crucial fact emerged: Trump is a strong, consistent plurality candidate. The issue between Cruz and Rubio looked like one of coordination — if they could combine into one candidate, that candidate might well beat Trump.
There have been only a handful of contests in which “Med Crubio” wouldn’t have defeated Trump or come pretty close. That number includes New Hampshire, where Kasich took second place. And tonight’s race in Michigan shows Kasich and Cruz competing for second while Rubio is way down; you could combine his votes with either Kasich or Cruz and it wouldn’t touch Trump’s lead in that state (based on the last numbers I saw, 10:20 p.m. EST). However, Kasich and Cruz together would beat Trump.
So here’s the game as I see it now: Kasich may come to replace Rubio as the “establishment lane” candidate who can at least deprive Trump of a delegate majority. Cruz will probably remain the best bet to win a plurality in his own right. If Cruz comes in second place, the establishment will have a choice to make.
I said in our Slack chat today that I had a “gut feeling” that Sanders could beat his polling in Michigan. I also said, for the record, that you should mostly ignore that gut feeling. But it wasn’t a total shot in the dark. There were a few things that made me think a closer-than-expected result was possible:
- Our demographic model, as opposed to our polling model, suggested that Michigan could be relatively competitive. It had Sanders winning Michigan by 4 in an even national race. The national race isn’t even — instead, Clinton is up by 13 percentage points in our national poll average. But still, that would extrapolate to a high-single-digit or very-low-double-digit win for Clinton, and not the blowout pollsters were expecting.
- As I wrote earlier today, complacency was a risk for Clinton voters in Michigan, especially with an open primary with voters potentially casting ballots in the Republican contest instead. “In sports, we’d call Michigan a ‘let-down game,’” I said.
- Michigan has a history of polling upsets, such as John McCain winning the Republican primary in 2000 and John Engler beating Jim Blanchard in the gubernatorial race in 1992 despite being way behind.
Basically, I’m not sure that Michigan was ever really a 20-point race, as polls had it. Based on the demographics of the state, it probably narrowly favored Clinton. But then, perhaps some of her voters didn’t show up, or voted in the GOP primary instead, because it didn’t look like Clinton needed their vote. That might potentially be enough to push Sanders over the top, although it will be very close.
By the way, this is part of why we try to approach the primaries from multiple perspectives. Our polling averages are a useful tool, but not the only one we look at.
It’s possible that Clinton closes this gap, but people should understand that a 33,000+ vote deficit is a lot with 74 percent of the precincts reporting. Clinton’s going to need a lot of heavy lifting from the remaining votes out.
Sanders is up 4 points with 73 percent in … probably still a lot of vote out in Detroit, Flint, Grand Rapids, etc.
I just got back from walking my dog; where do things stand in Michigan?
We’re still waiting on the Michigan Dems’ side of the results, where the race is looking unexpectedly close between Clinton and Sanders. Exit polls coming out of the state are showing that about four in 10 Democrats said that electability and experience were most important to them — those people leaned Clinton — and six in 10 said honesty or empathy were most important, and those are people who have tended toward the Sanders side of the equation.
This, from our friends at ABC News, is interesting: “Eight in 10 voters in the Democratic contest in Michigan were more interested in an experienced candidate than in an outsider,” and although in the past Clinton has won this group by seven out of 10, her share in Michigan of this demographic is “just more than half” right now.
Could it be that some “main line” Michigan Dems — the exit polls say that seven out of 10 voters in the primary identify this way — decided that Clinton probably had things sewn up and decided to go with bit of a “heartstrings vote” in the form of Sanders? Maybe.
By this time in 2008, 42 states had held a primary or caucus (or both, in Washington’s case). The 2012 and 2016 elections have not been so front-loaded. So who’s choosing the Republican nominees in these early contests? The proportion of “swing” states that had voted by now has been about the same this year as it was in 2012.
What’s different is that twice as many strong Republican states will have cast their nomination votes by tomorrow morning — Southern strongholds such as Mississippi, Alabama and Texas, alongside Western states such as Idaho (and some states with nonbinding caucuses). And fewer states where Republicans lost big in the previous election are among the states voting up to this point. This shift is partly because states such as Washington took later slots this time and partly because states like Maine and Michigan moved from Democratic strongholds to the competitive column from 2008 to 2012.
I said earlier that this was a boring election night. I changed my mind. This is very exciting. Thank you to those of you who voted in the Michigan Democratic primary!
One thing helping Sanders tonight is a comparatively strong performance with African-American voters. He’s losing them only 65 percent to 30 percent in Michigan, according to exit polls, which doesn’t sound great but is much better than in other states, where he’s lost them as badly as 91 percent to 6 percent.
Could that bode well for Sanders in Illinois, Ohio and other states with black populations similar to Michigan’s? It certainly can’t be a bad sign for him. But some caution is required since we haven’t gotten much data on how black Democrats are voting outside of the South. (Exit polls were conducted in states such as Massachusetts and Iowa, but there weren’t enough black voters there for the exit polls to estimate the results.) That makes it harder to say whether Michigan is part of a trend or some sort of fluke — but obviously it’s something we’ll be analyzing in the days ahead.