People sometimes apply the term “tossup” a bit too broadly, using it to refer to anything close enough that they don’t want to render a prediction about it.
In Michigan, however, the term is appropriate. Rick Santorum, who once trailed Mitt Romney badly in the state, then surged to a clear lead there, then saw Mr. Romney regain his footing and pull back ahead, appears to have some late momentum in the race — perhaps just enough to win, and perhaps not.
In fact, after adding in a poll from Public Policy Polling that showed Mr. Santorum ahead by five points in interviews conducted on Monday night, our forecast model of the state briefly showed a tie, with Mr. Romney and Mr. Santorum each at 37.6 percent of the vote.
A second poll was then published, from Foster McCollum White Associates and Baydoun Consulting, which showed Mr. Romney with a narrow two-point lead. That was enough to put Mr. Romney back ahead in the forecast model — by the whopping margin of 0.7 of a percentage point. The forecast now gives Mr. Romney a 55 percent chance to win Michigan and Mr. Santorum a 45 percent chance.
However things turn out on Tuesday, this has been a dramatic enough sequence that it demands some explanation. It is unlikely that Mr. Santorum’s last-minute rebound is purely a statistical fluke. There is a fairly rich amount of polling in the state and, importantly, Mr. Santorum has gained ground in consecutive polls issued by the same survey firms. In the Baydoun Consulting poll, for instance, which had him down by two points on Monday, he had trailed by eight points just days earlier. So the rebound is probably real.
There are about eight plausible explanations for what might have caused it:
1. Mr. Santorum has the better closing message. His campaign has been more positively oriented, although not uniformly so.
2. Voters are rallying to his side after he took some harsh treatment in news media coverage.
3. His voters are more enthusiastic and starting to come home as likely voter models become more accurate.
4. Mr. Santorum is picking up support from Newt Gingrich supporters who have concluded that Mr. Gingrich is not viable.
5. A set of minor gaffes by Mr. Romney, related to the staging of his Ford Field speech and a remark he made about Nascar, hurt him at the margins, as well as the fact that he took part of Sunday away from the campaign trail to attend the Daytona 500.
6. Mr. Romney had some temporary momentum from last week’s debate — in my view it was a “win” for Mr. Romney but not an overwhelming one — which has since evaporated.
7. Mr. Santorum, whose “super PAC” bought a fair amount of advertising inventory in Michigan about a week ago, has equalized the advertising gap in the closing days of the campaign, having been disadvantaged by it before.
8. Mr. Santorum is benefiting from Democrats, some of whom are crossing over to vote in an effort to create chaos in the primary, and some of whom are responding to robocalls that were placed by Mr. Santorum’s campaign.
Take your pick from this menu — I tend to think that Nos. 1, 6, 7 and 8 are probably more salient factors than the others, but there is no way to tell.
I also don’t think there is any way to say who has an edge above and beyond what the top-line numbers in the polls say. On Monday, when I examined a set of 15 “intangible” factors that I sometimes use to break ties and get a feel for a close election, I found that they came out about evenly. Mr. Santorum’s most tangible advantage is probably that he has outperformed the polls in several previous states; Mr. Romney’s is that he has banked an advantage from early voters.
One word of warning: factors like early voting, crossover voting and the relatively large amount of demographic diversity within Michigan will make it tricky to call the state based on exit poll results and the first few precincts that report. For instance, if early and absentee results are reported before those cast on Election Day, as is common in some states, Mr. Romney could initially emerge with a lead that proves ephemeral. I generally take the view that the news networks are too quick to call a race — shouldn’t have we learned something from Florida in 2000 or Iowa this year? — but there is reason to be especially cautious here.
Let’s briefly examine the race from the standpoint of each of the four candidates competing. The values you see represented below reflect the confidence intervals from the FiveThirtyEight forecast model, enough to cover 90 percent of all possible outcomes.
Mitt Romney
FiveThirtyEight forecast (most likely outcome): 39 percent
High end of forecast range: 45 percent
Low end of forecast range: 30 percent
How much will winning or losing Michigan — by any amount — matter to Mitt Romney and his chances of winning the Republican nomination?
This is a separate question from how much winning or losing should matter. Fundamentally, we learn essentially nothing different from a state if a candidate wins it by 1,000 votes or loses it by the same margin. And there are only two delegates awarded to the overall winner in Michigan; most of them are allocated at the Congressional district level instead. (Although several Congressional districts are close enough that a swing of a point or two could make the difference.)
Still, the news media narrative would undoubtedly be negative were he to lose by any margin. Mr. Romney would not be the first candidate to lose his native state — George W. Bush lost Connecticut in 2000 but won the nomination, and Gerald Ford lost Nebraska in 1976 and captured the nomination that year. (Mr. Bush and Mr. Ford did carry Texas and Michigan, the states where they made their homes at the time, as Mr. Romney will probably carry Massachusetts.) But Mr. Romney did win Michigan in 2008 and was thought to be a clear favorite to win there again — at least until the polls started to show Mr. Santorum ahead.
If Mr. Romney loses, you’ll begin to hear questions asked, like why Mr. Romney has had trouble closing out victories, whether his organizational strengths are overrated and whether voters are souring on Mr. Romney as they get to know him better — not to mention all the chatter about late-entry candidates and brokered conventions.
None of these questions would necessarily lead you to the conclusion that Mr. Romney is likely to lose the nomination even if he loses Michigan. (If he also performs badly on Super Tuesday, you could make a more credible case for that.) The point is simply that there are narrative stakes, as well as delegates, on the line — and the narrative can eventually spill over into the delegate allocation in a variety of ways.
Rick Santorum
FiveThirtyEight forecast (most likely outcome): 38 percent
High end of forecast range: 44 percent
Low end of forecast range: 30 percent
The flip side is that Mr. Santorum has also taken some lumps in Michigan, and he will have some of his own questions to answer should he lose. And if Mr. Santorum’s campaign has done a more careful job of managing expectations in Michigan, he was ahead in the polls for much of the past two weeks.
Among the questions that Mr. Santorum might need to address is whether his success is limited to caucus states, why he has had trouble securing support from party leaders, whether he is focused enough to avoid being nicked by Mr. Romney through advertising and debates, and whether he can continue to make controversial remarks about social issues without damaging his standing among general election voters.
Again, we should probably be asking these questions about Mr. Santorum win or lose, just as we should be asking the questions about Mr. Romney — unless, perhaps, one candidate emerges with a reasonably clear win.
But the fact is that the set of questions about Mr. Santorum are likely to receive much more emphasis if he drops the state and much less if he prevails. Perhaps the treatment will be slightly more equivocal if the result is extremely close, but I wouldn’t count on it: both Mr. Romney’s initial apparent win in Iowa and the subsequent reversal of that result had a fair amount of impact on the tenor of news coverage, even though they came by the tiniest of margins.
Ron Paul
FiveThirtyEight forecast (most likely outcome): 13 percent
High end of forecast range: 20 percent
Low end of forecast range: 7 percent
I could imagine Mr. Paul outperforming his polls in Michigan — he had one of his better debates last week, and he has seemed to draw the largest crowds in rallies around the state. But Mr. Paul is nowhere close to the front-runners, and unlike them, he is applying an almost pure delegate-aggregation strategy that does not rely on momentum.
Mr. Paul has devoted some time to campaigning in Michigan, suggesting that his campaign sees at least some hope of getting on the delegate scoreboard. His best shot might be in the 12th Congressional District, where the college town of Ann Arbor is located and Republican turnout is likely to be thin, and perhaps also in the 13th, which is mostly in Detroit and where there are fewer Republican voters still. The Eighth Congressional District, although it contains Michigan State University, is a less likely target, since turnout there should be higher and more mixed.
Newt Gingrich
FiveThirtyEight forecast (most likely outcome): 10 percent
High end of forecast range: 15 percent
Low end of forecast range: 5 percent
Mr. Gingrich has spent almost no time in Michigan, instead focusing on Super Tuesday states. He might have a rooting interest in a Romney victory, especially a narrow Romney victory that would cool off Mr. Santorum without creating much momentum for Mr. Romney either. That could give Mr. Gingrich an opening to prevail in Tennessee and Oklahoma on Super Tuesday, where he now trails Mr. Santorum — and perhaps take one last turn as the “anyone but Romney” candidate.