On Friday, Mitt Romney pulled into the lead in our
polling-based forecasts of the Michigan Republican primary. The forecast model has now become more confident in that conclusion, in fact, because two new polls released on Sunday night also showed an advantage for Mr. Romney.
The new polls, from Public Policy Polling and We Ask America, gave Mr. Romney leads of two and four percentage points, respectively, over Rick Santorum. They join two other polls released after Wednesday’s Republican debate, from Mitchell Research and Rasmussen Reports, that gave Mr. Romney leads of two points and six points, respectively.
There is less probability of an election night surprise when the polls are fairly consistent with one another, as they are in this case. That’s why our model makes Mr. Romney a 77 percent favorite to win Michigan even though he has a relatively tenuous 4-point advantage there.
Still, if the forecast holds, the Michigan result would be the closest of any primary to date (Iowa and Maine were closer, but they were caucuses) — close enough that it is worth considering whether there are any factors that might not be picked up by the polling but which could work for or against either candidate.
In 2009, I developed a set of 15 clarifying questions for close elections that were meant to be applied in cases like these. The questions cover various hypotheses for why a candidate might overperform or underperform in polling or why it might not tell the whole story. Oftentimes, the questions balance out, but they do not always — in the New Jersey gubernatorial election in 2009, for instance, they tended to favor Chris Christie, which is why I said the odds were in his favor even though the polling showed a dead heat there.
Although the questions were phrased with the general election in mind, they can be easily adopted to a primary. Let’s work through each question quickly. If most of them tend to favor Mr. Santorum, we might conclude that the race is more of a tossup. If they favor Mr. Romney instead, then he could be the favorite to beat the four-point spread that the model establishes.
1. Which candidate’s supporters are more enthusiastic?
A week or two ago, polls were finding that voters who were most certain or excited to vote in Michigan tended to favor Mr. Santorum, giving him a potential advantage especially in the case of low turnout. Now the surveys are more equivocal on that point. The Rasmussen Reports poll, for instance, showed a slightly larger lead for Mr. Romney (nine points) among those who said their choice was certain than it did overall.
Still, there has been enough overall evidence of the relative lack of enthusiasm for Mr. Romney throughout the nomination process to score this as a point in Mr. Santorum’s favor.
2. Which candidate is liable to have the better turnout operation?
Count this factor as a point in favor of Mr. Romney, who is widely regarded as having the better organization in the state. It should be noted that all the candidates’ “ground games” are fairly weak by historical standards, and if Mr. Romney does not win Michigan, I think we’ll need to ask whether this factor is quite all that it’s cracked up to be. But having more boots on the ground can’t hurt.
3. Is one of the candidates a challenger to an incumbent, who might benefit from the ‘incumbent rule’?
The incumbent rule is the notion that undecided voters tend to break toward the challenger in a close race. This was once regarded as a savvy piece of political wisdom, but the empirical evidence for it is lacking in recent years.
More important, there isn’t an incumbent in this race. Perhaps Mr. Romney is somewhat closer to being one, in that he is the more familiar face to Michiganders and won the state in 2008, but that’s too tenuous to tally a point against him.
4. Do the demographics of the undecided vote, or the weakly attached third-party vote, favor either of the candidates?
Most of the polls are no longer showing very many undecided voters. But in the survey that had the most, from Mitchell Research, they seemed more like the types of voters who would support Mr. Romney, as the undecideds were more likely to be female than male, and somewhat conservative rather than very conservative.
The rule also refers, however, to third-party candidates. Often (as in New Jersey in 2009), these candidates underperform in polling because voters conclude they are not viable and switch to one of the front-runners, not wanting to waste their vote.
There is an obvious analogy here in primaries, which is candidates who are well off the pace — in this case, Newt Gingrich and Ron Paul. One of the more heartening numbers for Mr. Santorum comes from the Rasmussen Reports poll, which found that he would defeat Mr. Romney by two percentage points in a true two-way race. That might suggest that he could gain a percentage point or two if Mr. Gingrich’s and Mr. Paul’s supporters drift away from them.
Still, since Mr. Gingrich and Mr. Paul no longer have all that much support in Michigan, and since the demographics of the undecided vote seem to slightly favor Mr. Romney, we are calling this a wash. If we are seeing on Tuesday night that Mr. Gingrich is in the single-digits, however, Mr. Romney’s advisers might have some reason to be nervous.
5. Which candidate got the more favorable coverage in the morning newspaper, or on the local evening news, on the Sunday before the election?
Although Mr. Romney generally had the stronger news cycles over the course of the last week, that should be priced into the model by now, since two of the polls were conducted on Sunday night and since the model also makes an effort to account for a candidate’s momentum. So this category stands to consider more late-breaking developments that were too late to be accounted for by the polling.
Mr. Romney did draw some negative headlines for his speech at Ford Field, which was inartfully staged and held before a mostly empty stadium. Still, the most recent polls postdate the Ford Field speech, so any impact should be accounted for.
6. Which candidate has the better “elevator pitch,” particularly as encapsulated by the commercials they’re running in the 48 hours before the election?
Neither candidate, frankly, has done much to drive a message, with Mr. Romney’s campaign (and his “Super PAC”) running mostly negative ads, while Mr. Santorum has pivoted awkwardly between economic and social issues and competing critiques that he is too much of part of the Republican establishment and too conservative to be electable. No point awarded here.
7. Which candidate has a wind at his back from the national political environment, or has a message that squares better with the national political mood? Which candidate’s party is liable to have the better election night nationwide?
This question, like the others, is phrased in the context of general elections, but there is an intuition here that can also be applied toward primaries. Is there evidence from other states — or from national polls — that one candidate has the momentum?
In this case, there is, enough to score a point for Mr. Romney. He has erased Mr. Santorum’s lead in the Gallup national tracking poll and he has also gained ground in Arizona, where our model now projects a blowout victory for him. This data makes it less likely that the shift toward Mr. Romney in Michigan is some sort of statistical fluke.
8. Does either candidate begin with a built-in lead from early or absentee voting?
Yes, Mr. Romney does, and so he scores another point. In the Public Policy Polling survey, Mr. Romney led by 33 points among those who have already cast early or absentee ballots. Although these voters represent only 17 percent of the projected turnout in the survey and although some of this is because Michigan’s rules allow voters 60 and older to cast absentee ballots for any reason (a group with whom Mr. Romney performs strongly anyway), it’s always nice to have a few votes in the bag.
9. Which candidate, if any, stands to benefit from up-ballot or down-ballot races?
This factor applies to cases were there are multiple races on the ballot and there could be coattails. However, the presidential primary is the only race on the ballot on Michigan on Tuesday, so it has no relevance here.
10. Which candidate stands to benefit from cellphone-only voters, or other voters who may not be represented in the polls?
Many of the polls in Michigan are automated surveys that do not call cellphones. But there isn’t much evidence on how this might bias the results against a candidate. The closest thing to an exception is probably Ron Paul, whose demographics are more cellphone-heavy, but he is not within striking distance in Michigan.
However, there is another group of voters that could be helpful to Mr. Santorum, which are Democrats who might cross over to vote for him in an effort to create turmoil in the Republican primary. Indeed, Michigan has a history of this type of tactical voting, which may have helped John McCain to a relatively clear victory there in 2000.
In theory, these voters should be captured in the polls like any others, but they may be screened out by likely voter models, who may regard them as dubious if they have not voted in past Republican primaries. I’m a little skeptical about how much of this will materialize, but it’s enough to score a point for Mr. Santorum.
11. Has the polling in previous elections in the state, or in similar elections in similar states, tended systematically to underestimate the performance of either candidate’s party?
Yes — polls throughout the primaries and caucuses so far have underestimated Mr. Santorum’s performance. Mr. Santorum scores a point here and it is perhaps the best reason to be optimistic about his chances.
12. Which candidate drew more people to their campaign appearances in the state over the last two weeks of the campaign?
With rare exception, Repubican candidates have drawn small-to-middling crowds throughout the nomination campaign, and Michigan seems to be no exception, with most reports describing audiences of perhaps 100 to 1,000 people. Estimates of crowd sizes at some Ron Paul events have been larger than that, but neither of the front-runners deserve a point here.
13. Which candidate got more contributions from voters within the state over the last six weeks of the campaign?
Mr. Romney has raised much more money than the other candidates in Michigan throughout the campaign — $1.6 million in itemized contributions, versus only about $50,000 for Mr. Santorum.
However, this category is meant to focus on fund-raising close to the election, and Mr. Romney’s advantage was surprisingly tenuous in January. In that month, he raised $51,246 from 179 itemized donors in Michigan, according to his recent Federal Election Commission filing, versus $38,743 among 101 donors for Mr. Santorum. That’s still a slight advantage, so we’ll score a point for Mr. Romney, but the slowdown in fund-raising could be problematic for Mr. Romney in the longer term.
14. Which candidate has run the more positive, optimistic campaign, and will leave voters feeling better about themselves as they exit the polling place?
Mr. Santorum has run the more positive campaign and scores a point in this category. His “Super PAC” has run some preemptive and negative ads in the state, but they have been counterbalanced by positive spots — more than in Mr. Romney’s case.
15. Which candidate’s party controls the state’s electoral and judicial apparatus, in the event of a recount or otherwise disputed election?
Once again, this question is framed in terms of a general election, but can be applied to a primary by considering which candidate has more support from statewide officials. In this case, that is Mr. Romney, who scores a point because he was endorsed by Michigan’s governor, Rick Synder, along with a series of other current and former elected officials. In the event of a disputed vote-count in Michigan — and remember, that has already occurred in two Republican states so far — they could weigh in favor of Mr. Romney.
If you’ve been keeping score, you’ll find that we’ve counted five categories (No. 2, 7, 8, 13 and 15) in favor of Mr. Romney, and four (No. 1, 10, 11 and 14) in favor of Mr. Santorum. There was also a case for scoring point No. 3 and No. 4 in favor of Mr. Santorum, but we came down against doing so.
Overall, that is not much of an edge for either side — which leads me to conclude that we should trust the polls, which give an edge to Mr. Romney but also leave Mr. Santorum with plausible upset chances. If the result does not abide by the polls — say, Mr. Romney wins by 10 points, or Mr. Santorum wins by any margin — we can revisit these factors to consider which ones might have played the largest role in the unexpected result.