Just days ago, Mitt Romney appeared to be on a glide path to the Republican nomination. Coming off a big victory in New Hampshire and an apparent win in Iowa, he had just received the endorsement of Jon M. Huntsman Jr., and had as much as a 23-point lead in some national polls of Republican voters. He also led in South Carolina polls, sometimes by double digits, and attacks by Republican rivals on Mr. Romney’s tenure at Bain Capital did not appear to be harming him.
The past several days, however, have seen one of the most shocking reversals of momentum ever in a presidential primary.
Newt Gingrich, in second place in South Carolina polls, performed well enough in Monday night’s debate in Myrtle Beach, S.C., to set off a double-digit swing in some polls, literally overnight. Then on Thursday, Gov. Rick Perry of Texas dropped his presidential bid and endorsed Mr. Gingrich, and Mr. Romney’s victory in Iowa disappeared as the state double-checked its totals and found that Rick Santorum had received the most certified votes. Meanwhile, Mr. Romney struggled through another debate in North Charleston, S.C., on Thursday, while Mr. Gingrich delivered a thunderous soliloquy to the moderator, John King, on a question about allegations made by one of Mr. Gingrich’s ex-wives, seeming to turn the story to his advantage.
Although there is still considerable uncertainty about the outcome in South Carolina tonight, the most recent polling data shows that Mr. Gingrich’s lead there continues to grow in the closing hours of the campaign, setting up what could be an impressive margin of victory. An American Research Group poll, conducted Thursday and Friday, gives Mr. Gingrich a 14-point lead over Mr. Romney. Another survey, from Public Policy Polling, pegged Mr. Gingrich’s lead at 9 points — but found that it had continually increased as the poll was in the field, with Mr. Gingrich ahead by 14 points based on interviews conducted after the Thursday night debate.
The FiveThirtyEight forecast model, which heavily weights the most recent polling and a candidate’s momentum, now projects Mr. Gingrich to win South Carolina by just under 10 points, forecasting him to get 38.7 percent of the vote to Mr. Romney’s 29.3 percent.
The forecast model also presents a vote range or confidence interval for each candidate, something that is especially important to consider in this instance because of the volatility in the polls. (By design, the range is wider the greater the level of disagreement in the polling.) Mr. Gingrich, for instance, could plausibly finish with as little as 26 percent of the vote tonight or as much as 49 percent, according to the model.
Following our past practice, we will consider how the range of possible outcomes for each candidate could play out in Florida and in subsequent states.
538 forecast (most likely outcome): 39 percent
High end of forecast range: 49 percent
Low end of forecast range: 26 percent
There is still some chance that Mr. Gingrich could lose to Mr. Romney — the forecast model pegs the odds at about 13 percent, while the betting market Intrade puts them at 10 percent. The South can be a difficult region to poll: in 2008, polls badly underestimated Barack Obama’s margin of victory in South Carolina and in other Southern states, while some polls incorrectly projected Mike Huckabee to win the Republican primary that year. And Mr. Romney still led in some polls conducted after Monday night’s debate, including some by high-quality polling firms like Marist College.
That polls conducted after the second debate on Thursday night hint at continued for momentum for Mr. Gingrich is not a good sign for Mr. Romney. Still, Mr. Gingrich’s projected margin of victory in South Carolina is not dissimilar to the one Mr. Obama held over Hillary Rodham Clinton in the New Hampshire primary in 2008. The forecast model “knows” that massive errors sometimes occur in the polling, and that these errors do not always abide by the prevailing news media narrative about the race.
But the uncertainty in the forecast could also work to Mr. Gingrich’s benefit: he is roughly as likely to win by 20 points or more as he is to lose the state.
The margin of victory is likely to matter some because Florida is a stronger state for Mr. Romney. As this article by Stephen F. Hayes explains, Mr. Romney’s campaign has long placed a priority on Florida, and his advertisements should dominate the state’s 10 media markets. Moreover, much of the voting in Florida takes place early, and Mr. Romney’s campaign has encouraged his supporters to vote as soon as possible. Ed Morrissey of the Hot Air blog estimates that Mr. Romney has already banked a 37,000-vote advantage based on this early balloting.
538 forecast (most likely outcome): 29 percent
High end of forecast range: 39 percent
Low end of forecast range: 19 percent
Mr. Romney could still win South Carolina if he finishes at the top end of his forecast range. If he finishes at the bottom of the range instead, with no more than about 20 percent of the vote, there is an outside chance that he could finish in third place. But this would require either Representative Ron Paul of Texas or Mr. Santorum to considerably beat their forecasts in addition to a big underperformance by Mr. Romney. Meanwhile, surveys suggest that, although Mr. Romney has lost some backing, most of his renaming supporters are firmly committed to him.
If the results come in close to the model’s expectations, with Mr. Romney losing to Mr. Gingrich by perhaps 9 or 10 points, it will be important to check the exit polls for signs of whether the result seems to reflect more of a rejection of Mr. Romney or growth in Mr. Gingrich’s support.
Initially, it had seemed that Mr. Gingrich’s gain in the polls had come mostly at the expense of other conservative candidates like Mr. Santorum and Mr. Perry. That appears to be less the case, however, based on the most recent polls, which have shown Mr. Romney’s numbers declining in addition to Mr. Gingrich’s gaining. Between the forecast the model put out on Monday and the one it has now, Mr. Romney lost six percentage points, while the numbers for Mr. Santorum and Mr. Paul roughly held steady.
538 forecast (most likely outcome): 16 percent
High end of forecast range: 24 percent
Low end of forecast range: 8 percent
Getting about 16 percent of the vote in South Carolina, as our model now projects, would in some ways be a good result for Mr. Paul. The South’s political culture has historically been more friendly to populism than libertarianism, and Mr. Paul received only 4 percent of the vote in South Carolina in 2008. Nor does Mr. Paul have the impressive turnout operation in South Carolina that he had in states like Iowa and New Hampshire. If Mr. Paul has a strong finish, it will be a sign that his campaign can compete in almost any region of the country, possibly allowing him to pick up enough delegates to have some influence at the Republican convention.
Whatever the result in South Carolina, Mr. Paul’s focus is likely to shift to low-turnout caucus states like Nevada and Colorado, where his relatively strong organization and enthusiastic base could enable him to win a larger share of the vote. Mr. Paul may root for Mr. Romney to do poorly tonight, since he is also likely to be strong in the caucus states.
538 forecast (most likely outcome): 14 percent
High end of forecast range: 22 percent
Low end of forecast range: 7 percent
South Carolina has in some ways been frustrating for Mr. Santorum. He received the backing of a key group of evangelicals last week, and he performed well in the Thursday night debate. However, he is only projected to win about 14 percent of the vote in our forecasts, and is a slight underdog to Mr. Paul to finish in third place.
If the result goes according to the forecast, it will testify to the fact that it is difficult for more than one candidate to have favorable momentum in the same state at the same time, and that Mr. Santorum has been overshadowed by Mr. Gingrich. It might also provide a reminder that the political culture is complex in South Carolina. Although Mr. Santorum has strong credentials as a social conservative, he is not a Southerner, and he is a Catholic rather than an evangelical Protestant, factors that contrast with Mike Huckabee, who finished with 30 percent of the South Carolina vote in 2008.
The chain of events in which Mr. Santorum remains most relevant to the campaign, in my view, is one in which Mr. Romney’s campaign seems to be in grave trouble — but where Republican elites are worried about the possibility that the party could nominate Mr. Gingrich instead, who has extremely poor favorability ratings with the general electorate.
That could possibly lead to Mr. Santorum’s being viewed as a compromise choice. He seems to be aware of this: in recent debates, he has arguably done the best job of appealing to a general election audience, taking time to explain the rationale behind his positions without resorting to hyperbole, and occasionally even attacking his rivals from the left. Finishing in a clear third place tonight could keep this possibility alive for Mr. Santorum. A last-place finish, however, would put him under pressure to drop his campaign.