The volatility of the South Carolina primary race over the last week made it an especially hard contest to track. Just about every poll had Newt Gingrich in the low 30s from the middle of January until a few days before Saturday’s vote. Meanwhile, Mitt Romney remained stuck in the mid-to-upper 20s with Ron Paul pulling 11 to 15 percent of the vote.
In the final two days before the primary, Public Policy Polling and American Research Group both showed Gingrich surging into the upper 30s (40 percent in the case of ARG), while Romney and Paul were holding steady and Rick Santorum was between 13-16 points.
Most polls had Gingrich in the lead over the past week, with the exception of Politico and NBC/Marist. Politico released most of their numbers. They had the ratio of men and women close to the turnout of 50-50, but they were very low on voters under 30—exit polls showed 9 percent, but Politico had them at just 1 percent. The publication also had 65-and-over voters at 37 percent of survey respondents, but the exit polls showed turnout of this age group at 27 percent.
According to the exit polls, Gingrich did better among the 65-and-over crowd than the under-29 voters, and he beat Romney among both. NBC/Marist does not release their demographic data so it’s hard to tell where they were off.
In the end, Gingrich won with 40 percent of the vote followed by Romney with 28 percent, Santorum with 17 percent and Paul with 13 percent. ARG and PPP deserve full credit for picking up the large distance Gingrich put between himself and Romney in the closing stages.
More than 50 percent of voters said they made their decision in the final days of the race. That means the last two polls from ARG and PPP have a distinct advantage over the others. YouGov/Polimetrix also had a poll released in the last days before the election, but with a sample of just 339 respondents. They had Gingrich winning, but by only a few points.
Both PPP and ARG were very close on both Gingrich’s winning margin—ARG actually hit it right on the nose—and Romney’s numbers, which PPP nailed. But only PPP made the right call on Santorum in third with Paul a close fourth.
According to the exit polls, Santorum did best among very conservative voters—36 percent of electorate—and those who feel “it is very important that a candidate shares your religious values.”
As most pollsters won’t release their full data, it’s tough to say where they missed Santorum’s numbers. My guess would be they had under sampled these very conservative, religious voters. PPP, which does release its data, had very conservative voters at 41 percent, most likely the key to getting Santorum’s finish correct.
As for correct projections, on January 18th CNN released a poll with Romney up by 24 points. It will be interesting to see where the polling has his numbers as Florida’s contest nears.
Stefan Hankin is founder and president of Lincoln Park Strategies, a Washington D.C.-based public opinion firm. Follow him on Twitter at @LPStrategies