While most pollsters undershot Mitt Romney’s support, their numbers were close enough.
In the end, Mitt Romney didn’t have to sweat too long into the night. The numbers ahead of Tuesday’s vote appeared troubling for his campaign. While Arizona was considered a foregone conclusion, polling showed a dead heat between Romney and Rick Santorum in Michigan.
Most polling, admittedly, undershot Romney’s support, but the numbers were not far off. Pollsters did well predicting the four candidates’ finishes, avoiding many of the problems we’ve seen in past states with, say, Ron Paul.
Over the last couple weeks, public polling showed a volatile race. This was after Romney appeared to be on cruise control with a comfortable lead. That eroded, though, at the beginning of February when Santorum enjoyed a tailwind of support after his caucus sweep. He was then leading Romney until about a week ago when the former Massachusetts governor caught and then passed Santorum.
The weekend before Tuesday’s vote, every public survey had the race within the margin of error but, with the exception of Public Policy Polling and American Research Group all showed Romney with a slight lead. Ultimately, Romney won 41 percent of the vote, with Santorum at 38 percent -- give or take a few decimals -- Ron Paul at 12 percent, and Newt Gingrich at 7 percent.
None of the public polling had Romney over 40 percent in Michigan. Baydoun/Foster and Rasmussen both had Romney at 38 percent and PPP had Santorum at 38 percent in their last poll, which had the former Senator ahead. Also noteworthy was that half of the polls over the last three days had the number of undecided voters in the 9 to 11 percent range, while the other half had the number of undecided in low single digits or at zero in the case of We Ask America’s survey.
Given the close nature of the contest it stands to reason that the higher number of undecided voters was correct and they broke in the last 24 hours toward Romney.
Arizona was a bit less interesting and, as a result, had considerably less public polling. Still, the surveys again showed a fairly consistent pattern. Romney was enjoying a comfortable lead through early February when Santorum made a run and moved the race to within three points. Around Feb. 20 Romney started to pull away again.
In the last few polls, Romney was enjoying the support of 39-42 percent of respondents while Santorum was in the upper 20s. With the votes tallied, Romney received 47 percent of the vote followed by Santorum at 26 percent, Gingrich with 16 percent and Paul at 8 percent.
As was the case in Michigan, polling over the last few days before the election was slightly low on Romney’s level of support but accurate on Santorum’s. Just about every poll had Paul higher than he ended up, which has been the opposite of what has happened in most states when it comes to predicting the Texas congressman’s vote tally.
The one outlier was ARG which had Romney up but only leading 39 percent to Santorum’s 35 percent. Without exit polls and no crosstabs to look at it is impossible to see where exactly they were off, but they also showed Gingrich and Paul virtually tied, which also turned out not to be the case.
With Super Tuesday a week away watch for whether Romney’s numbers move or if Santorum’s close second place in Michigan keeps Republican voters holding steady.
Stefan Hankin is founder and president of Lincoln Park Strategies, a Washington D.C.-based public opinion firm. Follow him on Twitter at @LPStrategies