The polls missed the mark ahead of Tuesday’s primaries in Alabama and Mississippi. Not counting Georgia, which wasn’t competitive, public pollsters have actually gone zero for three in Southern contests since South Carolina voted in January.
The problem is that many surveys undersampled evangelical voters who make up the core of the GOP electorate in Dixie. Was this the case again on Tuesday? Indeed, the results from Alabama and Mississippi showed a similar discrepancy.
Two firms released polling in the days leading up to the primaries and both were badly off the mark. American Research Group (ARG) had Mitt Romney up in Mississippi at 34 percent, followed by Newt Gingrich at 32 percent and Rick Santorum at 22 percent. Public Policy Polling (PPP) had Gingrich in the lead at 33 percent, followed by Romney at 31 percent and Santorum in third at 27 percent.
Santorum won Mississippi with 33 percent of the vote, followed by Gingrich and Romney at 31 and 30 percent respectively.
Looking at the PPP crosstabs (kudos to them for being one of the few firms that release those; ARG only releases a few of their data points), they had evangelical voters making up 70 percent of the electorate. In fact, their numbers were at 80 percent, according to the CNN exit polls. That 10-point difference was substantial.
Moreover, it was also interesting to see PPP have Gingrich winning evangelical voters with 33 percent, followed by Santorum (31 percent) and Romney (28 percent). According to the exit polls, Santorum carried this cohort with 35 percent followed by Gingrich at 32 percent and Romney at 29 percent. The numbers weren’t terribly off on Romney and Gingrich, but the undersampling of evangelicals generally and the 4-point drop for Santorum appears to be the cause of the polling getting it wrong ahead of Tuesday.
In Alabama, the public polls also missed the mark. ARG had Gingrich winning with 34 percent followed by Romney (31 percent) and Santorum (27 percent). PPP had Romney winning (31 percent) with Gingrich (30 percent) and Santorum (29 percent) following. Santorum ended up winning with 35 percent of the vote, with Gingrich and Romney both coming in at 29 percent.
This time PPP was closer to where exit polls had evangelical voters — 75 percent in exit polls versus 68 percent for PPP — but the internal numbers on this question showed some interesting results. Santorum won evangelicals with 35 percent, according to the exits, and his level of support dropped to 31 percent among non-evangelicals. In PPP’s last poll, they also had Santorum winning evangelicals (34 percent), but only had his support among non-evangelicals at 21 percent — a 10-point drop from the exit polls.
Missing the mark on evangelicals in Mississippi and then missing non-evangelicals in Alabama makes for some interesting Wednesday-morning quarterbacking. Still, the Southern states seem to be causing some issues for public pollsters. It’ll be very interesting to see if the pollsters can redeem themselves in Louisiana later this month, but also keep an eye on Missouri given the higher concentration of evangelical voters there.
Stefan Hankin is founder and president of Lincoln Park Strategies, a Washington D.C.-based public opinion firm. Follow him on Twitter at @LPStrategies