Another election cycle, another opportunity to dissect the performance of political polling. On November 11, the Marketing Research Association Mid-Atlantic Chapter hosted a panel discussion digging into what went right and wrong in the 2014 election polling and what we learned about voters and the Democrat and GOP campaigns.
Mark Blumenthal, senior polling editor for Huffington Post, noted that the National Exit Poll (NEP) found overwhelming economic anxiety. The NEP also found about 44 percent approval for President Barack Obama’s job performance, while 55 percent disapproval. According to Blumenthal, presidential approval/disapproval in the NEP correlates extremely well to how you voted.
How did the GOP and Democrats perform at meeting the interests of voters. Polling found that the most important issues in the 2014 elections, measured in final week, were the economy, healthcare and immigration. Respondents said that the issues Republicans were talking about with voters matched those priorities perfectly. Democrats talked mostly about social issues, which ranked lower.
Although overall turnout in the 2014 elections was down overall, noted Blumenthal, but turnout increased in 14 states. The most notable outliers were Virginia and Maryland: both states ended up being competitive, but turnout was down compared to 2010.
Polling error was a major focus of discussion. As Blumenthal noted, every survey has a series of errors which are part of the process and baked into the design. Those survey errors are usually random, small and off-setting. “But not this time.” Likely-voter modeling in particular suffered in the 2014 election. Getting such models right has “always been the hardest part” and “always will be,” said Blumenthal.
Jason Boxt, managing director of the Glover Park Group, remarked that "this was a very predictable election,” so much so that across the last three midterm elections (2006, 2010 and 2014), exit poll data looked similar.
Boxt called out voters for inattention in 2014. "More people paid close attention to the Ebola virus" than to 2014 election, he said. Boxt also lamented that some campaigns didn’t live up to their potential in 2014: "Democrats just ran the wrong races in a number of places."
Hillary Clinton, based on polling data, starts out as a formidable Presidential candidate bringing about 52% of the vote already baked into the electorate, said Boxt. “The GOP will have to take votes away from her.” Asked about polling results showing Clinton behind significantly against “generic Republican,” Boxt joked that he has yet to meet that particular Presidential candidate.
Jim Hobart, vice president at Public Opinion Strategies, looked deeper at voters’ interactions with the 2014 campaigns. Almost 20 percent more likely voters, for instance, were contacted by campaigns in 2014 than in 2006. However, Hobart noted, while voter contact was up, political campaigns still matter: Democrats may have done great voter contact work, but their messaging was all wrong.
While the 2008 and 2012 Obama campaign technology edge has been oft discussed, and indications were that the Republicans have made vast improvements in that area just in the last two years, Blumenthal pointed out that technology is nice, but it doesn’t move results by more than one or two points. Fancy technology works only on the margins for voter turnout. However, Hobart responded, investments in voter turnout brought results. Where Democrats did not dump money into voter turnout – Maryland and Virginia in particular – turnout tanked, and cost the party seats. California may end up similarly for some House races.
Bob Moran, head of Brunswick Insight and president of the MRA Mid-Atlantic Chapter, reiterated Boxt’s contention that the election was eminently predictable. He looked back at the “six-year itch” for midterm Presidential elections and found that the record for losses goes all the way back to Ulysses S. Grant. The only exception to the rule is Bill Clinton.
Pictured, from left to right: Jim Hobart; Jason Boxt; Mark Blumenthal; Bob Moran.