Stung by a slew of late-breaking defeats in recent governors races, Democrats believe they’ve come up with a new turnout model that will better predict how those contests will break in the final days.
The Democratic Governors Association on Tuesday convened its polling, analytics and media consultants in downtown Washington to unveil a model designed to give the party better intelligence about where, and whom, to target in the closing days of an election. The idea is to bolster campaigns’ understanding of undecided voters, but also to identify which voters might switch from one party to the other at the ballot box.
It’s the latest step in the party’s effort to revamp their polling operations following now-President Donald Trump’s upset victory last fall. There’s a sense of urgency attached to the project: The 38 governorships up in 2017 and 2018 may represent the party’s best shot at returning to power before the 2020 presidential election.
“This kind of redefines, at some level, undecided voters — and the difference between an undecided and a persuadable voter. Just because someone says they’re undecided in a poll doesn’t mean they’re persuadable,” said DGA Political Director Corey Platt. “And just because someone says they’re with you doesn’t mean they’re not persuadable. And so this is a better tactical tool to help identify who the people are that we need to persuade at the end.”
Republicans now control 33 of the nation’s 50 governorships — the most since 1922 — in large part because GOP underdogs eked out victories over the past four cycles against Democratic favorites in a number of states, despite polling that led Democrats to believe they were headed to victory. Kansas Gov. Sam Brownback, Maryland Gov. Larry Hogan and Maine Gov. Paul LePage are among those who trailed in many of the polls before the 2014 election, but ultimately prevailed.
The harshest blow came the year after, when most observers expected Democrat Jack Conway to win the off-year race in Kentucky. But Republican Matt Bevin defied the polls and defeated Conway easily.
Last year, Democrats similarly fell short in Indiana and Missouri, even though their pre-election surveys suggested each race was winnable. Similarly, exit poll data indicates late-deciding voters broke sharply for Trump over Hillary Clinton in the three northern states that provided Trump with his Electoral College majority: Michigan, Pennsylvania and Wisconsin.
Following last year’s races, Democratic pollster Jefrey Pollock, president of Global Strategy Group in New York, led a project to re-contact voters in Indiana and Missouri — along with two states where Democrats won the governor’s race, Montana and West Virginia — to figure out what happened at the gubernatorial level.
The research, shared with POLITICO, suggests that while each state was different, Republicans won undecided voters in Indiana and Missouri by margins that likely proved decisive in those close races.
In Indiana, the research shows, half of voters who said they were undecided before the election said they voted for now-GOP Gov. Eric Holcomb, compared to only 28 percent for Democrat John Gregg. Moreover, 10 percent of voters supporting Gregg before the election said they ended up voting for Holcomb, while only 6 percent of Holcomb voters backed Gregg.
The late break in Missouri was even more stark. Republican Eric Greitens won 55 percent of undecided and third-party voters, according to the analysis, while Democrat Chris Koster won only 21 percent of these voters.
“Missouri has the greatest polling error of any state in the country,” Pollock said.
While part of that error was systemic — the polls overestimated Koster’s “support across the state, but particularly in places like Springfield,” Pollock said — that was dwarfed by the undecided who broke toward Greitens by wide margins.
But Pollock — who said his analysis factored in poll respondents’ tendency to overstate their support for the actual winner — also noted that the Democratic candidates in Montana and West Virginia held the line among late-deciding voters.
The model the DGA hopes to implement uses other polling, demographic and commercial data to identify the universe of truly persuadable voters — the specifics of which the DGA and Pollock say are proprietary.
“It’s not just the undecideds. It’s also that there’s a bunch of people who may very well switch,” Pollock said. “Seventy or 80 percent of voters aren’t switching. But that 20 percent — who are they?”
For now, the DGA is hopeful that party operatives working on gubernatorial races will embrace this new model, though there’s no guarantee it will gain widespread acceptance.
“This isn’t about the DGA imposing particular standards,” said Elisabeth Pearson, the DGA’s executive director. “Everybody, especially after 2016, is looking for ways that we can improve some things, change some things. … I think it’s going to be great.”
While the new approach helps identify persuadable voters, it doesn’t in and of itself solve Democrats’ most significant problem: From last year’s presidential race back to the 2014 midterms, more late-breaking elections broke toward Republicans.
“I think there’s no question that there’s been a Democratic headwind that we have had to fight against,” said Pollock. “That’s true on the governors’ race level, that’s true in Senate races. That’s why we are where we are. And that’s why national politics do matter.”
But, Pollock added, Trump’s current unpopularity could flip the script for the party if it holds — and make Democrats the beneficiaries of any late movement.
“What I fear, though … is that we would overlearn that example and go into what looks to be right now a tremendous potential year thanks to Donald Trump and the backlash [against him],” Pollock said. “And, all of a sudden, we would have overlearned all the wrong lessons.”
Pollock hopes the new model will help the party solve a problem that has dogged it for the past three year — one that he knows all too well.
“[Fellow Democratic pollster] Fred Yang and I sat in this room [last year] and told a whole bunch of people, ‘That’s what happened in 2015 in Jack Conway’s race,’” Pollock said Tuesday in an interview at DGA headquarters, citing the late movement toward Republicans in the Kentucky race.
“And it felt like bulls—. Even as I was presenting it: It’s like, this all feels like a massive cover-up for the pollsters. To be like, ‘No, no, no, no. We swear all these people moved.’ And yet, that’s actually exactly what happened in these [2016 governors races], and it happens to be what happened in the presidential.”