At Republicans headquarters in Washington, top strategists predicted days in advance that the GOP would win the Senate and gain House seats despite polls suggesting many races could go either way.
Iowa’s razor-thin Senate race? Won, Republican National Committee Political Director Chris McNulty forecast, during an interview with the Washington Examiner last Thursday. The Colorado Senate race featuring the state’s first-ever all-mail balloting? In the bank, McNulty asserted. The party’s uphill climb in the North Carolina Senate race? On the cusp of a come-from-behind victory, he said. The Republicans took all three races on their way winning a 52-seat Senate majority that could increase to 54 by the time Alaska is called and a Louisiana runoff is decided in December.
“A lot of these races that common punditry is saying nip and tuck coming down the wire — more than our share are going to break our way,” McNulty said.
How could McNulty confidently project a sweeping Republican advance so far ahead, and just two years after the party’s failure to read data properly cost it the White House and congressional seats? The party points to the $100 million it has spent overhauling the way it campaigns, finally giving it a data and analytics-driven operation that can compete with the Democratic machine built by President Obama.
Just as importantly, McNulty and his staff decided to do what the Democrats did and focus their energies almost entirely on getting out the vote among groups highly likely to be Republican but that don’t usually vote in midterm elections.
Final statistics are still not in, but McNulty predicted that the GOP would maintain its historic midterm turnout advantage because the party didn’t waste resources reaching out to people who were already reliable Republican voters.
Speaking at RNC headquarters in Washington last week, with computer screens full of get-out-the-vote data updating in real time in front of him, McNulty said, “A volunteer was asking a question — was really confused with our program, because they were passing up doors that they knew Republicans were in. Yep, that’s exactly right, because it’s probably Chris McNulty’s wife, or it’s Chris McNulty’s father-in-law, who’s going to go vote.”
After watching what Obama’s re-election campaign accomplished in 2012 — particularly in Ohio — McNulty, a Buckeye State native, took seriously the Democrats’ promise to expand the 2014 electorate. So did the rest of the RNC leadership, including Chairman Reince Priebus, the key force behind the committee’s operations overhaul.
Accordingly, the strategic decision was made to direct virtually all of the committee’s resources toward turning out roughly 1.3 million high-propensity Republicans who rarely vote in some nine Senate battlegrounds. They were identified by the RNC’s new data-analytics voter-scoring program. The approach was applied across 23 states with competitive Senate or gubernatorial races, and in competitive House districts.
That meant direct mail, phone calls and the anchor of the GOP’s revamped ground game: door-knocks by volunteers who now had apps for their smartphones that allowed them to update the party’s central database in real time. McNulty said 71 percent of the data the RNC put into its system this election cycle has come from smartphone walk apps built to the committee’s specifications and used by a volunteers going door-to-door. Last cycle, GOP volunteers relied on paper bubble sheets.
The intense focus on door-to-door canvassing instead of phone banks was taken directly from the Obama campaign playbook, because the president proved how well it worked, McNulty said. The RNC began deploying a volunteer army in important House, Senate and gubernatorial battlegrounds at the beginning of the cycle. On the even of the midterm elections, it numbered 20,000, including 2,000 who were paid party organizers.
Data analytics and beefed-up voter scores, not just on Republican voters but on Democrats and independents, allowed the RNC for the first time to measure progress across the electorate and monitor how successful Democrats were in their attempt to bank early and absentee votes from their low-propensity midterm voters. It was low Democratic turnout a few days before the election, and the projected behavior of the Democrats who were voting, that left McNulty confident of victory in Colorado, Iowa and, cautiously, North Carolina.
McNulty said press accounts of early vote statistics were simplistic and did not reveal what kind of voter was showing up, and who they were voting for.
“The public stories are often too simple, they just say this many Democrats voted, this many Republicans voted, this is your deficit, based on registration. When the fact of the matter is not all those Democrats are voting Democrat. Voter scores tell us that that is the case,” McNulty said.
The RNC hasn’t had this level of data before and hasn’t been able to make credible projections. In 2010, when Republican Senate nominee Ken Buck was hemorrhaging GOP votes to Democratic Sen. Michael Bennet in the final weeks of the Colorado Senate campaign, Republican strategists had no way of knowing it was happening. Bennet, now chairman of the Democratic Senatorial Campaign Committee, was hoping to help Sen. Mark Udall reprise the success he had in 2010. It didn’t work, as GOP Rep. Cory Gardner ousted the Democratic incumbent handily.
RNC Communications Director Sean Spicer likened the committee’s data and field program before 2014 to flying a small plane without any instruments. “We added instruments this cycle, so we can fly at night,” he said. “That’s literally the difference, to be able to fly at night and know exactly where you are at all times. It’s been that dramatic of a shift.”
Party officials said they are confident that their data gave them accurate picture of what the electorate ended up looking like on Tuesday. The RNC provided the Washington Examiner with some examples of what the committee’s data analytics team was looking at that led to the Republicans’ optimism:
• Expanding the GOP electorate. The focus on early voting and people who didn’t vote in 2010 is very important. We are winning with first-time registrants and people who haven’t voted since 2006 in nearly all states. In some states we’re winning in early vote where we shouldn’t be and in others we should be down by much larger margins for the Democrats to feel safe going into Election Day. Based on our predictive analytics we have produced a lot of “new votes” from people who had a certain unlikelihood of voting. Lastly, we’ve looked at the electorate — both with those who have already voted and projecting where the unaffiliated voters are going to land — based on our predictive analytics (voter scores) to find out where we stand. When we weight unaffiliated voters to where we think the actual electorate is, we’re feeling very, very good.
• Arkansas. Our voter scores have Southern Democrats going GOP and we’re up considerably with low propensity and new voters.
• Alaska. Republicans are outperforming their registration numbers by 8.9 percent, Democrats are outperforming theirs by 4.8 percent. Democratic and Republican ballots are even in rural Alaska. This is significant because they have touted their rural ground game and it isn’t showing in the numbers.
• Colorado. We’re leading in the ballot returns and based on where these voters are in our universes we’re very happy with what we’re seeing. Our social pressure messaging is working — a lot of the people who filled out commitment cards (pledge to vote) earlier in the process have mailed in their ballots. We have more of our low propensity, presidential year only voters turning out. 18-25-year-olds breaking for [Rep. Cory] Gardner.
• Iowa. For the first time in modern Iowa early voting history, Republicans passed the Democrats last week in total ballot returns/early vote. This had never, up until that point, happened in the state. Our weekly share last week of non-2010 voters was 26.4 percent, compared to only 18 percent for the Democrats. We are consistently growing this share, while the Democrats resorted last week to turning out a large bloc of 2010 voters to pad their request totals. They have over 21,000 ballots outstanding from this summer, over three times our outstanding ballots.
• Louisiana. Both parties outperforming registration by 6 percentage points and there are Democrats coming our way per our voter scores. Minority early vote turnout is not what they were saying it would be.
• North Carolina. Southern Democrats are voting Republican in our predictive analytics and historically we’re down based on party registration heading into Election Day. This shows that partisan registration isn’t a reliable indicator for North Carolina. After the early voting period in 2012, Republicans trailed in party registration by 422,238 votes (16 percent). The registration makeup of Election Day in 2012 was roughly 38 percent Democrat, 35 percent Republican and 26 percent unaffiliated, and Mitt Romney won North Carolina by100,000 votes. At present, Republicans trail in the pre-Election Day vote by party registration by 180,116 votes (14 percent). Things have come our way over the past couple of days with the numbers based on our predictive analytics closing in our favor.
This article was originally published at 5 a.m. and has been updated.
