How Republicans got polling right in 2016

Many pundits were surprised by Donald Trump’s win, prompting the question, ‘How did the polls get this so wrong?’ Indeed, particularly at the state level, a number of public polls missed the final results by a wide margin.

{mosads}But polls conducted for the National Republican Congressional Committee (NRCC) were among the most accurate in the country. Our final polls showed Republicans winning in 30-plus competitive House races by an average of eight points. Republicans ended up winning those districts by nine points on average.

So how did we get it right when so many other highly respected pollsters got it wrong this year?

In 2012, the last presidential election cycle, the NRCC’s polls missed the mark in several key races, particularly those in presidential swing states. Coming into this year, we knew it was imperative to learn from our past mistakes to ensure that our polling was accurate, in order to best serve our incumbents and candidates in close races. Under the guidance of our executive director, Rob Simms, we spent countless hours with Republican pollsters to determine what we got wrong and how to correct it.

We identified three key areas that would be vital to ensuring our polling would reflect the actual makeup of this year’s electorate. First, we realized our likely voter screens were too tight, excluding many low propensity and newly registered voters. This year, we removed that screen and included all voters in our surveys.

Next, due to changing technology, more and more people do not use a landline phone, instead relying solely on a mobile device. To ensure we did not exclude these potential voters who skew younger, we doubled the cell phone percentage of our samples.

Finally, working with the top Republican polling firms in the country, we took a deep dive into district-specific data to determine what the makeup of the electorate would be in individual districts. Using both historical and consumer data to structure our samples, we were able to come up with a set of assumptions that accurately projected the demographics of the 2016 electorate.

This is where Democrats and many media-sponsored polls got it wrong. Democrats were shocked on election night because they based their polling assumptions off of the 2012 electorate without looking closer at what was actually happening this year. They believed their base would turn out at the same record rate it did for President Obama, while Republican turnout would be depressed.

But this was not the case. Even among Democrats, Hillary Clinton was unable to approach the level of enthusiasm that President Obama generated. And her bruising primary fight left lingering resentment that caused many Democrats, particularly younger voters, to either stay home or vote for a third party candidate. At the same time, far from being depressed, Republican turnout, particularly in rural areas and smaller cities in the Midwest, surged.

These faulty assumptions created a false narrative that a Hillary Clinton victory was all but assured and that Democrats would win control of the Senate, while making big gains in the House. Just weeks before the election, House Democrats told the media that they were within striking distance of winning the 30 seats necessary to retake control of the House and install Nancy Pelosi (D-Calif.) as Speaker. With few public polls being conducted for House races, Democrats were able to show their wildly inaccurate polls to reporters to generate numerous stories to that effect.

But House Democrats were not content with the rosy polls conducted by reputable Democrat pollsters. Realizing that there were few outside polls to push back against them, the Democratic Congressional Campaign Committee (DCCC), and House Majority PAC, a Democratic Super PAC focused on House races, began publishing methodologically dubious polls as part of a communications strategy to create a narrative that safe Republican districts were in play.

These polls were nothing more than junk science that no reputable pollster would ever publish, relying on bizarrely optimistic turnout assumptions, conducted in a single night using mixed-mode online panels with live or automated calls, sometimes with no cell phone sample.

In Minnesota’s 3rd Congressional District, DCCC released an internal poll showing their candidate winning the race by 2 points. She lost by nearly 14 points. In New York’s 19th Congressional District, a DCCC poll showed the Democrat winning by 5 points. She lost by 9 points. House Majority PAC got in on the action, publishing a poll showing their candidate in Indiana’s deep-red 9th District losing by just 2 points. She went on to lose by 14 points. Democrats unethically used these junk polls to influence the media and create a narrative more favorable to their candidates.

So what can be learned from polling in 2016? First, not all polls are created equal. Before trusting a poll, we must examine the methodology and assumptions behind it.

Second, while past elections are an indicator of what to expect in the future, they must not be our only guide. Each election is unique, with different issues motivating voters, and different candidates appealing to different segments of the population. To understand who is going to turn out on Election Day, we first have to understand how voters feel and what they want.

Finally, when political organizations release their internal polls, the public should approach them with increased skepticism. Reporters should not allow obviously flawed polls, like those we saw this year from House Democrats, to color their coverage of elections.​

Lauren Hutchinson is the Polling Director of the National Republican Congressional Committee. She is a Michigan native and a graduate of the American University.

The views expressed by contributors are their own and not the views of The Hill.