Conventional Wisdom and Polling Data Challenged in 2016 Election Aftermath

by Shawn Healy, PhD, Civic Learning Scholar

Six days removed from the toxic 2016 Election, we are still left with lingering questions, in desperate search for answers. In the next several posts, we’ll attempt to collectively discover them, and encourage you to send along topics you’d like us to discuss.

One prominent question is how did so-called political experts, myself included, get the presidential race so wrong?

There is no single explanation, of course, but clearly the strength of Donald Trump’s candidacy was underestimated from day one of his campaign. Like President Obama in 2008 and 2016, Trump reshaped the electorate, consolidating the traditional Republican base while also extracting working class whites from the Democratic coalition, perhaps once and for all. In so doing, Trump altered the electoral map, weaving the Solid South with the Rust Belt, thereby neutralizing the bi-coastal strength of the Clinton candidacy.

In spite of Clinton’s well-funded and functioning campaign apparatus, her establishment credentials were anathema in another change election. The late-breaking revelation of further FBI scrutiny of her emails certainly didn’t help, and the all-clear arrived too late as the damage was already done. Clinton underperformed Obama among key demographic groups and turnout from the Democratic coalition was depressed in key cities like Milwaukee, Detroit, and Philadelphia.

A related question centers on perceived polling inaccuracies. True, Clinton underperformed national polls that had her ahead by three-plus points, but her narrow popular vote margin falls within standard polling error. The most glaring misses occurred in the Midwest, where polls showed Clinton with comfortable leads in Minnesota, Wisconsin, Michigan, and Pennsylvania. She lost the latter three and narrowly carried the Land of 10,000 Lakes. They did pick up on Trump’s traction in Iowa and Ohio, and this correlated with his strength across the Rust Belt.

Polls are calibrated with predicted turnout models and some failed to account for depressed Democratic turnout in tandem with robust voting patterns among working class whites in rural areas.

Moreover, polling in the modern era is rife with difficulty on account of a max exodus away from landlines and a reluctance to answer calls from unknown numbers. There is also the possibility that some Trump supporters refused to reveal their preference to a pollster, but later cast ballots in his favor.

Finally, there was an unusually high percentage of undecided voters going into the final stretch of the campaign when the last polls were in the field. These voters broke decisively (3-2) to Trump on Election Day.

It’s far too simple to say that data lost alongside the Democratic (and perhaps Republican) establishment last Tuesday. Polls are mere predictions and were successful in capturing late momentum towards Trump. And the only poll that counts involves actual ballots.