Unskewing polls isn’t just for Republicans anymore. After a CNN/ORC poll showed Donald Trump narrowly leading Hillary Clinton nationally, MSNBC adjusted the numbers to look like the 2012 electorate instead, giving the Democrats a 4-point lead.
Trump supporters in particular have a complicated relationship with the polls. When the Republican presidential nominee is up, the numbers are cited triumphantly, not least by the candidate himself. When he’s down, they protest biased polls that are as “rigged” as the election.
Beneath partisan conspiracies about the polls lie real disagreements over what the 2016 electorate will look like. Those differing assumptions heavily sway the results, especially now that there is broad consensus that the race is tightening.
“[T]he CNN folks assumed an electorate that is not impossible for Trump, but it would be a historic shift if it occurred,” MSNBC’s Chuck Todd said in explaining why he re-weighted the poll to reflect the same percentage of non-college educated white voters as the electorate that gave President Obama a second term. Other skeptics noted that the sample was plus-4 Republican.
Throughout the 2012 campaign, many Republicans complained that the polls were oversampling Democrats and wrongly assuming Obama would be able to replicate his 2008 turnout among key demographic groups like millennials and minorities. This wasn’t limited to partisans and conspiracy theorists, however.
Gallup is perhaps the country’s most respected polling firm. Its final pre-election survey showed a dead heat with Mitt Romney edging Obama 49-48 percent. The Obama campaign pushed back against Gallup’s polling, showing especially tight battleground state races near the end of the campaign.
The Obama team’s assumptions about the electorate turned out to be correct and he won re-election. Gallup sat out the 2016 primary polling.
“The reason we were doing this so-called unskewing of the polls, which turned out to be a huge mistake, was that we were convinced the electoral model had changed from 2008 to 2010,” Going Red author Ed Morrissey told the Stream, a conservative Christian publication.
Yet there remain questions about what electoral model applies in 2016. Younger and minority voters don’t like Trump, but they don’t have Obama-like enthusiasm for Clinton. College-educated whites similarly like Trump less than any Republican nominee for whom data is available, but they are also distrustful of Clinton.
Can Clinton recreate the Obama coalition, at least to the degree that she can win a plurality — current polling shows third-party candidates running stronger — against Trump? Or does Trump have an unorthodox path to the White House? NBC News’ Dante Chinni calls this the “Trump hope.”
“What if he can bring out a larger-than-usual group of his non-college educated white base, while at the same time a large group of college-educated whites and minority voters stay home because they are not strongly behind Clinton,” Chinni wrote.
“Doug Schwartz, the poll director, says we don’t have specific data on that, but obviously higher turnout for non-college whites would be helpful to Trump,” Quinnipiac assistant polling director Peter Brown told the Washington Examiner after the release of results showing Clinton ahead in Pennsylvania and North Carolina while Trump led in Ohio and tied her in Florida.
In a close race, modest shifts in turnout can have a big impact. They can also lead to arguments about whether polls are skewed and how best to unskew them.
