In October 2015 the most venerable name in public opinion polling, Gallup, announced that it would no longer conduct horse-race election surveys or predict the outcomes of presidential contests. Gallup had been the gold standard for such polling since the 1930s, but, after calling the wrong winner in 2012, the organization’s leadership decided that capturing a representative voter sample during the final volatile phase of a national election had become all but impossible. Gallup was thus spared the humiliation endured by many pollsters in 2016. Most of its counterparts, however, remain in denial about the limitations of their obsolete methodology and are consequently producing wildly inaccurate 2020 results.
The most obvious symptom of this intransigence is their refusal to consider the possibility that their models should contain some mechanism to account for the “shy Trump voter.” The latest excuse for failing to do so involves a rhetorical device that uses the defining characteristic of these voters to “prove” they don’t exist. Geoffrey Skelley, for example, writes at FiveThirtyEight that a recent Morning Consult poll “found little sign of shy Trump voters.” He reinforces his point by quoting another report from the American Association for Public Opinion Research that also found no evidence that such voters exist. This is a classic “appeal to ignorance” (i.e., the absence of proof is proof of absence).
Shy Trump voters are, by definition, hidden. Many are people who rarely answer calls or texts from numbers they don’t recognize. These voters are very real but quite invisible to pollsters. Some shy Trump voters are willing to participate in some public opinion surveys, yet remain reluctant to level with pollsters. A recent IBD/TIPP poll found the following: “Overall, 20% of registered voters say they’re uncomfortable revealing their preferred candidate.” This isn’t the only survey to reveal such reticence. A recent Cato Institute poll found that the reluctance of conservatives to share their political views has increased from 70 percent to 77 percent since 2017. CloudResearch asked specifically about shy Trump voters:
Such concerns were more often than not expressed by Republicans and Independents, and also by those who said they would vote for Donald Trump.… 11.7% of Republicans say they would not report their true opinions about their preferred presidential candidate on telephone polls.… 10.5% of Independents fell into the “shy voter” category, just a percentage point lower than how Republicans react to phone polls.… 10.1% of Trump supporters said they were likely to be untruthful on phone surveys — double the number of Biden supporters.… The results could have implications in terms of the true accuracy of phone polls.
Indeed they could. Are October 2020 polls that purport to show former Vice President Biden ahead of President Trump significantly different than the mid-October 2016 surveys that indicated Hillary Clinton had all but won the election? For the comparable polls, there is little difference. The NBC/WSJ poll, for example, had Clinton up by 11 points on October 10, 2016. The NBC/WSJ poll showed Biden up by 11 points on October 12, 2020. Clinton was up by double digits in five polls between October 9 and October 16 of 2016. Biden has been up by double digits six times during the same seven-day period this year. At about this point in October 2016 Hillary Clinton’s poll numbers began their fatal decline.
op:
