© Josh Sager – October 2016

2016 has been an extremely strange year in politics.

A proto-fascist oompa loompa with a history of groping women managed to foment a civil war within the Republican Party, insult essentially every demographic group outside of white men, and ended up running away with the GOP nomination. On the Democratic side of the ticket, an ancient Jewish Socialist who looks like he uses a hurricane as a hairdryer nearly managed to derail Hillary’s presidential bid, despite the fact that the Democratic establishment, media elites, and moneyed interests were aligned against him (as we are now seeing in the Podesta Leaks).

In line with the bizarre nature of this election season, the last month of the election has turned out to be an absolutely brutal slog through the worst politics most of us have ever seen. Trump was outed as a groper who may have sexually assaulted as many women as Bill Cosby, while Hillary’s internal campaign communications were leaked, proving essentially every attack from Bernie supporters correct (e.g. Hillary secretly supported self-regulation for the banks and TPP, her staffers and DNC staffers worked together against Bernie during the primary, etc.).

Both candidates are disliked by a majority of the nation (Links: Trump and Clinton) and, if either were running against virtually any other opponent, this race would be over. Fortunately, it appears that significantly more people hate Trump than Hillary—especially since his sexual assault allegations have been played 24/7 over CNN and MSNBC over the last two weeks—allowing her to open a fairly significant polling lead nationally.

Unfortunately, several pieces of evidence lead me to worry that Trump’s voter base isn’t gone, merely too ashamed to publicly claim him. If this is the case, it is possible that Trump will significantly outperform the polls on election day, perhaps even enough to win the election.

Trump and the “Bradley Effect”

Political science research has identified situations where social pressures can lead polling to be skewed against controversial candidates—this is called the Bradley Effect, after the first political campaign in which the bias was identified.

Tom Bradley was the first African American candidate who had a chance of being the Governor of California. Despite the fact that Bradley finished the 1982 race with a strong lead in most reputable polls, he lost the vote tally. Researchers found gaps between the polling and the voting numbers for white voters, suggesting that many said that they would vote for Bradley out of a fear that they would be seen as racist if they didn’t, but they felt more comfortable in the privacy of the voting booth.

Trump is being lambasted as a sexual predator in the media and I have no doubt that some of his previous supporters are ashamed to be associated with him. This shame could prevent them from publicly claiming Trump while talking to reporters or pollsters, but would be negligible in the voting booth—nobody would know who they voted for and the underlying issue that first led them to support Trump was unlikely to be affected by his sex life.

In the GOP primary, Trump polled an average of 6% better in anonymous, but scientific, online polls when compared to comparable polls conducted over the phone. A study of this found that that the largest difference was in educated voters, who supported Trump in far higher numbers if they were totally anonymous and not interacting with an interviewer. This is particularly bad news if the polling is, in fact, being skewed, as educated voters are the ones who are least likely to publicly support Trump, thus gaps in this demographic could be very important.

Indirect Indicators

If Trump’s supporters are hiding due to social pressure and not really changing their minds, how would we identify this before the actual vote? One way could be to look at indirect factors that indicate voter preferences

For example, after the release of the Trump tape (“grab them by the pussy”), the overall support for a “generic Republican” actually increased by 2%. This is strange given that the ostensible leader of the Republican Party is Donald Trump. This same poll found that Republican support for generic Republicans was unchanged during this time period, indicating that the positive movement was in independents and Democrats.

One explanation for this lack of harm against the rest of the GOP’s polling after Trump’s scandal could be that voters are willing to admit that they are voting GOP in 2016, just as long as they don’t have to explicitly state that they are voting for Trump. These voters are still conservative-leaning independents or Republican partisans and will vote GOP across the ticket (split-ticket voting is rare) when nobody is looking.

If there is a polling bias impacting Trump’s presidential bid, it will become apparent during exit polling on voting day—that said, this will be too late to fix the situation. Democrats need to avoid complacency and continue all campaign activities as though Trump were tied with Hillary in the polls (as he was before the scandal). There is no cost to this strategy and it very well may prevent a terrible upset in the 2016 race.