Despite the problems created for Hillary Clinton by the recent FBI announcement that it was reviewing a new batch of emails, the Democratic candidate remained in the lead of national and swing-state polls.
But, what if the polls could not measure the real levels of support for a candidate as unusual as her Republican rival, Donald Trump?
That question, fueled in part by Trump's own campaign, is based on the presumed existence of voters who favor the GOP candidate but are ashamed to reveal their preference to pollsters, because they don't want to be linked to the polemical candidate
The issue of a systemic error in all the polls has become more important as Trump narrowed the gap to Clinton and stood less than 3 percentage points behind the former secretary of state in some polls taken just days before Tuesday's election..
Those voters do in fact exist, but they are not enough to push Trump ahead of Clinton in the popular vote, according to a report published last week by Morning Consult and Politico.
The study gave the voters polled the option of answering questions about their preference on the internet or on a telephone interview.
On the telephone poll, Clinton won by a 5 point margin, 52 percent to 47 percent. On the online poll, she won by only 3 points, 51 percent to 48 percent.
That difference is not enough to affect the results of Tuesday's election. In comparison, a similar study by the same companies during the Republican primaries showed a difference of 8 to 9 percentage points between the two types of polls.
Morning Consult said the smaller gap in the more recent polls reflected the fact that primaries tend to draw more adult voters with higher levels of education and income – voters more likely to be concerned about the public images of the candidates.
What's more, voters in primaries are generally asked about which candidate would best represent their party, making it more likely that social desirability plays a bigger role in their replies to the surveys.
There are also other signs that the hidden vote for Trump is limited or even nonexistent. A comparison of the results of online and telephone polls on Clinton and Trump show no significant difference.
“If Trump sympathizers are reluctant to admit they support him, he should be doing better in the online surveys,” Alan Abramowitz, a columnist with Sabato's Crystal Ball, told Univision Noticias. “My opinion is that the hidden Trump vote does not exist.”
No embarrassment about Obama
Pollsters have long studied the distortions that the social desirability of candidates could introduce into telephone surveys.
One case that increased the concerns came in 1982 in California, when Los Angeles Mayor Tom Bradley, an African-American, lost his bid for the governor's mansion to a white candidate even though the polls had given him a significant advantage.
The resulting theory, known as the “Bradley Effect” holds that some voters favored the white candidate, George Deukmejian?? but told pollsters that they favored Bradley or were undecided.
But the Bradley Effect was knocked down in 2008 by Barack Obama's candidacy. Pollsters were concerned that white voters were saying they favored the African-American candidate but would not vote for him in the end, said Courtney Kennedy, director of survey research at the Pew Research Center.
“The fact that the polls were right in 2008 is another reason to believe that this year we are not failing to detect a hidden vote for Trump,” said Kennedy.
Patrick Murray, director of the Polling Institute at Monmouth University, said the claims about a hidden vote in the current presidential campaign has been fueled by Trump's own campaign.
“Pollsters only analyzed that possibility after they talked about it,” Murray told Univision Noticias.
Trump campaign director Kellyanne Conway has been pushing that line for months, apparently to keep his followers from being discouraged as Trump continually wound up behind Clinton in the polls, sometimes by a broad margin.
“It's become socially desirable – especially if you're a college-educated person in the United States of America – to say that you're against Donald Trump,” she told CNN in August. “The hidden Trump vote in this country is a very significant proposition."
Murray said that any doubts about a hidden vote and other questions about polls in this campaign are the result of the great volatility that the surveys have recorded. “People see sudden changes in the polls, and they believe anything is possible,” he said.
But the volatility has nothing to do with voters who are ashamed to say that they support Trump, he added. It's because support for each candidate is so fragile that many voters have changed their preferences based on negative news reports about Clinton or Trump.
“In past elections, it didn't matter if they said they favored (Mitt) Romney or Obama.” he said. “The voters' intentions were more unmovable.