June’s CNN/SSRS national poll was one of the first to report that the summer political grounds were shifting, with Democratic presumptive candidate Joe Biden opening a 14-point lead over President Donald Trump among registered voters. Other polling organizations were soon to follow, also reporting double-digit or near double-digit margins.
Of course, events play a role, and polls are snapshots in time. These recent polls followed protests that erupted after the death of George Floyd at the hands of Minneapolis police officers. Covid 19 cases were also spiking up again, with the economy continuing to suffer. In addition to the President’s slipping horse race numbers, the CNN/SSRS poll also found that the President’s approval rating, at 38%, had fallen to its lowest point since January 2019.
Political pendulums can swing back forth quickly and so can polls. But one issue that has surfaced in the news media after some state polling failures in 2016 is the claim of the “shy Trump voter.” This is the assertion by some pundits and politicians that polls undercount Trump’s support because “shy” Trump voters mislead pollsters about their vote intent or feel intimidated to admit that they are supporting Trump.
For example, after the spate of earlier polls found that the President’s support was fading, veteran GOP pollster Neil Newhouse opined, “I’m still convinced there is a shy Trump supporter, a hidden Trump vote. I’m convinced that number is at least 2 to 3 points.” Other pundits pointed to anecdotal evidence that a few people said they lied to reporters. Therefore, if true, Trump’s support is broader than the polls report.
Little Evidence of “Shy Trump Voters”
So SSRS researchers took a deep dive into this serious accusation, reviewing research that followed the 2016 election. The findings: the “shy voter” hypothesis could not be substantiated. Here are a few of the findings:
A thorough review by the American Association for Public Opinion Research (AAPOR) concluded that late-deciding voters for Trump, not shy voters, were a major factor in Trump’s victory and the failure of some pre-election state polls to capture the late movement. These Trump late-deciders, particularly in key battleground states, greatly outnumbered the late-deciding Clinton backers. The AAPOR analysis also pinpointed poorly conducted polling in some battleground states, which undercounted lower educated males.
In fact, in what appeared to many to be a polling debacle, the 2016 national polls were pretty much on target with Clinton’s almost 3 million edge (2.1 percent) in the popular vote. The average Clinton support in the national polls was 3.2 percentage points, well within the error margin.
Many Late Deciders Swing to Trump
Why were there so many late-deciding voters in 2016, particularly in four critical battleground states? The AAPOR report notes that many voters disliked both candidates, so many voters apparently waited until the end to decide. Then, there were the late breaking issues near election day as pre-election polling was underway – Wikileaks’ release of hacked Hillary emails, Russian intervention on social media, and the FBI’s unprecedented letter to Congress announcing the reopening of the investigation into Hillary’s email server.
The AAPOR researchers further compared results from pre-election live-interviewer phone polls to self-administered (IVR, web) interview modes. They hypothesized that estimates of Trump supporters would be lower and non-disclosures higher in the live interviewer mode compared to self-administered modes if the “shy voter” effect was confirmed. They compiled a data set of 208 battleground and 39 national polls conducted by both live interviewers and self-administered during the final 13 days of the campaign. They concluded that “the results are inconsistent with expectations of the Shy Trump theory.”
In another election review, respected Columbia University researcher Andrew Gelman points out that Trump outperformed the polls the most in deep red states such as North Dakota and West Virginia where we assume respondents would have had little embarrassment in declaring their support for him. But he did no better than the polls’ predictions in solidly Democratic states, where the purported shy voters might reside.
Additionally, even if you buy into the unsubstantiated shy voter theory and subtract the 2 or 3 points that Newhouse offered, that still leaves Biden with a double-digit or near- double-digit lead in the national polls.
Campaigns Would Pay a Price for “Shy Voters”
From a tactical viewpoint, a Trump voter who misleads pollsters on the horserace question is actually causing great pain to the Trump campaign. Guided by the President’s falling poll numbers in traditionally red, but now battleground states, such as Texas and Ohio, the campaign has to allocate vital dollars into protecting its flanks in those states.
Moreover, big money donors, reading the betting odds, may well place their political and super PAC bucks into Biden’s campaign coffers. And, fearing a wipeout, Trump’s party will need to buttress down-ticket candidates with money and support diverted from the national ticket.
Caution: Summer Polls Often Have Little Shelf-Life
Despite Biden’s current edge in the polls and the lack of evidence of the “shy” Trump voter, I caution you, as a longtime election pollster, not to let these summer poll numbers mislead you into thinking that the outcome is preordained. Forget both the “shy Trump voter” thesis and the overconfident politicos on the other side who talk of “tsunamis.” Elections are full of drama, twists and turns. While some gleeful Democrats envision a “tsunami,” we’re only in the fourth or fifth inning of this game, with lots more to come. And “October surprises” often erupt out of nowhere.
With early balloting just a few weeks away, there’s also the turnout wildcard: who’s sitting the election out? A new Knight Foundation report notes that the 100 million eligible Americans who did not cast a vote in 2016 outnumbered the vote totals for either Trump or Clinton. Most national polling so-far has been based on self-reported “registered voters,” not “likely voters.” Covid 19 pandemic health risks facing voters may depress turnout at what may be fewer in-person polling places. Primary elections so far this year have seen a surge in mail balloting. But no-excuse mail balloting is under legal threat in many states.
Of course, in the end, the biggest decider of all is the Electoral College, which each election year further increases the probability that the popular vote will not decide the race. Whatever the national polls find, the outcome in just a few battleground states will determine the victor. Two of the last five Presidential contests were decided not by the popular vote, but by the Electoral College.
Mark Schulman, Ph.D., a veteran of the polling industry, has worked extensively with the news media and served as Time magazine’s pollster for almost ten years. He is a former president of the American Association for Public Opinion Research and has been interviewed by many news media outlets including CNN, Fox News, Vice Media/HBO, Time Magazine and the BBC.
Additional Insights From Mark Schulman
SSRS Launches 2020 Election Polling Coverage
With the 2020 presidential campaign heating up, SSRS’s non-partisan national and key state polls are providing the news media, candidates, political pundits, and researchers with snapshots of Americans’ views on critical issues, where voters stand on these issues, candidate preferences, and approval ratings of elected officials.
Polling Beyond the Horse Race and Trial Heats
The reason politicians and media focus on polls is that polling provides a vital link between voters and candidates. Surveys give a representative sample of citizens a voice to communicate with their elected officials and candidates seeking office. And, of course, polls create news.