They wrong. They’re rigged. They’re partisan. They’re worthless.
Polls took as vicious a drubbing as civility and real news in the latest election. And they’re still drawing withering abuse from Donald Trump and his supporters, who maintain that surveys showing that he is highly unpopular are lies, perfidious lies. Trump, in fact, continues to promote a narrative that we are living in a post-poll world, that polls are not only inaccurate but passé; nobody cares about them.
That, perhaps, is not necessarily the worst strategy, considering that the most recent Gallup Poll found Trump is the first elected president with an initial approval rating under 50 percent, making him the most unpopular new president in the history of the survey. But polls do matter and people do pay attention to them, says UC Berkeley political science Associate Professor Laura Stoker. Moreover, reputable surveys are generally pretty accurate. Even the recent national presidential polls, which most people feel utterly missed the boat, were pretty much on target.
“Unfortunately, national polls mean nothing in a national presidential election”
“Trump lost the popular vote by almost three million votes, which essentially confirmed the national polling,” says Stoker, who, along with co-author, Berkeley graduate Andrew McCall, is polishing a paper on the challenges of securing representative survey samples. “Unfortunately, national polls mean nothing in a national presidential election,” says Stoker. “It’s the state-by-state polls that count, and those are more problematic. There can be slight inaccuracies in state regional polls that are insignificant when considered singly, but collectively they can add up.”
Also, says Stoker, the big data employed by modern polls are predicated in large part on what people did in the past. And during the last two presidential elections, millennials and minorities were greatly motivated by Barack Obama’s candidacy. That enthusiasm, as it turned out, didn’t manifest to quite the same degree for Hillary Clinton.
“Millennials and minorities are less inclined to turn out to vote generally, and the ‘likely voter’ algorithm is very difficult to figure out in any case,” said Stoker. “Those factors combined to produce relatively slight survey inaccuracies, but they were enough to give Trump the election.”
Finally, election polls shouldn’t be confused with general survey polls. During an election, pollsters, as noted, focused on “most likely” voters. The “tracking polls” that result from such data are snapshots, skewed toward particular cohorts, and can be affected by changing technologies, e.g., the internet and non-landline phones.
Polls taken on subjects unrelated to elections, however, usually sample the public at large, not the relatively narrow and unpredictably evolving “most likely voter” segment. Nor are they usually conducted in the short time frame of a tracking poll. They strive to capture a panorama of the American political and social landscape, not a quick shot of a particular vested group. And far more often than not, they reliably reflect citizen sentiment.
“The growth in polling has been steady, it has continued after the election, and I don’t expect that to change.”
But for those of us who don’t want to be confused by the facts, does all that matter? Didn’t the election convince the public that polls are bunk?
Not really, says Henry Brady, the Dean of the Goldman School of Public Policy, and a political scientist whose métiers include political polling, political participation, electoral politics and statistical methodology. “Has Trump somehow undermined public faith in polling? No doubt he’s raised some doubts in some minds,” says Brady. “But I think most people understand that polls are conducted pretty well and that election results are generally reliable, including members of Trump’s own party. They’re telling him not to go down this road. They understand that there’s absolutely no evidence of fraudulent polling.”
In fact, says Stoker, “The growth in polling has been steady, it has continued after the election, and I don’t expect that to change. The media relies on poll results to frame their stories. The public is interested in surveys. There hasn’t been any sign of a widespread loss of faith in the veracity of surveys.”
Nor does Stoker expect Trump’s fulminations to have any real impact on how surveys are conducted and considered.
“Trump has no real bully pulpit except where his core supporters are concerned,” Stoker says. “The people who adamantly support him pay attention to what he says, no matter what he says. So from that context, this ‘distrust’ in polling is very partisan, and it’s from a specific group of people.”
A group of people that includes Trump, of course. And Trump, ultimately, is the wildest card in his own administration’s deck. To a certain degree, his post-election popularity seems correlated to how he behaves on social media. While his 140-character diatribes against enemies real and imagined worked for him during the election, that may not be the case now. Angry 3:00 a.m. tweeting bolstered Trump the candidate; it may undermine Trump the President.
“He’s lashing out at Gallup, at Meryl Streep, at almost everything and everyone,” says Stoker, noting Trump’s antics will likely play a greater role in determining his popularity than public perception of the surveys that track him.
“But the big unknown is his tweeting,” says Stoker. “The media is in a crisis because they’re unsure of how to cover his tirades. I could be wrong, but I don’t think his handlers are going to be able to rein him in.”
The real problem, says Brady, isn’t Trump’s sniping at the pollsters for delivering results he doesn’t like. Larger issues are at stake.
“there are some indications that people may be less likely to respond to polls if their candidate is in a poor news cycle, or is getting a lot of bad publicity”
“The question isn’t whether he’s fundamentally hurting the public’s faith in the veracity of the polls,” Brady says, “it’s whether he’s hurting the fundamental legitimacy of the United States as a democratic society, and how many times people are willing to give him a pass. The things he’s pushing for could have truly disastrous results.”
Brady cites Trump’s proposal for a 20 percent tariff on Mexican imports to cover the costs of building his much-trumpeted border wall as an example of impending calamity.
“What it really means is that consumers will pay far more for their goods, a large number of factories will close in Mexico, and more people, not fewer, will find their way over the border, looking for work,” Brady says. “It will be a lose/lose scenario, except for Trump, of course, who can claim that he fulfilled a campaign promise. At a certain point, I suspect he’ll begin wearing thin even for his core of supporters.” Still, no one should assume the President is down for the count just because he’s down in the polls.
“There’s been a lot of talk recently [among political scientists] about the relative willingness of people to respond to surveys,” Stoker says. “There has to be much more research on this, but there are some indications that people may be less likely to respond to polls if their candidate is in a poor news cycle, or is getting a lot of bad publicity, and more likely to respond if the candidate is doing well. So in any given random sample, the response may reflect the fortunes of a preferred candidate at the time of the survey more than a change of mind [by the respondents].”
Bottom line: Trump may not be doing as bad with the public-at-large as his current poll numbers indicate
In any event, few pollsters are liable to get pink slips over the next four years due to a dearth of public interest.
“I’m sure that no matter how much Trump attacks the polls, they’ll still be relevant,” Stoker says. “People will still follow them.”