2020 Election Polling: The Science & Psychology of Polls in a Partisan Age

Show Notes, Transcript & Electronic Sources

September 29, 2020

During this 2020 election season, not a day goes by without another new Presidential poll result blaring from the cable networks and rebounding across the Web. But how accurate are these polls? What has been learned from the 2016 polling fiasco where many polls predicted a large Clinton victory? And how much more difficult is election polling in our highly partisan age?

The Purple Principle asks these questions of veteran pollster and Director of the University of New Hampshire Survey Center, Dr. Andrew Smith. In Episode 10, Dr. Smith details lessons learned from 2016 currently in force for 2020 election polling. But he also describes some  ongoing challenges faced by pollsters, such as the “spiral of silence,” when large numbers of respondents fail to reveal true preferences, and the difficulty  reaching voters through mobile phones with call blocking.  

What about aggregating the many dozens or hundreds of polls for a more accurate average result? Dr. Smith explains why that only muddies the waters. He also analyzes the factors contributing to our highly polarized polity while doubting some of the proposed dysfunctional democracy, such as campaign finance reform. 

Instead, Andrew Smith issues a provocative challenge to independent-minded American voters this 2020 election. To hear that challenge, and learn about the rigors of election polling in a partisan place and time, tune into Episode 10  of The Purple Principle with Dr. Andrew Smith, Associate Professor of Political Science and Director of the University of New Hampshire Survey Center.  

Original music composed and created by Ryan Adair Rooney.

Electronic Sources

Andrew Smith

UNH Survey Center

William Frey (5/18/17). “Census shows pervasive decline in 2016 minority voter turnout.” Brookings. 

Jonathan Martin (9/4/16). “Young Blacks Voice Skepticism on Hillary Clinton, Worrying Democrats.” The New York Times. 

Stanley B. Greenberg (9/21/17). “How She Lost.” The American Prospect. 

“Who Will Win North Carolina” FiveThirtyEight 2016 Election Forecast. 

Thomas Peterson (9/6/16). “Spiral of Silence.” Encyclopedia Britannica. 

Courtney Kennedy (8/5/20). “Key things to know about election polling in the United States.” Pew Research Center. 

“An evaluation of 2016 Election polls in the U.S.” AAPOR. 

Courtney Kennedy et. al. (Spring 2018). “An Evaluation of the 2016 Election Polls in the United States”. Public Opinion Quarterly 82:1. 

Kyle Plantz (2/17/17) “UNH Pollster Makes Adjustments to Surveys in Age of Trump.” NH Journal. 

Liz Farmer (12/21/18). “Who’s Your Governor? 1 in 3 Americans Don’t Know.” Governing Magazine. 

“Who Will Win New Hampshire?” FiveThirtyEight

“New Hampshire Results: 2016 Election” (8/1/17) The New York Times.

Nick Hatley & Courtney Kennedy (8/18/20). “A Resource for State Preelection Polling.” Pew Research Center. 

“Weighting.” American Association for Public Opinion Research. 

“Election Polling.” Pew Research Center. 

Richard H. Pildes (11/18/19). “Small-Donor-Based Campaign-Finance Reform and Political Polarization.” Yale Law Journal. 
Lauren Gambin (3/10/19). “‘Not the billionaires’: Why small-dollar donors are Democrats’ new powerhouse.” The Guardian.

Transcript

Andrew Smith:

The presumption, unfortunately, for the poll aggregators is that if you take really good quality data and bad quality data, somehow that will average and you’ll have better stuff coming out of it. 

Robert Pease (host):

That’s Andrew Smith, Director of the University of New Hampshire Survey Center1, a veteran pollster of the all-important New Hampshire presidential primary as well as general elections. 

Andrew Smith:

And I say, well, if I have a bottle of spring water and a bottle of water that I scooped up out of a mud puddle on the street outside and I poured them together, would you want to drink it? 

Robert Pease (host): 

What polls to drink and not drink this 2020 election season on this episode. I’m Robert Pease, host of the Purple Principle. 

Emily Crocetti (reporter): 

I’m Emily Crocetti, staff reporter. And there’s an awful lot of polls out there these days.

[Archival audio collage]

Emily Crocetti (reporter):  

Plus the memory of some very misleading polling back in 2016.

[Archival audio collage]

Robert Pease (host):    

My interview with Dr. Smith begins on the touchy subject of the 2016 election polling.

Andrew Smith:

I think everybody was surprised that Trump won that election in 2016. Both our polling in the state was showing different things. Although, you know, if you go back historically – and it’s always a matter of going back and looking at things, the clues that you missed – it’s in things you learn afterwards, that the Clinton campaign was anticipating that African American turnout would be the same as it was for Obama in 2008 and 2012.2 Why would you think that you are going to get the same coalition of voters that Obama did? You know, those sorts of things, when you hear about it afterwards, you’re just stunned that that was the logic. The Clinton campaign stopped polling two to three weeks before the election was over.3 They relied on big data going in and the big data said they’re going to win in a walk. For example, the polling in North Carolina, there were four polls that made up the averages in the week before the election. Three of them were robo polls. One of them was a web-based poll.4 So there’s bad data in, bad data out.

Robert Pease (host):  

That’s great info, on polling methods. But we do also live in a very polarized media environment. And there’s a lot of misleading, if not downright false information out there. Does that make it more difficult to poll accurately?

Andrew Smith:

I think there are a number of other reasons that make it much more difficult to poll people. First, the hardest thing to do is get a hold of them with cell phones and caller identification and call blocking and a number of things like that. It just makes it hard to find people. That’s a bigger issue. But the psychology that we have to understand about voters has always been there. When people talk to a survey researcher who doesn’t know them, they still want to make themselves look good in the eyes of the person who’s asking the questions. So there’s a great social desirability to look good, and the political scientists call it the spiral of silence,5 that if you favor a candidate that you don’t think is popular, you’re not only less likely to put a sign in your yard or have a bumper sticker on your car, you’re less likely to say that you’re going to vote for that candidate. That’s the hardest problem.

Robert Pease (host):  

And was that a factor in the failure of Presidential polling in 2016?

Andrew Smith:

Well, I should first say that the national polls were very accurate in 2016.6 They were pretty much right on predicting the size of Clinton’s victory. It was statewide polls that were much less accurate. And what we saw in New Hampshire specifically to that is there was a segment of the population that was not willing to talk with us. And these were people that were Trump voters. These were typically men who had some college education. They weren’t college graduates, they weren’t high school graduates, they were kind of like blue collar, small businessmen. And there’s quite a bit of research around the country that if you didn’t weight your samples by level of education, you missed that Trump vote. But the interesting thing was, we went back and looked at elections going back to 2000 and including an education weight, and it didn’t make any difference in any of those elections. It was unique to the Trump election.

Emily Crocetti (reporter):   

Interesting point about the spiral of silence and voters afraid to say who they’ll vote for. Seems like that must be  worse during polarized elections, like 2016 and possibly 2020.

Robert Pease (host):  

Exactly. You’re too young to remember, but if we look back to elections like Bush (as in the father) vs. Dukakis in 1988, or Clinton (the husband) vs. Bush or Bob Dole, the two sides were just not very far apart. 

Emily Crocetti (reporter):    

And it reminds me of what Dr. Jay Van Baavel said about how difficult and painful it is for people to change political identity or voting behavior.

Robert Pease (host): 

Even with a lot of evidence to do so. Here’s neural scientist Jay Van Bavel on why change is so hard.

Jay Van Bavel (previously recorded audio):

And for most people, the notion of letting go of a belief system or a party identity that they’ve held really closely to who they are is horrifying, because if you’ve been a party member of the Democrats or Republicans for 10 or 20 years, the notion that you’re just going to abandon that after – you know, you have friends who are members of that party, you’ve posted signs on your lawn or stickers on your car – for you to completely abandon that is deeply threatening to a lot of people. So there’s lots of incentives that people have psychologically to just simply ignore contradictory information. It’s actually the easiest thing you can do in that situation.

Emily Crocetti (reporter):   

Which gets back to that spiral of silence in polling.  And Dr. Smith seems to know what went wrong in the 2016 election with that segment of Trump voters. But how are we correcting for this in the 2020 polls? 

Robert Pease (host):  

We’re getting to that. Here’s more on changes in polling from the 2016 to 2020 elections.

Robert Pease (host):  

Let’s just provide a little more explanation for our listeners. So your prediction in 2016 without the educational weighting was for a bigger Clinton margin?

Andrew Smith:

Yeah, we had Clinton up by about seven points in New Hampshire, and she won by about a percentage point. But once we included education weight, what that really means what we’re doing is adjusting the sample of people that we talk to so it reflects what the census data says the level of education is among people. When we adjusted our sample to account for that, that’s when our prediction would have been right on what the actual results were, if we had included an education weight. It just has never been a factor in New Hampshire before. But we certainly have included education weights since 2016.7

Robert Pease (host):  

I see. So your methodology in 2020 will have changed from your original method in 2016. 

Andrew Smith:

That’s one of the major reasons that survey researchers like to use data from elections, because it’s the only time that you really have an opportunity to predict something based on the results you get to a survey. And historically, what survey researchers have done is adjusted their general methods based on the ability of a particular methodology to accurately predict an outcome of an election. So that’s happened. 

Robert Pease (host): 

And has there been a nationwide adjustment, similar to your own here in New Hampshire, based upon the 2016 results?

Andrew Smith:

After 2016, the American Association for Public Opinion Research conducted a fairly exhaustive analysis of what happened with polling in 2016, and they concluded again, the national polls were by and large right, that it was the state polls that were much less accurate. And it was more the models that were being used by the media to predict what was going to happen in the election, which relied on those state polls that were way off and gave a misleading count of what to expect on election night.8

Emily Crocetti (reporter):  

So changes have been made to polling since the 2016 fiasco. 

Robert Pease (host):  

But we have to remember polls are always a snapshot in time and can’t be perfectly accurate. 

Emily Crocetti (reporter):   

So there’s always a margin of error.  

Robert Pease (host):   

Plus, the media can cherry pick the polls they report and how they frame that coverage. 

Emily Crocetti (reporter):    

Exactly. Let’s not forget what Dr. Dominik Stecula told us about the trends in network news over the past few decades.

Dominik Stecula (Previously recorded audio)

What happened is that back in the late seventies, early eighties, if you are following the news, roughly a third of the time you would encounter reference to a politician or a quote from a politician, so one in three stories roughly. Now, fast forward to the mid-nineties, we’re now at a 56% majority of content that is now partisan. Fast forward to 2016, now it’s two thirds, it’s 67% so now essentially it’s not just people who are in the echo chambers who are exposed to hyper partisanship; it’s everybody that even residually follows the news. And you know, that kinda turns off a lot of people, especially the kind of pure independents who are going to get turned off by politics.

Robert Pease (host): 

That is a huge factor, and we’ll discuss more of that in a minute. First though, I asked Dr. Smith for advice on all those calls that seem like polls, except for one little thing,

Emily Crocetti (reporter):    

The way they trash opposing candidates and positions? 

Robert Pease (host): 

Exactly.

Robert Pease (host): 

Well, like my colleagues here at the Purple Principle, I’m a lifelong independent in New Hampshire myself, a swing state. So  I get a lot of calls, mostly push polls, and sometimes I don’t answer them. What’s your recommendation for our listeners?  

Andrew Smith:

The first thing that I would do is ask the name of the organization that’s conducting the survey, ask the interviewer who calls you, what’s the name of your organization? What’s the name of your company? Ask who is paying for the poll. Most of the push polls, or what you’re calling push polls, are campaigns testing their messages on the strength of their candidate and the weaknesses of their candidate and the strength and the weaknesses of their opponents. That’s perfectly legitimate campaign research, but it’s not what we would think of as media polling. So if they get a call from an organization that says they’re doing research for a newspaper or a television network or something like that, I would answer those surveys. That’s an opportunity to legitimately have your position represented in a poll. 

Robert Pease (host): 

So, there used to be a little more of an American center than there is now. Let’s say if you go back to the 1980’s or 90s. At one time there were, you know, let’s say six or seven moderate New England Republican senators; we’re now down to one, hanging by a thread. There used to be an equal number of conservative Southern Democrats, there might be one and a half of those left. Why do you think that’s happened, and is it reversible in any way?

Andrew Smith:

I think there’s a couple of reasons why it happened. The gerrymandering that’s gone on over the last several decades, although actually, gerrymandering is as old as the Republic; it pre exists the Republic. But the gerrymandering that’s going on has become much more sophisticated and both parties use it to make sure that the great majority of their candidates really don’t face any general election competition. That means the competition comes in primary competitions where the activists are, which causes the candidates to run further to the left and to the further to the right. The media have exacerbated this because they figured out the ESPN model that arguments sell advertising. So if you watch MSNBC, Fox, CNN – it doesn’t matter which one of those shows – the goal is to get an argument between people, because people want to watch through the commercial break to see the next argument. And that helps the parties raise money, because you can go to folks and say, “did you hear what so-and-so said on TV today? Can we stop them? Send your 5,000 or a thousand dollars in today?” And now with the internet, we can keep tapping those people. So it’s been a boon to political parties in raising money by having a polarized electorate.9 So to a certain extent, it’s working for everybody except the American people.

Robert Pease (host): 

Well then, what can we do about that? What about some of the proposed reforms to make the presidential election a little more rational or possibly not as lengthy and expensive? 

Andrew Smith:

I don’t think campaign finance reform is an issue. I think that’s kind of a moot point, because candidates are now able to raise so much money independently from small donations on the internet. They can get more than enough money. The length of time that the campaigns go on is a bit much, but I don’t know how you’re going to stop that. I’d go a different direction. I’d say that in this case, politics is too important to be left to the people. I think that we may have been able to choose better candidates for president when they were controlled by the smoke-filled rooms, by the political party bosses who understood who was the most electable candidate, who had the support among various institutions across their party and other parties as well and could make a better candidate. The amount of the explosion in money and the protractedness of our presidential campaigns occurred after we went to an all primary and all open caucus system. And I think it’d be difficult to argue that the type of candidates that have come out of that system are especially better than the candidates that were put forward by parties during the previous 50-75 years.

[Archival audio collage] 

Emily Crocetti (reporter):     

So even if polling has improved – And the jury’s still out on that in 2020 – Dr. Smith clearly says our democracy is just not getting the best and brightest of candidates. 

Robert Pease (host): 

And our earlier guest on an earlier episode, former Congressman, Jason Altmire, explained why this is so often the case in Congressional elections.

Jason Altmire (Previously recorded audio):

It’s all about the way we handle elections in this country. We have a system that is designed to elect and protect people on the political extreme, on the fringe. And that is because of what happens in our primary process. So you are seeing great disgust in the country with the polarization that we see all around us. Some people have chosen to disengage from the political process and just not vote and not participate. That is clearly not the right answer.

Emily Crocetti (reporter):   

But does Dr. Smith have any prescriptions for our democracy, other than going back to smoke filled rooms?  

Robert Pease (host):   

No easy ones. But a more educated electorate sure would help. So at the risk of mangling Shakespeare, we could say the fault, dear voters…

Emily Crocetti (reporter):    

Lies not in our stars?

Robert Pease (host):   

In our stats, actually; lies not in our stats.

Emily Crocetti (reporter):  

But in ourselves…

Andrew Smith:

I think it is kind of a vicious circle that we’re in right now, but you know, the voters still control their destiny. It’s not like that we are just pawns in this game and we have to do whatever the political parties want to do. To me, again, as a political scientist, I am just stunned by the civic ignorance of our country, the lack of awareness of history, the lack of understanding of the institutions of the United States, their lack of interest in politics. We have the ability now to read newspapers from all over the world, follow politics all of the time, but fewer people actually follow politics now than did 20, 30, 50 years ago. We only get about 20% who are willing to say that they could name their state Senator. Only about 40% in the state can name their Congressman.10 

I think one of the biggest problems we have in politics is that we presume that the public really pays a lot of attention and cares as much about politics. And certainly people in the press do. And I do because that’s the kind of people I’m interacting with. But you know, in the U.S. we hire our politicians and we send them off to conquer Washington and then we go away until the next election comes around. 

Robert Pease (host):    

That was our featured guest today, Dr. Andrew Smith, Director of the Survey Center at the University of New Hampshire, challenging you, the independent-minded American voter, to get informed this election.

Emily Crocetti (reporter):    

We’re trying to assist in that effort, but you may also want to check out “Democracy Works” from Penn State.

Robert Pease (host):     

And “Democracy Matters” from James Madison University.

Emily Crocetti (reporter):     

And the archive of Civics 101 podcasts by New Hampshire Public Radio.  

Robert Pease (host):  

As well as important fact based media out there in print form.

Emily Crocetti (reporter):   

Such as The Economist magazine, reporting on economics and politics for, oh, just about 180 years or so.

Robert Pease (host):   

And from our nation’s capital, the non-partisan Cook Political Report. It might be cliché by now but a wise if cranky Englishman is believed to have said…

Emily Crocetti (reporter):    

Between puffs on a disgusting cigar…

Robert Pease (host):    

Democracy is the worst form of government…

Emily Crocetti (reporter):    

Except for all the others. 

Robert Pease (host):    

Here at the Purple Principle we have some questions about our imperfect democracy leading up to the 2020 elections and beyond. How did we get so partisan? How could we get less partisan? And can independent-minded Americans play a role in bridging the divide? Tune in next time as we ask comedian Shane Mauss whether partisanship has dulled America’s sense of humor.

Shane Mauss

I don’t think partisanship has ruined comedy. I think that cell phones have ruined comedy. 

Robert Pease (host):    

This is Robert Pease for the Purple Principle team: Emily Crocetti, Staff Reporter; Kevin A. Kline, Audio Engineer; Janice Murphy, Marketing & Outreach; Emily Holloway, research and fact checking. All music on today’s episode was composed and created by Ryan Adair Rooney. More info and connect with us via social media and on our website at purpleprinciple.com.

1994
2017

In 1986, 34 Democratic and 29 Republican Senators signed the Immigration Reform and Control Act, creating a citizenship path for illegal immigrants while making it a Federal crime to employ illegal workers. The House also voted in favor, 238 to 173. The bill was signed into law on November 6, 1986. 

congress.gov

In 1972, 17 Republican Senators joined 35 Democratic Senators to override the Nixon Administration veto of the Clean Water Act, a landmark piece of environmental legislation. The House override vote was 247 to 23, including 96 Republicans and 151 Democrats in the majority.

govtrack.us

  • Question for Team Purple?
  • Want to offer Feedback on our Podcast?

GET IN TOUCH