What do you think? Leave a respectful comment.

What went wrong with polling in 2020

Political polling is under scrutiny once again. While pre-election surveys set expectations of a Democratic landslide, the presidential contest and many Senate and House races came down to the wire instead. In fact, Republicans gained ground in the House. To understand what went wrong, Judy Woodruff talks to pollsters J. Ann Selzer of Selzer and Company and Chris Kofinis of Park Street Strategies.

Read the Full Transcript

  • Judy Woodruff:

    Political polling is under scrutiny once again.

    While pre-election surveys suggested Democrats were in a strong position to make gains across the board, the presidential race came down to the wire, and Democrats were disappointed by congressional and Senate results.

    To help us understand what happened, we spoke to two veterans of the industry, J. Ann Selzer, Iowa pollster and president of Selzer & Company, and Chris Kofinis. He's a Democratic pollster and the CEO of Park Street Strategies.

    Ann Selzer, Chris Kofinis, thank you both for being with us.

    So, look at the results. Joe Biden won, President Trump lost, but the president did better than a lot of polls suggested he would. He did better in Texas, better in Florida, certainly better in the national polls.

    Ann Selzer, do you have a theory about what went wrong?

  • J. Ann Selzer:

    Well, I have a couple of theories. I will share one.

    And that is that it appeared to me that the Democratic surge, its arc, peaked a little early, around the time the early voting was starting and that their push was to get people requesting absentee ballots and getting those filled out. And the surge on the Republican side happened closer to Election Day, that their big push was to get people to show up that day.

    And it could be that there was just more enthusiasm and excitement, more recruitment of getting people to vote who maybe were going to sit this one out. I don't think it's so much changing minds at the last minute, but I do think the decision on whether to vote may have happened late for Republicans.

  • Judy Woodruff:

    Chris Kofinis, what about this? How do you see how the polls got it wrong in so many cases?

  • Chris Kofinis:

    Well, I don't think there's just one reason.

    I think one of the factors, to be frank about it, was the so-called silent Trump voter. I mean, we had been seeing that in our research for a long time. The difficult part about what this was trying to figure out, what — how big was it? What percentage of the electorate was really a silent Trump voter?

    And — but it was revealed. And I think what was happening was, one, I think some pollsters were probably discounting it. I think there was a lot of noise. We started seeing in some of our polling for our clients that strength amongst Trump was among Hispanics, which was really, I think, something that people didn't expect.

    And I think part of — I think the challenge in this election, compared to maybe '16, but even prior to '16, is, Trump is such a difficult candidate to poll because of the perception around him, which makes, I think, certain voters reluctant to say that they support him.

  • Judy Woodruff:

    Ann Selzer, you did — in fact, your last poll had it close to right in Iowa and elsewhere.

    Do you think there's a special difficulty polling around President Trump?

  • J. Ann Selzer:

    I don't know that I saw anything like a shy Trump voter in my data.

    We asked not only, who are you planning to vote for? And we had Trump ahead by a large margin. But we also asked, who did you vote for in 2016? And we found that, in many, many states. And we don't find that there's a reluctance, an apparent reluctance, to tell us about that.

    I think there's something else perhaps happening, and that has to do with enough representation among rural voters who go strongly for President Trump.

    And, Judy, I think the question about whether this is unique to Trump, the future will tell us. He certainly has been someone that it was far more a cult of personality and understanding why people felt so strongly about him. I think we have yet to still figure that out.

  • Judy Woodruff:

    What about that, Chris Kofinis? Are pollsters and are — and I know there's so many pollsters, it's hard. We can't lump them all together.

    But particularly those who work in the media, work for media organizations, are they looking hard enough at people, whether it's rural areas or people in the middle of the country? How do you see that?

  • Chris Kofinis:

    I think there are some predisposed notions that kind of both frame what pollsters do, as well as how the media analyzes those polls.

    I mean, I think we have to be really honest in the polling profession that there were some real mistakes made in this election and big misses, both at — in the Senate level, as well as the presidential level. I think we have got to step back and be a lot more critical about what polling is there for.

    There's this tendency, I think, to want polling to be kind of a predictor of the future, we're Nostradamus, we're going to tell you exactly what the outcome of the race is. It's art and it's science.

    And I think we have got to do a better job of listening and asking much more critical questions to understand, why are people thinking about voting one way or another?

  • Judy Woodruff:

    Ann Selzer, are we expecting too much of pollsters and the work that they do, that you do?

  • J. Ann Selzer:

    Well, certainly, when the polls lead the public to think of a certain outcome, and then that has to be adjusted, it's a bit like Shakespeare. First thing let's do, let's kill all the pollsters, instead of lawyers.

    And there is sort of a sense. People get very invested in polls. And then it becomes like they almost think it's a public utility, that they are owed correct polling, when it's just far more complicated an industry than that.

    I do think this is a time for reflection and really unpacking, pollster by pollster, what assumptions are being embedded through the methodology, through the way that they are deciding who's a likely voter, for example, how they're managing the telephone numbers that they're calling.

    There's lots of things in there, very technical, but I think some things will be revealed, I hope.

  • Judy Woodruff:

    Chris Kofinis, how much of this is just not being able to reach enough people, frankly, to have a representative sample?

  • Chris Kofinis:

    I mean, part of it.

    To be honest, I think what we have seen is the response rates were actually really good this time, strangely enough, primarily because of the pandemic has more people at home more willing to answer phones.

    I don't think it's the response rate that's the challenge here. I think the challenge here is, are we asking the right questions? And I always get disturbed when I see polls, especially in media polls, that tend to ask what I call very basic, cursory questions, and don't try to dig beneath, what are — why are people thinking one way or another?

    I think the more important thing is, how do we get to the point where I think the media starts asking much more critical questions, and we're not cheerleading polls simply because we like who's losing or who's winning? That is not the job of a pollster. It is not my job to tell the client what they want to hear. It is not the job of, I think, a pollster to tell the media what they want to hear.

    It's trying to understand what is going on in the electorate. In this election, there was a lot more complexity going on that I still don't think we have fairly understood.

  • Judy Woodruff:

    A lot of conversation around polls certainly leading up to elections and always afterwards as well.

    Ann Selzer, Chris Kofinis, we thank you both.

  • Chris Kofinis:

    Thank you.

Listen to this Segment