What do you think? Leave a respectful comment.

Pollsters largely got it wrong in 2016. How are they polling in 2020?

The national polls show Democratic nominee Joe Biden leads President Donald Trump at around the same levels as he did before the pandemic and the mass protests in the wake of George Floyd’s murder. Special Correspondent Jeff Greenfield joins Hari Sreenivasan to dissect the polling numbers and what pollsters have learned since 2016.

Read the Full Transcript

  • Hari Sreenivasan:

    Pennsylvania is one of the so-called battleground states where polls show voters split almost evenly between the two parties and where a tiny margin of victory in the popular vote can mean an electoral college win in the race for president.

    For more on what's developing in the final weeks of the campaign I recently spoke with Special Correspondent Jeff Greenfield who joined us from Santa Barbara, California.

    Jeff, it's the season when people see stories about polls that Biden is up in one poll by eight points and then Biden is dead even in another poll. Are they measuring the same thing, the same electorate? How is it possible?

  • Jeff Greenfield:

    Basically, there are a whole bunch of assumptions that any pollster would make. Do I have the right mix? Do I have enough college educated whites? Do I have enough rural folks so I have enough people of color? The other thing is that they're basically guessing also about who's likely to vote, because if they're not talking to people who are voting or if people talk to them and give them their preference and it turns out they don't vote, then the poll is really no predictive value.

    And shouldn't you think polls are predictive value anyway? So it's really not that much of a mystery. It's just that every pollster has a different set of assumptions about whether they got the sample right. Which is why Biden can be up by five points nationally in one poll and 12 in another.

  • Hari Sreenivasan:

    So is there any kind of a real clear picture?

  • Jeff Greenfield:

    Here's my advice. Take the average. And the average basically tells you that Biden is up nationally, by about seven points.

    I should mention, on March 28, Biden was up by about seven and a half points. So if he had gone into hibernation on March 20th and awaken yesterday, you would have assumed nothing happened. Not a pandemic, not the George Floyd killing, not the economic collapse.

    It's really astonishing how unmoving that national number is. And as we all knew last time, Hari, how that national number doesn't necessarily tell you the significance.

  • Hari Sreenivasan:

    Right. So when we go look at the state by state picture, what do we learn? I mean, we. What are we? What did we learn last time?

  • Jeff Greenfield:

    Last time the state polls were — in Wisconsin and Pennsylvania particularly — were way off. The national poll is kind of right. They had Clinton winning by about two or three points nationally, and she did. And pollsters have been scratching their heads about how they got this so wrong.

    One of the explanations is they did not measure the intensity of support for Trump and the antipathy for Hillary Clinton among rural voters. That kind of let's think about who you're talking to or the assumption about what people will do is partly what made those numbers wrong.

    The other theory is that Trump voters are less likely to respond to pollsters' phone calls or online questions than others. And there is something to that.

  • Hari Sreenivasan:

    How is that possible, that the racial reckoning post-George Floyd, the pandemic, the economic catastrophe that so many people find themselves in that are not participating in the upside of the stock market? How does that not have an effect?

  • Jeff Greenfield:

    We've seen this now, I think, for five years. And Trump was very candid years ago when he said to Lesley Stahl, I attacked the press so then when my focus here, negative stories, they don't believe them.

    And so I think what's happened is nothing has moved the Trump base. But Trump has been unable to expand that base. That's why the numbers are so consistent. He's the first president in recorded memory never to have hit 50% in approval. And the test that we're going to see over the next couple of months, I think, is whether or not Trump can get his base passionate enough so that they turn out an unexpected numbers the way they did four years ago.

    And once you can render the pre-election polls on well, and I need to say one other quick thing about this whole business of predicting, which is that people don't seem to get that when someone like Nate Silver, the guru, says that Biden has a 3-1 chance of winning, that's not a prediction. That's a probability estimate. If you ran this campaign a hundred times, Biden would win 70. These numbers don't tell you with any kind of certainty what they seem to tell you. I'm suggesting that everybody try the decaf, take a deep breath, and we'll know in 7 weeks or so.

  • Hari Sreenivasan:

    Alright, Jeff Greenfield, joining us from Santa Barabra, California. Thanks so much.

  • Jeff Greenfield:

    Good to see you, Hari.

Listen to this Segment

The Latest