Stony Brook University Professor of Political Science Helmut Norpoth, author of the forthcoming book Commander in Chief: Franklin Roosevelt and the American People, joined SiriusXM host Alex Marlow on Monday’s Breitbart News Daily to discuss his critique of polling methodology. Although most mainstream media polls give Hillary Clinton a commanding lead in the 2016 election, Norpoth has forecast an 87 percent probability that Donald Trump will win.
“The polls rely on reaching people by phone. Most of them do that. It’s very difficult these days, as everybody knows, because people don’t answer their phones any more – landlines, cell phones, et cetera,” Norpoth explained. “You reach a very small fraction of the people that you try to reach. The numbers aren’t quite so well-known, but I think it’s pretty low. It’s probably below ten percent. So that’s one issue.”
“The other one is, I mean, in the end, it doesn’t matter what people tell you in a poll. They have to go out and vote,” he continued. “We know that just about over half of the people who are eligible to vote actually do vote, so the problem is, how do you figure out exactly who’s going to turn out and who is not?”
“These are pretty big issues that have bedeviled some of the polls in the past. Gallup, for example, in the last election, 2012, had Mitt Romney winning with a final poll, and, of course, that wasn’t the case – and Gallup is no longer in the horse race business. The gold standard had to quit the business. I think that should tell you something,” he said.
Marlow asked about the Investor’s Business Daily poll, regarded as one of the most accurate surveys in the 2012 election, which currently shows the 2016 race as a statistical tie nationwide.
Norpoth said he had no details about this poll’s methodology, surmising that “they would have the same problems as all the others.”
Where Do Clinton and Trump Stand?
“It’s interesting that you have quite a spread right now. I mean, you have that poll, and then you have I think the ABC poll had Clinton up by 12,” he observed. “You can see that there’s quite a range, and I’m not enough of an expert on the details of these things, because I don’t know them, to make a judgment about maybe who is right or wrong. I’m just saying it’s very uncertain.”
Norpoth noted that polling companies “do a lot of weighting after the fact” to compensate for the low response rate for phone surveys.
“They get what they get, and then they check against the Census distribution,” he said, citing the example of the L.A. Times tracking poll, which does not use the same weighting assumptions as the other surveys.
“The L.A. Times poll has usually been a poll that’s showing Trump doing quite well, being ahead or at least tied, when the other polls are showing him way behind. Somebody did a re-weighting of the polls based on some of the others and found that if they used the same weights as the others, the polls would come out very similarly. So a lot of it depends on how the weighting works, and that’s a big problem,” he contended.
Norpoth and Marlow also discussed the famed example of 1948, in which polling was halted a few weeks before the election, causing news organizations to miss how “the race tipped” in the final days, as Norpoth put it. The resulting “Dewey Defeats Truman!” headline has become the iconic example of polling malpractice.
“I would never rule that out, that you have some changes, and especially with a candidate like Trump, who is trying to establish himself, and often sort of shoots himself in the foot, and then he suffers at the polls. The question is, can he recover from that, and I think that’s sort of the big problem for him right now,” Norpoth said of the 2016 race.
“I think we learned a long time ago, when Richard Nixon put out this notion of the ‘silent majority,’ that it’s very risky to sort of judge things by signs of overt protests in those days, and maybe enthusiasm this time,” he said of the “enthusiasm gap” in 2016, which appears to strongly favor Trump.
Norpoth recalled seeing an article about Google searches, which have become “another predictor of the election that people have used,” and based on the number of people searching for Trump or Clinton, “Trump is doing very well.”
“But of course, some of those searches may be because people are scared about him and trying to find out what he’s all about, so I’m a little skeptical of that too,” he added.
“I think we’re really in a new world with some of these tools and techniques that we have to fathom how voters are behaving, so we may be in for a surprise,” he proposed.
Norpoth said he takes the “long historical view that this election is poised to tip the scales toward the Republicans because of the swing of the pendulum, with or without Trump.”
“My feeling always has been, what I saw in the primaries, was that Donald Trump did very well. He beat a large field. According to my metric, about how primaries are shaping up, he did better than Hillary Clinton in the Democratic race, so that gave him a leg up,” he said, leading to his prediction seven months ago that Trump would win the election.
“I still feel that there are these factors at play that may be obscured right now, or trumped – to use a pun – by things that the candidate does himself. If he can find a way to help himself a little bit more, I think he would be able to capitalize on those advantages,” he said.
Norpoth explained that his model does not rely on opinion polling: “It’s real polls. It’s what happens in elections, past elections, general elections, and primary elections. This time, I simply based my prediction on what happened in New Hampshire and South Carolina, and Trump, as we know, won both. Hillary won just one of them. Relative to the strongest opposition, Trump came out ahead in that kind of a metric. That’s what I’ve used in past elections, since 1996, to make predictions, and it has worked in all of those five elections to predict the popular vote winner. That’s my sense and my confidence, that the prediction might be right.”
“If you go back to 1912, when we had presidential elections with a good number of primaries – that’s about a hundred years, quite a few elections – if you compare the candidates based on their performance in primaries, you’ll find that the candidate who was stronger in the primaries wins the general election,” Norpoth elaborated. “That was the case in 1912, with Taft against Wilson, and you find it in many elections since then. It’s that kind of historical irregularity, that I think is in play, that puts Trump ahead.”