'We Were Off In The Senate Race' — Maine Pollster On How Collins' Victory Wasn't Predicted

Nov 4, 2020

Updated 1:22 p.m. Wednesday, Nov. 4: 

Dan Shea, a politics and government professor at Colby College, oversaw one of the major polls that came out prior to this year’s election.

He spoke with Morning Edition host Irwin Gratz to reflect on what we know now that the real polls have closed.

This interview has been edited for clarity. It was conducted before Senate challenger Sara Gideon conceded to Republican incumbent U.S. Sen. Susan Collins.

Gratz: As you looked at the results today, how do you think they matched up with the polling results you saw before the election in Maine?

Shea: Not especially well. There are a couple of places where I think we were closer than in other spots. I think we had the statewide race at presidential level about where it is. I think we pegged the minor party vote share about where it’s going to be. But we were off in the Senate race.

Sen. Susan Collins will likely do much better than we expected in our polls. And I’ll add much better than all of the polls that were conducted in Maine, not just by Colby. OK, misery loves company, but we were all off by a good amount in the state of Maine.

If I could also add, this is a phenomenon that seems to be happening across the country. There are a number of Senate races that are much tighter than folks thought. A lot of polling, again, as we saw in 2016, did not seem to really find the sentiment of voters.

We see that in the presidential race too, where, regardless of what the outcome is like, clearly there was a surge for President Donald Trump that the polling in advance did not catch. And that’s, of course, what happened four years ago as well. Do you have any idea why?

Well, we’ve been thinking about it. We’ve been crunching the numbers like crazy this morning. There are a couple spots, I think. The first is that Collins is doing especially well throughout the rural municipalities in Maine. We knew all along that she would dominate in rural areas of Maine, in the 2nd District and so forth. But she’s not just winning some of these communities by 2 to 1, she’s winning by 3 to 1 and 4 to 1 and 5 to 1. These numbers are overwhelming. So there’s one piece: I’m not sure we captured the preferences of rural Mainers as carefully as we might have. But at the same time, there’s something else going on. It’s almost the opposite. And that is that Sara Gideon is underperforming in a number of the more urban areas throughout the state. She’s just not getting the numbers she needs in some of those areas. And it has to do with the issue of split-ticket voting.

Is that what happened in Lewiston? That’s apparently one of the places that Collins carried last night and it certainly has not historically been known as a Republican stronghold.

Exactly. Auburn, Lewiston, Wells, Sanford, Turner — a bunch of places — what happened was Gideon underperformed Joe Biden by a large number. So what we see, it seems to us, is that split-ticket voting, which many have thought had disappeared from America … So there was a study from the Pew Research Center just a few days ago that said only about 4% of Americans would split their vote. We see it in Maine, and that’s what happened in this race. So in a way, it’s almost, you might say, a celebration of Maine’s independence once again. We’re bucking that tide. There are a lot of split-ticket voters in the state of Maine. We didn’t expect that. We didn’t see that through rounds of polls. And that’s what showed up last night.

*We’re talking about voters who might have supported Biden for president but ended up voting for Collins, or who voted for Jared Golden for Congress in the 2nd District, who’s also ahead, but at the same time for Collins.

Precisely. When we talk about split-ticket voting, one of the great examples is 2008, where Barack Obama wins every county in Maine but one, and then Collins wins every county. In my classes, that’s a perfect example of a lot of split-ticket voting. What we had thought, what political scientists have been writing about and talking about, is the disappearance of split-ticket voters. I am convinced that it’s very rare — think about this. Maine now will be the only state in two presidential election cycles with a winner of the state — the party, the President’s party — didn’t also carry the Senate race. We’re the only state where we see this split outcome in two presidential cycles. It was very strange to see that, first time ever, in 2016. It almost happened again, with the exception of Maine.

When you did polling this year, I assume there were times when you polled people and asked them about the presidential race and the Senate race, which presumably would allow you to capture that split voting?

Yes. In fact, we did a lengthy analysis about the best predictor of the Senate vote — what factors would most likely predict a vote for Gideon or Collins. And the No. 1 variable that jumped up, it wasn’t gender, wasn’t political party. It was preference on the presidential election. So we felt really sure, to be honest, we felt really sure there wouldn’t be very many Biden voters that move to Susan Collins, and we were wrong.

I’m not gonna be too defensive on this. I think it’s important to simply say we were off the mark. But polls are also snapshots in time. And there were 4% who were undecided, and turnout matters. And there was a judicial vote that Collins stayed out of, and so forth. So things can change in the last week of an election. But back to where I started, yeah, you know, we didn’t hit this one as well as we thought.