Last week Emerson College released their second poll of Connecticut’s 2022 gubernatorial and senate races, and found both Ned Lamont and Richard Blumenthal with comfortable leads. The results weren’t too different from their first poll, done back in May, which suggests a very stable race. Democrats seem to have this one in the bag.
Except this is pretty much exactly what the polls told us about the 2018 governor’s race. The final Emerson poll of that race, taken a week before the election, showed Ned Lamont leading Stefanowski 46%-39%, with independent Oz Griebel in third with 10%. Lamont’s 7-point lead was outside the margin of error of 3.7%.
Other late-October polls were all over the place. Sacred Heart gave Stefanowski a narrow 40%-38% lead, Quinnipiac saw a closer Lamont win of 47%-43%, and a poll from Gravis Marketing had Lamont with a wide 46%-37% lead.
The final results, in which Lamont squeaked by 49%-46%, were most accurately predicted by Quinnipiac. The other polls, not so much. Nobody had Stefanowski doing as well as he did, though Quinnipiac’s poll had his final result within their margin of error of 4%. But even Quinnipiac overestimated the strength of Oz Griebel’s candidacy; almost all polls had him winning 7%-10% of the vote, while he ended up with only 4%.
That comfortable lead for Lamont in the latest Emerson poll might be more of a mirage than a certainty, then. This reveals an uncomfortable truth: public polling is very uneven right now, and has been treading water since 2016.
There are lots of reasons. The 2016 election blew up most pollsters’ turnout models, and they’ve struggled to predict who will be at the polls in the years since. Fewer people have landlines, and how many of us will answer when we see a call from an unfamiliar number on our cell phones? Pollsters have had to get creative in how they reach people, with somewhat mixed results.
If we look at what polls are supposed to do, which is to give us an idea of how people who will actually turn out are going to vote within a reasonable margin of error, only Quinnipiac was able to hit the mark in 2018. Okay, great. So we should take a look at the latest Quinnipiac poll, right?
Sure, except they haven’t polled this race since May. That’s the other uncomfortable truth: pollsters stopped caring about Connecticut when we stopped having competitive congressional races. We’ll likely get a few more polls before Election Day, but I wouldn’t expect much.
Another headache for pollsters this year is that nobody has any idea who is going to show up to vote. The Supreme Court’s overturning of Roe v. Wade has Democrats fired up, and recent special election results and the spectacular failure of a constitutional amendment to ban abortion in deep-red Kansas suggests that the president’s party may not suffer the usual midterm drubbing. Will this year’s electorate be more like 2016, 2018, or 2020? I’m guessing it will be its own thing, unlike any of the previous years, but that’s only a guess.
Emerson Poll Demographics
| Race | |
| Hispanic or Latino of any race | 7.2% |
| White or Caucasian | 80.2% |
| Black or African American | 8.1% |
| Asian American or Pacific Islander | 1.0% |
| Other or multiple races | 3.5% |
| Education | |
| High school or less | 20.6% |
| Some college but no degree | 30.3% |
| College graduate | 28.8% |
| Post-graduate school or advanced degree | 20.2% |
| Age | |
| 18-34 years | 15.0% |
| 35-49 years | 25.3% |
| 50-64 years | 32.1% |
| 65 years or over | 27.6% |
| Gender | |
| Male | 46.5% |
| Female | 51.6% |
| Nonbinary or other | 1.9% |
So we have a bunch of polls that show Ned Lamont and Richard Blumenthal with big leads over Bob Stefanowski and Leora Levy. What should we make of them?
Emerson is predicting an electorate that is whiter, more female, and more educated than the state’s population as a whole. This sounds like a good guess, but it’s worth remembering that the same poll underestimated Stefanowski’s support in 2018. Polls may be undercounting Republicans and Republican-leaning voters, which is something that has happened nationally in the past three elections.
I’m guessing that there are Stefanowski/Levy voters that the polls are missing. There’s good reason to think this; the kind of people who don’t trust polls are much more likely to vote Republican than they have in the past. The right wing is deeply distrustful of anything they perceive as “mainstream” media, and pollsters often fall into that category.
Because both Stefanowski and Levy are now running campaigns aimed at the GOP base, the chance of more of these voters increases. Levy was always running this kind of campaign, it’s just who she is. Stefanowski has recently pivoted from milquetoast moderate to culture warrior, and this is almost certainly why. There aren’t enough votes in the middle for him to win. But there may be more votes out there on the right, all he has to do is lure them out of the fever swamps.
Are there enough of these voters in Connecticut to swing these races? Absolutely not. But they will likely make it closer than the polls are saying right now. That is, unless a tidal wave of furious pro-choice women sweeps them away first.
So take the polls with a grain of salt!
