Be wary of political polling. It is often wrong.

By Matthew Gagnon

For most of the past year in casual conversations among Maine politicos, the consensus has been that the impending gubernatorial contest between Janet Mills and Paul LePage would be very close. Whether you thought Mills would win, or you thought LePage would, generally speaking you thought it was going to be a barnburner.

And for most of the last year, that has proven true. I’ve been privy to a lot of non-public polling for the last year or so, and for nearly all of that time the race between the two was more or less tied. All signs were that conventional wisdom was playing out.

This summer, though, some evidence began to emerge to the contrary. Several things began to converge at the same time, from the Supreme Court decision in Dobbs, to a deadlocked Congress passing part of the president’s domestic agenda, to Joe Biden’s announcement of some student loan cancellation. Most importantly, though, was the summer Mills ad barrage attacking LePage, while LePage has remained mostly quiet on the air.

Those things seemed, at least on the surface, to have had an impact on the race. Some of those same polls I was looking at earlier this year have shown movement in Mills’ direction, and in the last couple weeks we have seen some public polls that have shown Mills with a comfortable lead over LePage. 

Last Friday, Emerson College released a poll which claimed that Mills was up by 12 points, and the University of New Hampshire released one Tuesday putting Mills’ lead at 14.

But to what degree should we be putting our faith in public polling in this race?

You have to admit that it is a fair question to ask given the recent history of polling in Maine. In 2020 for instance, polling showed that Sara Gideon was beating Susan Collins in every single public poll conducted in that race — including one by Quinnipiac University in September that showed Gideon with a 12-point lead — and yet Collins ended up winning by an 8.6 percent margin. That same year, out of 12 public polls of the 2nd Congressional District, 10 showed Joe Biden beating Donald Trump there, often comfortably, and yet Trump ended up winning the district by a 7.5 percent margin. 

Similarly, in 2016 there was limited polling done in the 2nd District race for president, but Emerson College did publish a poll in late October showing Hillary Clinton beating Trump there, only to see Trump win by more than 10 percentage points. Similarly, UNH published a poll around the same time on the congressional race, suggesting Emily Cain had a 2-point lead on Bruce Poliquin. In the end, Poliquin won by 9.6 percent.

Even as far back as 2014, the last four public polls in that year’s gubernatorial contest showed that it was tied, or that Mike Michaud had a small lead, but LePage ended up winning by roughly 5 percent. 

There is only one time in recent memory where polling appeared to be reasonably accurate, and that was in the 2018 gubernatorial contest, when the polling matched the end result almost exactly.

So what gives?

This is a question that many in politics have been grappling with, particularly in the wake of the recent “surge” of support for Democrats nationally. The New York Times ran two pieces on this very subject recently, wondering aloud “are the polls wrong again?” Their conclusion: it looks likely, as Democrats are apparently polling well in exactly the places where surveys missed most in 2020. Politico also ran a story on Monday highlighting the fear among pollsters that they may be once again blowing it in 2022.

A closer look at the two most recent polls from Emerson and UNH suggests those doubts may be accurate. Emerson’s poll showed that the economy was the most important issue to 39 percent of voters, and of those voters, 69 percent said they were voting for LePage, yet he somehow is losing by 12 points. In the UNH poll, 46 percent of those sampled were Democrats, and only 14 percent were unenrolled, while the actual alignment in Maine is closer to 35 percent and 32 percent, respectively. 

The reality is, the fundamentals of political polling have changed, and though pollsters are trying mightily to adjust, they are failing miserably. It may no longer be possible to trust most horse race surveys like we once did.

Maybe they’re right this year, but they probably aren’t. Either way, you shouldn’t put your trust in them and as the old adage goes, the only poll that will ultimately matter is the one held on election day. Ignore the noise.

Gagnon of Yarmouth is the chief executive officer of the Maine Policy Institute, a free market policy think tank based in Portland. A Hampden native, he previously served as a senior strategist for the Republican Governors Association in Washington, D.C.