When a poll came out last week showing Amanda Janoo leading Gov. Phil Scott, it wasn’t well received by Team Scott. Campaign manager Dustin Degree called the University of New Hampshire Survey Center’s Green Mountain State Poll an “outlier,” which is factually incorrect on its face. It can’t be an “outlier” because there were no other polls to compare it to.

But I can understand the reaction. You thought you had it in the bag, and here comes an inconvenient poll poking holes in your Shield of Invulnerability.

Well, now we have a second survey, and what I can tell you is that each of them outlies the other. The new poll commissioned by Vermont Public is more favorable to Republicans across the board than the UNH survey. In the gubernatorial race, it puts Scott 10 points ahead of Janoo while UNH gives Janoo a six-point lead.

In the campaign for lieutenant governor, UNH has Democrat Molly Gray holding a decisive 19-point edge over incumbent John Rodgers while VP reports Gray with a lead of eight points. And there’s a huge divergence in the campaign for Vermont’s single seat in the U.S. House: Instead of incumbent Becca Balint’s 66-30 edge in the UNH poll, the Vermont Public survey gives Balint only a 15-point lead with, again, enough undecideds to turn the race around.

Those are big enough swings to cast doubt over the entire enterprise. I mean, each poll claims a margin of error under 3.5 points, but the differences between the two are waaaaaay bigger than that.

Funny thing, Team Scott wasn’t much happier with the Vermont Public results than with UNHSC’s. A written statement provided to Seven Days asserted that public polling “has always undervalued the governor’s support and this poll, while much better than the UNH survey, continues to do the same.”

Seems a bit churlish to me. I guess they aren’t happy unless they’re grinding the Democratic candidate into the dust.

Andrew Smith, head of the UNHSC, provided a written defense of his outfit’s methods to Seven Days: “We have accurately predicted elections in Maine, New Hampshire and other New England states using this methodology since we started using it in 2020.”

The partisan splits between UNHSC and VP are largely the product of a difference in undecideds. There are a lot more of ’em in the Vermont Public poll. Here’s how the decided/undecided breaks down in each poll.

• Governor, UNHSC: 94 decided to 6 undecided. Governor, VP: 80 to 20.
• Lieutenant Governor, UNHSC: 89 to 11. LG, VP: 81 to 19.
• Congress, UNHSC: 97 to 3. Congress, VP: 78 to 22.

The larger ranks of VP undecideds appear to have leaked away from Democratic candidates. Look at the two polls’ numbers for the three Dems:

• Amanda Janoo, UNHSC: 49%. Janoo, VP, 32%.
• Molly Gray, UNHSC: 54%. Gray, VP, 43%.
• Becca Balint, UNHSC: 60%. Balint, VP, 43%.

Brief pause on the last set of figures: VP’s number for Balint is simply unbelievable. Unlike Scott, Balint has high approval numbers. There’s no sign of antipathy toward her except among hard-core Republican voters. I’m certain that UNHSC will prove much closer to the mark in the Congressional contest when all the votes are counted. A partial observer might propose that if VP is that wrong about one race, it might be wrong about all three.

But the broader question is, why does VP apparently shunt so many liberal voters into the “undecided” category? Republican numbers are consistent between the two polls. It’s the Democrats who suffer from some factor in the VP methodology.

I don’t have a ready explanation. We won’t know which poll was accurate until Election Night.

So where does that leave us? Safe to say, as Seven Days’ Kevin McCallum wrote, that Scott “finds himself in his tightest campaign since he was first elected in 2016.” Hence the brusque reactions from Team Scott. They’re not used to being in an actual competition.

Here’s one big thing that ought to have them running scared: Scott’s support is below 45% in both surveys. His job approval is statistically the same: 47% approve in the UNHSC poll and 45% in VP.

That ought to be ringing alarms at Scott HQ. The voters have seen him in action for almost a decade. He’s one of the two most familiar politicians in Vermont (Bernie) and he’s facing a virtual unknown. Even so, less than half of the voters are on his side. Let’s say you give him the VP poll and its 10-point lead. He still might face a tougher task getting to 50% than Janoo. Her enemies are time and resources, and those are formidable foes. Scott’s enemy is his faded popularity. How does he turn that around in five weeks?

One obvious way is to jumpstart his campaign, and he appears to be doing just that. The next round of campaign finance reports isn’t due until October 1, but candidates are required to report mass media expenditures within a couple days of cutting a check. Well, the Scott campaign has filed five separate mass media reports since last Thursday. They total $159,355, more money than he’d spent in the entire campaign before then. Team Scott has invested $125,705 in TV advertising plus $33,750 for online ads. The online content touts the governor and a long list of Republican candidates for the Legislature. The TV spots mention Scott and no one else.

Who’s not on that list? John Rodgers. Two years after Scott practically tied Rodgers to his rear bumper and pulled him over the finish line, he doesn’t seem to be lifting a finger for his supposed right-hand man and presumed successor. That either means his campaign sees Rodgers as dead weight, or they’re too worried about Scott’s prospects to care about the Lieutenant Governor.

Or both. Both is an option.

It’s rare for two contemporaneous surveys to produce such different results. One thing’s for sure: These polls are finally bringing some vastly overdue media attention to the campaign. Maybe we’ll see a more consistent effort to cover this thing. Maybe Janoo will be taken more seriously as a Scott challenger than anyone since Sue Minter in 2016. And maybe that will contribute to the outcome both polls project: the closest gubernatorial election in a decade.