The 2021 elections end later today, November 2. For background on the gubernatorial elections in New Jersey and Virginia – including how I calculate polling averages – please see here and here. I made two minor adjustments since then: all polls with a field midpoint of October 1 or later are given a weight of 3.0, while all polls conducted for a campaign or by a clearly partisan pollster are given a weight of 0.5.
The bottom line is this:
- Democrats Eric Adams and Michele Wu are heavily favored to be the next mayors of New York City and Boston, respectively
- Democrat Phil Murphy will be reelected governor of New Jersey, defeating Republican Jack Ciattarelli, though the margin will be far lower than the “fundamentals” (state partisan lean, incumbency, generic ballot) suggest it “should” be.
- Republican Glenn Youngkin may be poised to upset former Democratic governor Terry McAuliffe.
In New Jersey, only the size of Murphy’s margin is in doubt. Eighteen polls of this election have been released publicly in 2021, 11 since Labor Day – eight of which were conducted entirely since October 1. Overall, these B+-average polls have a slight historic Democratic bias of 0.2 points, on average. They show Murphy ahead by an average of 7.6 percentage points (“points”) – a 2.3-point drop in two weeks – with about 7.8% choosing a third-party candidate or genuinely undecided; the lead barely changes (7.5 points) using only polls conducted since October 1. Figure 1 shows that after declining sharply over the summer, Murphy’s lead has stabilized to 7-9 points since mid-August.
Figure 1: Murphy polling lead in 2021
Depending on how the few undecided voters split (and assuming 1.5% third-party, cutting two 3% polling values in half), Murphy is “projected” to win by 5-9 points. Given the strong Democratic lean of New Jersey, my – hunch – is the few remaining undecideds will break slightly Democratic, say 55-45, for a projected Election Day margin of 8.2 points – something on the order of Murphy 53.3%, Ciattarelli 45.2%, Other 1.5%.
This brings us to Virginia, where the conventional wisdom is that McAuliffe is in trouble. Two weeks ago, I thought this was nonsense. Now, I am not so sure. In fact, a Youngkin upset seems increasingly plausible.
A total of 46 publicly-released polls have assessed this race in 2021, with 32 released after Labor Day – 23 of which were conducted entirely after October 1. These B-average polls have a Democratic bias of ~0.5 points, on average. Overall, McAuliffe’s lead is now only 0.4 points – down sharply from 2.5 points two weeks ago. The trend looks even worse for McAuliffe if only polls released in October are examined (Youngkin +0.5), while the 15 most recent polls show Youngkin ahead about 2.0 points. Figures 2 (linear regression model) and 3 (polynomial regression model) show how McAuliffe’s polls fell off a cliff about two weeks ago, though the values have stabilized over the last week.
Figure 2: Linear Model of McAuliffe lead in 2021
Figure 3: Polynomial Model of McAuliffe lead in 2021
Both models project Youngkin to be ahead on Election Day by anywhere from 1 to 5 points, well in line with the 15 most recent polls.
However, there is a significant caveat to these data.
On October 28, the third of three Fox News polls of this race was released; it was conducted October 24-27. Fox News is A-rated and has a historic Democratic bias of 1.8 points. The previous two polls, conducted September 26-29 and October 10-13, had shown McAuliffe leading by 4 and 11 points, respectively, among registered voters. The latter poll, meanwhile, showed McAuliffe leading by 5 points among respondents designated “likely” to vote (73% of registered voters). Already, we see two red flags: 1) the unusually-large six-point gap between registered and likely voters in the late-September poll, and 2) the 7-point increase among registered voters in just two weeks. The late October poll, meanwhile, showed Youngkin leading among registered voters by 1 point and by an astonishing 8 points among “likely” voters; this is by far the largest “topline” lead for Youngkin. Thus, in just four weeks, McAuliffe went from +4 to +11 to -1 among registered voters, which seems – implausible. Moreover, the mean gap of 6.5 points between registered and likely voters in the two October polls is much larger than the historic average of 1-2 points this late in a campaign. Still, this could suggest McAuliffe maintains an edge in the universe of all registered voters – and that the likely voter models are a bit too restrictive vis-à-vis Democrats, perhaps overcorrecting for the undercounting of Trump voters in 2020.
By comparison, the overall 10-poll rolling average went from McAuliffe+2.0 to McAuliffe+1.0 to McAuliffe-2.1 over this same four-week period. Much of this volatility (albeit matched in New Jersey and in generic ballot polls) can be attributed to these Fox News polls. Remove the most recent Fox News poll (or, for that matter, all three of them) and McAuliffe lead increases to 0.7 points overall, and to 0.2 points in October polling, while Youngkin’s lead over the last 15 polls drops to 0.8 points.
Furthermore, consider the highly-rated Washington Post (A+, D+0.9) and Emerson College (A-, D+0.8) polls. In early September, the former poll gave McAuliffe a 3-point led among likely voters, which dropped to 1 point in late October. At the same time, Emerson went from even in early October to McAuliffe+1 in late October among likely voters. All four outcomes are within the margins of error of the other three, making it difficult to say whether the margins changed or not. This consistency, though, contrasts strongly with the extreme volatility of the Fox News polling. In fact, taking just the average of the most recent Washington Post and Emerson College polls, along with one from Suffolk University (B+, D+0.9), shows McAuliffe ahead 0.3 points in the raw margins and Youngkin ahead 0.2 points after adjusting for historic bias. I think these three polls give the most accurate picture of the campaign – effectively a coin flip – at least as of October 23 or so.
On balance, I still think McAuliffe is the barest of favorites (on the order of 5:4) based upon his apparent strength among registered voters, a record number of early voters, and Virginia’s increasing Democratic lean. My – hunch – is the few remaining undecideds will break even for a projected Election Day margin of 0.5 points – something on the order of McAuliffe 49.5%, Youngkin 45.0%, Blanding et al 1.5%. That said, anything from McAuliffe winning by 4 points to Youngkin winning by 3 points remains plausible, if not probable.
We shall see shortly both who actually wins and just how good my projections are.
Until next time…please wear a mask as necessary to protect yourself and others – and if you have not already done so, get vaccinated against COVID-19! And if you like what you read on this website, please consider making a donation. Thank you.
 726 likely voters, 1,004 registered voters