Yesterday, a new poll came out (of the Virginia governor’s, LG and AG races, by Roanoke College polling) that was dutifully, stenographically reported by the political media, but that definitely shouldn’t have been for a bunch of reasons. See below for a screenshot of the headline, which is that supposedly “Spanberger lead over Earle-Sears narrows,” from 17 points (!) in the May Roanoke College poll to just 7 points now. And, according to Roanoke College, the LG and AG races have the Democratic candidates (Ghazala Hashmi and Jay Jones) up just 3 points over their Republican opponents (John Reid and Jason Miyares). Except…see below for a bunch of reasons why the media, and everyone else, should take this poll – or any poll by Roanoke College, really – with a HUGE grain, hell a PILLAR, of salt.
- First, as I wrote yesterday, “Dead giveaway that Roanoke College’s previous poll and/or this one is totally f’ed up: there’s NO WAY IN HELL that Trump’s approval went from minus 34 in May to minus 15 now in Virginia. WTF??? Just crazy.“
- Second, “There’s also NO WAY IN HELL that Youngkin went from minus 6 in May to plus 8 today. What would have caused that? Uhhhh…. “
- Third, “This also makes ZERO sense by Roanoke College polling: Hashmi leads Reid by 12 among independents, Jones leads Miyares by 9 among independents, yet Hashmi and Jones only lead by 3 points overall in a Dem-leaning electorate? WTF??? Not possible.” (As a smart Virginia politico told me, “Jones and Hashmi have wider lead among independents. Just Dems appear less decided which almost certainly isn’t ‘real’.
Overall this pollster is bad at (doesn’t) pushing undecideds.’)
- And “In the end, the problems w/Roanoke College polling are multifold: 1) internal inconsistency; 2) inconsistency btwn their polls; 3) absurdly high undecideds; 4) huge misses on polls, such as their Roanoke Mayoral poll (off by 24 points!); 5) wild swings for no good reasons; etc.”
- As Adam Carlson and Chaz Nuttycombe point out, this poll was weighted, for some bizarre reason, to the 2021 Virginia governor’s race exit poll. That, of course, was a very Republican-friendly electorate, given the usual pattern in which Virginia voters go hard against the president’s party in the White House, particularly when the president is very unpopular, as Joe Biden was in the fall of 2021 and as Donald Trump is now. As Chaz Nuttycombe says, it’s “insane” to do what Roanoke College polling did here, and I agree 100%. Totally indefensible, and probably means that we need to add several (5? 10?) points to Spanberger’s lead in this poll, just to compensate for that bizarre (“insane”) weighting.
- Roanoke College’s polling has had major problems in the past, such as its poll just before the November 2024 elections, showing Joe Cobb (D) with a 24-point lead over David Bowers (R) for Mayor of Roanoke. In the end, Cobb won by just 0.15% (37.26%-37.11%), so the poll was off by 24 points.
- Also see my post from August 2024, This New Roanoke College “Poll” Has Some Really, Really Bizarre, Non-Credible, Skewed-Towards-Republicans Sh** In It, Such as Having Tim Walz’s Favorability at -17 Points (WTF???). At the time, I wrote: “First of all, Kamala Harris is definitely going to win Virginia by more than 3 points; most likely, we’re talking 5-10 points in the end. So this just seems like a Republican-leaning poll from a pollster that’s been consistently Republican-leaning this cycle.” And “Tim Walz at -17 favorability (34%-51%) is absolutely unbelievable when Walz has consistently been at around +5 favorability nationally. That alone calls into serious question the credibility (or complete lack thereof) of this ‘poll.'” And “It’s also VERY odd to see Kamala Harris at -10 favorability (43%-53%) in this ‘poll,’ when she’s been at +2 to +6 favorability in polls since August 12. Why on earth would she be 12 to 18 points lesss favorable in Virginia than nationally? Who the hell knows.” “Also, why on earth would Trump’s favorability rating in Virginia have increased by 4 points since May? Crazy.” “The idea that Virginians are almost evenly split on whether or not Trump’s a threat to democracy isn’t credible, nor is the near-even split among Virginians who supposedly think that the switch from Biden to Harris was a threat to democracy. No chance any of that is even remotely accurate.”
Bottom line: Roanoke College’s latest poll, of the Virginia Governor/LG/AG races, is highly suspect, for the reasons noted above. And it’s not like this is a pollster with a great deal of credibility built up over the years, so it’s hard if not impossible to give them any benefit of the doubt. Having said all that, let’s just assume that the Spanberger vs. Sears, Hashmi vs. Reid and Jones vs. Miyares races are close – and donate, door knock, etc. accordingly. Then, if Dems end up winning by 10+ points or whatever in November, we can credit ourselves with helping to make that happen, regardless of what a methodologically flawed, internally consistent, etc. Roanoke College poll in August had to say…
