Virginia Poll: Calm Down
Stop Trusting Polls As If They Mean Anything,
The poll is wrong.
Here are two things that can be true at one time: Terry McAuliffe could end up winning the Virginia Governorship by 4% in November, and the Sunday-released Virginia Governor is total crap that should not be taken seriously.
What's the "right" margin, if the poll's D+4 is wrong? Fuck if I know, but the problem is, the poll is crap. The poll has Terry McAuliffe getting 61% of the Black vote and winning Black voters by 41%, while Joe Biden got 90% of the Black vote on route to a 81% victory with Black voters in November. If you think Terry McAuliffe will only get 61% of the Black vote, I have oceanfront property in fucking Wyoming for you, and that this poll got this very basic thing very wrong means it is unreliable.
If you take the poll's white vote and plug in that 81% margin with Black voters, you turn a D+4 into a D+11, because fucking up the Black margin by 40% in a 17% Black electorate (by the poll's stated compostition) is worth just under 7% statewide. Oh, but you can't do that, because what if the white number is also wrong? That's entirely correct, and why this poll is a steaming pile of dog shit.
I didn't write about the New Mexico 1st poll when it came out because, as I said to someone, whatever it says I'm not changing my mind. US polling is shit, and until such time it isn't shit, I'm not letting polling tell me what to think about a race. Thank God I didn't do that in New Mexico 1st, because that poll undershot the Democratic margin by 9%. It's also why I haven't written a single column about the (generally very good) Generic Ballot polls for Democrats right now, or at least one of two reasons. I have no reason to believe these polls are meaningful snapshots of the current landscape, and even if they are, there's no guarantee it means anything for 2022.
I think McAuliffe wins fairly comfortably, but I think this entirely on the basis that he is a popular former Governor running in a blue state without a deeply unpopular President dragging him down. If this poll had been a McAuliffe +14, I would be similarly unmoved, because I don't care, and neither should you. Polling is valuable because, in theory, it informs things about the races that are polled, and in the rest of the world, that value is clear. It's not clear given that US polling is about as clear or coherent as the result of a drunk playing darts. They might get a good score, but it's almost certainly luck.
Should I take this poll's white number - Glenn Youngkin +5, compared to Trump +6 in November - at face value? Fuck if I know, but this pollster had a Georgia runoff poll with Jon Ossoff winning by 8% and only losing Georgia whites by 31%. He won by 1% and lost Georgia whites by 46%, so the idea that we can trust the white numbers as solid is bad faith. Could we see a scenario where the white tab is too favourable to McAuliffe, the Black tab is way too favourable to Youngkin, and the errors cancel each other out? Oh, that's absolutely possible, but even if that happens, the poll is still trash, and it just got lucky.
The problem with this poll, and frankly every poll at this point, is that there can be no analysis of it beyond, "well, this could, maybe, maybe, be correct," and in this case we already know one part of it isn't correct. Like, I'm not trying to be a dick, but Glenn Youngkin is not going to get 20% of the Black vote, it's just not happening. Yes, I know, I said the same thing about Hispanics and was wrong, yes, you can dunk on all my old tweets if that makes you feel any better, but we literally did this two separate times in Georgia and that was the only state I seemingly had a good handle on all of 2020.
The way Youngkin wins is outrunning Trump with whites - specifically, whites with a degree in the DC suburbs. This isn't a controversial statement - that this is the path forward for Youngkin is undeniable. What is arguable is whether he can execute that path, but literally nobody, no matter how bullish or bearish on Youngkin's chances you are in the macro, thinks this result - match Trump with whites, massively outrun him with Blacks - is how this race would be close. If Youngkin actually gets 20% of the Black vote, he's winning in a landslide, and if McAuliffe actually only loses whites by 5%, he's outrunning Biden.
This is not me declaring McAuliffe will outrun Biden, or outrun the Northam 2017 result of a nearly 9% win. Yes, I am extremely confident that McAuliffe will win, but that's because I think the range of plausible results goes from like D+2 to D+12. It can get close, it can get uncomfortable for Democrats, but they're not losing the race, barring a huge shock, in my view. What I am saying, clearly and simply, is that this poll should not move you in any way. You should not take what this poll says seriously, because the chances it is right in any meaningful way are about the same as me suddenly becoming straight. This poll might have two errors which cancel each other out and then, after the fact, they try to spin that it was a success. If Youngkin wins whites by 12 and McAuliffe wins Blacks by 75, which would get you close to a D+4, that is not a victory for this poll. That is this poll getting lucky. We all know the Trump era record of polling, and the poll in New Mexico 1st doesn't bring any confidence to the table. Trusting any individual poll to be a reliable indicator of the state of a race is bad, but it is especially bad when the data doesn't pass the most basic of smell tests.
What's the current lead in Virginia? I have no fucking idea, and neither does anyone else. Yes, we've had a poll, and still, we have absolutely no fucking idea what the state of the race is right now. Could be close, could be a blowout for McAuliffe, could be something in between, hell, probably is. But I haven't the foggiest idea, because I'm not going to pretend that just because something is called a poll means it has any value. This poll in particular, and most US polls at this point, should be viewed for what they are - best guesses. To imbue any more meaning to this is to give them more power than they deserve.