Evan Scrimshaw: Georgia Notebook: SurveyUSA, Early Vote, And Analysis

(We're once again pleased to have Evan Scrimshaw (@EScrimshaw) write this column today for us as a guest writer. Evan has years of experience with global election forecasting and is an expert on Australian, American, British, Irish, Scottish, and Canadian electoral politics. We're excited to feature his work here.)

With the release of a SurveyUSA poll showing Jon Ossoff up 5% and Raphael Warnock up 7%, my self-imposed ban on writing on polls is over - at least for a day.

The main question everyone has - is this poll correct? - is worth asking, but the answer is I have absolutely no idea. The question we can ask is whether this poll makes sense internally, and whether the component parts of the poll all make sense together. Here, we can judge a poll.

For those who missed Emerson's recent poll, they had a white number quite similar to SurveyUSA - a small, 3% swing to Ossoff with white voters compared to Biden - and somehow had David Perdue getting 20% of the Black vote. That, categorically, is wrong, and even if that topline were to be correct, they'd be absolutely garbage pollsters for getting there the wrong way. SurveyUSA's work at least passes an initial smell test, under the hood, but again, I have no idea if they're right or not.

SurveyUSA has Ossoff doing 4% better with whites than Biden, the share of the electorate that's white down 6%, and Ossoff holding his own with non-white voters. The poll has a much smaller amount of white split ticket voters than SurveyUSA's last offering, suggesting that the unity ticket messaging Dems are using is helping - although, it could just be that David Perdue's corruption and pandering to the right has pissed off suburbanites who were already queasy about voting for him given their contempt for Loeffler. (Or it's just noise that means nothing, I guess.)

The main contours of the poll are broadly backed up by the Early Vote data - Black voters are making up a higher share of the electorate than November, and the white share is down - and, as of now, down more in rural and exurban Georgia than in white suburban Georgia. This poll is a fairly extreme picture of what a GOP turnout drop can do, but it is well within a reasonable range of outcomes.

What would have happened if the 2020 election had been fought with the 2016 electorate in Georgia? It's a weird question for me to pose, since hypotheticals are generally useless, but here it matters. If you use Joe Biden's margins by racial group - splitting whites by education, of course - and apply the 2016 electorate, Biden would have won the state by 3.9%.

Biden won by getting a huge, 21% swing his way with whites with a degree, and by doing just well enough with non-degree holding voters to withstand the surge of those voters who came out for Trump. In rural Effingham County, turnout was up 50% from 2012 to 2020, and somehow, Trump still lost. Plenty of talk about the state talks about what happens to the suburbs when Trump is gone, but little of it talks about what happens to the rurals when Trump goes away. According to SurveyUSA, the rurals aren't turning out this time, and the early vote says the same thing.

If Biden would have won Georgia by 3.9% if the electorate was 2016's and not 2020's, then what you're implicitly - or, in my case, explicitly - saying is that Trump did really, really well to keep the state close. And, fundamentally, he did. Suburban reversion is one hell of a theoretical drug, but for all the column inches and overdone tweets about that possibility (thoroughly examined last week), there's remarkably little about how fragile the GOP coalition could - and, I stress, could - be, if these Trump voters feel that conventional Republicans fail to deliver. If this election is any indicator, they're not feeling too enthused - but then again, it's David Perdue and Kelly Loeffler, two corporate shills with insider trading allegations. (By the by, Democrats would be in much worse shape if Doug Collins had won the 2nd spot on November 3rd, to be clear.)

The other lasting impact of this pair of races is they provide a data point on the role of campaigns - whether or not winning the campaign and winning office have much connection to each other. The GOP campaigns are a disaster, with the two Republicans barely appearing together and joining in on a war against the state's Secretary of State and Governor, who are *checks notes* Republicans. On the other hand, Ossoff and Warnock are hitting on all cylinders, with Warnock launching an ad campaign today targeting the ruby red south of Georgia where the entire ad is Fox News hosts attacking Loeffler for insider trading. Using clips of people with credibility to Republicans to depress GOP turnout? Brilliant. Not to be outdone, Ossoff's been the breakout campaign star of the cycle, with a passion, a fire, and a willingness to say what he thinks that was sorely lacking from his 2017 run.

If the GOP win, then the notion that campaigns matter should be put on ice. An Ossoff/Warnock victory would not be dispositive that campaigns do matter, but if Loeffler and Perdue win, then it's very clear that nothing matters anymore.


Will Ossoff and Warnock win? I think so, speaking solely for myself. I think it is hard, based on the early voting data and the SurveyUSA poll, to make the case the GOP are likelier to win right now than the Democrats, which is not to say that things can't change. The GOP could get the turnout they need on election day, rendering all of this chatter about early voting struggles moot. Even with SurveyUSA and the early vote data we have, I'm not going to say the GOP are anything worse than a 35% shot - solidly a Lean D race in my anti-tilt view, but absolutely not a sure thing. Nuance doesn't make for a great answer, but it's the only reasonable course of action. That said, if you were to come away from this thinking I'm the most optimistic of Chuck Schumer's chances of being Senate Majority Leader in 2021 that I've ever been, you're not guilty.


Popular Posts