Evan Scrimshaw: Can you say the GOP are favored right now in the Georgia Runoffs?
(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.)
Longtime readers of my work - here, and at past stops - will know that my work does not shy away from contrary opinions, and for those who don't agree, can sometimes (read: all the damn time) read as needlessly contrarian. Usually that's a natural impulse, but for once, let's tap into the case for why *actually* the GOP are favoured in Georgia and see whether it holds up under any amount of scrutiny whatsoever.
Perdue Almost Won Outright
It's probably fair to say that the GOP would be sizable favourites to win the Jon Ossoff/David Perdue battle if the electorate that voted in November voted again exactly. That seems fairly obvious, given Perdue's better outcome than Donald Trump's in November. This fact has made some people believe that Perdue is favoured in this race, and that's where the high comedy comes into play. The electorate won't be the same as November, and we know this at this point.
It is really, exceedingly clear that the electorate will be more educated than November, and that it will have a higher share of the electorate being Black than November - two trend lines that point to the electorate being more Democratic than November. I tweeted out at estimate at some point that a personal electorate model I built pointed to an electorate that would have voted for Biden by just under 3% - and now, that estimate would be more Democratic than it was then.
The GOP turnout machine - aided in large measure by the same sort of Trump surge voters that saw North Carolina stay red, and which saw Ohio move right - was actually really good in November, but suburban slippage was just too much for the Trump campaign to withstand. Trump and Georgia is a weird relationship - Trump is a horrid candidate for the metro, but an amazing one for the rest of the state. And now he's off the ballot.
Past Runoffs Have Helped The GOP
In 2008, Jim Martin went from losing by a few percent to getting walloped because his coalition - rural whites and Blacks - had turnout go down from the general election to the runoff. Now, the Black share of the electorate - on a like for like comparison - is up 3.8% from the November General Election. It could be that Black voters have tapped out, theoretically, but that is a fairly low probability outcome. Those same rural whites that failed to come out in 2008's runoff? They're now Republicans, and they're still not turning out, and those educated suburbanites who did vote in that runoff? They're now Democrats.
Desire For Divided Government
There's been one state in two Presidential cycles to split their tickets for Senate and the Presidency - Maine, which should be avoided as a topic to avoid a highly unacceptable level of foul language - and even in 2018, the Senate elections were mostly in line with a Presidential partisanship, with voters not looking to embrace a check on Trump, but actually the opposite - with the GOP making a net gain of 2. And even Alabama 2017, a partisanship breaking result, wasn't about partisanship, it was about Doug Jones versus Roy Moore's singular terribleness. Voters don't decide for divided government in their glory, they end up there in their indecision.
The Polls Have Perdue As A Slight Favourite
Yes, the 538 average has David Perdue up. The only credible pollster to poll the race publicly has Jon Ossoff up 5%, but either way, US polling is garbage and should be tossed out.
The argument that the GOP can credibly make is that their voters are all just going to show up on the day, which is why this site's official position (also known as Lak's position, from which I don't dissent) is Lean Democrat. That means that the GOP could still easily win, which is well within the idea that Ossoff is up by something like 2%, which is in line with the modelling that has been released publicly.
The problem for the GOP is that I'm not sure that modelling is correct - mostly, I think the county projections for how partisan the Early Vote will be is wrong, because the Black electorate share going up will help Democrats. I think it's a very solid attempt to project how this might go, and wariness of criticisms of a model from he without one is always a fundamentally correct position, but I do wonder if they're just a tad conservative - which, if they think Ossoff wins with overly Republican assumptions, the GOP are in trouble.
Now, the GOP path to victory is pretty simple - either in late Early Voting, or on the day, their turnout surges to get an electorate that looks pretty close to 2020. Maybe they pull it off - it would be a perfectly plausible for them to win, but not likely. That's why it's Lean D. You can quibble with the edges of the probabilistic distribution, but more often than not, Jon Ossoff wins, and the GOP arguments for why that isn't true don't hold up. They're answers looking for logic, and not finding it often. The case could have been made for Tossup in the past, but the early vote is just so bad for the GOP. They've got a shot, because Election Day is still to come, but they're the underdogs. Lean D.