I want to acknowledge a lot of the others on this #GASen modeling ride:
@ADincgor -- the model's co-creator @joe__gantt, who also models it and has given us the precinct adjustment for our model @thunderousprof and @dooraven, who track early voting reporting with us
[1/]
@EconWanabe, @Eggymceggerson3 and @umichvoter99, who track the backlogs and outstanding ballots and follow the data just as closely. @krazgreinetz, who's done a great job writing on/tracking this. @EScrimshaw, who was the first to yell at me about why our initial model was wrong
@OpenModelProj, who have really done a standout job in improving model transparency @DFMresearch, who's been modeling this on the side as well @nolesfan2011 and @elium2, who have walked me through several key GA metrics and voting procedures @JDubwub for his great graphics
Whatever happens, some of us will be wrong tomorrow.
That's okay. This is a race with a lot of uncertainty, and you should expect that some of the assumptions we make will break. After all, there hasn't been anything like this in recent history.
Whatever happens, don't go into the mentions of modelers who made *data-based takes* on why they thought the GOP could win and laugh at them if Democrats win.
Do not go into the mentions of people who thought Democrats would win and yell if the GOP win.
Most forecasters and modelers are keeping their models private. Anyone making their forecast public deserves a lot of credit for putting it out there instead of shame if they're wrong. All they do is follow the data. If things break, then they break, but they've been honest.
I'm almost certain I forgot some people in this tweet, so please just go ahead and tag them if I haven't done so already; any oversights are unintentional and are a function of me having accidentally had milk before taking lactase supplements and thus being in great pain recently
oh, yeah, some others who've done some great work:
@ryanmatsumoto1 -- wrote some great columns on this that I would highly encourage you to read @PoliticalKiwi -- first made the chart on enthusiasm's correlation to e-day @CautiousLefty -- has been tracking exurbs/rurals forever
And, of course, @breaking_et and @KrissieTX; you all already know what they do. Krissie's done a great job with Team Ossoff, as expected, and Breaking ET has done a wonderful job reporting on-the-ground turnout numbers.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Two months after Joe Biden carried the state, Democrats have pulled it off and appear to have won a trifecta by sweeping the double-barrel runoffs in Georgia. #GASen
Every single organizer, volunteer, campaign worker, and Democrat should be proud of what Georgia has just achieved. It's a roadmap for the rest of the South, and it's exactly how you capitalize on shifting political trends.
Jon Ossoff is 33.
Given his age and the shifting landscape of Georgia's politics, he could keep that seat blue for a long, long time.
From political punchline to savior. What a journey.
Given that Democrats still have most of their election day voters from November who haven't voted yet, the GOP probably needs 1.1M voters tomorrow, IMO, to have a fighting shot, and even then they need to drive up the margins.
Where does that come from?
The GOP has had a serious turnout problem during all of early voting. If your argument is that many have switched their vote to election day, fine -- I agree to some extent! But that ignores the fact that many just won't vote. And not everyone who plans to vote will.
So who does that help?
My gut feeling is that people here overestimate the red skew on election day in margin. There's an argument to be made that Democrats shouldn't expect their election day vote to crater as much. And white rural voters are awful with turnout in general.
Our final race rating is Lean D, and we will not be moving it from this regardless of any other EV developments. I don't believe the data favors Rs at all, but I'm not comfortable publicly moving a race beyond lean given the assumptions in the model.
We stand behind our model and our data, but there are implicit assumptions that don't allow for such a high level of certainty.
An example: what does turnout look like? We've never had a runoff like this, so we don't know.
Secondly, is it possible that election day skews more heavily Republican than 61-39? Sure! How much? IDK. I think the GOP would need a borderline miracle to hit the threshold needed right now, but that's assuming our model's EV split assumptions hold. And they might break!
In the precinct model made by me and @ADincgor, we've now accounted for the individual-level shift observed by NYT/Siena, thanks to precinct race and party adjustments made by @joe__gantt.
Okay, now, let's go into some of the methodology behind the race/precinct adjustment for in-person EV, made after a conversation with @Nate_Cohn. It's not random at all, and it is described in great detail in the spreadsheet, so I encourage you to take a read if you're curious.
Our original precinct projection weighted everything by geographic splits only, which captured a lot of the picture, but did not capture the reality behind individual voter attitude shifts, which did not show up in precinct or county level data. This fixes that.
@Nate_Cohn is right -- the early electorate is probably closer to D+13, as his tweet says, than D+6 (our model).
Here's the question: What's our model trying to do?
Are we trying to forecast the electorate? Or are we projecting the November electorate onto the current one?
If we're forecasting, we're going to be way off. We don't have the data for it, and I don't have the time to pay the $250 for the voter file and go through it and build a model by Tuesday.
If we're going to project the November electorate onto the current one, then some questions need to be asked.
As @Nate_Cohn so kindly explained to me, there is an individual level shift that doesn't show up in the precinct data, but let's think of it like this, in his words...