Thread by @blockedfreq

There's been a lot of conversation on the merits of analyzing the early vote this evening. Since I'm doing a fair bit of it for PA, allow me to offer a few thoughts about it. (1/x)

Disclaimer: I will say this - you do have to be careful about doing stuff like that. It varies wildly from state to state in terms of effectiveness. A big reason why I'm doing this revolves around testing a wider hypothesis I have about a hybrid model of election analysis. (2/x)

Subsequently, many of you quite rightly wonder how and where I came out of nowhere to start doing analysis for #PASen on an in-depth level. It just so happens that this year and an election that I'm heavily invested in finally allowed me the chance to test my thesis. (3/x)

So, let's start. The EV is not and should not ever be an indicator of how the final vote is going to look like. We had more than enough proof of that in 2020. What it can do, however, is give us useful insights which, combined with polling and other factors, can help with (4/x)

predicting the outcome, overall picture, and/or what electorates are going to have to look like on ED for candidates to win. Take Dr. Oz, for instance. Based on what we know in the EV *and* polling, we know he needs to win ED by at least 20, give or take a few pts. (5/x)

That in turn influences perceptions and knowledge of a whole host of other variables. By proxy, is it impossible for Oz to get that kind of margin? Of course not. But with Mastriano, no Trump on the ballot, trash favorables, etc? It becomes more of an uphill battle for him. (6/x)

Furthermore, it becomes even more of a struggle based on further supplemental information we can get from the EV. One key metric is the fact that we know R's are underrunning their turnout rate in a number of key PA counties v. Dems. Easy enough to make up on ED, right? (7/x)

And that's fair enough. But here's the problem with that. Imagine a race course with a start and finish line. On ED, R's are essentially starting from 600K votes behind to get to the start line and *then* start racing Dems. They'll win the actual ED stretch, of course. (8/x)

But here's where underrunning D's in vital counties becomes even more of an issue. Take that imaginary mark they're starting behind and move it even further back. That makes for an ever-widening gap. So R's have to not only turn out on ED at the expected margins they need (9/x)

they need to turn out even more beyond that to compensate for underrunning, thus making the path forward even more difficult for Mastriano/Oz. Who knows, however? Possible that they could fix the underrunning issues before ED and make that margin more acceptable. (10/x)

And this is where EV analysis can be useful! If, on ED, they're still underrunning their margins based on the final report, we know for a fact that Oz's not going to have to only win the ED vote by 20. He's going to have to do even better than that. (11/x)

This part is where things get a bit more individualistic, where people are freer to use polling, their priors, comparing past histories of a specific race, etc. What matters, however, is that we have actual data to base this off of and use it to shape our models. (12/x)

Essentially, that is my hypothesis, where, as early voting and mail-ins become more common, we're able to use them in conjunction with other metrics to paint a vivid election picture. Now, maybe all of this will be useless next Tuesday, but maybe it won't. Nevertheless, we (13/x)

have to continue to push the boundaries of what makes for good election predictors. It's how we get better, build each other up, shape the future of elections dialogue, etc. My hope is that I will be vindicated, but I also take responsibility and am willing to be wrong. (14/x)

Anyways, just as a side note at the end of all this since the #PASen race was the primary example of this explanation - because of the way I interpret my model/analysis, I firmly believe that Oz is the underdog. And yes (again), it's possible for him to win. (15/x)

But if you think he's starting ED with a magical advantage, you're deluded. He has a very steep hill to climb. Whether he can is up to him and his campaign. And if he does win, it's going to be by a fairly narrow margin. Simply put, he has a lot more going against him than (16/x)

for him. Underruning the EV, trash favorables, a real drag at the top of the ticket, etc... I could go on, but I've said it all already. You guys get the idea. We're at the point in the race where fundamentals start pressing their thumbs on the scale heavily and the one (17/x)

fundamental that may work in Oz's favor is the national environment if it ends up being redder than expected, which is a big assumption since nobody knows what's up with polling right now. Everything else? A dumpster fire for him. (End)

Technically, this is a 19 post thread with this one. Sorry, @battl2heaven. :(

@PNWPragmatist @Shane_maps @ChiCyph80