8 votes

Trump’s electoral college edge could grow in 2020, rewarding polarizing campaign

1 comment

  1. alyaza
    the actual article here is interesting and points to a probable, if unlikely scenario, but what interests me more is how this relatively average article has gone over in psephologist twitter,...
    • Exemplary

    the actual article here is interesting and points to a probable, if unlikely scenario, but what interests me more is how this relatively average article has gone over in psephologist twitter, where there has been a Robust Discussion for literally two days now about how accurate this article is, how things like this should be reported, and if they should be reported at all in the way this article does. (in other words, a fucking trash fire and a half.) gather around for yet another long tale of "what the fuck is even happening" as we peer into another heart of darkness at length.

    Part One: Nate Cohn, Professional Anxiety Inducer

    our story begins with this chain of tweets by nate cohn on his story here:

    Trump's Electoral College edge endures heading into 2020, and it could grow further--potentially allowing him to win the presidency while losing the national vote by 5 points or more. That huge split between the Electoral College and the popular vote would be unprecedented and it is not necessarily likely. But it is imaginable, given the president's resilience in Wisc. and Fla., and the potential for a historic turnout amid a racially polarized campaign

    The president's resilience comes down to two crucial but iconoclastic regions: Miami-Dade County and the Milwaukee sububrs. There, the president's approval rating seems to be holding firm or, in Miami-Dade, may be higher than his '16 vote share. The president's resilience in Wisc. and Fla. keeps his Electoral College edge alive, despite falling back farther in Penn and Mich.
    It could grow further with racial polarization or high turnout; either would pad Democratic margins in nationally, but not much in the battlegrounds

    It is important to emphasize that this isn't necessarily enough to make the president a favorite to win. I have no position on that, seriously. And for now, his ratings remain under 50 in states worth 270 electoral votes.

    this analysis, naturally, makes the rounds and sends a whole bunch of people into concerned panic. but not so fast, because there are some naysayers.

    Part Two: Nate Silver, Professional Objection Maker

    the other nate in this story, nate silver, has some objections--subtweeted, of course. he has his own thread, saying:

    The basic notion that Trump has an Electoral College advantage seems right, he obviously did in 2016, but I'm pretty skeptical that we can say much right now about whether it will be larger, smaller or nonexistent in 2020. History would say you should have a loose prior toward it being smaller since the Electoral College advantage is historically not very durable, e.g. it benefited Democrats as recently as 2012. Also you need to look at things probabilistically. If there are several states that look like the potential tipping point state, that's a lot more robust for Democrats than if they have just one path to 270, especially if those states are unique from one another demographically.

    now, because nate silver and nate cohn follow each other, it's not like this subtweeting is exactly subtle.
    cohn fires back with As I emphasize in the final section: '2018 isn't destiny.' There are many ways for this to change. But we can talk about the status quo--as you did in November, based on House results. And on balance, it suggests that today there's a larger gap than '16.

    silver responds in kind with a quote tweet (this is annoyingly common among these two): "The fact that the results of people actually voting in the 2018 midterms implies a rather different takeaway than 2018 polls suggests this line of analysis is not super robust though."

    Part Three: Literally What Is Happening

    at this point we are just in full fledged "twitter pedantry mixed with nerdery of the highest order" mode. finally these two decided to actually respond directly to each other instead of weirdly quote tweeting and making this shit impossible to follow, which goes like this:

    COHN: The results of people actually voting implies about the same takeaway as the 2018 polls
    SILVER: I don't really agree with you, although part of that is that I think it's a mistake, 16 months ahead of the election, to try to pinpoint the exact tipping point state instead of looking at the overall robustness of the map.
    COHN: If you don't agree, I'd be happy to hear evidence to the contrary. I think I've put forward a lot.
    And I do think trying to pin-point it matters, not necessarily for predicting the race but making sense of it and perhaps changing it, as Democrats mull their nominee
    SILVER: Conditional on Democrats winning the popular vote by 3 points, would you bet on Trump winning the Electoral College?

    elsewhere, they go on another tangent, like so:

    COHN: In Arizona/Wisconsin, after imputing uncontsted races, the House result is D+2
    In the US, it's D+7 [1]
    SILVER: There are a lot of ways to impute uncontested results and a lot of them are pretty bad. And if you're doing that, you probably also ought to adjust for incumbency, which generally hurt the D's in swing states since they were running against mostly GOP incumbents.
    COHN: But just about all of them--and certainly the one used here--are better than giving the Democrats 100% of the vote in one-eight of the state. And we also have estimates that account for incumbency, and that nudges things a little bit--to D+2.5. And we have a more sophisticated estimate, seen on the map as the 'result-based' estimate, which attempts to model the presidents approval as a function of House results, while ironing out candidate specific effects (like an outlier in IA-4). All of it lands on the same answer

    unfortunately, this then proceeds to become exponentially harder to follow with the intervention of a third party thread splitter, psephologist dave (nathan) wasserman. wasserman opines, in a quote tweet of his own to silver's "Conditional on Democrats winning the popular vote by 3 points, would you bet on Trump winning the Electoral College?" question, "fwiw I’d lean yes."

    silver, of course, replies with "I'd probably lean no. I certainly think Trump could win an election where he loses the popular vote by 3 points or even slightly more. That's basically what he did in 2016. (He won WI, the tipping-point state, with ~1 point to spare.) But I'm skeptical that he'd be the favorite." he then helpfully clarifies with this long take:

    The reason is that I see this as being a fairly dynamic process. To be honest this is sort of my critique of the @Redistrict/@Nate_Cohn Electoral College analysis; I think they focus on a more static, modal outcome rather than the range of plausible outcomes. The Obama-era coalition was fairly favorable to Democrats in the Electoral College. Obama would likely have WON an Election in 2008 or 2012 in which the popular vote was very close. This wasn't that long ago. But the coalition where the urban & suburban South becomes Democratic, at which point states like TX, AZ, GA, FL, NC have high potential to flip, is also pretty decent in terms of electoral efficiency for Democrats, even if they're often losing the Midwestern swing states.

    Basically, I think Clinton 2016 was probably close (maybe not exactly at) the local minimum for D's. So a coalition that either "turned back the clock" (Biden?) or turboboosted demographic change (Harris/Castro?) could be better than Clinton 16, which was stuck in between. p.s. A lot depends on TX and FL. The best case for pessimism for D Electoral College math is that TX is still a cycle or 2 away from being a tipping-point state, though it will be eventually. And that FL, despite being a state where demographic change should help Ds, is weird.

    silver gets long and slightly less long responses because fuck it. of interest is the first tweet cohn sends, though, "This is fair, in the sense the map can/will obviously change between now and election night '20. It is unfair in a different sense: the piece is plainly describing the president's current standing, and goes to unusual lengths to explain that it can change" because silver responds to that with a point about how weird it is to be surprised that people are taking away a point the article lingers on extensively even as it's supposedly not the takeaway of the article:

    Yeah, I think it's kind of BS to lean really heavily into a particular takeaway in top 80 percent of the article, not to mention the headline/lead/social promotion/etc., and then to introduce the caveats in (literally!!) the 42nd paragraph.

    we then get a series of quote tweets:

    COHN: I think it's kind of BS to not be able to tell the difference between describing the president's standing today, with section heds like "The state of the Electoral College, 2018" and a prediction for the result in 2020, but that's just me
    SILVER: Dude, the headline is literally "Trump’s Electoral College Edge Could Grow in 2020, Rewarding Polarizing Campaign" !!! Of course that's what readers' takeaway is going to be, that it's a prediction about 2020.
    COHN: Of course it could! It already has! That's important! Maybe you don't think it is, because you think it could change--which it could.

    at some point cohn decides to also actually respond to one of these tweets, which brings us to a conclusion, as follows:

    COHN: It could grow! That's really important!
    MATT YGLESIAS: Can I host a beer summit, guys? This is a lot of fighting over a difference in emphasis.
    SILVER: Actually it's about ethics in data journalism. [2]
    COHN: lolol. here's what's not in dispute, afaik: the president's approval rating in the tipping-point states is meaningfully what it is nationwide. If you want to dispute it, go ahead. If you want to say it can change, I agree. If you want to say it's irrelevant, that's preposterous. If you want to say it has 0 effect on p(ec-pv) split, I think you're probably wrong about that. If you want to say it's unethical to disagree with you, that's laughable.

    as of now, there has been no further reply, suggesting that maybe this argument--which, for the record, transpired over nearly two full days of actual time--over basically nothing finally ended.

    Epilogue: Is There Any Moral To Be Had?

    i hope we've learned a few things from this all, namely that:

    • twitter sucks as a discourse platform
    • nerds are gonna nerd
    • the discourse is literally inescapable
    • why
    • stop quote tweeting people when they're literally just going to reply to you, you fucking animals

    i hope you have enjoyed this glimpse into the void, and remember: we are still a fucking year and a half away from any actual election and already having long ass discourse like this that goes nowhere!

    Notes for the Unaffiliated

    • [1]: most of you are probably aware or can infer what this means, but for those of you who are not this is shorthand which refers to the partisan lean of voting in an election; in 2018 for example, the nationwide vote was D+8.6, or 8.6 points in favor of the democratic party (but when accounting for uncontested races, it was around D+7). please however that, in other cases, this same shorthand may refer to the Cook Partisan Voter Index, which compares a given polity's "average Democratic or Republican Party share of the two-party presidential vote in the past two presidential elections to the national average share for those elections", and is thus not comparable to pure vote share.
    • [2]: yes, that is a fucking gamergate reference. and yes, it did apparently woosh nate cohn.
    12 votes