Princeton Election Consortium

A first draft of electoral history. Since 2004

Jul 13: Biden 380 EV (D+6.4% from toss-up), Senate 53 D, 47 R (D+5.6%), House control D+8.0%
Moneyball states: President AK AR IA, Senate MT KS ME, Legislatures KS TX NC

Reanimating… Meta-margin Biden+4.6%, most effective GOTV states NV/AZ/FL/NM/NH/MI/NC

May 24th, 2020, 4:15pm by Sam Wang


Just a quiet Sunday, reanimating some code…minus some problems from 2016.

As of today, survey support for Biden is 4.6% above threshold in terms of Electoral College mechanisms. Median snapshot of current polls: Biden 343, Trump 195 EV.

Today’s poll-based modal outcome (i.e. the single most probable combination):

Here’s what it looks like if just 2% of voters flip from Biden to Trump uniformly:

Not exactly a landslide situation yet.

As always with the Electoral College, voter power varies tremendously by state. Here is per-voter influence over the Presidential outcome. In these units, my “jerseyvote” is worth about 0.5.

This valuation is key to our mission in 2020, which is to help readers focus their efforts where they will be most effective. In the coming weeks, we’ll expand the valuation from its past focus (Presidential, Senate, and House) to include state-level races, including legislative races. These last races are especially important, since redistricting will happen in 2021 – and redrawing the political playing field for a decade.

Tags: 2020 Election · President

14 Comments so far ↓

  • James McDonald

    Any chance you could provide a map allowing users to add/subtract uniform percentages and see the effect as a re-shaded map and a histogram of expected EV results?

    You did it for us for +2 Trump, but what happens with +2 Biden, or +1.4 Trump, or whatever…

    (Or maybe that option is already there and I was too clueless to figure it out.)

    Thanks

  • Brad Davis

    Sam, I saw an interesting article today (https://www.politico.com/news/2020/05/26/2020-election-democrats-281470) that said that it’s possible (probable?) that there will be a major economic turn around enough before the November elections that Trump will be able to tout as his successes, and that this will make it much harder for Biden to win, because people still trust Trump more on the economy than Biden.
    I’m not convinced that the economy will be in a turn around mode yet. I think as states re-open, there will be some new outbreaks, especially when we get into the fall (September, October) that will also likely cause a second wave. But if we assume it’s true that the economy will rebound significantly by November, is there anyway we can look at historical data to get an understanding of how much this might sway voters? How much does the economy that they are experiencing in that moment vs recent history affect their voting decisions?

  • A New Jersey Farmer

    Hello Sam. I took your advice after 2016 and got involved in local and state politics, and read the posts about your gerrymandering project.

    I clicked on the links for the data, but I don’t see updated Pollster numbers. Is there someplace else I should be looking?

    Thank you.

  • Partha Neogy

    A 2% shift in the popular vote produces ~ 20% shift the the electoral college vote. Apart from the other things that are wrong with it, this amplification of the already knife-edge sensitivity of the election process would seem to add to its overall instability, uncertainty and vulnerability to manipulations.

    • James McDonald

      That’s not inherently due to the EC.

      If you’re looking for a majority in a very large sample divided near 50/50, even small uncertainties in the raw ratio will amplify to large variations in the expected winner.

      But the EC does make thing worse by concentrating the controlling variation into a few small states, among other things enhancing the ability of corruption in two or three states to swing the outcome.

  • Matthew J. McIrvin

    Resist the temptation to calculate a probability, and you’ll be fine.

  • Adam

    While I didn’t feel like posting it in the aftermath of the 2016 election, I did run the figures for which election predictions were most reliable, and in fact, despite your overconfidence in the overall outcome of the election, you were actually a very close second place to 538’s predictions on the state level.

    That is, your per-state predictions and certainty levels scored better than anyone else I looked at, using a proper scoring function. For that reason I’d definitely still be interested is seeing your predictions, per-state if nothing else.

    I think you could provide a valuable contribution. Even after four years, I haven’t found myself able to trust “black box” pollsters. I like understanding what I’m being given.

  • Bill Sternberg

    The new color choices for the EV graphs make them very difficult for my colorblind eyes to look at. The dividing line between “Trump Wins” and “Biden Wins” becomes effectively invisible in the graph depicting a 2% flip towards Trump. It’s just lost in all the data. I hope you will consider going back to the old colors.

    • Joe B.

      @ Bill:

      “The new color choices for the EV graphs make them very difficult for my colorblind eyes to look at.”

      True. However, the dividing line is at the same place in both graphs. Should make it pretty simple to “see” it’s position in the second graph, even if you’re color-blind.

  • Paul

    I have a question. In the past you had highlighted specific races, Senate and House in which donating to a Democrat would have the greatest impact. Will you be providing that information again this election cycle and, if yes, approximately when?
    Despite the ‘.ca’ on my address I am a US citizen (duel actually). Thanks

Leave a Comment