US election minus two days

Biden is the clear favourite to win the US election, but there’s still a plausible chance for Trump, mainly due to a five-point Biden lead in Pennsylvania.

Guest post by Adrian Beaumont, who joins us from time to time to provide commentary on elections internationally. Adrian is an honorary associate at the University of Melbourne. His work on electoral matters for The Conversation can be found here, and his own website is here.

This is a slightly modified version of an article I had published for The Conversation earlier Monday.

Two days before Wednesday’s US election (AEDT), the FiveThirtyEight national aggregate gives Joe Biden an 8.6% lead over Donald Trump (52.0% to 43.4%). Biden’s lead has decreased by 0.3% since last Thursday. In the key states, Biden leads by 8.3% in Wisconsin, 8.2% in Michigan, 4.8% in Pennsylvania, 3.1% in Arizona and 2.2% in Florida.

Biden’s lead in Pennsylvania is almost four points below his national lead, and that gives Trump hope of pulling off an Electoral College/popular vote split, as occurred at the 2016 election. Pennsylvania is the most likely “tipping-point” state that could put either Trump or Biden over the magic 270 Electoral Votes.

If Biden loses Pennsylvania, but wins Michigan, Wisconsin and Arizona, he would have 269 Electoral Votes, one short of 270. Either Maine’s or Nebraska’s second Congressional District could in that scenario give Biden the narrowest of Electoral College wins. These states award one Electoral Vote to the winner of each of their districts, and two to the statewide winner. All other states are winner-takes-all.

In 2016, Hillary Clinton lost the tipping-point state (Wisconsin) by 0.8%, while winning the popular vote by 2.1% – a difference between the tipping-point and popular vote of 2.9%.

Analyst Nate Silver says while Trump can plausibly win, he would need the polls to be wrong by far more than in 2016. At this stage in 2016, the FiveThirtyEight forecast gave Trump a 35% chance; he currently has just a 10% chance. Trump only has a 3% chance to win the popular vote.

Trump had one very good poll result from a high-quality pollster: a Selzer Iowa poll gave him a seven-point lead in that state. But most high-quality polls have been far better for Biden: Siena polls for The New York Times gave Biden six-point leads in Arizona and Pennsylvania, a three-point lead in Florida and an 11-point lead in Wisconsin.

In FiveThirtyEight aggregates, Biden leads by 2.0% in North Carolina and 1.5% in Georgia. He trails by 0.3% in Ohio, 1.2% in Texas and 1.7% in Iowa. If Biden won all these states, he would win over 400 Electoral Votes. Florida is now in this group of states when it had previously been better for Biden.

Trump’s net job approval ratings have jumped two points since last Thursday. In the FiveThirtyEight aggregate, his net approval with all polls is -8.5%, and -7.0% with polls of likely or registered voters. The RealClearPolitics average has Biden’s net favourability at +7, while Trump’s is -13.

I wrote on October 22 that there are two key measures where Biden is doing far better than Clinton. First, Biden is over 50% in national polls, which Clinton never achieved. Second, he has a net positive favourability rating, whereas both Clinton and Trump were very unpopular in 2016.

The US election results will come through on Wednesday from 10am AEDT. You can read my wrap of when polls close in the key states and results are expected. A key early results state is Florida; most polls close at 11am AEDT, but the very right-wing Panhandle closes an hour later.

In the FiveThirtyEight Classic Senate forecast, Democrats now have a 79% chance to win control. The most likely outcome is a 52-48 Democratic majority. The 80% confidence range is 48 to 56 Democratic seats. All these measures are unchanged since Thursday.

206 comments on “US election minus two days”

Comments Page 4 of 5
1 … 3 4 5

  3. Trump and his allies are desperately trying to suppress postal and pre poll votes.

    Says all we need to know about who those votes will favour.

  4. Diogenes, I think he was worse than even people who knew he would be bad thought he would be. (if that sentence makes sense) You’re forgiven. 🙂

  6. Diogenes:

    Tuesday, November 3, 2020 at 8:37 pm

    [‘Dixville Notch, NH, voted Biden 5-0.
    I’m calling it!!’]

    Yeah, but in the hamlet of Millsfield, 12 miles to the south of Dixville Notch, Trump scored 16 to Biden’s five. It’s close to being over though.

  11. Soon, people across the US will wake up and some who haven’t voted yet will vote. Thankfully, most of this will be overnight for us, so sleep will spare us a lot of anxiety and election day tea leaf-reading.

    Anyway, I don’t really know how to post an image here, so just imagine I posted a screenshot of the “Dawn of The Final Day” screen from Legend of Zelda: Majora’s Mask.

  13. Rational Leftist @ #161 Tuesday, November 3rd, 2020 – 9:13 pm

    Soon, people across the US will wake up and some who haven’t voted yet will vote. Thankfully, most of this will be overnight for us, so sleep will spare us a lot of anxiety and election day tea leaf-reading.

    Anyway, I don’t really know how to post an image here, so just imagine I posted a screenshot of the “Dawn of The Final Day” screen from Legend of Zelda: Majora’s Mask.

  16. 3z,

    For all its flaws, our system of democracy is probably the best in the world. We do live in a lucky country and we are surrounded by water so we can keep the riff raff out.

  18. Thanks for that image C@tmomma.

    Honestly, I’d be tempted to say I’d prefer the Moon crashing down than another four years of Trump!

  21. It is highly likely Trump will lead in the vote count on election night, but the margin will narrow to Biden as the early voting is counted.

    I can see a few people having kittens tomorrow as they watch the on-election-night counting.

  22. Following on from Murray and his fellow travelers attempting to do a job on Palaszczuk, not to be outdone, they’re back again trying to do the same to Biden. Guests are Hanson and Bernardi. It really is pathetic.

  23. Two states that will report results early are Florida and North Carolina. If they’re going bad for Trump, it’s all over. Conversely, if Biden’s not doing well, it’s far from over.

  24. Rational Leftist @ #168 Tuesday, November 3rd, 2020 – 9:41 pm

    Thanks for that image C@tmomma.

    Honestly, I’d be tempted to say I’d prefer the Moon crashing down than another four years of Trump!

    That’s okay, RL, it’s one of my son’s all time favourite games as well. He has played every one in the series, though I remember that one, the moon and the beautiful music to this day.

    Sorry I didn’t choose an image from the game itself but I chose that image because it kind of reminded me of people lining up to vote. 🙂

  25. Mavis @ #172 Tuesday, November 3rd, 2020 – 9:54 pm

    Following on from Murray and his fellow travelers attempting to do a job on Palaszczuk, not to be outdone, they’re back again trying to do the same to Biden. Guests are Hanson and Bernardi. It really is pathetic.

    They are beginning to suffer from the law of diminishing returns. People now know that they are not giving an honest opinion but a biased one.

  26. From Politico:

    The final polls before Election Day show Joe Biden ahead in enough swing states to win.

    Polls favoring Biden: Mich. +8, Minn. +9.4, Nev. +4.8, N.H. +11.1, Pa. +4.8, Wis. +8.2.

    Toss-up polls: Ariz. +2.6, Fla. +2.5, Ga. +1, Iowa +1.4, N.C. +1.9, Ohio +0.7, Texas +1.

  28. Josh Wingrove@josh_wingrove·
    5h
    In Michigan, Trump gives a shout-out to Lil Pump, calling him “Little Pimp.”

    In some ways Little Pimp is better than Lil Pump.

  29. Late Riser:

    The problem with using statistics to predict a binary outcome is that someone always wins the lottery. Once the outcome is decided the chances cease to matter. A one in ten chance can happen. Maybe there’s a talented cartoonist out there willing to draw a picture of Russian roulette.

    The aim is for your probabilities to be “well-calibrated” over many events, meaning that your 90% calls come up 9 times in 10, your 60% calls 6 times in 10 and so on.

  37. Off to bed….will listen through the night on my pillow radio.
    Got a really bad feeling about how this turns out.
    Dr Blind Freddie from the Tommy Institute, Dept. of the Bleeding Obvious, hasn’t exactly filled me with confidence for a Democrat win.

  39. caf @ #179 Tuesday, November 3rd, 2020 – 9:26 pm

    Late Riser:

    The problem with using statistics to predict a binary outcome is that someone always wins the lottery. Once the outcome is decided the chances cease to matter. A one in ten chance can happen. Maybe there’s a talented cartoonist out there willing to draw a picture of Russian roulette.

    The aim is for your probabilities to be “well-calibrated” over many events, meaning that your 90% calls come up 9 times in 10, your 60% calls 6 times in 10 and so on.

    All true, but there is only one event. 🙁

  42. All true, but there is only one event.

    There’s a presidential election every 4 years – but also the probabilistic models tend to be based on forecasting individual state win probabilities, so you get a bunch of results in a single year and for some of the forecasts, hundreds of house district forecasts every 2 years. There should be enough data there to estimate reasonably well how well-calibrated the methodology is.

  44. Got a really bad feeling about how this turns out.
    Dr Blind Freddie from the Tommy Institute, Dept. of the Bleeding Obvious, hasn’t exactly filled me with confidence for a Democrat win.

    mundo, you really are a supreme bullshitter.
    Blind Freddie is now Doctor Blind Freddie?
    You want Trump to win, that’s the bottom line, and I’m going to judge you on that.
    He won’t, and neither will you.

  45. caf @ #192 Tuesday, November 3rd, 2020 – 10:41 pm

    All true, but there is only one event.

    There’s a presidential election every 4 years – but also the probabilistic models tend to be based on forecasting individual state win probabilities, so you get a bunch of results in a single year and for some of the forecasts, hundreds of house district forecasts every 2 years. There should be enough data there to estimate reasonably well how well-calibrated the methodology is.

    Yes, but I think we’re writing about different perspectives. I accept that the models are calibrated “reasonably well “. The 538 model, for example, gives us an overall probability based on tens of thousands of scenarios, but there will be only one outcome for tomorrow. The die is being rolled just once. There is no “best of three”. That’s where anxiety lives.

  50. a r,

    Actually, the “calming” 538 video you posted gets into the polling error. It gives hope. But what I’d like to know is, what is the 10% level for Biden? It has to be well over 400 EC votes, compared with 270 for Trump.

Comments Page 4 of 5
1 … 3 4 5

Leave a Reply

Your email address will not be published. Required fields are marked *