The previous analysis, last Saturday, saw former VP Joe Biden winning all 100,000 simulated elections earning, on average, 364 electoral votes and a mean of 174 for President Donald Trump. Today, some 75 polls later, Biden still maintains the lead, wins them all, but now with 362 to 176 votes, on average.
The little movements we have seen from last Saturday include several states with very few polls (Indiana, Mississippi and Nebraska’s 2nd CD). More substantively, one new poll in Iowa and with three old polls dropping out, changed Biden’s chances from 68% to 83% probability of winning an election held today. (Iowa…go figure!) A new poll added to Maine 2nd CD’s other two polls flips the district from Trump with a 60% probability of winning to Biden with a 60% probability of winning. Finally, three new Texas polls and the loss of two old polls boost Trump’s chances from 78% last Saturday to 85% today.
This race is a bit of a snoozer compared to 2016, when Hillary Clinton’s lead dropped precipitously during the last 10 days of the election:
Here is what it looks like for Biden with less than a week to go to the election (FAQ):
Clearly, if the election was held today, Biden would have a greater than 99.9% probability of winning.
The single most likely outcome from the simulations had Biden winning 357 electoral votes (with a 22% probability). Here is the full distribution of electoral votes [FAQ] from the simulations:
Ten most probable electoral vote outcomes for Biden (full distribution here):
- 357 electoral votes with a 22.04% probability
- 356 electoral votes with a 12.49% probability
- 358 electoral votes with a 10.67% probability
- 351 electoral votes with a 5.40% probability
- 395 electoral votes with a 4.16% probability
- 341 electoral votes with a 3.89% probability
- 340 electoral votes with a 2.96% probability
- 350 electoral votes with a 2.93% probability
- 375 electoral votes with a 2.77% probability
- 360 electoral votes with a 2.48% probability
The smallest electoral college vote going to Biden was 333, but the probability of anything under 340 votes is only 2.6%. Biden would have a 2% chance of receiving over 400 electoral votes.
Summary: After 100,000 simulations:
- Biden wins greater than 99.9%, Trump wins less than 0.1%.
- Average (SE) EC votes for Biden: 361.5 (16.3)
- Average (SE) EC votes for Trump: 176.5 (16.3)
- Median (95% CI) EC votes for Biden: 357 (339, 398)
- Median (95% CI) EC votes for Trump: 181 (140, 199)
Each column of this table shows the electoral vote total aggregated by different criteria for the probability of winning a state (Safe=100%, Strong=90%+, Leans=60%+, Weak=50%+):
|Threshold||Safe||+ Strong||+ Leans||+ Weak|
This table summarizes results by state. Click on the poll count to see the individual polls included for the state.
|6||2||Votes||polls||Votes||Biden||Trump||% wins||% wins|
* An older poll was used (i.e. no recent polls exist).
Details of the methods are given in the FAQ.
The most recent analysis in this match-up can be found from this page.
Elijah SFA McDotcom spews:
I keep digging into the crosstabs of many of the polls in the battleground states – Geez there are a lot more of them this year compared to 2016 – to look for evidence that key demographic predictors are not being weighted and so far, mostly I’m not seeing it.
Perhaps this year we’ll discover some new characteristic that is highly correlative and either over or under sampled. That seems to be the implication of all the Republican “don’t believe the polls” messaging. Some secret demographic sauce that helps Republicans over perform.
But the evidence was there in 2016. Late deciding voters were disproportionately less-than-college educated. They broke hard for Trump in the final weeks. They were under sampled in a number of key state surveys. And no correction was applied. But the universe of late deciding voters is much smaller this year. And so far every state poll I’ve dug into is weighting their results to more closely reflect the educational cross section of historic turnout.
If there is any group that is turning out in historically larger than expected numbers this year in some states it is probably younger voters, under age 50. I’m especially watching AZ, TX, and GA. And these are states where the GOP appears to already have lost some significant support among suburban female voters and among over-65. They are also states where COVID is blowing up, where the poll margins are narrow, but where the trend in the closing days is moving in the wrong direction for the GOP. And, though they may be under represented in polls, I wonder if some of these younger voters are also late deciders this year. And that might go either way.
COVID, postal rat-fuckery, poll closures, activist Republican judges, and armed Republican insurrectionists attacking polling places all make this an unusually complicated election to analyze. And I doubt the counting will be over before the end of the week. Trump will have lost, the Senate will have swung, and probably Florida will have flipped. But Trump will sue. And Republican governors and canvassing boards in states facing early canvassing deadlines like Arizona, Texas, Ohio, Florida, Georgia will seek to muddy the outcomes and force a court to intervene if the results are close enough. Think Florida 2000 multiplied by ten. Only this time they won’t be fighting to save Trump, but to save the Senate.
Thomas Jefferson spews:
I’m just looking forward to the end of the junk political mail, the asinine TV and radio ads and bloviating candidates and their surrogates.
You asked for evidence that key demographic predictors are not being weighted?
Perhaps the black vote should be underweighted if the black turnout isn’t there. Although SNL assures us that Black People Vote Late!
The above is satire. Whatever happens on Tuesday, happens on Tuesday.
But I wanna go play disc golf! spews:
The most recent MN poll included in this analysis is through the 27th, and has Biden up 10.4 points with MOE < 4.
Darryl's poll average has Biden up 8.4 in MN.
Four days before the election, Biden is visiting Minnesota.
[Update: Makes more sense in that on the same day Biden also is stopping in IA and WI.
Not quite my Emily Litella moment.]
Elijah SFA McDotcom spews:
You continue to indicate that you don’t understand how the math works.
Less-than-college educated voters were under sampled by state level polls in key battleground states in 2016. It’s believed that this is because college educated voters are more comfortable and perhaps even more enthusiastic about answering survey questions about a political campaign. The under-sample could have been corrected by weighting the responses differently according to educational level so that the resulting percentages more honestly reflected the historical turnout of voters in these various educational categories.
Trump (Bannon/Lewandoski) crafted a strategy that specifically targeted non-college educated whites in MI, WI, and PA, among others. They used a simple observation about available numbers of non-participating less educated whites and reasoned that the right kind of blanket racist appeal projected through tens of thousands of fake social media accounts (Parscale/CA) would be a cheap way to activate enough (barely) of those voters. And the state level polls missed it (especially in WI) because they didn’t weight for education.
State level polls in NC and VA are certainly taking race into account in weighting their responses. There are 30 NC polls included in this analysis. And they are taking ongoing changes in racial demographics of each of those states into account. Comparing early voting turnout rates to final results is of course an apples to oranges comparison, pointed out by Schweikart’s careful attention to late-early returns by Rapepublicans, who all told us they wouldn’t vote by mail. Worse, comparing racial turnout eight or twelve years ago, to racial turnout today misses important changes in demographics that have taken place.
Much of “whatever happens on Tuesday” has already happened.
Some of it happened in the last few weeks. And some of it happened in the last few years. And some of it happened four years ago when the Republican Party astonished the world by nominating a candidate who would go on to confess a long history of violent sexual assaults.