It is election day! I’ve added a couple dozen more polls since The yesterday analysis, but the results have not changed much. Yesterday, Vice President Joe Biden was leading President Donald Trump with an average of 362 to 176 electoral votes.
Today, I tightened up the “current poll” window to 10 days. This has the effect of adding uncertainty in under-polled states like Washington and Oregon. But doing this should better capture late trends in hotly contested states.
Today, after 100,000 simulated elections, Biden wins 100,000 times. Biden received (on average) 360 to Trump’s 178 electoral votes. The election is today and this model predicts Biden will win with near certainty. The most likely outcome (14.6% probability) is 351 electoral votes. Biden’s worst outcome in the simulated elections was 308 electoral votes, suggesting that Biden has almost no chance of an outcome below 308 votes. The highest electoral vote total was 418.
Here is the distribution of electoral votes [FAQ] from the simulations:
The trajectory of this race over the past year can be seen from a series of elections simulated every week including polls from the preceding ten days (FAQ).
Ten most probable electoral vote outcomes for Biden (full distribution here):
- 351 electoral votes with a 14.63% probability
- 350 electoral votes with a 8.32% probability
- 352 electoral votes with a 6.84% probability
- 389 electoral votes with a 4.08% probability
- 357 electoral votes with a 3.95% probability
- 335 electoral votes with a 3.62% probability
- 354 electoral votes with a 3.24% probability
- 356 electoral votes with a 3.11% probability
- 353 electoral votes with a 2.80% probability
- 336 electoral votes with a 2.16% probability
After 100,000 simulations:
- Biden wins greater than 99.9%, Trump wins less than 0.1%.
- Average (SE) EC votes for Biden: 359.6 (18.9)
- Average (SE) EC votes for Trump: 178.4 (18.9)
- Median (95% CI) EC votes for Biden: 353 (334, 403)
- Median (95% CI) EC votes for Trump: 185 (135, 204)
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.
Roger Rabbit spews:
The real margin of error in statistical analyses is this:
“For the fifth day in a row, the US Postal Service moved fewer ballots on-time in critical battleground states than it did in the previous day, according to new court filings.”
“Robocalls are falsely telling Michigan residents to stay home on Election Day and vote Wednesday instead, Michigan Attorney General Dana Nessel says. The robocalls to residents in Flint, Michigan, tell voters to cast ballots Wednesday due to long lines at election polling places, Nessel says. ‘Obviously this is FALSE and an effort to suppress the vote,’ Nessel tweeted.”
How do you statistically measure the legitimate votes that aren’t counted because of crap like this? Here’s my suggestion: Make it against the law, then count how many Republicans are behind bars, estimate the number of voters in each state they kept from voting, and that gives you the approximate numbers.
Biden up by 7 in PA. So the reason so say something like this on Election Day is…
I get it, he’s up 7 on average, and Marist has him up 5 and over 50% – to me that sounds solid for ol’ Joe. It’s just a curious thing for his campaign manager to say with that big a lead in PA and zero days until the election.
It may be an Election Day full of curious tweets and recorded comments from people who are privy to the inside numbers:
Re @ 1
It’s debatable whether your spew should be include in this thread.
Your spew approaches irrelevance if all of the D mail-in votes were sent in early, doesn’t it?
Looks like I’ll be dropping my ballot off at the Coupeville location. Haven’t opened it to vote yet.
Roger Rabbit spews:
@4 The fact you claim #1’s commentary on statistical analysis is of debatable relevance in a thread about statistical analysis doesn’t make it debatable. However, if you wish to appeal to higher authority, write your complaint here _ and send it here [ ].
N in Seattle spews:
I find it interesting that there isn’t a single state in the “Weak” category … for either side. Seems like the relative positions are pretty well fixed.
Unless, that is, turnout and/or demographic projections are off-base. Which, I think, would bode better for the D side than the R, as the unanticipated differences would surely be toward higher turnout and younger voters. It would be sweet as honey if TX or OH squeezed into the Democratic column.
Richard Pope spews:
Trump likely takes FL GA NC, but he is still 22 votes short if rest of your calls are accurate.