Final Map & Thoughts (Patreon)

BREAKING NEWS!  Texas Republicans lost in federal court in their efforts to toss out 120,000+ ballots in Harris County, after having lost in state court over the weekend.  Great news for Texas (and democracy)!

----------------------

Consistent with my last post, this is the map I'm taking with me into Election Night.  The changes since last week are to move Texas (38 EV) and ME-2 (1 EV) into the "uncertain" zone.  Note that I am NOT using unallocated "uncertain" states to mean "toss-up" states -- Texas is definitely not a toss-up state; Trump is narrowly ahead.

Instead, what I've done is to back out one standard deviation's worth of certainty from the best aggregator models (538 and The Economist's) to figure out what might surprise us on Election Day.

So what does that mean?  Here are my thoughts:

1.  You're not wrong to be (a little bit) nervous about the election.  If you include only states where Biden has leads outside the margin of error, you wind up with 252 EV, which is 18 shy of what Biden needs to win.  So yeah, it's not "in the bag" at this point.

2.  That being said, if you're betting on the election's outcome, it's obviously way better to be in Biden's position than Trump's.  As of today, 538 gives Biden a 90% chance of victory, while the Economist pegs it at 96%. 

To explain out what that means, let's look at Gelman et al.'s September 2020 peer-reviewed academic research, "Information, incentives, and goals in election forecasts" in the Journal of Judgment and Decision Making, Vol. 15, No. 5, September 2020, pp. 863–880.

Given the above assumptions, suppose the forecast of Biden’s national vote share is 54% with a standard deviation of 2%. Then the probability that Biden wins can be calculated using the normal cumulative distribution function: Φ( (0.54 − 0.517)/0.02) = 0.875. Now suppose that our popular vote forecast is off by half of a percentage point. Given all our uncertainties, it would seem too strong to claim we could forecast to that precision anyway. If we bump Biden’s predicted two-party vote down to 53.5%, his win probability drops to Φ((0.535 − 0.517)/0.02) = 0.816. Thus, a shift of 0.5% in Biden’s expected vote share corresponds to a change of 6 percentage points in his probability of winning. Conversely, a change in 1% of win probability corresponds to a 0.1% percentage point share of the two-party vote. There is no conceivable way to pin down public opinion to a one-tenth of a percentage point, which suggests that, not only is it meaningless to report win probabilities to the nearest tenth of a percentage point, it’s not even informative to present that last digit of the percentage.

In other words:  there is no meaningful distinction between "90%" and "96%" (and the authors would not want you to draw such distinctions.

3.  Remember that polling errors are math errors, which means that statistically speaking, it's just as likely that they will benefit Biden as hurt him.  Here's what that means, concretely.  Are you nervous that Biden will lose Wisconsin, damn the numbers?  You should know that Biden's lead in Wisconsin (8.1 percent) is larger than Trump's lead in South Carolina (7.9 points).

4.  There has been no significant movement ("tightening") of the polls in the past five days.

5.  At the end of the day, all the fancy math and academic research confirms the conventional wisdom:  if Biden wins Pennsylvania, Ohio, or Florida, this election is literally over.  If Biden wins North Carolina, Arizona, or Georgia, this election is probably over, even if he loses Pennsylvania.

So, this is what I'll be doing Tuesday night:  taking my map and watching to see how the actual results match up with what we expect.  The first full results come in at 7 pm Eastern, when we get poll closings in Georgia, South Carolina, Virginia, and Vermont.  Half an hour later, polls close in North Carolina, Ohio, and West Virginia.

That means that right before our livestream begins at 8 pm, we'll have hard data.  If the polls are right, it should look like this:

Called for Trump:  West Virginia (3 EV)

Called for Biden:  Vermont (3 EV)

Too Early to Call:  South Carolina (9 EV) and probably Virginia (13 EV)

Too Close to Call:  Ohio (20 EV), Georgia (16 EV) and North Carolina (15 EV).

Ultimately that means I'll be looking at South Carolina right as we prepare to go live.  If South Carolina is called for Trump before 8 pm, he's probably slightly outperforming the polls.  On the other hand, if it is "too close to call," then probably Biden is slightly outperforming the polls.

If there is any characterization of the races in Ohio, Georgia, or North Carolina -- that is, if they are described as "leaning X, but too early to call," that suggests a significantly good night for whomever X is.  I don't think this is a likely outcome, which is why I'm looking at South Carolina first.

I wouldn't be surprised if Virginia is called for Biden before 8 and I wouldn't read anything into that (other than a massive demographic shift in that state as part of a long-term trend).

See you tomorrow at 8 pm!