The Uneven Split of Toss-ups, October 21, 2020

Since the beginning of this website, I realized comparing forecasts was going to be problematic because of the toss-up vote. It looked like the toss up votes could make a real difference in who won the election.

A quick remedy was to give each party half of the toss-up votes. It wasn't realistic, because states can not divide their votes. It is winner take all.

It would be nice just to leave all the forecasts exactly as they are presented, but the toss-ups really act as holes in the forecast. I've decided instead of a 50-50 split, I will use an uneven split based on the existing consensus of which states might lean Democratic or Republican. Since we are 12 days from the election, I don't think this consensus will change. Forecasts which consider AZ (11), FL (29) and NC (15) as toss-ups, will be added to Biden's total. Forecasts which consider Georgia (16), IA (6), OH (18) and ME2 (1) as toss-ups will be added to Trump's total.

The 270 to Win consensus forecast has 290 electoral votes for Biden without the uneven split of toss-ups. The split adds 44 EV's to Biden's count for a total of 334 electoral votes for Biden. So Trump's total goes from 41 electoral votes to 204 EV's. In many cases, the differences between an even split and uneven one are small. In the case that Texas (38) is considered a toss-up, based on the existing consensus, I consider it to lean towards Trump.

Of course, there is another interpretation. The 270 to Win forecasted 290 EV's for Biden, plus there's another 85 unforecasted toss-up votes, which could all go to Biden or Trump. So, with this interpretation, Biden could get on election day (week?) anywhere from 290 to 375 EV's, and the forecast is still correct as stated. So, the 334 electoral vote is a mid-point estimate, based on the assumption that Biden will not get all the toss-ups and the 375 EV's would be a maximum estimate.

With this uneven split, a Biden victory with 334 EV's is likely in 11 of our 17 poll based forecasts (65%). Another two forecasts, Princeton Election Consortium and JHK forecasts predict 335 EV's as they consider ME2 as a lean Democrat. There are four more estimates above 335 EV's, and I would not really rule any of them out. As I noted in my last post, there is a "Rust Belt" dependence among states, and if Georgia goes to Biden, so might Iowa and Ohio. Or the entire rust belt including Michigan, Minnesota and Wisconsin can be won by Trump.

The broad consensus can be wrong, as in 2016. I will issue an updated forecast in a couple of days.

David Lord

Oct 21, 2020

270 to Win Forecast, with Florida and North Carolina assumed to be lean Democrat states, and Georgia, Iowa and Ohio to be lean Republican states

Click the map to create your own at 270toWin.com