If Trump needs enthusiasm to win FL, things aren’t looking good
All figures are from this google spreadsheet with links to original sources from polls and the Early Vote Election Project by UF Michael McDonald
The state of the race : Biden +2.4%
The state of the race in FL with just 24 hours before polls close on election night at 20:00 EST can be accurately described as tight. As of Nov 1st, the last day of early in-person voting, more than 62% of registered voters (RV) have already cast their ballot either by mail or early in-person, and the margin of those who have already voted is +3.7% Biden-Trump. The fivethirtheight adjusted polling average is +2.4 and predicts a final vote of +2.5%.
Criteria for a Trump victory in FL : over-perform Dem in turnout by >8%
As explained in an earlier post, Trump needs to over-perform Biden in turnout between Rep and Dem by appx +8%. This isn’t unimaginable, since that’s how he won in 2016 where he over-performed turnout by +6.8% by party registration ID and generally the characteristics of Rep voters overlap with those who have more regular intrinsic voting habits and also face fewer barriers to voting.
In a highly polarized election, both sides are potentially more energized so its reasonable to expect a higher turnout from Dep and Rep. In 2016 Dem turnout was 74.4%. Given a modest increase in total Dem turnout improvement to say 77%, that would imply the following turnout scenarios for a Biden win, and a Tie. Any higher turnout for Rep would mean a Trump victory.
On election day, Trump would need to turnout at least an additional 8.5% of RV in FL, or an additional +24% of registered Rep which have not already cast their ballot for a total turnout close to +90% of registered Rep. That would be a +9% increase in Rep turnout from 2016 level of 81%. The relative turnout by party of early voting so far is about even, at 66% Dem, 66% Rep and 52% Ind.
That’s a substantial amount of total enthusiasm boost required in order to turnout new voters that wouldn’t show up otherwise, but it’s not impossible. The question is — is there any evidence of Rep over-performing Dem, or Ind for that matter, in enthusiasm?
Partisan voting preference : mail-in vs in-person
One narrative that is popular in the media now to explain how it’s possible to see a sudden jump in Trump support on election day is that polls have been showing a consistent preference among Trump supporters of in-person vs mail-in, and also, more critically that they prefer to wait until Tue Nov-3 instead of early in-person.
Let’s do the math though.
A recent NYT-Sienna college survey of likely voters during the last week Oct 27–31 found 32% Dep, 34% Rep and 31% Ind, so based on what we know already that’s not enough for a Trump victory because he’s only ahead by +2% by party ID, but let’s continue and revisit this question as it’s the most important unknown factor.
Of those Rep who responded to the survey, 63% said they have already voted, and of those who had not yet voted 5% intended to return their mail-in ballot, 31% early in-person and 63% prefer to wait till Tue. Keep in mind though this was a blended survey taken 27–31 while early in-person was still open, it’s closed now so presumably those 36% who intended to either return their ballot or go to an early in-person polling location have done so already, so the only remaining unknown is the 63% who plan to wait until Tue.
63% sounds like a lot, until you do the math to estimate how much that is vs how much is required for a Trump victory. It works out to just 6.2% of registered voters, 17% of registered Rep for a total Rep turnout of 83%. Still a +2% turnout improvement from 2016, but short of the +24% required for a tie. So how do we arrive at +17% of registered Rep from that 63% in the survey?
17% = 14.4 mil reg voters / 5.2 mil reg Rep * 6.2%
6.2% = 78% total turnout * 34% Rep in the survey * 37% Rep not yet voted * 63% waiting until election day
The 78% total turnout is an estimate reported by Michael McDonald from the University of Florida who is championing the Early Vote Statistics project to compile early voting data. His turnout prediction is based on information he has received from variety of sources. I’m not sure anyone else is going to be in a better position to predict turnout.
The major uncertainty in this calculation is how well the poll captured the right composition of likely voters = people who actually vote one way or another. So in order for there to be a surprise upset of +24% of Rep instead of the predicted +17%, there would need to somehow be invisible Trump supporters that the poll systematically missed.
Looking for evidence from early voting of who is winning the turnout battle
So can we learn anything from the trends of early voting this past week? Is there any evidence either of Rep out-performing Dep on enthusiasm, or preferring to vote closer to election day vs early?
When it comes to mail-in ballots, overall there is a clear preference in Dem and also Ind and that is consistent with what the surveys have found. We can also see that the few Rep who decided to vote by mail did so early on and at this point if they haven’t already turned in their ballots, based on the trends and the survey it is unlikely that they will do so. As of now there are still ~ 23% of ballots unreturned for Rep, and about the same for Dem.
This 23% could be misleading though. More likely is to expect a similar low rate of returned ballots by all 3 parties as shown in this trend. Many of these ballots will never be returned because either they went to an inactive voter who decided to stay home (very few at this point), or the voter switched their voting preference and decided to surrender their ballot and either vote early in-person or vote on election day, which you can do. If you don’t take this assumption you end up with weird conclusions such as a total turnout close to or exceeding 100%.
How about in-person voting though? Supposedly the primary partisan divide was that Rep preferred to vote in-person because they have been receiving messages to be suspicious of mail-in ballots, and are less concerned about catching the coronavirus. Evidence of that partisan preference has been reflected in the surveys and also early voting behavior.
Now the trend is reversed, as expected and Rep overall show up in larger numbers than Dep for early in-person, but the concerning sign for the Trump campaign is that this trend was short lived. The most enthusiastic Trump supporters came out early, but as the week went on those numbers have been gradually dwindling from Rep, while the numbers from Dem and Ind has been steadily increasing as election day approaches.
Now think about that for a moment …
If you are an enthusiastic Trump supporter brought in and energized where previously you stayed away. Are you so exciting to keep seeing the news everyday of a tight election, seeing and hearing about people going to the polls, and yet you are so relaxed to wait it out until Tue? Why not go down and get it over with this past Sat? You can go in-person to a normal ballot machine and don’t have to worry about whether your ballot will be lost in the mail, etc.. and you can arrange it at a time that is convenient for you on a weekend, as opposed to taking off work on a Tue, also when you may be concerned of protestors and whatever at the polls.
If Trump does manage to pull off a surprise +24% of reg Rep turnout on election day, it would not be enthusiasm, but nostalgia for some unexplainable preference for voting on a Tue instead of a Sat.
We will see..