WARNING:  Right brained readers of this blog may want to skip today’s post as it is the perfect example of what happens at the end of an election cycle if you have three too many degrees in political science.  And had the good fortune of studying under two of the most noted behavioral social scientists in America.

Last night I had a dream only a political scientist who has studied voter behavior since 1968 could have.  I was back at the University of Virginia, except it was not 1968 when I penned my first electoral analysis in a term paper titled, “George Wallace Is Alive and Living in Boston.”  It was November of 2020.  In the dream my professor John Ellwood asks his students, “After 2016, is there anyone in this class stupid enough to think he or she can predict who will carry Florida this year?  If so, now is the time to make your case.”  I timidly raised my hand to which Dr. Elwood replies, “Why am I not surprised?”  Below is the analysis I would have shared with him. (NOTE:  Dr. Ellwood, who earned his Ph.D. from Johns Hopkins University, wrote the recommendation that was the sole reason I was accepted to his alma mater despite a less than exemplary undergraduate academic performance at UVA.)

One of the advantages of the graduate program at Johns Hopkins University was the opportunity to take classes outside one’s own department.  At the time, the political science department at JHU was among the best in political theory, but short on faculty with experience in empirical research.  To make up for that deficiency, I took a class in game theory with James Coleman, best known for his study of the learning gap resulting from years of “separate but equal” policies in public education, and created an independent study with psychology professor Warren Torgerson whose research focused on typing abnormal behavior (a skill I found most helpful for my dissertation on voting blocs in the U.S. Senate).

The major lesson I learned from both of these academic icons was no different than most situations in life.  Filter out the noise and look for the signals.  And noise has certainly been the order of the day this election cycle.  Equally important, the numbers are only as good as the assumptions used for any predictive model.  So let me start there.  The model I created for predicting the Florida outcome on November 3rd depends on the following assumptions.

• When Democrats vote, Democrats win.
• Partisan defections affect the magnitude of the outcome, but do not determine the winner.
• Shifts in behavior by voters with no party affiliation (NPA), i.e. independents, can make the difference.

You may be surprised there is nothing related to demographics, either seniors or college educated suburban women, the focus of much media punditry.   Nor does it depend on early versus election day balloting.

Ready?  Put on your green visors.  Here come the numbers.  First, to test the validity of the model, I applied it to the actual 2016 results, initially relying on the second and third assumptions.

2016 Election

In 2016, 4.9 million registered Democratic voters in Florida cast a ballot of which 90 percent voted for Hillary Clinton and eight percent for Donald Trump.  On the GOP side of the equation, 4.5 million registered Republicans cast ballots with almost identical results–89 percent for Trump, eight percent for Clinton.  The 3.5 million independent voters broke for Trump by a 47-43 percent margin.  Despite the independent vote advantage, the model predicted Clinton should still have carried Florida by 177,230 votes.  Why?  Although independents made up 26.8  percent of all registered voters, they accounted for only 20.7 percent of all votes for president.

Either the model was flawed or another factor determined the outcome.  That is where partisan turn out becomes critical.   Just under 75 percent of all registered Florida voters cast a ballot in 2016.   When party affiliation is taken into account, Democrats made up 37.9 percent of all registered voters and 38.1 percent of 2016 Clinton ballots.  In contrast, Republicans outperformed.  Although only 35.3 of all registered voters, GOP ballots constituted 38.7 percent of actual voters.  With this additional factor, the model predicted Trump would win Florida by 90,000 votes, just 24,000 votes different from the actual margin of victory of 114,000 votes.

2020 Election

Which brings us to 2020 and the question, “What, if anything, has changed in the last four years?”  First, there has been an increase in voter registration by Democrats, Republicans and Independents.  The advantage, as his been widely reported, goes to the GOP with an addition of 427,000 registered voters.  Democrats added 299,000 registrations, while NPA/Independent registrations increased by 175,000.

But that is the only good news for the GOP.   According to a consensus of polling data, registered Florida Democrats will vote for Joe Biden by a margin of 96-2 percent, a six point increase from 2016.  However, Republicans remain slightly less united behind Trump with an anticipated margin of 91-7 percent.

The bigger news is Florida independents are now predicted to break for Biden by a 49-41 percent margin compared to the 47-43 Trump advantage in 2016 (Source:  October 7, 2020 Quinnapiac survey of Florida voters).  Without factoring in Democratic turnout, the model predicts a Biden win by 565,600 votes or 4.08 percent, slightly better than the FiveThirtyEight average of 2.4 percent Biden lead.

Therefore, it will again depend on Democratic voter turn out.  As of this morning, 51.6 percent of registered Democrats have already voted compared to 47.5 percent of Republicans and 35 percent of independents.  If the model is valid, Biden already has a 649,700 vote lead a week before election day.  Without a Trump in-person turnout of biblical proportions on election day, this margin will be hard to overcome, especially if Biden supporters also continue to vote in record numbers.

Just to make sure I do not make the same mistake twice, I went back to all my political posts in October and November of 2016.  It was a relief!  Fortunately, I never predicted a Clinton victory despite referencing polling data which suggested she would win.  In fact, I found that I have been pretty consistent.  Consider the following excerpt from the October 12, 2016 post titled, ” Why Trump Has Gone  Nuclear?”  It just as easily could have been written this morning.

As this election cycle comes to an end, and barring a major reversal of fortunes, Donald Trump will not be the next president of the United States.  Rather than exit with some modicum of dignity, he has elected to up the ante, trashing anyone and everyone who has abandoned his “crusade.”  Pundits attribute this to energizing his base of die-hard supporters.  However, is it more likely he views his constituents as potential consumers rather than voters?

Will Trump pull another inside straight?  He is betting the American voter is insane, and as Albert Einstein suggested, is doing the same thing over and over assuming the outcome is going to be the same.  I have more faith in the people of Florida and am putting my money on George W. Bush, who infamously tried his best to remind us, “There’s an old saying in Tennessee — I know it’s in Texas, probably in Tennessee — that says, fool me once, shame on — shame on you. Fool me — you can’t get fooled again.”

Hopefully, on November 4th, 5th or 6th, I will not have to change my alias from Dr. ESP to NostraDumbAss.

For what it’s worth.
TBD