About a month ago, I started digging into the relationship between the Florida governor and Senate races, which are probably two of the most important races in the country both politically and in terms of policy.
The basic theory of the relationship is simple: Andrew Gillum, a black progressive and the Democratic candidate for governor might be able to increase turnout across the Democratic base and win voters that Democratic Senator Bill Nelson wouldn’t. Rick Scott, the current Republican governor and Senate candidate, might be able to win some college-educated white and traditionally Republican Hispanics (many Cuban) that Ron DeSantis (a Trump-hugging Republican who has some real baggage around racial issues) might lose to Gillum. And a larger gap between these races (with Scott presumably outperforming DeSantis) would indicate that Scott achieved some separation from DeSantis and had a higher win probability than him. It doesn’t mean that Florida would elect a Democratic governor and a Republican senator, but it would show something about Scott’s chances.
It’s an interesting theory, so I devised a simple state to help track it: the Gillum Gap. It’s just the difference between Gillum’s margin in a poll and Nelson’s margin in that same poll. In most polls, the Gillum Gap seems real: The average poll taken between the end of the primary and now shows that Gillum’s margin has been about 2.7 points better than Nelson’s. That’s due to both overperformance by Gillum (he’s been getting about 1 percent more of the vote than Nelson) and overperformance by Scott (he’s been getting about 1.7 percent more of the vote than DeSantis in post-primary polls).
This gap (2.7 points) suggests that Scott is in a much better position than DeSantis, though he still trails – which supports the basic, intuitive theory of the election I spelled out a couple paragraphs ago. Gillum is doing a bit better than Nelson, DeSantis is underperforming Scott and things seem to make sense!
That being said, things get a little thornier when you chart out the Gillum Gap in individual polls and display it over time.
Each point in the graphic is a poll. The vertical location is the Gillum Gap (positive means that Gillum’s margin was larger than Nelson’s) and time runs left to right chronologically (specifically the middle of the poll’s field date is its horizontal location).
Some more recent polls show a big Gillum Gap, where his margin looks much better than Nelson’s, giving Scott a better chance of winning. Other polls show a low Gillum Gap—meaning that Scott and DeSantis perform similarly. And in a few oddball polls there’s a negative Gillum Gap—indicating that DeSantis outperforms Nelson when you look at the margin.
I don’t totally know what to make of this. The average Gillum Gap for the 30 days preceding the election is smaller than the Gillum Gap for 30 to 60 days from the election. That seems like good news for Nelson. Nelson has been locked in a tight race with Nelson for months, but he’s seemed to pull ahead by a bit recently, while Gillum’s advantage has been basically static. So the closing of the Gillum Gap reflects an improvement in Nelson’s fortunes.
But there’s a lot of noise in this data. I was hoping to see a more consistent Gillum Gap and thus be able to draw some deeper conclusions about what’s going on in this race and how it’s going to turn out. But for now, we have to settle for some ambiguity. Gillum seems to be the favorite, but winning is not a certainty. FiveThirtyEight puts Gillum’s win probability at 77 percent. The current iteration of THE WEEKLY STANDARD’s stats-based Senate forecast (SwingSeat) puts Nelson’s win probability in the same neighborhood. The official forecast has Nelson at a roughly 80 percent win probability, and the older version of the forecast (which uses only polls and doesn’t use “fundamentals” like presidential approval, which are much more favorable for Nelson) would give Nelson a roughly 66 percent win probability.
All those probabilities favor the Democrats by a significant amount, but none of them are certainties. The polls do seem to suggest that Scott is outperforming DeSantis, but Scott hasn’t achieved enough separation to become the favorite in his race. And we’ll just have to wait for the results to see if the demographic specifics of my Gillum Gap Theory (e.g. Scott does a bit better than DeSantis with Hispanics, college-educated whites) actually show up in the results
