With only three public polls on the 2018 Florida governor’s race since the first of the year, all yielding different results and using different methodologies, The Capitolist is proud to announce “The Capitolist HorseRace Index,” a weighted poll average that uses pollster accuracy rankings from fivethirtyeight.com, the age and obsolescence of each poll, along with news and election cycle adjustments to determine the true state of the 2018 Governor’s Race.
We’ve consulted with a statistical researcher and independent pollster to review our methodology. And to ensure that The Capitolist HorseRace Index is transparent, our formula for the HorseRace Index is posted publicly, along with detailed explanations for all value inputs. Experts, statisticians and political consultants are free to disagree with the methodology used, but ultimately, it’s the credibility of the index itself that is on the line, and we’re striving for accuracy above all else.
So what does the index show?
Since then, Gravis Marketing (rated B- for accuracy by FiveThirtyEight.com) conducted a pair of surveys exactly one month a part. The first, made public on February 20th, had wildly different overall results from Mason-Dixon, with substantially more undecided voters. The results had Putnam with a 2 point lead over DeSantis on the GOP side, while Democrat Gwen Graham and Phil Levine were deadlocked at 12 points each.
The second poll by Gravis showed DeSantis trading places with Putnam, and Levine opening up a 4-point lead over Graham, who also lost substantial ground to Andrew Gillum. The margin of error for both Gravis polls was 2.1 percentage points, but the methodology by Gravis, which surveyed the same panel of voters both times, differs dramatically from that of Mason Dixon.
The Capitolist HorseRace Index
We start by assigning a base weight to both polls. Mason-Dixon gets a 3.3 and Gravis gets a 2.7 by converting FiveThirtyEight.com’s letter grades into a grade-point scale. Next, we factored in the age of each poll. Here, Mason Dixon takes a significant hit to the base weight, since the firm’s only poll is now 44 days old. The first poll by Gravis also takes a hit because it is 28 days old, but the second poll is brand new, it’s base weight remains unchanged. Next, we apply an obsolescence penalty to the weight of the first Gravis Poll, as that data has been superseded by the newer poll, which remains unchanged. Finally, we apply a small accuracy adjustment to the first Gravis poll because it was conducted over a lengthy span (18 days) that included a major inflection point (Parkland school shooting) that likely yielded different attitudes from voters depending on which day they took the survey.
The results show Adam Putnam maintaining a slim lead over Ron DeSantis, 20.4% to 19.9% for DeSantis, while Richard Corcoran, still not an official candidate, pulls down 4.3% of all GOP voters.
On the Democrat side, the big news is that Gwen Graham is no longer the frontrunner, yielding the lead to Phil Levine, 14.2% to Graham’s 13.1%. Equally noteworthy, Andrew Gillum jumps from a 10-point underdog to within striking distance, at 10.4%.
Here are the full results:
| Poll / Date | Mason Dixon 2/6 | Gravis 2/20 | Gravis 3/20 | |
| Base Weight (538 Letter Grade Value) | 3.3 | 2.7 | 2.7 | |
| Poll Age (Days) | 44 | 28 | 1 | |
| Time Decay Factor (Min = 0, Max = 1) | 0.51 | 0.69 | 1.0 | |
| Obsolescence | 1.00 | 0.50 | 1.00 | |
| News / Election Cycle Adjustment | 1.00 | 0.83 | 1.00 | |
| Poll Weight | 1.69 | 0.77 | 2.67 | |
| GOP Governor | Weighted Avg. | |||
| Putnam | 27.0 | 18.0 | 17.0 | 20.4 |
| DeSantis | 23.0 | 16.0 | 19.0 | 19.9 |
| Corcoran | 7.0 | 3.0 | 3.0 | 4.3 |
| Undecided | 43.0 | 61.0 | 60.0 | 54.6 |
| DEM Governor | ||||
| Graham | 20.0 | 12.0 | 9.0 | 13.1 |
| Levine | 17.0 | 12.0 | 13.0 | 14.2 |
| Gillum | 10.0 | 9.0 | 11.0 | 10.4 |
| King | 4.0 | 2.0 | 2.0 | 2.7 |
| Undecided | 49.0 | 63.0 | 64.0 | 58.9 |
To view the underlying calculations, click here.