Q&A of the Day – Are any Florida pollsters using credible methodology?
Each day I’ll feature a listener question that’s been submitted by one of these methods.
Email: brianmudd@iheartmedia.com
Twitter: @brianmuddradio
Facebook: Brian Mudd https://www.facebook.com/brian.mudd1
Today’s entry: The Florida polls are becoming more ridiculous by the day! Biden up 11 in Florida, Ha! I bet you not even Biden believes that poll. Are there any Florida polls that are reliable?
Bottom Line: With Election Day only a few weeks away the pollsters are in overdrive mode in Florida. We received seven new polls this week alone. Of those, the results ranged from Trump +3, in the Fox35/Insider Advantage poll, to the eleven-point lead for Biden you mentioned in the Quinnipiac poll. A few weeks ago, I highlighted how poor Quinnipiac's polling was and how they’ve had the worst track record of any pollster in Florida over the past decade (Quinnipiac’s Election Day polling has only accurately reflected the outcome in about a third of Florida’s statewide races starting with the 2010 election cycle). So, about the reliability of sampling...
Florida’s current voter registration is as follows:
- DEM: 37%
- GOP: 35.7%
- Minor: 1.3%
- NPA: 26%
A perfect sample would mirror those results. Quinnipiac’s missed the mark on all accounts. They oversampled Democrats relative to Republicans by 1% while oversampling NPA’s by 8% and remarkably “other’s” by 6%. That’s a sample that’s off by a net 15%. Clearly, not reliable - yet somehow or another they assigned it a 2.8% margin of error. Not all pollsters released complete methodology, but of those who did, the closest sample came from USA/Today Suffolk which near-perfectly matched Florida’s electorate. Here’s who they sampled:
- DEM: 37.2%
- GOP: 36.6%
- Minor: 2.8%
- NPA: 23.4%
That happened to be the poll which showed Florida to be a tie. If you were to look at just one poll this week, that would be the one. It also happens to be the one which always reflects commonsense. What this exercise also illustrates is the flaw with simply relying on polling averages across all pollsters. Many will cite, for example, the RealClearPolitics polling averages as being a reliable way to account for variances in polling. All it takes is one absurd poll like Quinnipiac’s skew the averages meaningfully as well. Averaging the best sampled and worst sampled poll in Florida this week equals a 5.5% lead for Biden in Florida. That’s simply not the case. So back to where we started with your question. There is a reliable Florida poll this week. It’s the USA/Today Suffolk poll. There’s no guarantee that will remain the case. Polls are only potentially accurate in context. The context is the determined in the sampling which means every pollster is only as credible as the sampling used in their next poll.