You’ve heard the news by now that Donald Trump has won Florida, giving him a much clearer path to the presidency than the Democrats had hoped, and that the polling errors coming out of Florida are just as bad, if not worse than they were in 2016. So yeah, in retrospect, our polling methodology of surveying fifteen random people outside of a Publix in Sarasota a couple of weeks ago might not have been the most accurate way to conduct a statewide poll.
But hear us out.
Critics say that collecting responses from fifteen grocery shoppers in Southwestern Florida was a terrible way to get a reliable sampling of the entire state. And sure, when the fifteen people we surveyed included a former Sandinista living in exile, a Black man wearing a MAGA hat, and thirteen white women in the same bible study, I can see why people think those results might not reflect the entire electorate. But this Publix was a bellwether for Florida elections going all the way back to 1984. As the old adage went, “As the Publix off of Route 41 goes, so goes the Sunshine State.” So why mess with a good thing just because it was wrong once? Okay, admittedly, twice, but everyone in Florida has collectively repressed the 2000 election results from memory, so that one doesn’t really count.
In our defense, we learned a lot from the 2016 election. After our poll was off by +12 points in favor of Hillary, we had a real moment of reckoning. It was clear that we needed to completely rethink the way we approached statewide election polls. So we spent weeks brainstorming new and innovative ways to survey our state, taking into account changes in technology, demographics, and expanded ballot access this year, to get a clearer picture of what was in the hearts and minds of our fellow Floridians. That’s why, in 2020, instead of conducting our survey on a Wednesday, we did it on a Thursday.
Does that seem like an absurdly modest adjustment? Sure. Could we have gone further? Of course. But we had the data to back up our decision, and we were confident we were going to get it right this time. We stood outside of that Publix for weeks on end, each of us sweating buckets in the July heat, counting the number of suburban-looking white women that went in and out of there, and we found that suburban white women were 3.2% more likely to shop on a Thursday than on a Wednesday. And as it turns out, not factoring that in was exactly the sort of miscalculation that would’ve gotten us the correct outcome in 2016. Yes, maybe that was just correlation, not causation. And yes, there could have been some projection bias in our methodology. Also, we can’t be certain our data was reliable since most of us passed out at least once a day from heatstroke, so sometimes the numbers were less a rock-solid figure and more of a gut feeling. But make no mistake, we did everything we could to make our poll more accurate, even if the 2020 results say otherwise.
Plus, we weren’t the only ones to re-think our methodology. SuperClearSurverys took their survey outside of a Piggly Wiggly this year instead of the more traditional Winn-Dixie. Total game changer. They went from being off on Hillary’s numbers by -9 to being off on Biden’s numbers by +13. Talk about a paradigm shift. And Samplerama Polling went to a Florida State football game and counted all the MAGA hats, which is a complete 180 from 2016 when they went to a Florida State football game and counted all the people not wearing MAGA hats. So don’t think for a second that we’re some dinosaur industry unwilling to reckon with a changing electorate.
And besides, our entire state will be underwater in 30 years anyway, so is this really the time to be reinventing the wheel? That said, those thirteen church ladies seemed to think the jury’s still out on climate change, so maybe it is worth rethinking things after all. I’ve heard the Dollar General off of 301 is much more representative of Florida these days.