A brief (somewhat funny) history of mine, to give you some background:
In 2008, as a freshman in college, I attempted to make my first forecast. I used poll averages, some very amateurish models, to calculate the win probability of each state.
I calculated Obama's chance of winning as about 100%. Obama's comfortable victory aside, that was bad math, and I knew it.
It didn't take me long to realize my error.
The states, while they are "independent" in terms of how votes are counted - are far from independent as it pertains to how voters act.
If, for example, I thought Obama would win Ohio by 2 and Indiana by 1, then if my "Ohio" forecast is off by more than 2 in McCain's direction, it follows that Indiana (similar geographically and demographically) would also likely go that way.
If he *doesn't* win Ohio by 2, then he probably doesn't win Indiana by 1…and so on.
The states are “independent” but highly correlated.
I knew what math needed to be done, but I knew I wasn't capable of doing it then. After no more than a couple hours of working on it, I smiled, shut my laptop, and went to play basketball.
This correlation - best expressed as a matrix - is a lesson I'm not ashamed to admit I learned from 538.
The ability to understand something well enough to know that you don't fully understand it is a valuable one.
This is why 2016 was so infuriating for me: there was zero data to support Clinton being 98%, 99%, or 99.99% to win (as real experts asserted at the time). I didn't consider myself an expert by any means, but the math didn't make sense.
The only way they could reach that conclusion is if they made the mistake I was smart enough NOT to make as a freshman in college with very little relevant experience.
And they did.
Fast forward to more recent history: my forecasts have gotten much better.
Part of the reason my forecasts have performed better is understanding which correlations matter (say, a same-year election in the same state) and ones that don't (but…the margin!). And also, historical trends!
Here's why that matters for Florida.
As many people know, states with abortion as a major issue have seen large Dem “overperformances.”
I use quotes because I define performance here as relative to public expectation, not necessarily what good data/forecasts would suggest.
Florida just passed the nation's strictest abortion law, after their Supreme Court ruled that “abortion” doesn't fall under “right to privacy.”
The amendment overturning this strict ban requires 60% to pass. It is, thankfully, highly likely to pass. (I haven't gotten to forecasting yet, but I'd be surprised if it got under 65% support).
That likely means we can expect some turnout from some voters who might not have voted otherwise, and infer their political preference. This favors Dems.
Now to the math
The national popular vote, as you know, doesn't matter in US Elections. Each state counts independently. But the national popular vote does correlate highly with state-to-state outcomes.
This swing is not always uniform, however. Here are some swing states from 2016 and 2020, with their relative shift in election results.
Dems did better in the most important states in 2020 than they did in 2016. That makes sense.
Indeed, the net shift was positive almost across the board. Almost.
While Nevada was effectively flat, Florida got redder!
There's a lot to be said about Florida 2020.
For one, Trump won it by less than 3.5. It wasn't a blowout.
Second, almost everyone considered Biden a *favorite* there, 538 included.
Third, once it was called (relatively quickly) on election night 2020 that Trump had won Florida, analysts rightly assumed that other states would also probably get a little redder than 2016 (thus, he'd probably win again)
But the correlations were imperfect.
Florida was an outlier.
*Vibes* and not *data* seem to be driving the myth that Florida isn't in play for 2024.
Now, it's highly unlikely that Florida is decisive in 2024. If Biden wins Florida, it's likely he's won other more pivotal states like PA and MI, too.
But highly unlikely does not mean impossible. Importantly, even if Biden does win Florida (which would be called early) that doesn't mean we should expect a popular vote blowout.
The concept of “elasticity” was discussed by FiveThirtyEight a decade ago, and my work tries to take it a bit further: different factors can impact different states, even different races.
Candidate quality, for one, became a buzzword after the 2022 elections in which the historical trend (party in power tends to do well in midterms) wasn't followed. Even though I said that's what would probably happen, and why, in advance.
Abortion has become a bigger issue, which pushes Florida closer to Dems than might be “expected.”
If you, like me, think it moves the needle “more than a little” pricing in a little better turnout than normal for Ds, and a little better performance among swing voters for Ds, then yes, Florida is in play.