Brief thread: here's why I don't put a ton of stock into the *overall* pro-R registration trend in FL from '16 to '20. Take the 33 least populous of FL's 67 counties, which are mostly rural and located in the north/panhandle (aka the Deep South)...
In November 2016, these 33 counties collectively had a narrow 42%-40% GOP voter registration advantage. But here's how they *voted:*
Trump: 369,045 (67%)
Clinton: 170,871 (31%)
Btw, Trump's statewide margin was 112,911 votes.
Fast forward to 2020: the final data shows the GOP voter registration advantage in these 33 counties has grown to 48%-32%.
These aren't new Trump voters. These are existing Trump voters formalizing their party change from D to R.
In FL's other 34 counties (where 94% of FL's voters live), Democrats' narrow reg. edge has remained pretty stable, ticking down from 38%-35% to 37%-35%.
And Dems have seen modest encouraging growth in FL's suburbs, including counties like Seminole, Osceola, Duval, Orange, etc.
Bottom line: Rs can rightfully take pride in their recent net registration gains (March-Nov. '20), which far outpace the breakdown over the comparable period in '16.
But the *overall* '16-'20 pro-R shift isn't indicative of a sea change in vote intention.
Just in: final FL voter reg. data. Since this March's primary, here are each party's net gains:
GOP: +344,465
Dem: +197,821
NPA/Other: +197,818
Compare to the same period in '16:
GOP: +274,207
Dem: +307,961
NPA/Other: +220,857
Trump's '16 FL margin: 112,911 votes
For reference, here's the story I wrote for @NBCNews a few weeks ago putting the GOP's recent registration success in FL/NC/PA in context: nbcnews.com/politics/2020-…
Here's the final Nov. 2020 FL voter breakdown by party and race (14,441,869 registrants):
New Marist/PBS & NBC/WSJ polls bring us to a critical mass of October live-interview national polls, and to me Biden's gains appear to be pretty even across the board (except for Black voters, where there wasn't much room to grow).
To be clear, "final 2016 polls" are via @Nate_Cohn/@UpshotNYT's estimates and this chart is my running average of 2020 national live-interview poll crosstabs, including NYT/Siena, CNN, ABC/WaPo, NBC/WSJ, Fox, Marist/PBS, Quinnipiac, Monmouth, Suffolk and Grinnell/Selzer.
One fascinating takeaway is that Biden's 12% lead over Trump is the same size as in June/July, but:
1) far fewer undecideds today (54%-42% vs. 52%-40%) 2) Biden's average lead has expanded by 4% since June/July among seniors & Hispanics
New @pewresearch poll: Biden leads 69%-27% among those who plan to vote by mail, Trump leads 63%-31% among those who plan to vote on Election Day. pewresearch.org/politics/2020/…
Also interesting: @pewresearch finds that of the 5% of RVs who support Jorgensen/Hawkins, they would lean Trump over Biden 54%-42% if forced to choose - suggesting their presence might actually aid Biden?
Sorry, should have noted Biden also leading 55%-40% among early in person, per @pewresearch.
My new @nytimes op-ed: meet the 10 counties in America that hold the key to President Trump's fate - and illustrate why he's in such deep trouble. nytimes.com/2020/10/06/opi…
Today, I'd rate all ten of these top bellwether counties as either "toss ups" or "lean Biden:"
1. Maricopa, AZ 2. Pinellas, FL 3. Peach, GA 4. Marshall, IA 5. Kent, MI 6. New Hanover, NC 7. Wood, OH 8. Erie, PA 9. Collin, TX 10. Sauk, WI
These places don't fit one stereotype: they run the gamut from meatpacking hubs to military bases and peach orchards to yacht-dense retiree havens.
But one thing they have in common: their votes will matter a whole lot more than most Americans' votes.
The key to the stability of Biden's lead to date? He's basically stayed out of the way and ceded the spotlight to Trump. It's October, and this is still a referendum (vs. choice) election.
Hearing a few people ask: then why is Trump's approval up to 44.4% at @FiveThirtyEight? My answer: we're pretty close to Nov. and if the election were today, I wouldn't be surprised if Trump got ~44.4% of the vote.
This might be a good opportunity to explain why I will *not* be saying "I've seen enough" on Election Night 11/3 (besides the main reason, which is a longstanding contractual obligation w/ a network desk)...
My general approach to "seeing enough" on Twitter is to make that declaration at a ~98% personal confidence interval, whereas a media organization might prudently wait to project a race at a ~99%+ confidence interval.
In other words, I'd expect that for every 98 times "I've seen enough" ends up being correct, there might be two times it ends up being incorrect.
I'm comfortable w/ that level of risk on most nights. But on 11/3, the stakes are way too high to mischaracterize results.