I feel about this the way I felt about those Minnesota polls that Ed posted this morning: They can’t be reconciled with other data, so someone somewhere is wrong. It could be that the Minnesota polls are right — but if they are, then the national polling is wildly, wildly off. There’s no plane of reality where Joe Biden wins the national popular vote by seven points but loses a heartbreaker in Minnesota.
Likewise, it’s all but impossible to believe that Biden is leading in Florida right now — slightly, according to the RCP average — if he’s turned a 27-point Clinton lead among Hispanics in 2016 into a five-point lead. The pollster who conducted the poll for Telemundo said as much to Politico: “If Biden is going to flip Florida, he has to at least match Clinton’s numbers among Hispanics and that looks like it’s not going to happen.” I’m not sure that that’s strictly true, as the story of the campaign to date has been Trump overperforming with Latinos and Biden overperforming with whites, a much larger group. In theory, Trump could win Hispanics in Florida outright and Biden could still take the state by cutting deeply enough into the president’s advantage with white voters.
But that would be a heavy lift.
This isn’t the first poll of the race to show Trump doing remarkably well with Florida Hispanics, either. It’s why I think he has to be favored in the state on Tuesday.
Biden leads Trump among Hispanics by just 48-43 percent, numbers that could mean trouble for the former vice president in the nation’s biggest swing state…
Part of Trump’s improved numbers are due to strong support from Cuban-American voters in the state. Trump has long touted his support within that community after winning 54 percent of the Cuban-American vote, according to 2016 exit polls. Now, he is leading among them by 71 percent to 23 percent, according to the new poll.
“That might be the thing that saves Donald Trump in Florida,” Coker said.
Coker said Venezuelans and Colombians polled similarly to Cubans in offering strong support for Trump. However, Biden leads with Florida Puerto Ricans, 66 percent to 23 percent. Puerto Ricans traditionally have had lower turnout rates but are growing in numbers in the state.
Is there reason to believe this polling is accurate, more or less? Most definitely. Democrats in Florida have been warning for months that Biden is weak among Latinos there and isn’t doing enough to bring them home. Read this from early September noting a different poll showing Trump *leading* statewide among that demographic. A few weeks later brought more bad news for Team Joe. It would be shocking at this point if Trump didn’t top the 35 percent he pulled in 2016 among Florida Hispanics next week.
But there’s reason for doubt too. Per Politico, not every recent poll of FL shows a race this tight among the Latino minority. A Univision poll had Biden up 20 among that group, still off Clinton’s pace but much less dramatically than in Telemundo’s data. Another poll from CBS found him leading Florida Hispanics by 27 — exactly matching Hillary’s margin. If Trump *doesn’t* overperform among that demographic next week, he’s probably cooked in Florida. And as goes Florida, so goes the presidency.
…Well, for Trump at least. Biden doesn’t actually need FL if he’s cleaning up in the Rust Belt. And this same Telemundo poll shows that he’s faring more strongly with Latino voters in Arizona, Nevada, and Texas. Clinton won Hispanics in AZ by 30 points four years ago according to the exit polls. Biden leads there by 36 according to Telemundo. He’s also slightly ahead of Clinton’s pace in Nevada and only a few points behind in Texas.
I gave you a shot of optimism with the Telemundo poll. Now here’s a shot of pessimism via the LA Times:
More than 6.8 million people ages 18 to 29 had voted early or by mail in the national election by midweek, a 2 ½-fold increase over their voting level at this point four years ago, according to the Democratic data firm TargetSmart. Texas has witnessed a particularly robust turnout among younger voters. As of Wednesday morning, voters under 30 had cast more than 900,000 ballots in the Lone Star State, nearly three times the number as around this time in 2016…
Other battleground states mirrored that outsized turnout by young people, TargetSmart data showed: Young Florida voters have cast 569,000 ballots, a 42% increase over their 2016 total; some 404,000 young North Carolinians have voted, a 63% jump; and young people in Georgia have expanded their vote by 82% to more than 340,000.
You know which way younger voters lean. None of the above proves that they’ll be out in greater numbers this year than they were four years ago; it could be that the early vote is simply cannibalizing Election Day turnout, not adding to it. But if a youth wave is showing up — and if it’s also true that Biden is winning seniors this year — then the math starts to look very grim for Trump.
Which, of course, is not to say that he can’t win. Kristen Soltis Anderson, a pollster at Echelon Insights, reminded her Twitter followers of that this morning, with the caveat that a Biden landslide scenario is also on the table.
Current polling averages in the Sun Belt states – FL, NC, AZ, TX, GA – all have Trump in striking distance. They are tied or within just a point or two. Not enough for me to say confidently which way they'll go. If they all go Trump, that's ALMOST enough…
— Kristen Soltis Anderson (@KSoltisAnderson) October 29, 2020
There's a scenario where the map looks like this. And this scenario DOES NOT REQUIRE THE POLLS TO BE WRONG. pic.twitter.com/CZuREaQRoz
— Kristen Soltis Anderson (@KSoltisAnderson) October 29, 2020
Is it true that her map of a Trump win doesn’t require the polls to be wrong? What she means by that is that even in states where Biden leads narrowly right now (e.g., Florida), the race is so tight that his lead is within the margin of error. When a pollster has it Biden 50, Trump 48 and the outcome is Trump 50, Biden 48, that’s not a “miss.” The margin of error — let’s say three points in this case — means that the poll also envisioned an outcome where Trump did three points better and Biden did three points worse.
But it depends on which data you look at. Here’s the New York Times’s latest update on where things stand and what it will look like on Tuesday if the polls miss by exactly the same margin they did in 2016. This is not a “Trump upset” scenario:
If the margins in the middle column are what the margins on election night look like, we’re headed for a Biden electoral college blowout. His margin of victory in each individual battleground would be modest, four points or less in all of the big ones. But Trump won by a whisker in those states in 2016 and that hasn’t stopped him from treating the 306 electoral votes he got as a thunderous victory, never mind that he lost the popular vote nationally. All of which is a long way of saying that if the Times is right about where the race stands right now then Trump needs an even bigger polling error than in 2016 to pull the upset. And if we’re willing to believe that pollsters have adjusted their methodology since then to make sure they’re overlooking fewer Trump voters this time then a bigger polling error this year seems unlikely.
Update: Twenty minutes after I finished this post, Monmouth dropped its final Florida poll of the campaign. They have it Biden 50, Trump 45 among registered voters, with Biden gaining a point in a high-turnout scenario and losing a point in a low-turnout scenario. As for Latinos:
A 26-point Democratic advantage, basically identical to Clinton’s margin within that group four years ago. Five days until we know whether Monmouth or Telemundo was right.