Thread by @magi_jay

This piece is not great. The data analysis is problematic. And not just in terms of white women, but also in terms of minority voters, which is an even more serious issue.

https://twitter.com/PostOpinions/status/1332692621052039172

Let's look at the more serious issue first: The claim that Trump made gains with Black and Latino voters. The evidence that we have at this time is spotty, but I have seen very little that suggests Trump made gains with Black voters.

Trump does appear to have made gains w/ some sub segments of Latino voters, specifically voters of Cuban descent in Florida. Lenz correctly notes that Latinos are not a monolith, but then goes on to treat them as a monolith by extrapolating the S. FL gains to all Latinos.

Further, it's all well and good to say that Dems should listen more to Latino voters, but listening to concerns re: immigration reform from some subgroups of Latino voters is different than listening to concerns about socialism and/or US relations w/ Cuba from other Latino voters

And, again, our data is spotty at best at this point and this is especially true for minority voters. Just as we know there were weaknesses in FL w/ some Latino voters, we know there were game-changing strengths in GA & AZ.

Finally, onto white women. But, first, a caveat, when I talk about this subgroup, I am not defending them on a moral level. I care about precision in data analysis b/c I care about data. I also care about electoral strategy. I am not interested in defending white women. . . .

If 55% of white women voted for Trump, that is slightly over half. If 45% voted for him, that is slightly under half. There is no real moral difference for me in these numbers. It's far too many white women who are either devoted to racism or happily complicit in it.

So, let's now look at the article. These two paragraphs should have stuck out to Lenz's editor, regardless of what they know about polls/data. Lenz first states that 2016 exits were off by quite a bit. . . .& then goes on to compare the better numbers to. . . .2020 exits.

A couple of things:
-Exits always have issues
-Exits have more issues in 2020
-Comparing between exits in diff. years to infer shifts makes sense (even if absolute values are off), but you have to exercise extreme caution if the method radically shifts, as it did btwn 2016 & 2020

-It is numerically almost impossible for Trump to have won 55% of white women and to have lost the election. You'd have to have other groups making up a larger share of the electorate for that to happen. And that's not what we've seen.

-A comparison of 2016-2020 CNN exit polls shows that Trump gained w/ every group (white women, Black women, Black men, Latina women, Latino men, etc). This is extremely unlikely, given he lost the election. So that should be a clue as to something goofy.

https://twitter.com/magi_jay/status/1328034679388315650

So, those are just a few points. Overall, I think Trump's gains w/ minority voters are vastly overstated by the media and no one should perpetuate these broad non-empirical* statements w/out more nuance.

(the evidence in S. FL is empirically supported, at least initially)

And, it is a matter of empirical fact that Democrats have been able to flip college educated white women. We've got loads of data on this. Importantly, we've been able to flip them *w/out* throwing POC under the bus. Why someone would want to misrepresent this is beyond me.

Finally, it is not up to me to say if the Biden team did enough to reach out to Black, Latino, Asian, or Indigenous voters. But broad statements about people in these groups flipping to Trump is an act of erasure. The actual data shows a lot more preliminary nuance. . . .

There are the difficulties for Dems in S. FL. But there's also the case of Dems holding numerical ground in SW TX. And record-breaking Black, Asian, & Latino turnout in GA. As well as record breaking Latino & Navajo turnout in AZ. This statement captures none of that.