Tuesday, May 21, 2024

A quick note on Ezra Klein's latest exercise in meta-panic

As previously mentioned, the definitive rebuttal to Ezra Klein's NYT opinion piece is Josh Marshall's "A Quick Survey of Numbers, Vibes and the Inner Lives of Campaigns" which actually came out two days before Klein's piece (making it technically a pre-buttal). At the risk of piling on, I am going to add a few comments starting with this.

Klein buys into the Biden is in denial narrative with the following:

From Seven Theories for Why Biden Is Losing (and What He Should Do About It)

The polls are wrong. This appears to be Biden’s view. “The polling data has been wrong all along,” he told CNN last week. Axios reports that polling denial is pervasive in Biden’s campaign.

There are two things to say about this. The first is that it’s false. Even as pundits predicted a red wave in 2022, the polls showed Republicans falling short, and they were right. “The polls were more accurate in 2022 than in any cycle since at least 1998,” FiveThirtyEight reported.

Putting aside the problems with Axios's story, there are at least three problems with this claim.

1. It ignores the clear reality of how candidates spin news. If everyone who was behind in the polls and suggested they were wrong was delusional, then the great majority of politicians of the past 75+ years were at some point mentally imbalanced.

2. The polls actually have been wrong in a substantial and consistent way regarding the primaries, particularly the performance of Nikki Haley. The NYT's own Nate Cohn actually wrote a column offering different explanations for the repeated fails. (These explanations are not aged particularly well, but that's a topic for another post)

3. And this is the biggie. The 538 link that Klein provides does not in any way support his point. Time frames matter. The 538 piece, which is quite good, clearly states they are talking about how closely the average of the final 21 days matches the actual vote. This is a very different question than do polling numbers six months away from the election correlate strongly to the final outcome. That is an answerable question and there has been some research into it, but it is not, as far as I can tell, addressed at all in this piece.

We'll come back to this later. 

No comments:

Post a Comment