Harry Enten on how Harris' debate showing may affect Trump's chances - CNN

[TexasTowelie]

Registered voters who watched the debate between Kamala Harris and Donald Trump broadly agreed that Harris outperformed Trump, according to a CNN poll of debate watchers conducted by SSRS. CNN senior data reporter Harry Enten weighs in on the potential impact the debate may have on the 2024 race.

[AmBlue]

1. Add in the Swift endorsement

...and the bump will be even higher, tho not very measurable since that age group probably would never even be polled, nor answer their phones if they got a polling call. I know I don't answer #s that I don't recognize, and have never once spoken to a pollster.

[GopherGal]

2. I'm trying to think like a researcher...

about how you could attempt to separate the effects of the Swift endorsement from the debate.

The news media will be filled over the next week with comparisons of polls before and after the debate, but most will not be able to distinguish causality between the two near-simultaneous events. Even the "dads, Brads, and Chads" who dismiss or dislike Swift will have heard of her endorsement,, with sometimes unpredictable influence. ("that b&*# that I have to see pop up on screen for 5 seconds during my football game every week wants me to vote Harris? She needs to stay in her lane. I'll show her and vote Trump" )

The way I would want to design the study is as a "longitudinal study" that follows the change with time (before vs after the debate). (There will probably be examples galore in the media over the next week or so where they interview different groups of people before and after but likely only a few tracking the same group of individuals). An ideal study would interview the same subject before and after the debate/endorsement and have data on Swifties vs non-fans (I'm not expecting a "haters" subcategorization) But even a non-ideal study might add questions about that categorization to the "after" queries. Someone is sure to ask voters directly if the endorsement affected their opinion, but who knows how reliable a self-assessment of that effect would be. At a minimum, I'd want to ask my voter sample if they had watched the debate, or at least read/watched the coverage thereof.

The longitudinal studies where you follow the same individuals with time are the gold standard (i.e. what's used for medical trials) because they can give much better estimates of effects. But they're methodologically much more difficult to implement. And to make one ex-post-facto when you didn't know in advance that Swift's endorsement would be coincident with the debate is just not likely. But that's frame in which I'll try to think about the poll stories that will be coming out this week.