Our editor-in-chief has a gripe:
Oh good another sweeping analysis piece based on early exits that are even more massively misleading than normal early exits.
— Clara Jeffery (@ClaraJeffery) November 28, 2020
Seriously, people, just don’t do this. The polling this year was unreliable and there’s every reason to think the exit polls are unreliable too. But analysts are gonna analyze, so we get a lot of opening sentences like this:
Take all this with a grain of salt, but….
No. Just stop there. I know it’s frustrating for numbers-oriented folks to have no reliable numbers to analyze the election with, but right now we don’t. And this is doubly true if you’re comparing 2020 to 2016, which had some reliability issues of its own. Like it or not, this year’s exits just aren’t good enough except at the broadest possible level (for example, shifts of six or seven points or more, which will probably hold up even if the precise magnitude of the shift changes when we get better numbers).
I myself have taken an even more extreme tack: I haven’t even looked at the 2020 exit polls. Even if I swear to myself that I’m just curious and I won’t let them influence me, they will. I’d rather have no opinion than one based on bad data.