i think polling cross-tabs kinda create the discourse in a weird way?
we talk about the latino vote and the asian-american vote partly bc, in many polls, it's grouped that way (which is often unavoidable given sample sizes)
just kinda a weird way for journalism to work
• • •
Missing some Tweet in this thread? You can try to
force a refresh
i have this feeling that shy trump is about to take on like 8 different definitions
i've seen a couple things floating around today that are like "shy trump is real!" but it meant non-response or people feeling personally sheepish about their vote (regardless of what they're saying to pollsters)
the OG definition is people lying to pollsters b/c of social desirability bias
but it's a catchy term so feels like all theories are gonna get filed under it, whether they're that or not
basically my thinking is that some chunk of the conventional wisdom jumped straight to "amash is running oh my god he's going to spoil it for biden"
which could happen but there are quite a few, completely not certain at all, things that would have to happen first
the election would have to be close enough for him to matter (not clear)
he'd have to maintain a big enough coalition to swing it (not clear, given downward general trend of 3rd parties as elex approaches)
he'd have to disproportionately take from biden (not at all inevitable)
Sooo I have an exciting new experimental thing to show you guys! It’s the post opinions simulator. It’s a model-powered Build Your Own Election interactive for Iowa (w/other versions later)!
Also some deets about how I think about this, math, etc
1/
Here’s how I think about it: everyone has simplified representations of how elections work in their brain -- qualitative or quantitative. IOW everyone is a modeler.
Everyone can sorta run through different scenarios on their model: if X happened, how would it change things
2/
So we thought it might be fun to lay out one of these models, let people play with it within some responsible guardrails as a way to help people think through elections and learn about it via that sort of tactile experience
3/
My meta theory on Trump is still that he never really updated his playbook/personality/strategy from the GOP primary.
It worked well enough to barely win a general election against HRC, but it hurt him in the midterms and makes him unpopular now.
And imo that explains a lot
Like the strategy is
Be super restrictionist on immigration
Be loudly "politically incorrect" on tons of stuff
Dominate media coverage all the time, even if it means doubling down on your negative qualities
Outsource issues you don't care about to existing GOP politicians
And that was good enough to a) give him a plurality in the GOP primary and b) help keep the already-fractured non-Trump wing of the GOP from consolidating --> win the nom
Then in the general he didn't change much & had the good luck of running against HRC, who was also unpopular
1) Cruz has been ahead this whole time 2) This election is way more normal/generic than coverage suggests 3) We're obsessed w/Texas b/c we believe that it shows us the future 4) It actually doesn't
I got polling data on over 3000 Trump tweets and found
1) They're are generally not super popular 2) Some the least popular tweets were insults 3) Some of the most popular tweets were boring ceremonial things 4) Rs, Ds sort of of agree on a lot of them
They've been getting national samples to rate Trump's tweets for like a year and a half ish. The scale is "Great" "Good" "OK" "Bad" "Terrible" for every tweet
That gets translated into a numerical score --> Tweet Index!
One basic finding is that the average (and median) Trump tweet is not super popular. Also notice the bump in the histogram to the left of zero -- lots of negative scores in this dataset.
There's a big spread though. Theoretical max/min is -200, 200 so some tweets are popular