Educator, Scientist. Sports & Political Data Analyst. Other social media: RealCarlAllen 📖 Book: *The Polls Weren't Wrong* out now
Ohio, USA
Joined January 2013
- Tweets68.8K
- Following5.5K
- Followers5.6K
- Likes56.5K
Pinned Tweet
I fully expect DMs, EMs, etc to blow up in the next 48 hrs
If I don't respond, don't worry. I know who has supported my work and I won't lose sight of that
If you'd like to join me in developing a set of unified scientific standards for this field, my book is available in bio
Here's the breakdown by state, and 74% chance of Blansas isn't a misprint
Full forecast
realcarlallen.substack.com/p…
My biggest (interesting) election positions:
ME-02, PA-01: D at slight +money
Senate D NC+NH+MN+AK+MI at even money
Senate D GA + NC + NH at implied 87%
TX-23 R at slight -money
Sen KS R at even money
ME Sen D implied 63%
KS within 5 points implied 65%
Carl Allen retweeted
1/x: re: sports data, politics data, and my predictive models
I've noticed that most of the arguments/debates I've had with people, mostly about sports (but elections too) have a major common underlying theme:
That past results are the best predictor of future performance.
Made in 2020 in early stages of my research:
Why do candidates who poll consistently 50%+...never lose?
This is an old version of what I call, scientifically
The "circle of never lost"
Has never happened doesn't = can't.
But if your forecast spits out 5%+, it's shit
Ty yw
Friend messaged me about @FiveThirtyEight's double-digit chances for Trump in NH (11%) & ME (10%)
I've circled here what I call, scientifically, the circle of never-lost.
WI, MI, PA at the edge of never-lost circle. Could you convince me Trump is 10% there? Sure
NH and ME? Foh
When I cash out the $1000 parlay for $1300 seconds after the game has started but nothing has happened.
Imagine a pollster in MN sees D47-R44
They'll be judged on how closely result is to D+3
That 9% undecided likely lean D? Doesn't matter. If they split 70/30 D, well, I assumed they'd split evenly, which means your poll was wrong
If you see why that's insane, welcome to my hell
Happy Election Eve
Polls are not predictions of election outcomes or margin
The myth - "accurate polls predict the margin" - operationalizes the assumption that undecided voters always split evenly
This assumption is not neutral. It's a bad guess.
Polls are not predictions.
This happens every election cycle, and this year is no exception
Regardless of whose poll averages you look at (also not an objective scientific process to judge what "the polls" say):
We can say with certainty that undecided voters in many states strongly favor one party.
I see my content get much more attention when it's presented in a certain way
"Guy who nailed 2022 midterms..."
And it's taken as a harmless piece of promotion. A little innocent editorializing.
No.
It promotes poor science, probabilistic reasoning, and deserves no respect
To my knowledge (though I'm probably wrong on this): neither @NateSilver538 nor @gelliottmorris has openly promoted their work by saying "I called 99% of states right" or whatever.
They let others do it without pushback. A problem for a separate discussion.
But @lxeagle17 lol
Hey @NateSilver538 if you're going to cite my research you should at least cite it
routledge.com/The-Polls-Were…
Carl Allen retweeted
This is some outstanding work which gets into the REAL spot the (real) analytics folks should be doing:
Probabilistic distribution of polling errors.
Eg 90% chance polling error was at least 1%
70% chance at least 2%
50% chance at least 3%
Etc.
This is how it works irl
I'm not from Michigan, and I definitely preferred El-Sayed over Stevens in the primary, but I didn't follow the race closely.
But man, we dodged a bullet with this one. Well done Michigan.
@tbonier