←
Carl Allen69,478 posts
Carl Allen avatar

Carl Allen

@RealCarlAllen

Educator, Scientist. Sports & Political Data Analyst. Other social media: RealCarlAllen 📖 Book: *The Polls Weren't Wrong* out now

◷ Joined January 2013▧ 16,972 Photos and videos
Carl Allen avatar
Carl Allen@RealCarlAllen·4 Nov 2024

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

Post attachmentPost attachmentPost attachmentPost attachment
Carl Allen avatar
Carl Allen@RealCarlAllen·3h

Oh, and most importantly: 3) What is broken in the field of poll data, analysis, and why it matters.

Carl Allen avatar
Carl Allen@RealCarlAllen·3h

I picked up a lot of new followers again, presumably for election-related analysis. Welcome, nerds. This article is probably one of the longest I've written. A good 15 minute read. But it gives you a feel for 1) My background 2) My approach to science realcarlallen.substack.com/p…

Carl Allen avatar
Carl Allen@RealCarlAllen·4h

It's actually remarkable how willingly people in field engaged with me until they realized how right I am. Most of all, that I will not let them get away with being wrong. You might find this abrasive - fair. But remember, these people claim to be STATISTICIANS and ANALYSTS.

Post attachment
Carl Allen avatar
Carl Allen@RealCarlAllen·4h

Them: "what if you're wrong?" Me: "if there is sufficient data to show that any belief I hold is wrong, I will change it immediately. No exception, no equivocation, no defense mechanisms, no excuses. You?" Them: "Your tone bothers me. You're not engaging in good faith. Blocked"

Carl Allen avatar
Carl Allen@RealCarlAllen·5h

"Carl why are you so rude to other people in the field?" Have you ever seen a biologist debating a creationist? An astrophysicist debate a flat earther? Now imagine the creationist and flat earther pretend to care about science then ignore data showing they're wrong. Eso

Carl Allen avatar
Carl Allen@RealCarlAllen·5h

Claim: "undecideds can be assumed to split about evenly" Carl, in 2020, within a few hours of researching the topic "But what if they don't?" My research is now worth literally millions of dollars to myself and prediction market traders. Take your morsels when you can.

Post attachment
Carl Allen avatar
Carl Allen@RealCarlAllen·8h

A lot of forecasters hedge because of public perception. Whether intentionally w fudge or indirectly w model choices, doesn't matter If they say something is 99%, and it doesn't happen, they're TERRIFIED If I say 99%, and it happens 199/200 times, I'm PISSED We're not the same

Carl Allen avatar
Carl Allen@RealCarlAllen·8h

I've gotten a lot of hate for my v high win % for NC, GA (99.6%, 99.2%), far above others Clearly, you have no clue what I know, how I know it, or how to solve these problems Things I give a 99.5% probability will "cost" me ~1/200 times. IDC Because I'm better at math than you

Post attachment

With Mississippi and Alaska now officially called Here's a look at where different forecasts disagreed This is the kind of work I do Post-Election, to figure out where and why we disagreed. Being directionally correct means nothing if it's not durable and repeatable.

Post attachment
Carl Allen avatar
Carl Allen@RealCarlAllen·8h

The nice way to put it is that polls are not predictions, so searching for which variables to weight (or not weight) in order to retroactively match the margin is Ptolemaic Astronomy. I'll get rich off them misinforming people instead of getting paid pennies to educate lmao

Carl Allen avatar
Carl Allen@RealCarlAllen·9h

Analysts and pollsters are arguing about whether weighting by past vote would have made polls "more accurate" I am, once again, blessing their hearts for the free EV. I thought it was education? They're always looking for the magical weights to convert polls to predictions.

Post attachment
Carl Allen avatar
⟳ Carl Allen retweeted
Carl Allen@RealCarlAllen·Nov 16

You: you were partially wrong Me: about what You: you just shout. Find someone to give you advice. Get therapy. Rage. Me: right so you can't show me anything I was wrong about, partially or otherwise. Just to be perfectly clear about what happened here. 🤏

Carl Allen avatar
Carl Allen@RealCarlAllen·12h

The rightmost column tells you the range of win probabilities for each state. These numbers are far too different to be explained by chance. Some are predictably wrong in a predictable direction for reasons I was happy to write about in advance. It's the rails, stupid

Post attachment
Carl Allen avatar
Carl Allen@RealCarlAllen·12h

With Mississippi and Alaska now officially called Here's a look at where different forecasts disagreed This is the kind of work I do Post-Election, to figure out where and why we disagreed. Being directionally correct means nothing if it's not durable and repeatable.

Post attachment
Carl Allen avatar
Carl Allen@RealCarlAllen·12h

With Alaska and Mississippi both now official, my forecast proved to be the most (and only) profitable one. No, this doesn't prove it *was better* because sample sizes of 1 prove nothing Just a nice data point.

Post attachment