🔥Kareem Carr | Statistician 🔥 Profile picture
Stats PhD student @Harvard • My Timeline is a cool, refreshing breeze of nerdy content. Follow me and enter the nerd space safe.
18 subscribers
Nov 11 • 4 tweets • 1 min read
Here is a problem I see with modern liberalism: if you tell a certain kind of liberal that there are two kids drowning and that they can only save one, they would immediately declare that they can save them both, and then act completely surprised when both drown shortly after. If that same liberal could magically go back in time with all the knowledge of what had happened, that person would do the exact same thing again, and then be just as surprised when both kids drowned for the second time.
Nov 10 • 6 tweets • 2 min read
People are getting the wrong end of the stick here. Nobody stopped her from experimenting on herself. The hold up is on *publishing* the results. Here's the quote from the article Image
Nov 9 • 10 tweets • 2 min read
I honestly get a lot of value out of ChatGPT. It feels built for people like me. I find identifying and correcting its mistakes pretty easy because I'm used to grading student assignments, but I also do things that minimize mistakes like: I input:
- examples of past solutions to similar problems
- a high-level sketch of the solution to the current problem
- background information if needed
- warnings about any potential complications or pitfalls
Nov 5 • 7 tweets • 2 min read
This Musk meme speaks to something true which is America is splitting culturally between the college educated vs the non-college educated.

There is however a third group. People who went to college but who think and act like people who didn’t. Image Basically you have these people who went to elite schools like Harvard or Stanford or Yale, who have law degrees and doctorates in many cases, telling the non-college educated that there’s no point to college because it’s not great job training.
Oct 24 • 4 tweets • 1 min read
there is clearly a force or mechanism that causes the US electorate to balance at precisely 50% democrat 50% republican no matter what either candidate says or does.

if i was a social scientist, i would be absolutely obsessed with this. my economics brain says maybe it's that each party is more extreme than the general electorate and is only willing to concede the minimum number of policies necessary to win which is exactly 50% + 1 vote.

this leads to a powerful finetuning mechanism on both sides.
Oct 18 • 6 tweets • 2 min read
For every ten likes, I'll post a new unhinged mathematician quote. Image if it's you or logical consistency, you know which one they're picking. Image
Sep 11 • 4 tweets • 1 min read
This common criticism of Nate Silver's methods represents a deep misunderstanding of statistical modeling. 🧵 Image We *should* be representing the outcome of the election as a probability because the outcome is *inherently* uncertain, and probabilities are how we communicate that scientifically.
Aug 12 • 7 tweets • 2 min read
One of the things I hate most about the cult of IQ is it leads to lot of magical thinking about how the brain works. There’s absolutely nothing shameful about relearning things you use to know. Image Research shows forgetting is a normal part of human cognition. Image
Aug 9 • 12 tweets • 3 min read
How to convince a skeptical colleague that you're right and they're wrong (thread) Image STEP 1: Show respect for their point of view by making a good faith effort to understand what they're doing and how it compares to your approach
Aug 5 • 10 tweets • 2 min read
Infographics of this dataset have been kicking around on the internet for years. It is an insult to real scientists everywhere. For every 10 likes, I will post a new ridiculous fact about how fake and ridiculous this "data" is. They report data on 185 countries but *104* of those numbers (more than half!) are based on *zero* data collected from people from that country. ZERO.

Rather than acknowledge this lack of data, they decided to guessimate based on surrounding countries.
Jun 26 • 8 tweets • 2 min read
People are getting thousands of likes for spreading this misinformation about sex differences. Let me explain why this interpretation of the data is wrong. 🧵 Image If you think 100% accuracy is too good to be true, trust your instincts.

The version of the model shown in the plot was basically fed the sex of the participants. That’s why it’s achieving 100% accuracy. Image
May 10 • 12 tweets • 3 min read
I keep seeing this Huberman clip all over my timeline so let’s use it as teachable moment to learn some statistics. The basic mistake is not taking the people who are already pregnant out of the pool of people who could be pregnant the next month. Of the starting 100, fewer and fewer will remain each month. Image
Apr 3 • 9 tweets • 1 min read
are you always busy but never seem to get enough done? i recently learned a very important lesson about focus: it's extremely powerful when all your projects fall under one overarching goal such that they feed into and enhance each other.
Mar 26 • 6 tweets • 1 min read
Racist propaganda in the form of statistical charts is extremely effective because most people are unaware of the decisions being made behind the scenes. Image What is typically presented to us as “the data” has often been meticulously pre-chewed for our easy consumption.
Mar 24 • 12 tweets • 2 min read
This plot earned a "Wow" from Elon so I guess we need to talk about it. Here's a thread of basic mistakes I see over and over again in these terrible race science plots: Image 1. Inference using crude or unadjusted rates

In a science experiment, we want to compare two groups that are identical in every single way except for the factor we care about.
Mar 23 • 7 tweets • 2 min read
I honestly think 99% of US politics can be reduced to this one projection by Pew Research. The desperate attempts to force women to have more kids, the terrified rants about immigration, the absolute panic about being “replaced”. Image The part of the great replacement theory that’s true is the white population is decreasing as a percentage of the total US population, but it’s a lie that this is somehow brown people’s fault.
Mar 21 • 10 tweets • 2 min read
We often want to know "What are the chances my hypothesis is true given my data?" The answer depends on the baseline plausibility of different kinds of hypotheses which is usually unknowable.

Frequentist and Bayesian statistics are two responses to this lack of information. Image In Frequentist statistics, the response is to accept the fundamental lack of information and try our best to proceed without it.
Mar 9 • 6 tweets • 2 min read
Here's something you might not know. Almost 2000 years before anybody had heard of Newton, Archimedes could do calculus. Image The only problem was Archimedes was a genius so him being able to do it wasn't of much help to the rest of us. Image
Mar 2 • 6 tweets • 1 min read
My sense of tech bros is they tend to simplify philosophical problems until they're solvable. For instance, the whole spectrum of human relationships gets reduced to a "follow" button. Image They then commodify their simplified solution which makes them rich, and them being rich and powerful makes people think they must have really solved the original problem.
Feb 18 • 6 tweets • 1 min read
This tweet is rightly being condemned for the disgusting racism it represents, but notice how the poster couches his racism as scientific inquiry. This is increasingly what racism looks like on Twitter. Image It’s easy to see where he’s going with this. If fertility rates are in any way different then this must mean these two people are different species, with all the implications that might have for who deserves human rights.
Feb 16 • 4 tweets • 1 min read
Bloom’s Taxonomy is one of the most powerful frameworks I’ve ever encountered. It defines what it means to think deeply. Image For every concept you know, your level of comprehension can be placed along this spectrum.