Crémieux Profile picture
I write about genetics, 'metrics, and demographics. Read my long-form writing at https://t.co/8hgA4nNS2A.
47 subscribers
Aug 26 7 tweets 3 min read
Retatrutide is more effective at generating weight loss and other benefits for the people who take it, but there does seem to be a cardiac safety signal.

This could potentially be important *for people who have preexisting atrial fibrillation and CVD.* Image Keep in mind, only one of these events became serious and the rest just passed: "Reported cardiac arrhythmias were mild to moderate in severity with the exception of one severe adverse event."

The reason for this signal seems to be a dose-dependent increase in heart rate.
Aug 26 24 tweets 9 min read
Gerrymandering is the death of centrism in America.

The more districts are uncompetitive, the less hope there is for moderate candidates, and the less value there is in moving to the center.

Gerrymandering means a more divisive and polarized America, with poorer governance🧵 Image A recent paper in the American Economic Review made the value of competitive elections clear using data from America and France.

Looking at American elections, when candidates are in their primaries, they're more radical. When they compete with the other party, they moderate. Image
Aug 25 10 tweets 4 min read
The World Health Organization frequently adopts irresponsible positions.

For example, they recommended against using non-sugar sweeteners (NSS)—zero-calorie stuff like aspartame.

Why?

Because of non-causal evidence. But all the causally-informative evidence said it was good! Image On the one hand, you have causal evidence screaming about one direction of effects that are theoretically expected. Among this evidence, there's one bad sign, but it's marginally significant (p = 0.012) among a bajillion effects examined.

AND THEY GOT THE EFFECT SIZE WRONG.Image
Aug 25 20 tweets 8 min read
I'm going to speculate about why von Neumann was omitted from Oppenheimer. Thread🧵

"With the Russians it is not a question of whether but of when."

"If you say why not bomb them tomorrow, I say why not today? If you say today at 5 o'clock, I say why not one o'clock?" Image Before the Senate, when asked about his ideology:

"I am violently anti-Communist and much more militaristic than the norm." Image
Aug 25 18 tweets 9 min read
The White House just issued an executive order to end cashless bail.

Over the last decade or so, many jurisdictions have adopted a policy where criminal offenders can walk free without having to pay bail.

They then routinely go on to commit more crime. This targets that🧵 Image Briefly, the way bail works in the U.S. is that the court allows a pretrial defendant—someone accused of a crime—to leave jail before their trial date if they can put up cash which they'll earn back if they appear for court. Image
Aug 25 8 tweets 3 min read
It's true.

Semaglutide works worse than Tirzepatide and Retatrutide, and it has much worse side effects. Image When you compare Tirzepatide and Semaglutide head-to-head on side effects, it's just no contest, and we know exactly why this is.

It should be noted that Retatrutide has a side effect profile like Tirzepatide's. It too manages to beat Semaglutide.

Image
Aug 24 14 tweets 5 min read
Newsom just signed a new plan for California's Congressional districts.

It's an aggressive gerrymander intended to exactly counter Texas' redistricting.

But I don't think this is wise. Go check the next picture. Image Presumably if California does this, they will galvanize the Republican states that weren't already on board with aggressive gerrymandering.

So be it, I guess. If he does this, he'll turn Congress even more one-sidedly Republican.

And there's more! Image
Aug 24 21 tweets 8 min read
"Root cause" discourse is silly; it's also wrong.

The root cause of obesity is eating too much.

The total amount of 'eating too much' needed to explain the obesity epidemic is one McDonald's Cheeseburger a day.

No poisons or toxins or diseases, society just got richer.Image Plenty of people have put in the effort to decompose the change. We know the required consumption increase is small.

We also know that more people have shifted down to lower levels of activity. Jobs are more sedentary as we've developed, so fewer calories out; also, more hunger! Image
Aug 23 11 tweets 4 min read
Kids whose parents earn top-1% incomes are much more likely to be admitted to elite universities, all else equal.

If you want to eliminate this bias in admissions, it won't do to eliminate legacy preferences.

The most effective and fair method is to just use test scores! Image Q: I can see they're the most effective, but why are they the most "fair"?

A: Because money doesn't buy you a higher IQ, nor can it purchase meaningfully higher test scores at a given IQ.

Test scores are not biased by social class like every other part of admissions portfolios. Image
Aug 21 25 tweets 8 min read
One person on this list loves IQ tests.

He tests all of his executives with a modified test like the Wonderlic. That means that you need speed and accuracy.

Let's see if you're smart enough to work at Vista Equity Partners!🧵🧵🧵

First question, 15 minutes on the clock:Image Question 2.

These questions are easy, but you have to go fast.

Because of the time constraints, very few people get all 50. Image
Aug 21 8 tweets 4 min read
Most men are entirely heterosexual, and we can prove this by measuring genital arousal. Image This methodology works in the general population and it works within twin pairs.

Amazing, no? Image
Aug 20 6 tweets 2 min read
I've seen tons of posts showing hypothetical male and female IQ bell curves that show women having extremely small variances relative to men.

Men are reliably more variable, but the empirical difference is not that extreme.

The ACT results from 2024 are representative:Image Men are ~15-17% more variable (remember to square your SDs!) in this sample. If you match the levels (since, often, the VR and the \delta are correlated), you get maybe +20% male variability.

That's not nothing, but it's not devastatingly huge.
Aug 18 22 tweets 7 min read
Having a child changes people.

It makes them far less rambunctious, and pushes them to be more stable.

Notice what happens to men's criminal offending rates in the lead-up to childbirth.

They become far less criminal!🧵 Image This isn't just about taming dads though.

The effect for women is also incredibly substantial, albeit with a small difference: there's a larger reduction in crime during pregnancy, with some rebound after.

But the rebound is only partial.Image
Aug 16 23 tweets 9 min read
One concept I wish more people were aware of is the Tocqueville Effect.

Named for Alexis de Tocqueville, this concept describes the curious phenomenon by which people become more frustrated as problems are resolved:

As life gets better, people think it's getting worse!🧵 Image You go to a supermarket and it's time to get some fruit.

Of course, when you go to pick your bananas and your berries, you want to pick the freshest stuff.

But if what's on display is a little less fresh than ideal, you might consider a speckled banana or squishier grapes OK. Image
Aug 14 14 tweets 5 min read
This is a great way to visualize the effect of divorce on children's success as adults🧵

Children whose parents went through a divorce while they were aged 0-5 ranked about 2.4 percentile points lower in the income distribution when they were 25 years old. Image The other effects—on teen birth rates, mortality, college attendance, and incarceration—are all relatively large while being absolutely small in effect.

In order, those are +73%, +35%, -40%, and -43%.

But here are those absolute effects:Image
Aug 12 13 tweets 5 min read
Why do GLP-1 drugs like Ozempic seem to fix so many different outcomes?

I just read a clever study that might help us find the answer🧵

The method they used was good ole Mendelian Randomization! Image Mendelian Randomization is a causal inference technique that uses genetic variants affecting some trait as instrumental variables (IVs).

To grok IVs, consider an example: how do you estimate the effect of smoking during pregnancy on birth weight?

One way is sibling comparison: Image
Aug 11 4 tweets 2 min read
This is real!

Generally people who say they were sexually assaulted with sedatives involved are incorrect about being sedated.

In the study referenced here, the prevalence of sedatives among cases was minimal (~2%). For comparison, antidepressants were detected 375% more often. Image Other details from the study worth mentioning:

1. This matches up with estimates from elsewhere. Women who think they were drugged generally were not.

2. Rohypnol is not common in general, nor is it commonly involved. Image
Image
Aug 9 8 tweets 3 min read
What happens when you provide students with subsidized or free meals?

Lots of studies have been published on this topic, but somehow the field hasn't reached a consensus.

Why?

Maybe because there's clearly publication bias. When accounting for it, effects fall towards zero: Image If you just look at all the effect sizes in the literature, you might start seeing the issue.

Notice the long tail of positive results? Image
Aug 7 15 tweets 5 min read
The Trump administration has officially taken a stance against debanking.

That means that, soon enough, no more Americans will be deprived of being able to hold a bank account because of the opinions they hold.

Americans will be free to think independently again🧵 Image The executive order begins with some background:

Americans, often at the behest of government officials, have been subject to the loss of access to financial services.

That often meant having no access to bank accounts, debit and credit cards, investment tools, and so on. Image
Aug 7 14 tweets 5 min read
Trump is about to sign a historic memorandum.

The memo is going to change how Republicans operate in a way that is almost unprecedented.

In a few short hours, Republicans will start embracing the power of DATA to undermine their political enemies.🧵 Image What do I mean?

Let today's memo be an example.

Today, the Trump administration is going after universities that have feign compliance with the Supreme Court's ruling in SFFA v. Harvard.

By that I mean, they've continued practicing affirmative action like NYU and Columbia.Image
Aug 6 20 tweets 9 min read
There's been a long COVID-related rise in self-reported disability.

Notice how the rise starting in mid-2020 mostly has to do with an increase in difficulty remembering things?

That's the brain fog symptom everyone became aware of. Image Importantly, in both the ACS—which lacks specific long COVID questions—and in the Household Pulse Survey—which added them in 2022—there's a curious demographic concentration of, first, new disability, and second, long COVID reports:

Young, female, Hispanic, and poorly-educated. Image