Tweet

Crémieux

Mar 24 • 6 tweets • 3 min read

Texas seems to have produced a disproportionate number of interesting natural experiments. Here are a few examples.

When Texas tried to get around an affirmative action ban by exploiting de facto school racial segregation, people changed schools so they could get in the top 10%.

People moved; home prices adjusted accordingly.

Among the students who could exploit the policy, at least 5% switched to a different, typically less-competitive high school, and they tended to displace minorities from the top 10% of that school's graduating class when they did.

The implementation of the 10% rule improved the enrollment and graduation outcomes for high-ranked students at low-ranked schools, and it did not harm college enrollment, graduation, or subsequent earnings among students who lost access, and it didn't affect earnings for either.

This sits well with Dale & Krueger/Ge, Isaac & Miller's findings regarding the returns to selectivity being in large part returns to student unobservables.

Texas also had randomized ballot orders, so they provided data that shows that moving from the last to the first position on a ballot has a massive effect for down-ballot races. In this study, it was nearly ten percentage points.

Sources:
journals.uchicago.edu/doi/10.17310/n…
sciencedirect.com/science/articl…

aeaweb.org/articles?id=10…

jstor.org/stable/23799087
journals.uchicago.edu/doi/abs/10.108…

link.springer.com/article/10.100…

• • •

Missing some Tweet in this thread? You can try to force a refresh

This Thread may be Removed Anytime!

Twitter may remove this content at anytime! Save it as PDF for later use!

More from @cremieuxrecueil

Crémieux

@cremieuxrecueil

Mar 23

New data from London shows that if you benchmark on a group's responsibility for homicides, Blacks are both less likely to be subject to police brutality, and less likely to be murdered.

Other groups are more likely to be murdered given their homicide rates.

Why did I use murder as a benchmark? Because it's highly legible: a murder victim is less likely to go unnoticed than any other crime category. And, as the Casey Review confirms, 9/10 murder cases are solved. That's more than any other crime category by far.

Here's another way to look at this data, and a breakout of the different forms of police use of force.

Read 8 tweets

Crémieux

@cremieuxrecueil

Mar 23

@epkaufm

@epkaufm made a good post touching on campus viewpoint diversity a while back and he provided some data in it that was very interesting.

Turns out, the colleges with the most viewpoint diversity were ranked lower.

The relationship is driven by the places that take in lots of conservatives being lower ranked. Moderates - hardly related; Liberals - positively related.

The ivory tower selects for liberal people and they keep aim to keep it liberal while keeping conservatives out.

Moderates can stay, but they're largely irrelevant. Universities averaged 54% Liberal, 26% Moderate, 19% Conservative.

The relationship holds among national universities, but only holds up directionally in the much smaller and more homogeneous sample of liberal arts colleges.

Read 5 tweets

Crémieux

@cremieuxrecueil

Mar 22

Have you ever heard that Beethoven was black? Well now his hair has been sequenced, his DNA is out there, and it turns out, this is where his ancestors were from:

And Beethoven's Y-haplogroup, I1a-Z139, is one that may have originated in Scandinavia

What about his disease polygenic scores?

Beethoven was in the 8th/20th quantile for Crohn's disease, the 13th for ulcerative colitis, the 2nd for IBS, the 19th (!) for lupus, the 6th for biliary cirrhosis, the 12th for sclerosis cholangitis, and the 20th (!) for cirrhosis.

Read 4 tweets

Crémieux

@cremieuxrecueil

Mar 21

https://twitter.com/dffeldon/status/1638196537590435840

If you wondered how bad meta-analyses can get, just take a look at this one.

A huge problem in the lit on GRE validities is conditioning on colliders. This study misses the work that doesn't do that and instead seems to just cite work that does.

Their first page of studies:

https://twitter.com/dffeldon/status/1638196537590435840

This matters because once you've conditioned on getting into a program or graduating, etc. you have already selected the sample on the GRE, and if you haven't selected them for their GRE scores, you selected them for compensating factors, like evidence of being conscientious.

@dingding_peng

Here are two greater preprints on this topic and a blogpost from @dingding_peng

arxiv.org/abs/1709.02895

arxiv.org/abs/1902.09442

the100.ci/2017/03/14/tha…

Read 10 tweets

Crémieux

@cremieuxrecueil

Mar 19

It's become a truism that if every woman had all the kids she wanted, we would beat the replacement rate. Here's the Institute for Family Studies showing just that:

In this blogpost, I've asked and answered a related question: what would dysgenic fertility look like if every woman had all the kids she wanted?

cremieux.substack.com/p/fertility-in…

The answer? Pretty good actually. As it turns out, the biggest gap between ideal and realized fertility is among the smartest women.

Read 8 tweets

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Share this page!

Crémieux

People who liked this thread also liked...

Try unrolling a thread yourself!

More from @cremieuxrecueil

Crémieux

Crémieux

Crémieux

Crémieux

Crémieux

Did Thread Reader help you today?

Don't want to be a Premium member but still want to support us?

Send Email!