Crémieux Profile picture
Jan 8, 2024 • 10 tweets • 3 min read • Read on X
Achievement gaps do not simply reflect geographic sorting.

This fact has been known for more than a century now, but some people still don't get it.

Short🧵

First: achievement gaps exist within the same schools.Image
This is not just true for race, it is also true for class. Image
When it comes to district-level family incomes by group, we see that intercepts differ.

In other words, even when lower-performing groups are wealthier, they tend to do worse than higher-performing groups. Image
If you're reading this, you probably know that this replicates the results for rearing household income (1/3):
Consider Reardon, Kalogrides & Shores (2019). They found that even in districts where lower-performing groups had higher SES than Whites, they still tended to perform worse.

In this picture, 79% of the mean Black-White gap is independent of SES. At the metro level, 62% was. Image
It is a pernicious myth that test scores and achievement gaps are mere reflections of geography and that geography plays a major causal role in test scores.

The truth is that their association primarily reflects selection.
I neglected to mention this: results from Moving to Opportunity suggest no significant effects on test scores.

• • •

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

Keep Current with Crémieux

Crémieux Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

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

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @cremieuxrecueil

Jan 26
Thread.

On the left, you can see a map of corruption indexed by the number of mob crimes per 100,000. On the right, you can see corruption indexed by how much people steal from the public purse.

And in the middle, a map of inbreeding.

Clannish people do clannish crimes. Image
Though it's noted in the image, I want to reiterate that the corruption measure on the right is reverse-coded, so higher values indicate lower corruption.

The correlations with consanguinity are 0.65 and -0.52, and they hold up splitting the country in half and in other specs.
Outside of Italy, in the wider world, corruption perceptions also relate to consanguinity.

The correlation is high, and far from perfect, but both measures contain error, so keep that in mind. Image
Read 26 tweets
Jan 26
The largest price-fixing operation in U.S. history took place when @tevapharm hired a woman to do "price increase implementation."

Through LinkedIn & Facebook, she organized a multi-billion dollar cartel, singlehandedly increasing generic drug prices.

There are lessons here🧵 Image
When the cartel started, the companies in question started filing ANDAs, the FDA's "Abbreviated New Drug Applications" to start selling a generic version of an existing drug.

You can see that the involved parties started filing and getting approvals en masse.Image
When the log(price) hikes are stratified across markets, we see that the cartel was either better able to or more greatly desired to keep prices elevated in smaller markets.

Which makes sense! When the drug is rare, it's easier to successfully collude. Image
Read 14 tweets
Jan 25
Multifamily rental management companies have recently begun adopting algorithmic pricing tools to adjust their rents.

In recent years, about a quarter of buildings and a third of all units are managed with these tools.

What's the effect?🧵 Image
The answer to this question depends on when you ask it.

Early adopters who picked up the technology in 2007, for instance, saw small decreases in asking rents, as well as small increases in occupancy rates.

Algorithmic pricing was making the market better for customers! Image
But, later adopters who picked up the technology in 2013 saw increases in rents and declines in occupancy.

The impact of algorithmic pricing had gotten less consumer-friendly over time.Image
Read 14 tweets
Jan 18
The end of affirmative action should have brought about major changes at American universities.

But did it?🧵

One piece of evidence it did is that the Black student share declined: Image
Another piece of evidence is that the Hispanic share declined: Image
And, consistent with Whites and Asians no longer trying to bilk the system, the multiracial share also dropped: Image
Read 10 tweets
Jan 17
It's well-known that a very small portion of the total criminal population is responsible for the overwhelming majority of all crime.

A new study shows that this is also true of prison misconduct:

Just 10% of prisoners are responsible for more than 70% of misconduct in prisons! Image
The above numbers were for males. Here are the numbers for female prisoners.

The numbers are eerily similar. Image
Misconduct overrepresentation holds adjusting for time served in prison, and being a high-misconduct prisoner is predicted by being younger, Black, having a more extensive criminal history, being a violent criminal, being in a state facility, using drugs, and mental disorders.
Read 5 tweets
Jan 16
I used to like this chart, but now I think it's too misleading and we should leave it behind in 2024.

🧵 Image
The key issue is how household size is adjusted for.

In the OP image, they divide by the square root of household size. This is problematic because it means Gen Z incomes are being inflated to the extent they live with their parents.
Generally, when I hear that the younger generations are more successful, what I think is that they're more successful in the stereotypical ways:

They've got relatively better jobs, relatively bigger homes, relatively faster cars and all that.

But the OP graph isn't that.
Read 8 tweets

Did Thread Reader help you today?

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!

Follow Us!

:(