The Assimilation Myth — new post published!. Using data from a variety of countries, I show that ethnic disparities in human capital, economic performance, crime and cultural values tend to show substantial persistence across generations.

Newton is probably one of the most extreme volcels in history, at least relative to his accomplishments. He once wrote to Locke that “Being of the opinion that you endeavoured to embroil me with women . I answered 'twere better if you were dead.”

It's still difficult for me to understand that a dean would simply openly say:. “Diversity of the candidates has to be as high a priority as the scholarship.”. That is so crazy to me that I would've thought that even extreme activists would try to put it in milder terms.

If you want to read more, check out my full piece on Africa's poor numbers.

My conclusion is that almost all Sub-Saharan African homicide rates are dubious. We don't know the true rates, but they are systematically underestimated due to underreporting. One possible exception is South Africa, which has the highest statistical capacity.

The Ghana Police Force has also been criticized for “poor data and poor criminal record keeping, slow and sloppy investigations, and insufficient and oftentimes unqualified personnel” as well as “widespread cases of bribery and corruption”.

But is this at all trustworthy? No, of course not. Ghana has very poor death registration, not to mention poor cause-of-death registration.

Let's consider a different African example: Ghana. UNODC reports the homicide rate to be just 1.8 per 100,000. This is based directly on the homicide count provided by the Ghana Police Service with no adjustments. (UNODC and GPS numbers are identical, see below)

The situation is even worse in Africa. When the American homicide rate was calculated, it only used the population within the death registration areas. For African homicide rates, they just divide by the total population, despite there being extremely poor death registration.

The bias was very large. Eckberg's estimates suggest that the true homicide rate was >3x higher than official counts in 1900 suggested. Gradually, as death coverage became more complete, the two rates converged.

Eckberg (1995) shows that at each step the coverage increased, the American homicide rate revealed itself to be higher than less complete data suggested. This is because coverage was not just incomplete, it was unrepresentative.

We even have evidence from the West showing that incomplete death registration is problematic for homicide rate estimates. In the U.S., the coverage of death registration gradually improved, until it covered the entire population in 1933.

It's important to understand that such lack of data leads to a *systematic* bias in homicide rates. Lack of death registration (and cause of death registration) means that homicides are systematically *undercounted* or underreported.

It is difficult to overstate how lacking death registration across the continent is. Deaths are not just not registered, there is also incredibly poor cause-of-death registration (which is needed for accurate homicide counts). (Graph below shows quality of death registration)

However, even these data quality scores are *much* too optimistic for Africa. Why?. Because Africa systematically lacks the most important factor underlying accurate homicide counts: systematic death registration.

UNODC, for the most part (except, e.g., Nigeria), use official homicide counts provided by the relevant government agency. But these are extremely unreliable. UNODC have a national data quality proxy, and Africa systematically ranks poorly (or worse, lacks data to even score).

How do they estimate it? They use a socio-demographic regression model. They basically say “the country is this poor, this unequal, has this many young males, etc, therefore the homicide is probably about such-and-such.”. But we have no idea how accurate such predictions are.

There are two major sources that compile national homicide rates across the world, including African countries: WHO and UNODC. WHO, recognizing the unreliability of African homicide statistics, do not even use the official homicide counts and try to estimate it by other means.

African homicide rates are completely dubious. That's what I argue in a recent piece. Consider, for example, Nigeria's homicide rate. An organization that monitors homicides from news reports find >3x homicides than official counts, and household surveys suggest >10x.

This study is much less informative than the claim here suggests, and especially less relevant for Western contexts. 1) A polygenic score only captures part of the genetic influence, so it's a fallacy to suggest the intervention erases genetic disadvantage. 2) This problem is.

He even hid my reply lol