Discriminations and Disparities

[Amazon Link]

I have a number of Thomas Sowell hardcovers on my shelves, but since I am now ElderlyOnAFixedIncome, I decided to get his latest, Discriminations and Disparities, via the Interlibrary Loan services of the University Near Here. And it (eventually) came all the way from the University of Wyoming! My deep gratitude to all involved along the way.

Dr. Sowell recently celebrated his 88th birthday, but this book shows he hasn't lost a lot of speed off his fastball. (Sorry about that metaphor, I've been watching a lot of baseball this summer.) This book is relatively short (127 pages of main text), and concentrates on the fallacies people routinely fall into when looking at statistical disparities between groups of people. Those fallacies lead to probably-erroneous conclusions: that either the disparities are due to discrimination or genetics.

There are a number of ways people can go wrong, intentionally or otherwise.

For example, Sowell makes an early point that I haven't seen others make explicitly: outcomes are nearly always a product of multiple factors, all of which are necessary prerequisites. If just one of the prerequisites is blocked or missing for a group, the result is a highly skewed distribution for that outcome. Example: China was a technological leader, coming up with many inventions before the Europeans. But at some point China's leaders decided to impose a substantial isolationism; while China's people remained just as talented as before, this single factor doomed China to backwardness for centuries.

Sowell goes on to slice-and-dice the concept of "discrimination". There's "fact-based" discrimination, the kind we make all the time when making judgments based on objective qualities and empirical evidence—Sowell calls this "Discrimination I". And then there's old-fashioned bigotry, based not on fact but on usually-invidious perceptions and stereotypes: "Discrimination II".

But "Discrimination I" can be broken down further: while we would prefer that decisions be made on totally accurate knowledge about individuals ("Discrimination Ia"), it's also possible for people can base their judgment on individuals from statistical facts about the group they belong to. Sowell's example: suppose 40% of the people in Group X are alcoholics, while only 1% of the people in Group Y are alcoholics. And you are hiring for a position in which having an alcoholic would be ineffective or even dangerous. Deciding to hire from Group Y is "Discrimination Ib": a reality-based call not based in animosity, just a knowledge of probability. Even though that might be bad news for the 60% of Group X who aren't alcoholics.

Sowell's point: don't lump "Discrimination Ib" with "Discrimination II". It's easy (especially for people with no skin in the game) to pontificate about what's "fair": sure, we'd like to judge all individuals on an individual basis. But when that's impractical, and it often is, how can you blame people for making a relatively safer bet?

I've just scratched the surface. Sowell continues to talk sense in an age where people resist that sort of thing.