The book’s own web site

Mismatch—How affirmative action hurts students it’s intended to help, and why universities won’t admit it.

This is a book that reports a number of not-well-known facts, and offers some suggestions about what to do. The book has its technical side, but most of the detailed proofs and evidence have been cached at the book’s web site.

The authors are a former Chicago community organizer and civil rights activist turned UCLA law professor, and a former NYT Supreme Court reporter who is now a fellow of the Brookings institute.

The key facts are that

(1) The best predictors of success in college are standardized test scores and grades. (Call a weighted combination of these your “academic index”.) Holistic factors don’t trump these. Standardized test scores are valid predictors for Black, White, Hispanic, and Asian alike. Grades too, though with grades, one may need to take into account whether the school awarding the grades grades hard or easy.

(2) Most selective universities use racial preferences when deciding on admissions, and use them with a fairly heavy hand. There is a “cascade effect” in which schools at the top enroll most of the students who would be a good fit for the next tier of schools, and those enroll most of those who would be a good fit for the next tier down, and so on.

(3) There is a considerable “test score gap” situation. The average score on the SAT or the ACT, average grades, etc. run lower for Blacks than for Whites, who in turn run lower than Asians.

(4) Students generally learn best when they’re in classes pitched to their own level. If you’re a good student with a lot of ability, but you’re put into a fast-paced course aimed at honors students with finely honed skills and exceptional talent, you’ll learn less from that class than you would have learned from a class aimed at students like you, students who want to learn and have the ability, but need more guidance and more approachable homework assignments.

So it’s not a good thing to be admitted to a school where most of the students have a considerably higher academic index. Your chances of staying in school, completing your intended major, and passing gateway tests into a profession are all improved by attending a school where your academic index is roughly equivalent to that of most of your peers.

In other words, large racial preferences in admissions hurt the students who get them.

Special attention is given in the book to the most clear cut cases of this: law school, and STEM (science, technology, engineering, and mathematics.) One particularly telling analysis deals with how things played out before and after Prop 209 came into force in California.

This proposition outlawed racial preferences in admission to the University of California schools. For a while after it was passed, the law was to some extent observed. During this span of time, Black admissions to the most selective schools in the UC system fell. But graduation rates improved, and the absolute number of Black students completing their degrees from those schools rose. These results were in perfect alignment with the analysis given by one of the authors before the event, and in sharp contradiction to the predictions of most others.

When it comes to law school, academic index is the key predictor of success in law school. In turn, success in law school, as evidenced by good grades, and then, secondarily, as evidenced by the quality of the school, is the key predictor of success at passing the bar and then success in the profession.

In other words, these indices and grades are valid predictors.

Again, when it comes to law school, students who attend a school at which their academic index is comparable to the bulk of their peers learn more, in absolute terms, and are more likely to pass the bar and succeed in the profession, than students with the same academic index who attend a school for which their scores indicate they are mismatched.

It is, in short, a bad thing for the student to get a large racial (or legacy, or athletic) preference and be admitted to a law school where the courses will be pitched over their head.

Now, as to suggested solutions:

First, it would be nice to improve K-12 education. A test-score gap is apparent even at the start of K, but it grows over time and the problem of that growth is something the schools can address. There are schools that do far better with Black students than the typical public school, among them, some at which Black students outscore the White average. So, it can be done. These schools do not enjoy superior funding. They do enjoy superior teachers and superior student discipline. They do not cherry pick students for high prior grades or standardized test scores.

If it were possible to dismiss substantially below-average teachers, to attract and retain (with targeted raises) substantially above-average teachers, to expel intolerably disruptive students, and to shift into special, separate, classes students who have disabilities that prevent them from coping with mainstream K-12 instruction, then we could expect that the K-12 gap in public school results would at any rate narrow as it has in these charter schools, Kipp academies, and so on.

Second, and because narrowing the gap is a long term project, it would be good to at a minimum inform prospective students of the extent of the racial preference in admissions they are being offered (if any) and of the range of outcomes and odds of success for previously admitted students who got that level of admissions preference and who have already completed, or failed to complete, their studies at that school. In other words, schools should be required by law to be transparent in the way they use preferences and in the outcomes that result.

The authors also recommend that racial preferences be limited to no more than the SES preference, if any, that a school uses in deciding on admissions. They reject the idea of outright bans on racial preference in admissions, partly because experience has taught that such bans can always be circumvented. Partly, also, because a straight preference generates less of a mismatch problem than circumventions that achieve the same intended preferential effect.