A Foolish Decision on Testing

In November 2021, the Board of Regents of the University of California voted to permanently abolish using SAT and ACT scores as factors in determining admission to the university. The stated reasons are: (1) standardized tests are biased against students from socioeconomically disadvantaged backgrounds (essentially meaning low-income households where the parents do not have college degrees) and (2) the SAT does not substantially improve the university’s ability to predict college grades beyond just using the student’s high school grade point average.

These stated reasons are factually wrong. Furthermore, the Board of Regents displayed a complete misunderstanding of how students were actually admitted before the ban on using the SAT took place.

With regard to the relative importance of the SAT vis-à-vis grades, here are the data published in 2020 by the University of California Office of Institutional Research. In 2015 (the last year the study covers), grades alone explain 13 percent of the variance in first-year GPA, SAT scores alone explain 22 percent of the variance in first-year GPA and when added together, grades and SAT explain 26 percent of the variability in first-year GPA. So, adding SAT scores to grades doubles the predictability of college performance. This is a major improvement.

Next, I turn to the issue of bias. We first need to know what it means for the SAT to be biased against students who are low income and/or have attended high schools that are predominantly filled with low-income students. The SAT is biased against socioeconomically disadvantaged students if the SAT underpredicts their performance at the university. The standard measure of performance is first-year GPA. In over 50 years of testing and in hundreds of peer-reviewed scientific articles, the results are clear—the SAT is biased in favor of socioeconomically disadvantaged students and biased against socioeconomically advantaged students. In particular, if two students have identical SAT scores and grades or just SAT scores alone, the socioeconomically disadvantaged student will have lower university grades than the socioeconomically advantaged student will have. Furthermore, the student from a lower socioeconomic background will have a lower college grade than predicted by SAT scores and grades, while a student from a socioeconomically advantaged background will have higher college grades than predicted.

Given the above paragraph, why do people believe that the SAT is biased? Here is one reason given by one member of the Board of Regents: students from wealthier families can take courses to improve their SAT scores. I have no doubt that there is some truth in this. But how important is it? Evidently, not enough to overcome the inherent bias against socioeconomically advantaged students shown in the previous paragraph. Otherwise, socioeconomically advantaged students would do worse in college than predicted. There is another reason why people might believe that the SAT scores are biased—low-socioeconomic-status students have lower SAT scores. But it is uncontroversial that, other things being equal, the higher the student’s SAT score, the better the student is likely to do in college. It is unfortunate, but true. The problem does not lie with the SAT. Do not blame the weather forecaster when the prediction of a hurricane is correct.

The Board of Regents appears to be also misinformed regarding how students are actually admitted to the university. First, junior transfers are not required to submit either SAT scores or their high school grades. With regard to beginning college students, the University of California does use the SAT score and high school grades as a guideline for admittance, but particularly at the lower end of the SAT and grades range, the university strays away from these guidelines and targets individuals from socioeconomically disadvantaged families in order to pursue a more nuanced goal. Essentially, UC admission committees gave and still gives points for being from a low-socioeconomic-status background and/or attending a high school with a high proportion of low-socioeconomic-status students. This is perfectly legal, and it is not just a small number. For those who think that groups are not targeted, that the numbers of students involved is small or that SAT scores must be high in order to be admitted to UC, the following may alter your preconceptions. At my campus (UC Santa Cruz), about 32 percent of the first-year students entering in 2006–12 were poor. The bottom 25 percent of the first-year class had an average SAT score that was in the 32nd percentile of everyone taking the SAT. In contrast, students in the top 25 percent of the first-year class had an average SAT score in the 90th percentile.

UC wants to admit more socioeconomically disadvantaged students than would be the case if admission were solely based on expected performance at the university. Therefore admission committees award points for being socioeconomically poor. But, among the disadvantaged, the university would like to admit those who are most likely to do well. And the best predictor of how academically successful a student will be at the University of California is to consider the student’s SAT scores and grades.

Surprisingly, SAT scores are actually better predictors and grades are worse predictors of how well socioeconomically poor students will do at UC than SAT scores and grades are for students who are not poor.

The reason for this apparent paradox is that socioeconomically poor students with low SAT scores will have a much lower expected university GPA than socioeconomic advantaged students with similarly low scores, but as the SAT scores rise, the difference decreases (and at the very highest SAT scores, the relationship may be reversed). The appropriate policy is clear: when choosing among socioeconomically disadvantaged high school students, SAT scores are even more valuable for predicting University of California GPA than when the students are not disadvantaged. Eliminating the SAT will result in the university accepting academically weaker students who are socioeconomically advantaged as well as accepting academically weaker students who are socioeconomically disadvantaged. That is just a dumb policy. In a nutshell, if the Board of Regents wanted the University of California to admit more socioeconomically disadvantaged students, rather than getting rid of the SAT, they should have encouraged the campuses to admit more students from disadvantaged backgrounds. And the easiest way to accomplish that would be for admissions to weight being disadvantaged more heavily in any holistic review. Of course, the arguments raised here are not just confined to the University of California.

While some might not agree, a legitimate argument for increasing the number of students from socioeconomically disadvantaged backgrounds at a public university like UC is that the income distribution of the students’ families should mirror more the income distribution of families in the state, even if this means that academically weaker students will attend the university. From a slightly different perspective, it is reasonable to admit socioeconomically disadvantaged students who have faced a more challenging learning environment, even if their college grades will be lower. But if your goal is to increase the number of students from disadvantaged backgrounds at UC, then claiming that the SAT is biased against these students provides a more emotionally compelling story of right versus wrong—even though it is not true. It is an unfortunate characteristic of our times that both the right and the left cast policy in Manichean terms rather than as a trade-off between competing goals. I suggest that each individual determine the proper balance between their notion of merit and their notion of equity. Unfortunately, this balancing act is impaired when the facts are distorted, which is especially disconcerting in a university setting.