Getting rid of the SAT hurt Male UC Applicants

And benefited female applicants, of course

Last week’s post looked at how getting rid of the SAT affected the academic quality of UC’s applications and admits. I had intended to stop there but a reader asked if getting rid of the SAT had any effect on the gender mix and the answer turns out to be yes. I was so surprised at the size of the effect that I thought it worth another post.

Academic Comparison

Although the UC publishes figures on the number of applicants who have taken various numbers of honors classes or A-G classes, it does not break down those figures by gender. It publishes no data with which to compare the average quality of male and female applicants. Let’s do a quick recap of what is known from other sources about the relative abilities of the two genders.

Grades

Girls get better grades than boys. I can’t find California data but, among 12th graders nationally who graduated in 2019, females had an average GPA of 3.23 while males had an average GPA of 3.00¹. I also couldn’t find data on the distribution of grades². Is girls’ outperformance purely because so many more boys do really badly or are there more girls than boys even at the highest achievement levels?

Here’s what happened to the cohort of students who entered public high schools in California in 2020-21. 19% of male students and 26% of female students ultimately applied to a UC.

Even though the average female high schooler has higher grades than the average male high schooler, it doesn’t necessarily follow that the average female UC applicant has higher grades than the average male UC applicant. Do the top 26% of female students have better grades than the top 19% of male students? At the very least, any female academic edge will be much less than the overall high school GPAs would suggest.

SBAC Scores

SBAC scores measure attainment in ELA and Math. Girls score higher in ELA and boys score higher in Math. This is true whether we’re talking about all students or whether we restrict the comparison to just those who Exceed Standards, the highest of the four bands into which results are grouped and the one most likely to apply to a selective UC. In grade 11, males account for only 45.3% of those who exceed standards in ELA but 57.8% of those who exceed standards in Math. If we lower the bar a bit, males are 46.9% of those who meet or exceed standards in ELA and 53.8% of those who meet or exceed standards in Math.

AP Tests

Overall, males take only 44.3% of AP tests but they score 0.20 points higher per test. For a sense of scale, the difference in the average scores obtained by Asian and White students was 0.12 and between White and Latino students was 0.77. The pass rate (i.e. the percentage scoring 3 or higher) was 63.7% for males and 57.1% for females.

There are differences by subject area. Males both take more AP tests in STEM subjects than females and score higher on them. In humanities subjects, females take more tests but males score higher. This was a surprise to me. On the other hand, women take more AP tests in languages, Arts, and Music and score higher. This was not a surprise to me.

SAT Scores

Males score slightly higher than females on average but the difference is greater at higher percentiles. Among those who score 1400 or higher, about 60% are male. If we lower the threshold to 1200, about 53% are male. (For a sense of scale, back in 2019 when the UCs were requiring SAT scores, the average SAT score at Berkeley was 1430 and at Riverside was 1230).

To summarize, females have better grades, males score better on SAT tests, and the evidence provided by SBAC tests and AP tests is mixed. You could call that a tie.

Unfortunately for the males, the UCs only consider those factors where females score higher. SBAC tests were never part of the admissions process while SAT scores are no longer considered. The two academic criteria which are considered Very Important at every UC are Academic GPA (i.e. high school grades) and Rigor of Secondary School Record (i.e. how tough was your schedule). As for AP scores, “students should be aware AP test scores lower than 3 will not adversely affect their chances for admission”. Whether you took an AP class (a factor that favors females because they take more AP tests) is thus treated as Very Important at every UC but your actual grade on the AP exam (a factor that favors males because they score higher) is of much less importance. You can fail the test, in other words, and still get credit for taking the class

UC Applications By Gender

The first chart showed that far more women than men applied to UCs from California public schools. The gender mix among applicants from private high schools in California is not as heavily female. Nevertheless, systemwide, over 30% more women than men from California schools apply to a UC. Men actually apply to more campuses than women so, at each campus, there are typically between 20-30% more applications from women than from men.

In the last post, we saw that applications jumped at every campus after the UCs went test blind with the more selective campuses seeing the biggest jumps. We also saw that the more selective campuses saw jumps in the number of weak applications while the less selective campuses saw jumps in the number of strong applications. I attributed these phenomena to the perceived increased uncertainty surrounding who would be admitted where. Weak applicants thought they might now have a chance of getting in to UCLA and UC Berkeley while strong applicants feared they might not get in anywhere unless they applied to Riverside and Merced.

Interestingly, when we break out the growth in applications by gender, we see that the numbers of female and male applicants grew by roughly the same amount at the selective campuses but that the growth in applicants at the least selective campuses was driven largely by the male applicants.

Admissions Rate

Despite the fact that so many more women than men apply, women are admitted at a higher rate than men. This was true back in 2017-19, when the SAT was still in use, but getting rid of the SAT made the female advantage much greater. At Berkeley, women are 49% more likely to be admitted than men; at UCLA they’re 30% more likely to be admitted. San Diego and Irvine used to admit both the sexes at about the same rates but women now have about a 30% advantage at both. This is not attributable solely to greater weight being placed on high school grades: San Diego stood out last week because of the increased number of weak applicants it admitted.

The results at Davis don’t fit the standard pattern. Women had a 35% advantage back in 2017-19, the biggest of any campus, but only have a 33% advantage in 2022-24, after going test blind. At Merced, there’s hardly any difference between the sexes but that’s because Merced admits nearly everyone.

Given that so many more women apply in the first place, I’d be very surprised if these differential admission rates could be explained purely by academics. It’s probable that men are more likely to apply to the Computer Science and Engineering schools, which have lower admission rates than the other schools, but it’s mathematically impossible for this to explain all the difference in admission rates³.

My suspicion is that the way the UCs evaluate all the non-academic components of an application also favors women. Would it surprise anyone if women were better at storytelling i.e. essay writing? Perhaps their extracurricular activities are also more valued. I imagine that the application readers have trouble distinguishing all the football and soccer players.

Admission Numbers

Women apply in higher numbers and they are admitted at higher rates. Combine the two and you get a a huge difference in the number of women and men admitted. At seven UC campuses, there are 50% more women than men admitted, which is equivalent to a 60:40 gender split. At four campuses, there are 60% more women than men admitted. That’s equivalent to a 62:38 gender split.

Private High Schools vs Public High Schools

Applicants from private high schools tend to be stronger than applicants from public high schools. We saw last week that they were more likely to have completed 10 or more Honors classes. They also tend to score much better on AP tests⁴. In 2019, the last year for which we have data, private school students scored 0.46 points higher on average than public school students⁵. Compare this with the 0.20 points male-female difference or the 0.12 Asian-White difference. Of course, AP scores receive limited weight in the admissions process.

We saw last week that getting rid of the SAT had hurt the applicants from private high schools⁶. The private school applicants retained a slight edge at LA and Berkeley but admission rates at the other campuses were now significantly lower for private school applicants. This week, we’ve seen that getting rid of the SAT has benefited women and hurt men.

The following chart uses the admission rate of female applicants from public schools as the baseline and expresses the admission rates of other groups as a percentage of that. A number over 100% means the group members are admitted at a higher rate than women from public high schools. A number under 100% means that the group members are less likely to be admitted than women from public high schools.

Berkeley and Davis don’t have much of a difference between public and private schools but do have significant differences in admission rates by gender. San Diego has both a gender difference and a public-private difference and they get compounded. A male applicant from a private school is barely half as likely to be admitted as a female applicant from a public school. In 2022-24, the admission numbers were 15.4% of male private school applicants and 28.4% of female public school applicants.

1. U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics, National Assessment of Educational Progress (NAEP), 2019 High School Transcript Study (HSTS).

2. A couple of years ago, SFUSD published some data on grades at each school. They showed the number of ‘A’, ‘D’ and ‘F’ grades for each subject in each grade in each school and broke them down by ethnicity, by economic status, by English learner status, by special education status, and by homeless or foster care status. They didn’t give a breakdown by gender. The only data I can find is indirect. Historically, about 56% - 60% of the incoming 9th grade class at Lowell High School has been female. Since at least 70% of the seats at Lowell are awarded using a points system based on grades and a standardized test, it has been reasonable to conclude that there are more girls than boys in the top 15% of the achievement spectrum. I note, however, that only 50.3% of the 9th graders in 2023-24 were females. I know the district switched to a new standardized test for that year (which was the first after the 2-year lottery experiment was abandoned) and tweaked the points system a bit but I don’t know if that is responsible for the new equality or if this was just a one-year aberration. It’ll be very interesting to see the 2024-25 numbers.

3. In 2022-23, Berkeley received 64,190 applications from women and 58,313 applications from men and 5,707 applications from other gender identities. 13.8% of women and 8.6% of men were admitted. The Engineering school received a total of 29,672 applications and admitted 7.1%. (All these numbers include out-of-state and international applicants because no California-only breakdown is available.) No gender breakdown of the engineering applications was provided but let’s assume that 90% were from men and that they were accepted at a 7.1% rate. The admission rate for men applying to all other schools would then be 9.8%, still far below the observed rate for women.

4. In 2019, private school students took about 11% of the AP tests but I believe private schools are less likely to offer AP tests (relying more on either IB tests or their own honors classes) so I’m reluctant to make any statement about whether public or private applicants take more AP classes.

5. This, by itself, says nothing about the relative quality of public and private high schools. Most, if not all, is going to be due to selection effects. You don’t get as many private school parents who didn’t go to college or who don’t speak English.

6. Admissions is a zero sum game so a penalty for applicants from private schools is a benefit to applicants from public schools but there are about seven applicants from public schools for every applicant from private schools so the average public school applicant doesn’t gain very much.