A Foolish Decision on Testing

In Inside Higher Ed, Donald Wittman (professor emeritus of economics, UCSC) makes the argument that the University of California (President’s office and Regents) made the wrong decision in eliminating the use of the SAT and ACT exams in admissions.

The basis of his argument is that the SAT scores were better predictors of college performance for disadvantaged students than high school GPA is (and the combination was better than either alone)—and SAT scores are much better predictors for disadvantaged students than they are for students from advantaged backgrounds. Part of the problem is runaway grade inflation in high schools, where “A” is coming to mean “average”, rather than “unusually good”, so that any non-A grade makes a student look bad. It is very hard to use GPA to distinguish among students if they all have almost the same score, and small amounts of noise bounce their ranks around substantially.

There were mechanisms in place for increasing the fraction of students from disadvantaged backgrounds without running afoul of Proposition 209, which bans the use of race-based or sex-based criteria. Those mechanisms could have been turned up a bit, if the goal was really to provide more opportunities for disadvantaged students.

He makes the point that selecting students based on GPA rather than the combination of SAT and GPA means that admissions will do a poorer job of selecting among the disadvantaged students—setting the ones chosen up for failure and denying the positions to equally disadvantaged students who have a better chance of using the opportunity well.

His overall conclusion is simple:

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. 

I agree with most of what Prof. Wittman says, but I think he is missing a key point: the policy was never intended to help disadvantaged students. It was intended to give Admissions carte blanche to discriminate against the “over-represented” students: Asian-Americans and Jews, who generally score higher on the exams, but who generally get poor scores on the fuzzy “holistic” criteria that Admissions uses to select the students they like.

I blogged about this anti-Asian bias a couple of years ago, but I have little hope of seeing it change—discrimination against Asians seems to be acceptable to our legislators and Regents, and discrimination against Jews has a long history of being acceptable to almost everyone who is not Jewish.

UC is implementing anti-Asian policies

Two decisions made by the Regents of the University of California this year initially made no sense to me.

Eliminating SAT/ACT.

In January 2019, at the request of the President of the University of California, the Standardized Testing Task Force started studying whether the known disparities in outcome for SAT tests was hurting low-income students, first-in-family students, or students form underrepresented minorities (Hispanic, Black, or Native American) from gaining acceptance to UC.

Their report came out in January 2020 and is available at https://senate.universityofcalifornia.edu/_files/underreview/sttf-report.pdf.

The report looked at the data very carefully and came to some surprising conclusions:

It is true that the racial mix of students admitted into the freshman class at UC is quite different from the racial mix of high school seniors from the same year. Consider the groups that UC collectively refers to as Underrepresented Minorities (URM): Latino, African-American students and Native American students. In 2017-18, these groups constituted 59.1% of high school seniors in California, but only 37.0%of UC admits among California residents.

The second question is whether this 22.1 percentage point gap arises due to UC admissions decisions. The Task Force considered this in detail. Figure 3C-7 shows that about one-quarter of the gap is due to the admissions decisions of the UC campuses, but the remaining three-quarters of the gap relates to outcomes that occur before UC admissions officers read files. The single biggest factor is that relatively few of the students in the three underrepresented racial/ethnic groups complete the A-G coursework that both UC and CSU require for students to become eligible to apply. The other key factors include differences in the rates of graduation from high school and in applying to UC.

The fact that admission decisions explain only about one-quarter of the disparity in racial mix of high school seniors and admitted UC freshmen is important, but a reasonable person could wonder whether this contribution, although relatively small, might indeed indicate bias of some sort against applicants from some groups relative to others in admissions itself. We will look into this in responses to some of the questions below.

So there is some imbalance in who gets admitted, but only about a quarter of that is due to the UC admissions decisions—and even less is due to the standardized tests:

To re-state this more simply, large inter-group differences in SAT scores do not translate into major differences across student groups in admission rates at UC. This is probably the most important finding the Task Force has made in its data analyses.

Relying more on high-school GPA after eliminating SAT scores is likely to hurt minority students, rather than help them:

Given the Task Force’s findings that the SAT scores are evaluated in a way that effectively renormalizes scores to take disadvantage into account, SAT scores do not appear to play a big role in differences in admission rates between disadvantaged and advantaged groups. Other admission factors play a role in reducing the share of disadvantaged groups that are selected in the admissions process (as well as in reducing the share of disadvantaged groups in the admissions pool). A prime example is high school GPA. The report shows that variations in GPA matter more than variations in SAT in explaining admission rates.

Of course, the Office of the President then proceeded to ignore the data and the report and push for eliminating the SAT, even knowing that doing so would hurt minority applicants.  They convinced the Regents to go along with this.  Why would they do this?  Were they so convinced of their prior beliefs that they ignored every evidence that contradicted them, or were the never interested in helping minority students in the first place, but just using them as an excuse for a policy that had an entirely different motive?

If it is not to assist the admission of under-represented minorities, then perhaps the point is to reduce the admissions of the group that has the highest average SAT scores—the Asian-Americans.

Asking to repeal Prop. 209

UCOP also persuaded the Regents to call for a repeal of Prop. 209, which had added the following text to the California constitution:

SEC. 31. (a) The state shall not discriminate against, or grant preferential treatment to, any individual or group on the basis of race, sex, color, ethnicity, or national origin in the operation of public employment, public education, or public contracting.
(b) This section shall apply only to action taken after the section’s effective date.
(c) Nothing in this section shall be interpreted as prohibiting bona fide qualifications based on sex which are reasonably necessary to the normal operation of public employment, public education, or public contracting.
(d) Nothing in this section shall be interpreted as invalidating any court order or consent decree which is in force as of the effective date of this section.
(e) Nothing in this section shall be interpreted as prohibiting action which must be taken to establish or maintain eligibility for any federal program, where ineligibility would result in a loss of federal funds to the state.
(f) For the purposes of this section, ”state” shall include, but not necessarily be limited to, the state itself, any city, county, city and county, public university system, including the University of California, community college district, school district, special district, or any other political subdivision or governmental instrumentality of or within the state.
(g) The remedies available for violations of this section shall be the same, regardless of the injured party’s race, sex, color, ethnicity, or national origin, as are otherwise available for violations of then-existing California antidiscrimination law.
(h) This section shall be self-executing. If any part or parts of this section are found to be in conflict with federal law or the United States Constitution, the section shall be implemented to the maximum extent that federal law and the United States Constitution permit. Any provision held invalid shall be severable from the remaining portions of this section.

The reason was so that the University could re-institute racial preferences (which they refer to as Affirmative Action).  How badly are racial preferences needed to rebalance the student body?

According to https://www.universityofcalifornia.edu/infocenter/fall-enrollment-glance in Fall 2019, the undergraduate enrollment across all the UC campuses was

category	ugrads@UC	percentage of non-international
International	29754	—
Unknown	5855	2.98%
White	48433	24.66%
Asian/Pacific Islander	75676	38.54%
Hispanic/Latinx	55971	28.50%
Black	9371	4.77%
Native American	1065	0.54%

How does this compare with California population?  According to https://www.census.gov/quickfacts/CA, the California population consists of

category	percentage	UC%/population%
Unknown or multiple races	4%	0.75
White	36.5%	0.68
Asian/Pacific Islander	16%	2.41
Hispanic/Latinx	39.4%	0.72
Black	6.5%	0.73
Native American	1.6%	0.34

So all groups except Asian/Pacific Islander are underrepresented, and whites are more underrepresented than Blacks or Hispanics.  (The Native Americans are the most underrepresented—they have always been treated the worst by the American educational system.)

So the push for racial preferences clearly has a single target—reducing the representation of Asian-Americans and Pacific Islanders.

Update 2020 Aug 4:  John W pointed me to statistics on the ethnic distribution of California high-school graduates from 2014 (https://www.universityofcalifornia.edu/infocenter/ca-hs-pipeline), a somewhat better baseline to compare with because of the rapidly changing demographics of California.

category	HS number	HS %	UC%/HS%
Unknown or multiple races	10,314	2.45%	1.22
White	120,855	28.66%	0.86
Asian/Pacific Islander	57,687	13.68%	2.82
Hispanic/Latinx	203,894	48.36%	0.59
Black	26,056	6.18%	0.77
Native American	2,830	0.67%	0.80

The adjusted numbers show underrepresentation of Hispanics, more overrepresentation of Asian-Americans, but whites and Native Americans are no longer showing strong underrepresentation.  The overall conclusion—that the target of racial preferences is predominantly Asian-Americans—is not really changed.

The whole point of both admissions policies being pushed this year is anti-Asian sentiment.  It is not to help the under-represented minorities, but to discriminate against Asian-Americans.

This interpretation makes perfect sense of both decisions.  UCOP and the Regents knew that they couldn’t get away with saying that they wanted to reduce the admission of Asian-Americans, nor that the group they wanted to help was the white students, so they had to pretend that they were helping the under-represented minorities.

They didn’t care what the conclusion of the STTF report was—they already knew that SAT scores helped Asian-Americans, and they did not care what happened to the under-represented minorities.  They were hoping for a fig leaf to cover their naked anti-Asian attitudes and were no doubt disappointed that the data did not provide them one.

The whole mess looks a lot like the biased admissions of the 1950’s, when various non-academic criteria were added by many elite universities, in order to exclude the Jewish students who were performing the best on all the academic criteria.  We are seeing the same game being played out today, with Asian-American students in the role of the Jewish students.

 

Misleading by UC’s President

It is clear in Janet Napolitano’s latest e-mail to the faculty that she is first and foremost a politician and not an academic leader.  It is a good thing that she is retiring, though I have no faith that the recruitment process for her successor will find anyone better.  In her email, she says:

Changing the standardized testing requirement for undergraduates

Earlier this month, the UC Board of Regents unanimously approved my recommendation that UC suspend the standardized test requirement (ACT/SAT) for all California freshman applicants until fall 2024. This is the culmination of an intense, two-year, research-based effort to evaluate the value and use of standardized tests in admissions, beginning in 2018 when I asked the Academic Senate to evaluate the issue. I am grateful for the dedicated and diligent efforts of the members of the Standardized Testing Task Force, and all involved in this effort.

What she doesn’t say is that her recommendation was the exact opposite  of what the task force recommended.  She completely ignored the “research-based effort” and went with her gut.  One would almost think she was a Republican.

The task force studied the data carefully and found that the way the SAT was being used for UC admissions resulted in it being a useful predictor of retention and college completion, that it was more predictive for under-represented minorities, and that it helped under-represented minorities gain admission.  Essentially all the arguments against using the SAT turned out not to be supported by the data. (Disclaimer: I’ve only read summaries of the report, not the report itself, so I may have gotten a distorted view of it.)

But Janet Napolitano ignored the task force report and went for a purely political gesture—one that makes the UC admissions process more opaque and more subject to manipulation by admissions officers to admit students based on their prejudices and whims.  If the elimination of the SAT is not rescinded by the next UC president, we are likely to see even more selection for white students than currently, as most of the alternatives to the SAT (like extracurricular activities and essay evaluation) are even more correlated with socio-economic status than the SAT is.

UC salary numbers

UC posts their entire payroll (redacting names for student jobs) each year.  The 2016 numbers can now be found at https://ucannualwage.ucop.edu/wage/

I was curious about several things: who were the most highly paid at UCSC, how much coaches were paid (the top four payments systemwide were UCB and UCLA coaches), and how my pay corresponds to my colleagues.

Most highly paid

UCSC had 87 people paid $200,000 or more in 2016.  The most highly paid was Chancellor Blumenthal at $396,866 (though I don’t think tat includes the value of his housing and other perks). There were about 24 administrators in this group, though many of them are technically also faculty, even if they aren’t currently teaching.  All five of those who make over $300,000 are faculty, though only one of the five (Lederman) is listed as a professor, rather than by an administrative title.

A surprising number of those paid over $200,000 were astronomers—they get paid more than I expected.  The highest-paid faculty who are not also listed as administrators are Lederman, Madau, and Lin (all physics, astrophysics, or astronomy).

Although I think that a few of those making over $200,000 are overpaid, the numbers are not ridiculous (unlike the millions spent for some of the employees at UCB and UCLA).

Coaches

There are 671 employees across all campuses with “coach” in their title, with payments ranging from $125 to $3,577,299.  UCSC has 45 of them, but the pay range is only $1,708 to $74,902.  This does not count the 4 “ath trainer” positions at UCSC ($9,736–$43,447).

Coaches are not being paid generously at UCSC, so though I still think it unwise for students to be paying fees for supporting intercollegiate athletes (rather than physical education and recreation, which all can participate in), the coaches are not getting rich off the students (unlike UCB and UCLA, where 53 of the top-paid 60 UC coaches work).  If we add in the “ath mgr” positions, UCB comes out even worse.  A big chunk of UCB’s deficit comes from the stadium boondoggle, but UCB continues to pour money down the athletics rathole.

I’m glad that UCSC is not wasting money at the rate that UCB and UCLA are, but I do wish that UCSC would return to the days when student athletes paid for their own entertainment, rather than taxing other students.

Me

My pay is relatively modest—I came out 430th on the list for UCSC.  UCSC is listed as having paid 12,288 people in 2016, though many of those got only tiny amounts.  Of those getting $1000 or more, there were 10,480, of those making $21,000 or more (CA minimum wage at full time) there were 4,248, of those making $30,000 or more (UC’s theoretical $15/hour minimum at full time) there were 3,580.  So I’m estimating that I’m at around the 89th percentile for full-time workers at UCSC: a comfortable pay, but nothing extraordinary.  Among the professors at UCSC who are listed as professors (not administrators), I’m at 263 out of about 566: a little above the median (the total count includes faculty who were only there for part of the year or who had “visiting faculty” positions, but not “recall faculty” who have retired but are rehired to teach a course or two).

In the UC system as a whole, I’m at position 26,585 out of 141,138 making $30k or more (only about the 81%ile—the med-center campuses pay a lot more than UCSC does).

Postdocs

I was curious was postdocs get paid across the UC system and at UCSC.  The range is huge across the system from $14 to $255,950.  (The tiny amounts are probably not really pay—there are tiny reimbursements and honoraria that get counted as pay in the UC system.)  The huge amount is from UCSF, and probably comes from clinical work by an MD.

At UCSC the range is $557 to $70,833, similar to the range for coaches.  The median pay for postdocs at UCSC is $39,150.  This is just above what the City of Santa Cruz requires as a living wage (currently $16.21/hour plus benefits) and is reasonable for a single person, but not for someone supporting a child as well.

There are not many postdocs listed as such on the UCSC payroll (only 173), and many of them were probably there for only part of the year, so the number of postdocs on the payroll at one time is probably only 100–120.

Teaching Assistants

Graduate teaching assistants (“teachg asst” in the compensation database) are more numerous—there are 1003 listed (without names) for UCSC with payments from $91 to $41,927.  The median pay is $15,219.  Given that the median workload is 20 hours a week for 33 weeks, that is a respectable $23/hour, but it is not enough to live on in Santa Cruz.  MIT’s living wage calculator estimates that a single adult in Santa Cruz County needs about $27,779 before taxes (though the calculation probably needs to be fixed for grad students, as they do get some medical and transportation benefits that can reduce costs, but housing within reasonable distance of campus is more expensive than county-wide).

I was a little surprised to see the variation in how much TAs were paid at UCSC, as I thought that the pay scales were fixed.  Quite a few students got $14,995 (so that was probably the scale amount), but above that almost everyone had a different amount.  I wonder what made the differences?

Baking in inequity in funding

UCOP (University of California Office of the President) and the UC Regents recently decided to limit out-of-state students at UC campuses.  But they did not do so uniformly across all campuses.  They decided to let UCB, UCLA, UCSD, and UCI have more out-of-state students than the other campuses and to keep the extra tuition collected.

This continues a pattern that has been in place at least as long as the 31 years I’ve been at UCSC of ensuring that UCB and UCLA get more funding per student than the younger campuses.  I’ve seen no evidence that UCOP or the UC Regents have any intention of ever treating the campuses equitably.  Even when they are strongly pushed to do so, they find ways to weasel out and bake in extra money for UCB and UCLA (like the “rebenching” initiative, which was touted as redressing imbalances, but ended up making them worse).