College Admissions, Race, and Unintended Consequences

Note: I wrote this before the SAT changed–again.

The Big Reveal on UC Berkeley’s holistic admissions process created much fuss, most of it on behalf of Asians who are clearly the victims of discriminatory behavior.

Most people don’t completely understand how this “problem” came about, and why the UC and other universities are discriminating against Asians. It’s not so much “affirmative action for whites” as it is unintended consequences of being forbidden to use affirmative action for blacks and Hispanics.

The GPA Demographic Footprint

In November of 1996, the UC system was told by the people of California that it was not allowed to consider race in admissions anymore. The UC system, like all universities in this country, wasn’t about to listen to the rabble. So, as Saul Geiser writes, the system went looking for a reason to reduce the weight given test scores.

Such differences in the demographic footprint of HSGPA and standardized tests are of obvious importance for expanding access and equity in college admissions, especially at those institutions where affirmative action has been curtailed or ended. … at those institutions where affirmative action has been challenged or eliminated, admissions officers have been forced to reevaluate the role of standardized tests as selection criteria.
…
The result has been a de-emphasis of standardized tests as admissions criteria at some institutions. …
UC introduced “comprehensive review,” an admissions policy that more systematically took into account the impact of socioeconomic factors, such as parents’ education and family income, on students’ test scores and related indicators of academic achievement. [note: this is the process described in the NY Times article]. UC also revised its Eligibility Index, a numerical scale which sets minimum HSGPA and test-score requirements for admission to the UC system; the revised index gave roughly three-quarters of the weight to HSGPA and the remainder to standardized tests. … Under this policy [of Eligibility in the Local Context], which also took effect in 2001, students’ class rank within high school was determined solely on the basis of their HSGPA in college preparatory coursework, so that the effect of this policy, too, was to diminish the role of standardized tests in UC admissions.

So de-emphasize those evil, racist tests that traditionally represent, in the typical progressive’s mind, a means of reinforcing the institutionalized hegemony of the white man’s values. Grades, in contrast, reflected the school’s values, the school’s priorities. So majority URM schools, both charters and inner city, can put whatever grades they like on classes that can be called whatever they want. UC officials made the change, along with Eligibility in the Local Context, so that majority URM schools could lie about their students’ academic abilities properly reflect the students’ diligence and abilities in subjects simply not valued by the institutional racists at the College Board.

The problem is, alas, that UC admissions made changes to their policy based on the “demographic footprint” of tests, but they forgot about the demographic footprint of grades.

Namely: Asians, particularly recent immigrant Asians, kill whites on grades. The test score advantage is getting (suspiciously) worse, but the grade advantage is huge.

That wasn’t part of the plan. Look, universities know the game as well as anyone: grades are a fraud. That’s why, until relatively recently, all universities weighted test scores as high or higher than grades.

If high school grades were objectively accurate, why does the University of California have an entry level writing requirement?—and why is that writing requirement either a test or a college level course? (And I have my own doubts of college level courses, but more on that later.) Why is remediation a huge issue in state colleges? If high school grades meant anything, schools could just accept students with high grades and hey, presto. Problem solved.

But Saul Geiser is a good researcher, and his study finds that HSGPA is as good or better a predictor of freshman GPA as test scores. Sure. But that brings up another point: Freshman GPA is pretty worthless, too. It’s a metric that goes back to a time when everyone took the same classes. It’s ancient. It predates the growing disconnect between grades and ability.

I’d go further and argue that in total, college GPA is worthless for much the same reasons that high school GPA is. We hear constant stories about grade inflation at elite schools, while public universities are under tremendous pressure to pass as many wholly unqualified blacks and Hispanics as they can, given the huge number that can’t even get past the remedial classes. How can grade point average mean anything in a world that requires some students to pass calculus and while others only take remedial math (which they can skip if they got an SAT score of 600+)?

If college grades were objectively accurate, the “mismatch theory“, for better or worse, couldn’t even be conceived of. If an A at Harvard is the same as an A at Berkeley which is the same as an A at Florida State, then lower ability students couldn’t get higher GPAs at Florida State and Richard Sanders shouldn’t be pushing his mismatch theory. Besides, what do you suppose the renewed push for a college exit exam is about? And in a world where we did require an exit exam, what would be the best predictor of passing rates—college grades or incoming SAT scores? Everyone knows that answer: unless the exit exam was rigged, we’d find that passing rates were best predicted by SAT scores, which would show a distressing, racially uneven, distribution.

I understand that GPAs are a useful metric because the people who use them filter the data through context—race, school, major, and so on. That’s fine, but not when you have the University of California claiming that HSGPA predicts four-year college outcomes when the university in questions sending easily half of its URM admits into remediation classes.

UC knows this. The whole GPA thing is just cover. What did a little lie matter, if it allowed them to bring in more blacks and Hispanics and thwart the will of California voters? The only people hurt would be the kids who didn’t get 4.0s.

Except that turned out to be a whole hell of a lot of kids with really good SAT scores, a whole lot of them white. So let’s look at four UC campuses:

Campus	Ranking	3.75+ GPA	Non-Res	SATR		SATM		SATW600
				% 600+	% 700+	% 600+	% 700+	% 600+	% 700+
Berkeley	1	79	540	38	36	28	56	34	46
San Diego	3	86	1530	42	15	41	42	46	22
Santa Barbara	5	75	228	41	13	47	24	44	17
Santa Cruz	7	33	16	26	6	31	7	24	7

The reality of demographic footprints being what they are, the kids represented by these numbers are almost exclusively white or Asian. Any black or Hispanic getting scores above 600 are usually going to a higher ranked private school. Keep in mind that these stats are leaving out UCLA, Irvine, and Davis, technically ranked second, fourth, and sixth, although the difference between UCSB, Irvine, and Davis metrics are primarily in the demographics.

In other words, a hell of a lot of kids are getting 2000+ SAT scores who aren’t getting into the top 3 schools, while a whole bunch of kids with 1800-2000 scores are. And the difference, for the most part, is grades. In the days before California banned affirmative action, the UCs weighted grades and test scores much closer to evenly—a 3.8 GPA with excellent test scores and a demanding schedule could easily get into Cal or UCLA. No more.

So the GPA edge led to unanticipated consequences and a huge advantage for Asians. But that was just one of the problems.

Changing the SAT

The UC system wasn’t content with just devaluing test scores, so they tried another change that had terrible blowback, again in the Asian category. First, in 2002, Richard Atkinson called for an end to the SAT and a greater use of the SAT Subject tests. By 2005, the SAT had been completely revamped, the College Board having clearly understood the hint.

While I can’t be certain, I think the original changes were intended to increase the number of blacks and Hispanics eligible by making the test burden lighter. At this point, the UC doesn’t appear to have been thinking about Asian overrepresentation. I’m not sure this was caused by anything other than Atkinson’s pet fancy, which is sad, given the consequences—which have gone well beyond Caliornia.

I haven’t been a fan of the new SAT. My sense has always been that it became much easier, and more amenable to swotters. But not until I worked on this piece did I realize how much easier, and how much more apparently coachable it is, particularly for those who take hundreds of hours of test prep. The College Board releases percentile ranks by race and ethnicity, but I can’t find the original files for anything before 2005. (If anyone can, then you’re a hell of a googler and send them my way.)

However, I found a book that cites exactly the file I want.

So in 1995, 14% of Asians, 5.8% of whites, and .6% of blacks scored over 700 in math, which means that the percentile for 700 was 86%, 94%, and 99%. In 2010 (confirm here), those percentiles were 77%, 94%, and 1%.

Only Asians got a lot smarter? Weird. Not impossible. A lot more Chinese and Koreans are taking the test. Not my pick as an explanation, though.

In 1995, the verbal percentile ranks for 700 were (I think) 98.2% Asian, 98.7% for whites, and 99+% for blacks. While this article doesn’t mention it, the 2010 rankings show that the corresponding 700 percentile rankings are now 92%, 94%, and 99%.

Have whites and Asians have gotten a lot smarter in verbal? No. If you won’t take my word for it, check out GRE scores during that time, which was very similar to the 1995 SAT throughout the 90s and before and after, did not see a corresponding increase in scores.

Or–my pick–the reading test has gotten a lot easier, which I’ve been telling to anyone who will listen, and, in my opinion, it’s gotten easier in a way that allows it to be coached more effectively. And the coaching has become more effective over time. Let’s look at all the data together, with a couple more years from the new SAT:

Year	Math 700 %ile		Verbal 700 %ile
	Asians	Whites	Asians	Whites
1995	86	94	98	99
2006	81	94	93	95
2010	77	94	92	94
2012	75	93	91	94

(Cite for 2006, 2012).

Weird. SAT scores are generally pretty stable. Until I started researching this, I had no idea that the Asian increase was that dramatic, and it is part of a series of discoveries over past year making me wonder if the big gap between Asian and white test prep use (and time spent in test prep) is doing more than just giving Asians a slight edge. This piece is long enough without bringing up the ACT, but I believe that a 700 M corresponds to a 32—or it used to, anyway—and notice that a 32 is only 85%ile for Asians. I have been working on these two essays for ten days or so, and I haven’t yet been able to find ACT percentiles by race over the past ten years or so. Reading and English appear to be roughly the same. However, almost three times as many Asians take the SAT as take the ACT.

One other thing to keep in mind—the number of native Chinese and Koreans taking the test has exploded. Are they fluent in English? Ask any university freshman at an elite school with a Chinese or Korean grad student instructor. So by any stretch, the Asian mean should have been dropping slightly, shouldn’t it? Which means either Asian Americans have gotten phenomenally better, the Chinese/Korean nationals are also getting high Verbal SAT scores, or….what? What explains this jump?

Whites had increased scores in reading, which I believe supports my contention that the 2005 changes were easier. Why did white math scores stay stable? That’s part of a longer post that’s all intuition on my part. Suffice it to say that the SAT math section is both shorter and easier. This helps people with high attention to detail who aren’t as intuitively strong in math. (yes, I know, the SAT supposedly tests higher math since 2005. Too long for this post, but I would disagree.) However, I do understand there could be other factors at work.

I’ll write more about various things that might be the cause of the extremely strange boost in Asian math and verbal scores in the next post, but for now, let’s leave it at this: UC had a big problem. The changes they had hoped might improve scores for blacks and Hispanics (my interpretation) had instead led to unimaginable increases in high Asian scores in the SAT.

More Subject Tests! No, wait. No Subject Tests!

The College Board obligingly did UC’s bidding in making the test easier (my read), but UC continued to follow Atkinson’s directive by requiring two optional subject tests. Previously, applicants had to take two required tests that were much easier than the others (Writing and Math 1C), and then one optional test. Now, the Writing and Math 1C tests were disallowed, and students had to choose two of the “harder” subject tests to take. This requirement handed Asians an enormous advantage over whites. To see why, just check out the percentile rankings of the subject tests—but before you do, close your eyes and stereotype like mad. What subject tests do you think Asians are more likely to take? Okay, now look. Get a load of those Chinese and Korean tests! Roughly half of all takers get an 800–and check out how many of the testers are native speakers. My goodness gracious.

Notice, too, the skew on the Math 2c test. I’ve tutored students in both the Math 2c and English Lit tests for around a decade. It is far easier to prep students for the Math 2c than it is the English Literature test. The test is considerably more difficult as well, because high schools directly cover the math subject matter, but don’t cover the English Lit subject matter, which relies far more on, dare I say, innate literary analysis skills.

In general, subject tests aren’t subjected (hahahaha) to the same scrutiny as the SAT. It’s well known, for example, that the score distributions are wacky, and that in certain tests–Physics, I’m looking at you–the lowest score is higher than the usual 200. I’m not a psychometrician, but I’d love to have someone make the College Board explain this document. Why, if more people finish the English test and get the same amount right, are the scores so much lower? Why, given the much higher accuracy rate of the Korean and Chinese listening tests, aren’t the tests made more difficult? But I digress.

And so, just a decade after Richard Atkinson called for an end to the SAT as an admissions tool and complete reliance on the Subject tests, UC does a complete 180, ending the use of subject tests in admissions. Asians knew instantly this was All About Them, and there was little attempt to deny it:

It was also noted that white students seem to be the winners under the new guarantee; this should be of concern to Council. BOARS Chair Rashid acknowledged that the percentage of white students does indeed go up and the percentage of Asian students goes down. The reason for this is that Asian students seem to be very good at figuring out the technical requirements of UC eligibility. If the subject exam is removed, even more white and Asian students meet the requirements of eligibility. Political perception is another concern. This proposal should also not be viewed as the ultimate solution to diversity. It was also noted that the numbers of females goes up under either scenario.

(emphasis mine.)

Please note that bolded comment. They aren’t saying drats, Asians are smarter. Or drats, we need diversity.

Read the rest of the memo to see the various hoops they jumped through in order to get this cover.

Back where it all started
You know what would have been much easier? Require four Subject tests: English Lit, Math 2c, American History, and a Science. Asians would still do well, but it would have been harder. Dump the SAT, dump or devalue grades. If nothing else, we’d be giving smart kids of all races a chance to show their stuff purely through test scores, imperfect as they may be, rather than the vagaries of teacher assessment.

But that gets UC right back to the problem it started with, the reason it emphasized GPA over test scores in the first place, the problem that it created just to give them a cover story for ignoring the will of the California voters (and, eventually, the constitution as interpreted by the Supreme Court). A test-score only admissions process would eliminate almost all blacks and Hispanics from consideration. The problem: every attempt to bring in more blacks and Hispanics leads to more Asians.

Problem? Why is it a problem? Shouldn’t the universities just let the chips fall where they may? If the schools are overwhelmingly Asian, so what?

Well, for starters, relying exclusively on grades leads to Kashawn Campbell at the low end—hell, Kashawn’s story singlehandedly reveals the need for test scores, the fraudulence of high school grades, and the sketchy nature of college grades in one neat little package.

But more importantly, a huge number of the Asians admitted are either nationals or first and second generation Chinese, Koreans, and Indians.

None of what I’ve written or will write is intended in any way to rationalize the discrimination against Asians. Quite the contrary. Any fair admissions process would lead to overrepresentation of Asians. But I hope to persuade readers that college admissions in its current form, in both private and public schools, is so corrupt that getting outraged about discrimination for or against any one demographic is pointless. Any outrage you find is counterbalanced by another, and no, it’s not as if it all works out.

Why Most of the Low Income “Strivers” are White

So I was reading David Leonhardt’s story on elite colleges and low income kids with high test scores—not news, since I’d read Steve Sailer’s post on the study earlier—and was pleased to see that the reporter had at least mentioned race: “Among high-achieving, low-income students, 6 percent were black, 8 percent Latino, 15 percent Asian-American and 69 percent white, the study found. ”

Of course, while Leonhardt mentions race, he doesn’t mention that gosh by golly, those numbers are lopsided, aren’t they? and none of the posts or tweets I’ve read mention that tremendous imbalance (other than Steve Sailer, of course). Mokita–the truth we all know and agree not to talk about.

Steve said in his earlier post that he was “guessing” that the reserve of kids was white, and of course he was right. What I’d like to remind everyone, while they’re all ignoring the truth, is that Steve didn’t need to guess.

While Hoxby defined “high achieving” as 1300 SAT M-V, let’s be clear: no white or Asian kid without legacy parents or uncommon athletic or artistic ability has any shot at all at a top 20 school without a GPA of 4.0 or higher and SAT combined score over 1400.

According to the College Board, however, just 1500 African Americans scored 700 on either the Math or Reading SAT—which means almost certainly fewer than 1500 scored 700 on both.

The number of African Americans at the top 20 schools, using 2008 data (saved me looking up the individual common data sets), is 2,217.

Okay, a couple of the top 20 schools field football and basketball teams, but the steep SAT skews for athletes are usually found at the big public universites. So the entire reservoir of African Americans with genuinely competitive SAT scores (never mind grades) are taken up entirely by the top 20 schools and they’re already scooping into the scores below that marker. It goes down from there.

Hispanic admissions would tell a similar story, since only around 3000 Mexican, Puerto Rican, or other Hispanic students scored above 700 in either section (again, probably fewer achieved over 700 on both). Please don’t make me add up all twenty from the CDS—here’s six of the top 10 adding up to a bit over 1100 Hispanic admits in 2011 or thereabouts.

This article argues that elite schools recruit low income blacks and Hispanics as a two-fer—they are both poor and non-white, but it’s a mistake assume that the black and Hispanic admits are impoverished. Within races, SAT scores rise and fall with income, on average, and since so few blacks and Hispanics make top marks, it’s very unlikely that a noticeable percentage of low income blacks and Hispanics are hitting genuinely competitive scores (and I speak as someone who has coached low income black/Hispanic students in SAT/ACT, and even seen a few 600+ scores). Low income whites outscore high income blacks and tie high income Hispanics on every IQ-proxy test we have, and the SAT and ACT are no exception.

So no one needs to guess that the high scoring low income kids attending non-elite schools are a predominantly white population, and David Leonhardt didn’t need to mention it, although I’m pleased he did. The students in this category have to be predominantly white, as there aren’t enough high scoring blacks or Hispanics of any income level to fill the maw of top-50 universities desperate to pat themselves on the back for their “diverse” population; they are already granting a steep discount by the 20th school on the US News list.

Meanwhile, at 35th ranked NYU, 34-42% of their admits received 700 or higher on the Math or Reading SAT, while only 12-14% of the students were accepted with scores below 600 on either section. It’s probably just a coincidence that their Hispanic and black admits combined were 15%? So by 35th ranked NYU, they are reaching down into the 500s. Berkeley, at #21, accepts 3-5% of students with scores in the 400s, but then Cal has a football team.

None of this is news. But in presenting the problem as one of income, Leonhardt is coming perilously close to misrepresenting the story. It’s not gee whiz, how come poor kids are ending up at local community colleges and low-end state universities, but that poor white kids—and indeed, many middle class white kids—simply don’t have a chance at top-ranked schools because they are being actively discriminated against in favor of lower-scoring blacks and Hispanics of all income levels*. Most whites in both low and middle income categories know this full well, so they don’t bother applying—why waste the time or the application fee. Asians, of course, are also subject to discrimination, but as someone with seven years experience in the Asian test prep industry, I’m less bothered by the 100 point premium they pay against whites. Sounds about right, when compared to a white (or black or Hispanic, for that matter) kid of similar abilities who didn’t prep.

And as I’ve mentioned before now, the two tools the universities use to rationalize the discrimination are grades and course transcripts. Majority URM schools (both charter and comprehensive) can simply lie about their course content and grade based on effort. Unexpected consequence: Asians are overrepresented despite the discount, because white parents just don’t care as much about grades.

None of this will be resolved by the Supreme Court decision; universities have demonstrated unyielding allegiance to URM admissions and rich white legacy donors. But in my perfect world, college admissions would work something like this.

***************************************************
*I’m adding this later. Private schools are also discriminating against all non-legacy students in favor of “development” (wealthy or legacy or both) admits. I guess it’s too much to expect that, after their pursuit of money, their treatment of non-development candidates be even-handed.