“These tests are extremely flawed and very unfair.” So said Lieutenant Governor Eleni Kounalakis, a member of the University of California Board of Regents, which voted in May to suspend the ACT/SAT requirement for California students applying to UC schools. The news has been celebrated by scores of commentators who believe that admissions tests discriminate against applicants on “the basis of race and wealth.” Robert Schaeffer, interim executive director of FairTest, called the decision “a huge victory for both equity and academic quality.”
There are many compelling reasons to get rid of college admissions tests, but test bias is not one of them. The assumption that scrapping these tests will open the doors of educational opportunity for deserving, disadvantaged students is alluring -- but it’s a feel-good mirage, wishful thinking of the highest order that ignores the real gaps in the academic preparation of the haves and have-nots as well as the power of wealth to capitalize on virtually every dimension of the admissions process.
How large are the racial and economic disparities in test scores? For the SAT, out of 1600 total possible points, mean 2019 scores were 1223 for Asians, 1114 for whites, 978 for Hispanics/Latinos and 933 for African Americans. The most recent figures for income (derived from class of 2016 high school students) show an average SAT score of 1230 for students with household incomes more than $200,000; 1120 for those with household incomes between $80,000 and $100,000; and 1060 for students whose household incomes ranged between $40,000 and $60,000.
With such salient average score differences, it’s not surprising that so many observers see prejudice and discrimination afoot. Indeed UC regent Cecilia Estolano is far from alone in insisting that the SAT is a “racist test.”
I can certainly understand where this fierce opposition to standardized testing is coming from. The first generation of intelligence tests in the United States really were riddled with cultural and class bias. The architects of IQ tests were eugenicists who, through a combination of deep-seated prejudice and confirmation bias, finessed their data to reinforce existing social hierarchies, as they “proved” that the English, Dutch and Germans were more intelligent than the Irish, Russians and Italians, who in turn were brighter than Mexicans, Indians and “Negroes.” Lewis Terman -- creator of the original Stanford-Binet IQ test in 1916 -- and his colleagues advanced a fixed, hereditary theory of intelligence that was tightly linked to ideas about racial superiority and inferiority. (In the late 1910s and early 1920s, they were flummoxed by the much higher average IQ scores of northern Blacks compared to southern Blacks when the answer -- the Jim Crow South provided far fewer schooling opportunities for Black Americans -- was staring them in the face.)
After it was first administered in 1926, the SAT was marketed for decades as an “aptitude” test. Many of us continue to view test scores as indicators of innate abilities and natural-born potential. I guess African Americans just aren’t as smart as Asians, a casual observer might conclude after glancing at a table of average SAT scores by race. This kind of thinking is misguided and damaging in equal measure. And it fundamentally ignores a vitally important fact: tests like the SAT have always been achievement tests. Have you ever met someone with no formal education or instruction who can calculate square roots or fill in the blanks to complete a sentence?
Confusion about the nature of test bias is a prime motivator for charges that the SAT is a “biased instrument.”
Imagine a village where boys are prevented from attending school. If you administer a math test to all of the 12-year-olds in the village, the boys are going to bomb the test in comparison to the girls. Is the test “biased”? More likely than not, the results are revealing real differences in mathematical knowledge and skills as a result of divergent educational opportunities.
The question of test bias hinges on the question of accuracy -- does a test provide an accurate read on a student’s competencies in a particular academic domain? For a test to be biased, then, it would have to systematically underestimate (or overestimate) the skills and knowledge of a particular student population. Consider English language learners (students whose families speak a language other than English at home) -- math tests often underestimate their real mathematical knowledge as the students struggle to understand unfamiliar vocabulary and get hung up on decoding the text in word problems. The signal -- their genuine skills in math -- is obscured by the noise produced by language comprehension issues.
While claims that the SAT discriminates against racial minorities and low-income students are legion, specific examples of bias are in short supply. The prominent test critic Lani Guinier wrote an entire book decrying the SAT as nothing more than a “wealth test” that reflects the “values and culture” of upper-middle-class whites without citing a single instance of racial or class bias on the test. Test critics love to trot out an infamous example from a SAT analogies question that is over 25 years old -- "runner is to marathon as oarsman is to [regatta]" -- but you will be hard-pressed to find any evidence that this kind of class-coded vocabulary is more than a memorable anomaly. And I am still waiting for the memo that explains how exponents, angles, volume, surface area, slope and order of operations, all key topics on the math section of the ACT, are culturally or racially biased.
Test critics maintain that ACT and SAT scores do not provide “meaningful information about a student’s ability to succeed in college,” telling us “little more about students than what their zip codes already reveal.” They are, as UC regent Cecilia Estolano insists, merely a “proxy for privilege.”
There are three big problems with these claims. First, they are empirically wrong. The SAT is only a slightly weaker predictor of college performance than high school grades. (Combining the two provides an even more powerful predictive model.) When researchers control for students’ socioeconomic backgrounds, admissions tests retain over 90 percent of their predictive power. “The overwhelming conclusion across decades of research,” University of Minnesota psychologists Paul Sackett and Nathan Kuncel report, “is that tests are not biased against racial/ethnic minority group members in terms of their use in predicting subsequent academic performance.”
Second, the nothing-matters-but-wealth-and-privilege school of thought turns test score averages into deterministic, fixed points, as if all Black and brown students perform poorly and all Asians and whites are knocking it out of the park. This kind of facile categorizing masks huge variation within each demographic group. For instance, more than 72,000 Hispanic/Latino students scored above 1200 (74th percentile) on the 2019 SAT, while more than 47,000 whites scored below 790 (9th percentile).
Third, the proxy-for-privilege framework imagines that the educational advantages associated with wealth are exclusively artificial. The bugaboo of test prep looms large here. According to UC Regents chairman John Pérez, the SAT is essentially a measurement of the extent to which students are “able to avail themselves of prep material,” with access to test prep “disproportionately tied to family income.” This is a remarkably resilient myth. Research has consistently shown that test prep only produces incremental score improvements and is far from the decisive game-changer that both test critics and test prep companies claim it to be.
Test critics love to argue that tests “measure social, economic, and political capital” more than they measure “students’ academic ability or mastery of curriculum.” This is like saying that an annual physical is a better indicator of an individual’s social, economic and political capital than of her health.
Nobody should be surprised to learn that more affluent people tend to have better health outcomes. (Just take a look at how COVID-19 is ravaging Black communities.) The more money you have, the better access you have to nutritious food and high-quality health care, not to mention less polluted neighborhoods with leafy green streets ready-made for exercise. Similarly, the more money you have, the better access you have to well-resourced public or private schools with top-flight teachers, rigorous coursework and robust extracurriculars, not to mention books, museums and summer enrichment camps. It is a stubborn fact that different kinds of capital translate into better educational opportunities for more affluent students, making them, on average, better prepared to succeed in college than their less well-heeled peers.
To extend the health analogy to testing, ACT and SAT scores are a kind of educational version of a blood pressure reading. They are imperfect, partial measures that nonetheless provide useful information -- snapshots of cardiac and academic strength, respectively. What’s more, even though blood pressure tests are correlated with income (low-income individuals have higher rates of hypertension), we don’t simply set aside the blood pressure cuffs and ask patients for their zip code.
We need to think less about test bias and more about the structural biases embedded in society at large. A recent analysis shows that 20 percent of Black children and 19 percent of Hispanic children live in areas of concentrated poverty, compared to 6 percent of Asian children and 4 percent of whites. The UCLA Civil Rights Project coined the term “double segregation” to highlight the powerful nexus of race and class in the demographic composition of our public schools. Reflecting profound residential segregation throughout the country, three in four Black students and four out of five Hispanic students nationwide attend majority nonwhite schools. (About 15 percent of Black and Hispanic students attend so-called apartheid schools”, where whites make up less than 1 percent of the enrollment.) The typical Black or Hispanic student goes to a school with almost double the share of low-income students than the typical white or Asian student.
Teachers are widely acknowledged as the most significant factor influencing student achievement within a school’s walls, and they are the most inequitably distributed human resource across schools. By every measure of teacher qualification (certification, content expertise, selectivity of college attended and years of experience), students of color in low-income schools “are 3 to 10 times more likely to have unqualified teachers than students in predominantly white schools.”
In light of these dramatic racial and economic inequalities, it would be foolish to dismiss this observation from leading test measurement expert Daniel Koretz: “Any test that is working properly will show differences between advantaged and disadvantaged kids.”
Proponents of getting rid of standardized tests are almost always in favor of more “holistic” reviews, asserting that more weight should be placed on “less biased metrics” such as “academic achievements, personal qualities, extracurricular activities, community involvement and special talents” as well as high school grades and “qualitative evaluations by teachers and counselors.” Alas, advantages of wealth only compound when students are assessed by these more individualized criteria.
As sociologist Mitchell Stevens argued in his superb 2007 book, Creating a Class: College Admissions and the Education of Elites, information is the coin of the realm in admissions decisions. Admissions officers don’t evaluate applicants -- they evaluate applications. The detailed three-page letter of recommendation from a private school counselor is a lot more compelling than the brief boilerplate letter from a harried, overworked urban public high school guidance counselor. When it comes to personal statements, staying at home over the summer to take care of siblings generally makes for less interesting reading than tales of home-stays in France and volunteering to take care of sea turtles in the Galapagos Islands. It’s also a lot easier to showcase the community service you’ve performed if you don’t have to work a part-time job.
Scrapping admissions tests will not make a dent in two of the biggest advantages held by more affluent students: legacy status and highly specialized athletic skills. At selective schools, the children of alumni make up between 10 and 15 percent of the student body. (Many elite schools enroll more legacies than either African American or Hispanic students.) The most in-depth study of legacy students found that their legacy status boosted their chances of admission by a factor of three. In many cases, it makes sense to think about legacy applicants as their own distinctive applicant pool. Consider that Harvard University admitted 34 percent of legacy applicants and less than 6 percent of nonlegacy applicants between 2009 and 2015.
Like legacies, athletes make up another distinctive applicant pool that plays by its own rules. One study found that the odds of acceptance for athletes are four times higher than nonathletes. The Ivy League sponsors 35 different varsity sports, including a wide range of “country club” sports such as fencing, rowing, golf, lacrosse, squash, tennis and field hockey. At schools such as Williams and Amherst Colleges, varsity athletes make up 30 to 40 percent of the student body, most of whom have been “carefully scouted and recruited.” The “parents who can afford the lessons and league fees and summer camps, the cars for rides to practice and the houses in school districts with extensive sports programs” convey a massive competitive advantage to their offspring in terms of getting into selective colleges and universities. Atlantic writer Saahil Desai has aptly characterized college sports as “a quiet sort of affirmative action for affluent white kids.”
The cancellation of ACT and SAT admissions tests across the country over the past several months has prompted hundreds of schools to go test optional for next fall’s admissions season. Many test critics are downright giddy about this development.
That COVID-19 has forced schools to re-evaluate the role that standardized tests play in their admissions policies is a welcome development. I personally won’t shed any tears if the ACT and SAT are phased out -- it would save a vast amount of time, energy and money that is wasted on test preparation and could potentially dial down the intensity of the stress and anxiety that accompanies college admissions. And if it prompts us to rethink other aspects of the testing rigamarole, all the better. Too many hardworking high school strivers are caught up in the wheels of the Advanced Placement machine, stuck in tail-wags-the-dog classes where all of the teaching is geared to the test.
At the same time, we should refrain from celebrating this development as a blow for social justice that will usher in an era of fairer admissions. As last year’s Varsity Blues scandal made plain, money is a protean, cunning force that insinuates itself into every dip, bend and stream of the admissions pipeline. Nothing short of a random lottery would remove the admissions advantages afforded by wealth. With respect to expanding the educational opportunities for underrepresented student populations, the findings on going test optional are not terribly encouraging. The largest, most comprehensive study to date concluded that test-optional policies have “very limited effects” with “no effect on racial and socio-economic diversity.” This is one of the main reasons the UC Academic Senate recommended against UC making standardized tests optional, a recommendation the UC regents obviously ignored.
In fact, a special Academic Senate Standardized Testing Task Force found that test scores are better predictors of success (undergraduate GPA and college completion rates) than high school grades for underrepresented minority students, first-generation and low-income students in the UC system. (They attributed the weakening predictive power of high school GPA in part to the distortions introduced by grade inflation.) As the UC Academic Senate explained, “perhaps counter-intuitively,” the use of test scores “protects the admission eligibility of the very populations about whom there is concern, and ensures that under-represented, low-income, historically minoritized, and other similar populations are eligible for admission at UC."
“The major barrier to college access,” the Academic Senate concluded, “is not the SAT/ACT, but access to quality education and resources at the K-12 level.” Hear, hear. College admissions tests largely register rather than create socioeconomic and racial disparities. If we pretend otherwise, we will be abdicating our responsibility to ensure the least advantaged students receive a quality education.