The following is a rebuttal to a critique by Wake Forest professor Joseph Soares of the author’s previous article, “Class Warfare Comes to College Admissions.”
I am hardly shocked that Dr. Soares’ found so much fault with my article criticizing both his book The Power of Privilege and Wake Forest University’s decision to eliminate the requirement for standardized testing. That is to be expected—we are on opposite sides of a controversial issue. To begin my defense, I will cut to the chase with actual evidence.
Once the rhetoric is pared from his response, Dr. Soares makes four substantive claims or criticisms of my article. They are:
- SATs are poor predictors of college performance.
- SAT tests are biased, and the elimination of
standardized tests makes college admissions fairer.
- SAT tests and socioeconomic status (SES) are strongly correlated.
- His focus on class antagonism is a figment of my imagination.
For the first claim that SAT scores are poor predictors of college performance, I will begin my argument with the very study that forms the foundation of Soares’ argument, the 2002 statisitical analysis of the University of California system, authored by Saul Geiser with Roger Studley. I have reproduced the most relevant table of statistics below, in order to let readers judge for themselves.
Percent of Variance in UC Freshman GPA Explained by HSGPA, SAT I, and SAT II Scores
Predictor Variable Influence
HSGPA 15.4
SAT I 13.3
SAT II 16.0
SAT I + SAT II 16.2
HSGPA + SAT I 20.8
HSGPA + SAT II 22.2
HSGPA + SAT I + SAT II 22.3
Geiser and Studley used regression analysis and data from four years of entering freshmen in the University of California system to determine the percentage of influence that each of the three predictor variables above had on college freshman grade point averages (FYGPA). The predictor variables are high school grade point average (HSGPA), SAT I (designed to measure both aptitude and some achievement), and SAT II (directed at measuring achievement in specific subjects).
(A bit of confusion must be cleared up because of the two different types of SAT tests. Wake Forest has dropped the requirement for all standardized testing, both SAT I and SAT II. Soares defends that action on one hand, but sometimes favors achievement-oriented tests on the other. )
Soares and Geiser claim that HSGPA is the best predictor. Yet this table clearly shows that the SAT II was the best individual predictor at 16.0 percent, followed by HSGPA (15.4 percent), then the SAT I (13.3 percent). Obviously, the tests are significant predictors individually. It would be hard to argue against them on the basis of these numbers.
But the real question is not whether HSGPA is a better predictor of FYGPA than SAT scores, I or II. The real question, in light of the action by Wake Forest, is whether standardized tests significantly add to the ability to predict college performance, either individually or in combination with HSPGA.
The answer is easily obtained by observing the data. Both tests improve FYGPA prediction in conjunction with HSGPA as well: from 15.4 percent of variation explained to 20.8 percent for the SAT I (an increase of 35 percent in predictive ability), and from 15.4 percent of variation explained to 22.2 percent for SAT IIs (a 44 percent rise in predictive ability). Perhaps an argument can be made to eliminate one of the SAT tests, because they are closely related, but never both.
Soares also quotes from a recent study from the University of Georgia as evidence of the SAT’s ineffectiveness and of HSGPA’s superiority as a predictor. However, the economics professor who co-authored the study did not say that the SAT tests were ineffective. “Despite what critics say, the test scores really do matter,” he concluded.
Even if one accepts Soares’ and others’ claim that HSGPA is the best single predictor, the logic behind their rejection of the SAT tests is akin to a boxer who decides, because he hits harder with his right hand, to not throw punches with his left. The results are likely to be less than optimal in both cases.
When FYGPA is used as standard of college success, the answer to the question whether SATs provide valuable information must be a resounding yes. Smart kids get high scores—the SAT is hardly “a spurious number,” as Soares suggests.
(Note: for the sake of space limitations, I have not used the many studies more favorable to my case than Soares’ choices. Nor have I used the great amount of empirical evidence that shows SAT scores are related to graduation rates.)
As to the fairness question, the real debate is whether fairness is better served by an equality of outcomes, or an equality of opportunities. The latter method establishes standards and rewards all those who meet the standards regardless of their demographic or socioeconomic status. The former decides ahead of time what the results should be, and metes out advantages and disadvantages to the various groups in order to produce those predetermined results.
Soares is perhaps correct that I am unconcerned with “the discriminatory effects of one test,” because I am convinced that disregarding the results of these tests is more discriminatory than the effects of the tests themselves. I am more concerned about the prejudicial act of deciding ahead of time the share of accepted applicants that come from a particular demographic group. I am also concerned about the fundamental effects on society likely to result from policies that promote the less capable over the more.
Soares contends that the tests are biased against many groups. He does not, however, mention that some minorities get higher than average scores on the test, such as Jews and Asians. He also does not mention the numerous studies indicating that SAT scores overpredict the FYGPAs of black and Hispanic students. If the tests were truly biased against these minorities, their actual grades would be better than the grades predicted by their scores, not vice-versa.
This op-ed in the N.Y. Times by the chairman of the sociology department at CUNY also refutes claims of bias: “the minorities: (a) do as well on the allegedly biased questions as on other questions, and (b) give the same rank ordering of difficulty of questions as the allegedly favored group gives.”
I agree with Soares that there is a strong correlation between socio-economic status and SAT scores. However, to remove the effects of SES is to remove the effects of positive qualities passed by successful parents to their children, such as high native intelligence and a strong work ethic. Should we downgrade talented children because of their parents’ success? That is a confusing, divisive, and self-defeating message to send to society.
As far as Soares’ suggestion that I imagine his class warfare inclination, I will again refer to his praise of such ideas as “Berkeley-style socio-economics admissions policies” and “class-based affirmative action.” In The Power of Privilege, the model that he chose to describe Yale’s admissions policies is called “dominant class perpetuation,” from the French sociologist and left wing activist Peierre Boudieu.
Soares also champions the idea of elite colleges reserving at least 25 percent of the seats in each freshman class for students from the bottom 40 percent in family income. And, in his biography on the Wake Forest Web site, he recalls an adolescence spent “pursuing the eschatological myth of industrial society, Marxism.”
At some point, the evidentiary standard of “if it looks like a duck, and walks like a duck, and quacks like a duck, then it is most likely a duck” must be applied. This particular duck quacks in the language of class antagonism.
Soares’ and other SAT critics’ real objection is this: they don’t like the additional information the tests provide, that some demographic groups, on average, are less likely to perform well at higher academic levels. And they take a page from the playbook of arbitrary despots—if you don’t like the message, shoot the messenger. And that is a guaranteed way to make sure that bad decisions will be made on account of misleading information.
To read Joseph Soares’ critique of the original article, click here.
To read Schalin’s original article, click here.