The Dawn of a New Meritocracy

Higher education admissions have hit a watershed moment. Indeed, the recent public announcement of the University of Chicago no longer requiring SAT or ACT scores for undergraduate admissions is a major development that not only challenges the veracity of the instruments, but the current system of selective higher education admissions and operationalizing of meritocracy via standardized tests. But the question we all should be asking is toward what? What will replace the current hegemony of high stakes college admission tests?

The University of Chicago joins the hundreds of other colleges and universities across the country that are part of the "test-optional admissions" movement. However, up until now, some of the largest most selective institutions that have moved to test-optional admissions were Wake Forest University, American University and George Washington University. As one of the most selective and Ivy-Plus institutions, the University of Chicago is making a major move and precedent for highly selective and elite college admissions.

As demonstrated by many studies, the SAT has been found to be associated with parents’ education, family income and poverty, family wealth, high school quality, paternal grandparents’ education, extracurricular activities, group identity stereotypes, and cultural familiarity. Thus, the University of Chicago’s decision seems to take the implications of this research seriously, by giving the applicant greater agency in their college admissions process. While some institutions may follow suit, I suspect most will wait to see the outcomes of the University of Chicago’s decision.

I also am suspicious of what will replace the system of high stakes college admission testing, as many institutions have begun to move toward the use of predictive analytics for admissions. As a family of statistical and computational algorithms, predictive analytics are designed to predict a defined criterion such as college persistence and completion based on the ‘training’ of an algorithm on a set of historical academic, demographic and social data. With good quality and variety of relevant data to predict the criterion, algorithms can make reasonably accurate predictions of future events with some degree of error.

While this sounds promising, there are at least two concerns. First, given that the algorithms are trained on historical data that are situated in social systems, the algorithms will tend to reproduce or potentially augment structural inequalities in their predictions. For instance, with a criterion of persistence and completion, legacy applicants are likely to be weighted more favorably than first-generation or low-income applicants given existing associations of family financial resources with college persistence and completion. Second, with the necessary error in the algorithm, who will make up the false positives (i.e., predicted to complete but won’t) and who will make up the false negatives (i.e., predicted to not complete but will)? For some algorithms, the degree of false positives versus false negatives can be set in advance. Which error will be valued more or understood to be of greater concern? Statistically speaking, the marginal subject is often more likely to make up one of these marginal distributions. Thus, this statistical decision is an ethical decision about what applicants to privilege.

The system that replaces the high stakes college admission tests of yesteryear, may prove to merely reconfigure a new system of meritocracy. A sociotechnical system that promises to be a more ‘objective’, ‘efficient’, and ‘precise’ system of meritocracy, yet, trained on sociohistorical data that could render black, brown and poor applicants even more precarious in an age of information. Consequently, the future of fairness, equity and diversity in college admissions will be subject to calculability and contingency.