What causes a prospective student to matriculate at the College? A new working paper published by the National Bureau of Economic Research seeks to answer this question. The paper, titled “Students Choosing Colleges: Understanding the Matriculation Decision at a Highly Selective Private Institution,” was co-authored by Peter Nurnberg ’09; David Zimmerman, professor of economics; and Morton Owen Schapiro, president emeritus and former professor of economics.
The paper, which began as Nurnberg’s honors thesis for economics, uses data from students admitted via regular decision to the Classes of 2008 to 2012 to estimate a probit model with variables capturing a range of student attributes, such as academic ranking, race, contact with admission office, intellectual interests and extracurricular activity.
The study revealed a number of trends. Most significantly, “the race variables are among the strongest predictors of matriculation. When all other applicant characteristics are controlled for, minority applicants are substantially less likely than their white counterparts to enroll at Williams if accepted,” the paper states. The authors found that black students are 32 percentage points less likely to matriculate than white students. For Hispanic students, this figure was 23 percentage points.
In addition, the results suggest that the better an applicant’s academic ranking (a metric incorporating high school achievement, test scores and teacher recommendations), the less likely the applicant is to matriculate. However, applicants who “do not know what academic interests they want to pursue” are 10 percentage points more likely to matriculate than students with social science interests.
The model also found certain extracurricular inclinations to be significant. While admits who have strong studio art backgrounds are 13 percentage points less likely to matriculate, athletes classified as “top” and “middle” were 11 and 20 percentage points more likely to attend the College, respectively.
Admitted students who had visited the admission office were 12 percentage points more likely to attend Williams than students who did not. On the other hand, those who had been denied financial aid were 12 percent less likely to matriculate. Nevertheless, the study found that a $5000 increase in net price (for example, stemming from a $5000 decrease in financial aid) only decreased the likelihood of attending the College by 1.3 percentage points.
Apart from the Williams-specific implications of Nurnberg’s work, his model can be applied to admission data of other colleges. “Colleges can use it to predict individual yield probabilities more accurately and adopt positive policy implications and to better pinpoint target recruiting groups,” Nurnberg said in an interview. Furthermore, it can facilitate estimation of the total number of matriculants and the total financial aid budget given all admission data.
In an interview, Zimmerman explained the motivation behind the study. “While numerous studies address institutions’ decisions whether or not to admit a student, less time has been spent on who accepts the offer once it’s made,” he said.
Zimmerman also called Nurnberg “a wonderful guy.” Nurnberg served as College Council co-president with Jeremy Goldstein ’09 and was valedictorian for the Class of 2009. He first developed an interest in higher education economics while taking a tutorial on the subject under Schapiro’s instruction.