Predictive Modeling of Enrollment Yield for a Small Private College

Abstract

Recent examples of studies on predictive modeling of enrollment yield for the primary purpose of improving undergraduate enrollment management focused on large, public universities [Goenner and Pauls Research in Higher Education, 47, 2006; DesJardins, Research in Higher Education, 43, 2002; Thomas, Dawes and Reznik, AIR Professional File, 78, 2001]. This study differs in that it uses data from a small, private liberal arts college, one located in Wisconsin. Apart from its smaller size and relatively higher tuition rate, a private liberal arts college is also likely to have a different institutional mission compared to a public research university. And if the school has a religious affiliation, as this study institution does, the students it attracts may also be different. Supplied by the admissions office of the college, the data in the study consist of freshman applicants who were admitted (but before a financial aid offer was made) in the fall of 2001 (N = 541) and 2002 (N = 770), of which 221 and 250 enrolled, respectively. Following the methodological approach used in enrollment studies in the literature, a logistic regression model is fitted of the form ln [P/(1 ─ P)] = α + β1X + β2Y + β3Z + e where P is the probability of an admitted applicant enrolling, X is a vector of pre-college student personal and demographic characteristics including a measure of academic ability, Y is a vector of marketing or promotionrelated variables indicating where the applicant first heard of the college, and Z is a vector representing the applicant’s college preference and interest in varsity sports. α and βi , i = 1, 2, 3 are the coefficients to be estimated. e is the error term. Specifically, the predictor variables are as follows, with dummies coded as 1 or 0 otherwise: White (if white, non-Hispanic), Religion (if Catholic), HSGPA (high school grade point average), Internet (if first heard of the college through the Atl Econ J (2009) 37:323–324 DOI 10.1007/s11293-009-9177-7