Economics of Education Review 18 1999 375–390
An event history model of student departure
S.L. DesJardins
a,
, D.A. Ahlburg
b
, B.P. McCall
b
a
Planning, Policy, and Leadership Studies, University of Iowa, N491 Lindquist Center, Iowa City, IA 52242, USA
b
Industrial Relations Center, University of Minnesota, 321 19th Avenue S., Minneapolis, MN 55455, USA Received 1 September 1998; accepted 2 October 1998
Abstract
This study uses a modeling technique often used in economics and other disciplines but rarely applied to educational research. The technique, known as event history modeling, is used to examine the temporal dimensions of student
departure from a large research university. This approach allows researchers to remedy analytic problems often found when standard statistical procedures are used to study longitudinal events such as student departure from college. The
study confirms most of the substantive findings of earlier research but we document that key explanatory variables have differential effects over time. Pinpointing the times at which students are at risk of leaving college will permit
more efficient intervention strategies which could reduce the social, institutional, and individual costs often associated with leaving college before degree completion.
1999 Elsevier Science Ltd. All rights reserved.
Keywords: Event history modeling; Censoring; Time-varying coefficients; Unobserved heterogeneity
1. Introduction
When a student leaves college before degree com- pletion costs are imposed on society, the institution, and
the student. The educational research literature docu- ments these costs Pascarella Terenzini, 1991. The
social costs include reduced economic output if college graduates are more productive than students who drop
out. Society also may lose a source of public revenue, since college graduates tend to earn more than others and
college graduates have enjoyed a more rapid increase in their incomes in the last decade Psacharopoulos, 1985.
Persons with lower levels of educational attainment are more likely to need government transfer payments and
are more likely to become unemployed Kiefer Neum- ann, 1979. College graduates are less likely to be wel-
fare recipients Blank, 1989 and have lower rates of incarceration and anti-social behavior Haan, 1975.
Also, there may be adverse intergenerational effects if a person’s socioeconomic status is linked to the amount
Corresponding author.
E-mail: stephen-desjardins
uiowa.edu
0272-775799 - see front matter
1999 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 2 - 7 7 5 7 9 8 0 0 0 4 9 - 1
and quality of education of their parents Bowles, 1972. If intergenerational effects exist, high rates of dropout
exiting college and not returning by low-income and minority students may result in racial or socioeconomic
bipolarization in future generations.
Institutions are affected by student dropout because much of the funding provided to higher education is
based on enrollments. As public institutions grow, they receive proportionately more money. Since labor and
capital costs do not grow in proportion to enrollments, institutions often use the extra funds to implement new
programs and other kinds of development Bowen, 1980. Conversely, when enrollments are adversely
affected by student attrition, institutions may incur fund- ing decreases disproportionate to the loss of students.
Also, given the recent accountability movement in higher education, high dropout rates are considered by oversight
agencies and the general public as a sign of institutional failure. The relative scarcity of educational resources at
the federal and state levels makes it even more important that higher education institutions use public support
efficiently.
Given the individual, institutional, and societal costs associated with college dropout, institutions of higher
376 S.L. DesJardins et al. Economics of Education Review 18 1999 375–390
education must become better informed about how social, environmental, and individual factors affect stud-
ent departure decisions. This paper provides insights into how educational stakeholders can improve their under-
standing of the student departure process. The major pur- poses of this line of inquiry are first, to discuss a statisti-
cal modeling approach, event history modeling, that overcomes many of the deficiencies of previous models
of student departure and allows powerful insights into the dynamics of this process and, second, to present a
model of student departure that draws on data commonly available to institutions of higher education.
In Section 2 we discuss the previous research on stud- ent departure. In Section 3 statistical problems that have
plagued earlier studies of student departure are discussed and we present a conceptual and formal statistical model
of student departure. In Section 4 the model specification and sample attributes are presented. Section 5 details the
empirical results. In the final section Section 6 we dis- cuss the policy implications, the limitations of the study,
and possibilities for further research.
2. Previous studies of student departure