The Journal of College and University Student Housing - Volume 43, No. 1 - 15

Family Income, Campus Residency, and Student Retention

dents' race, sex, family composition, and their
parents' income and education. The analyses
also include high school GPA, SAT score,
and number of AP classes taken. Identiﬁers
of academic preparedness may predict who
lives on campus because the most-prepared
students tend to make college choices earlier
and respond to their institutional acceptance
letters faster, granting them a coveted place
in housing facilities with limited availability
(Thompson et al., 1993; Park & Eagan, 2011).
The ELS also captures students' rating of the
importance of living at home during college
or going away for college. Additionally, I controlled for social participation, which is very
likely to affect a student's desire to live on
campus, through students' average hours of extracurricular activities per week in high school
and their rating of the importance of making
friends at college. Finally, the characteristics of
the college attended are also likely to affect the
selection of on-campus housing. Colleges may
differ in availability of on-campus housing, as
well as in the emphasis placed on the importance of living on campus. College characteristics, such as institutional control (i.e. public
and private institutions) and institutional selectivity, also impact student outcomes and are,
thus, included in the model.
The key independent variable of interest
is the location of residency (on or off campus)
during the ﬁrst year of college, obtained from
self-reports in the 2006 survey data. I construct the outcome measure, retention into
the second year, using measures of month-bymonth enrollment from the second follow-up
survey. "Retained" students remained enrolled
throughout fall 2004 and spring 2005 and reenrolled in fall 2005. Ninety percent of the stu-

dents in the sample persisted into the second
year of college. This estimate includes parttime and full-time enrollment.

Analysis
Students who live on campus tend to differ
systematically from those who live off campus
(Schudde, 2011; Turley & Wodtke, 2010). Not
all students have the opportunity to live on
campus, and not all who apply are selected due
to housing shortages at college campuses. The
choice to live at home instead of on campus
may reﬂect ﬁnancial situations, family structures, and levels of college preparedness associated with poor college outcomes (Turley,
2006). While some of these differences may
be observable, and thus controlled through
statistical modeling, others are not. Because
highly motivated, higher-income students are
more likely to be placed in limited on-campus
housing facilities, research must attempt to
deal with students' selection into campus
housing (Schudde, 2011; Levin & Clowes,
1982; Turley & Wodtke, 2010).
In order to control for observed differences between on-campus and off-campus residents and to estimate variation in the effects
of campus residency across family income, my
analysis proceeded in two steps. First, using
regression analysis, I estimated the probability that a student would choose to live on
campus during their ﬁrst year, based on students' background (the variables that capture
student characteristics and experiences prior to
their decisions to live on campus). I used the
resulting estimate for each student, called a
"propensity score," to match students based on
their propensity to live on campus. Students
were matched to peers with similar scores

V O L U M E 4 3 , NO. 1 * 2 0 1 6

The Journal of College and University Student Housing - Volume 43, No. 1

Table of Contents for the Digital Edition of The Journal of College and University Student Housing - Volume 43, No. 1