Religion and College Attendance: Change among Students

Jenny J. Lee

The Review of Higher Education, Volume 25, Number 4, Summer 2002, pp.
369-384 (Article)

Published by Johns Hopkins University Press

DOI: https://doi.org/10.1353/rhe.2002.0020

For additional information about this article

https://muse.jhu.edu/article/30160

Access provided by Australian National University (29 Aug 2018 16:00 GMT)
LEE / Religion and College Attendance 369

The Review of Higher Education

Summer 2002, Volume 25, No. 4, pp. 369–384
Copyright © 2002 Association for the Study of Higher Education
All Rights Reserved (ISSN 0162-5748)4

Religion and College

Attendance: Change
among Students
Jenny J. Lee

A recent Los Angeles Times cover story reported that scholars are beginning
to “promote a new paradigm in academia: Religion matters” and that “until
recently, a long-standing academic bias against religion has blinded many
scholars to its powerful role in shaping both private lives and the public
culture” (Watanabe, 2000, p. A-1). For the most part, religion has been ne-
glected in higher education research, even though most would agree that
religion is a powerful force for many in shaping individual values, provid-
ing an overall sense of purpose, forming connections with others, and build-
ing a sense of community. On the individual level, religion provides the
overarching framework for one’s worldview and moral opinions, offering
insight into human drives, hopes, and reasoning, as well as perceptions of

JENNY J. LEE is a Ph.D. student of Higher Education and Organizational Change (HEOC)
in the Department of Education at the University of California, Los Angeles. A research
analyst at the Higher Education Research Institute, she studies faculty culture, organiza-
tional change, social constructions of belief and identity, and religion/spirituality in higher
education. She presented an earlier version of this paper, “Changing Religious Beliefs among
College Students,” at the annual meeting of the American Educational Research Associa-
tion, April 2000, in New Orleans. The author expresses appreciation to Linda Sax, Helen
Astin, Jennifer Keup, and other HEOC faculty and fellow students, Address inquiries to her
at Higher Education Research Institute, Graduate School of Education and Information
Studies, 3005 Moore Hall, Mailbox 951521, Los Angeles, CA 90095-1521; (310) 825-1925;
e-mail: j9lee@ucla.edu.
370 THE REVIEW OF HIGHER EDUCATION SUMMER 2002

the self in relation to others and the physical world (Berger, 1973b). On the
group level, religion offers resources and pathways for relational support
(Maton & Wells, 1995) and prosocial activity, including volunteerism
(Donahue & Benson, 1995), community access and awareness (Matton &
Wells, 1995), and social justice (Lebowitz, 1999). For these reasons, the study
of religion is both timely and necessary.
Inner belief frameworks like religion, are not fixed. The environment
and personal experiences within the environment can directly influence
individual beliefs. Hence, the study of religious change in higher education
is particularly critical since researchers have empirically demonstrated a
decline1 in religious activity during the college years (Astin, 1993b; Bowen,
1997; Cox, 1988; Levine, 1980). Astin (1993b) documents less attendance at
church, prayer, and grace before meals. These behaviors may reflect changing
religious attitudes as studies also reveal that students become less favorable
toward church, less convinced of the reality of God, less favorable toward
Sabbath observance, and less fundamental/conservative in their beliefs (Bowen,
1997). Others have identified college attendance as a leading predictor of
religious disaffiliation (Hadaway & Roof, 1988). Pascarella and Terenzini
(1991) also found a freshman-to-senior decline in doctrinaire religious be-
liefs. Given this phenomenon, the question of how college attendance in-
fluences changes in students’ religious beliefs is worthy of investigation.

S OURCES OF C HANGE IN R ELIGIOUS B ELIEFS AND I DENTITY

Past studies have identified specific individual characteristics and col-
lege environment factors as leading causes of religious decline (Condran &
Tamney, 1985; Nelson, 1988; Roof & McKinney, 1987). Roof and Hadaway
(1977, 1979) were among the earliest researchers to produce profiles of the
religiously disaffiliated. Using the 1973–1976 series data from the National
Opinion Research Center’s (NORC) General Social Survey, they found “that
those switching to ‘none’ [category of religious affiliation] tended to be
younger, predominantly male, more educated, more committed to the new
morality, less happy, and to have less confidence in American institutions”
(1979, p. 373). Although they studied only white respondents, their early
findings identified some other predictors of disaffiliation that were borne
out by subsequent quantitative studies (Roof & McKinney, 1987; Condran
& Tamney, 1985). In addition to being liberal, Roof and McKinney (1987)
confirmed that the disaffiliating were more likely to be young, male, edu-
cated, “committed to the new morality, and oriented generally to an ethic

1
This study, like others cited here, primarily focuses on monolithic religious faiths. By
decline or weakening, I assume a disassociation from the religion’s foundational views and
practices.
LEE / Religion and College Attendance 371

of personal fulfillment” (p. 35). Such qualitative terms as “less happy,” “more
committed to a new morality” (or more liberal), with “less confidence in
American institutions,” suggest that specific perceptions of the self may be
more strongly linked to religious disassociation than other self-perceptions.
Hadaway and Roof (1988) also identified such experiences as marital
problems, abuse, and social dislocations as predictors of decline in religious
affiliation and beliefs. Hence, these perceived experiences are more posi-
tively associated with religious change than others. They also found that,
while social status was an important factor, the effect was largely accounted
for by education. They concluded, “Higher education tends to expand one’s
horizons and may also mean greater exposure to counter-cultural values.
For many persons, such exposure has worked to erode traditional plausibil-
ity structures, which maintained the poorly understood religious convic-
tions that seem so typical of American religion” (p. 36). In other words,
Hadaway and Roof have found that college introduces belief systems and
values that may counter some religious beliefs. Those with lesser degrees of
religious convictions are especially susceptible to religious questioning.
This finding supports sociologist Peter Berger’s (1973a) claims that en-
countering contrary beliefs, particularly in college, likely result in question-
ing and possibly abandoning religious faith. An unchanging social experience
that confirms particular views validates meanings behind religious sym-
bols while, in pluralism, “the individual is forced to take cognizance of oth-
ers who do not believe what he believes and whose life is dominated by
different, sometimes by contradictory, meanings, values and beliefs. As a
result . . . pluralization has a secularizing effect. That is, pluralization weak-
ens the hold of religion . . . on the individual” (p. 80). Based on Berger’s
rationale, diversity-related experiences are particularly likely to weaken per-
sonal religious convictions. It would be expected that encountering diverse
worldviews and religions would give rise to questions and challenges to
one’s faith, hence leading to less religious commitment.
In sum, examinations by Hadaway and Roof show that for many, reli-
gious decline is at least partly attributable to higher education. Although
their early research identifies individual attributes and experiences com-
mon to most religiously disaffiliated, the interplay of these various aspects
during college remains ambiguous, nor is it clear whether college students
match the profile they established of religious disaffiliates. Moreover, their
study focused on factors that might lead to changes in religious identity and
only presumed a parallel weakening of religious beliefs.
The studies cited to this point dealt with switching to another denomi-
nation or withdrawing from religion. Yet, it can be argued that changes in
religious beliefs can occur separate from changing one’s religious identity.
Religious affiliation can be a combination of individual beliefs and cultural
identity. Moreover, identification with a religious group does not necessarily
372 THE REVIEW OF HIGHER EDUCATION SUMMER 2002

mean adopting its beliefs completely or committing to its theology; rather,

it may mean subscribing to a particular community, family heritage, and/
or culture (Caplovitz & Sherrow, 1977). A person may not practice or as-
cribe to the decrees of Roman Catholicism, for example, but still identify
herself as a Roman Catholic because it is part of her family and cultural
heritage. A Catholic who is nonpracticing has not necessarily abandoned
his/her religious beliefs, family heritage, and/or cultural practices. Nor is
denominational switching a good measure of religiosity. For these reasons,
this study uses changes in religious beliefs and convictions as a more accu-
rate measure of religious disaffiliation.
Although the studies cited above did not specifically sample college students,
their findings suggest the possible impact of the college experience on stu-
dents’ religious beliefs. First, the disaffiliated share particular personal charac-
teristics such as more liberal views. Second, higher education accompanies the
recognition of differences in beliefs. In particular, encountering diverse cul-
tures and values, particularly within the college environment may contribute
to a greater questioning and weakening of previously held religious beliefs.

Objectives
In an attempt to fill this gap in the literature, this study identifies factors
in the college environment that influence changes in students’ religious be-
liefs. Based on previous research on students’ religiosity and theories in so-
cial psychology, I hypothesized that:
1. Overall, students will experience a decline in religious beliefs and con-
victions.
2. The characteristics and experiences identified in earlier profiles of the
disaffiliated (e.g., being liberal and unhappy) would also appear among
college students who disaffiliated.
3. Greater involvement in diversity-related experiences would be strongly
associated with weakened religious beliefs.

Methodology
I derived the data for this study from the 1994 Freshman Survey and
1998 College Student Survey, conducted annually by the Higher Education
Research Institute (HERI) for the Cooperative Institutional Research Pro-
gram (CIRP) at the University of California, Los Angeles. The 1998 follow-
up survey was funded by the W. K. Kellogg Foundation. These combined
surveys provide a longitudinal database for a national cohort of 1994 fresh-
men who were surveyed upon entering college and resurveyed four years
later. The sample for this study included over 4,000 students attending 76
four-year institutions across the country.
LEE / Religion and College Attendance 373

Measures and Analysis

The primary variable of interest was a single dependent measure—self-
rated change in religious beliefs and convictions. The specific survey ques-
tion read: “Compared with when you entered college as a freshman, how
would you describe your religious beliefs and convictions?” This response
was measured on a five-point scale: (1) much weaker, (2) weaker, (3) no
change, (4) stronger, and (5) much stronger.
I applied two statistical procedures to the dependent variable: (a) fre-
quencies to observe the distribution of responses, and (b) a multivariate
design. A hierarchical regression analysis provided a clearer understanding
of how college affects changes in religious beliefs. Through this analytic
method, I examined the relationship between the college environment and
the outcomes, independent of student precollege characteristics. Blocked,
stepwise regression allows one to assess the significant predictors of the
dependent variable and to observe the change in regression coefficients at
every step, based upon each variable entering the regression equation.
I grouped 64 independent variables into 7 “blocks,” ordering them chro-
nologically according to Astin’s (1993a) Input-Environment-Outcome (IEO)
model of college impact. This conceptual model allows researchers to cor-
rect or adjust for incoming characteristics, or input variables, in order to
obtain a less biased estimate of the effects of the college or of environmen-
tal variables on the outcome being investigated. In other words, by control-
ling precollege student characteristics, such as religious affiliation, I can more
accurately assess the effect of various aspects of the college environment on
students’ changes in religious beliefs and convictions. Following IEO meth-
odology, I tested independent variables for stepwise entry (p < .005) into
the regression equation. The first two blocks consisted of precollege char-
acteristics: (a) student characteristics, including gender, parents’ religion,
and precollege religion, and (b) views, goals, and high school experiences.
The next three blocks were (c) student’s major, (d) institutional character-
istics, and (e) college experiences. The remaining blocks consisted of inter-
mediate outcomes, or concurrent measures, with the outcome: (f )
intermediate outcome views and experiences and (g) self-ratings. I entered
the self-ratings last because of their expected high association with the de-
pendent variable, which is also a self-rating. The order of these seven blocks
assures examining the impact of college on students independently of the
biasing effects of entering student characteristics. The following sections
detail the specific variables and rationale.
Precollege Variables: Personal Characteristics
I chose the following 64 independent variables mainly according to the
major findings from previous quantitative studies on religious change, as
well as studies on college impact. The first block, student characteristics,
374 THE REVIEW OF HIGHER EDUCATION SUMMER 2002

consisted of three variables: sex, student’s religion, and mother’s religion.

Earlier studies had found sex a reliable measure of control (Caplovitz &
Sherrow, 1977; Roof & Hadaway, 1979). I used four religion variables: Catho-
lic, Protestant, Jewish, and Eastern (i.e., Buddhist and Islamic) religions. I
employed the student’s and the mother’s religion. I then added the mother’s
religion, which had previously been found a strong determinant of family
religion (Roof & Hadaway, 1988), as a measure of control for the next blocks.
Precollege Variables: Activities, Views, and Goals
I also selected the second block—views, goals, and high school experi-
ences—as control measures for college’s influence on religious beliefs. Com-
prised in “students’ views” were their perspectives on politics, homosexuality,
abortion, premarital sex, use of marijuana, abolishment of racist speech,
and individuals’ effect on society. Based on past findings on the positive
relationship between liberal views and apostasy (Roof & Hadaway, 1988),
these variables identify which liberal views might account for weaker reli-
gious beliefs. Liberal views on homosexuality and abortion, for instance,
contradict most fundamental religious faiths. I hypothesized that more lib-
eral students would have weaker religious beliefs. I included one high school
experience, “attended religious services,” in the regression, expecting that it
would suggest greater commitment to religious teachings. I also added five
goals (help others in difficulty, influence social values, be well off finan-
cially, promote racial understanding, and develop a meaningful philosophy
on life) as pretest measures for the same intermediate outcome goals in the
follow-up survey. I included high school grade point average (GPA) as a
control measure of precollege academic achievement on religious beliefs.
College Environment Variables: Student Major
The college environment blocks provided my main variables of interest
to explore how college influences change in religious beliefs. I formed the
third block from six dichotomous measures representing students’ majors:
biological science, engineering, humanities, fine arts, social sciences, and
physical science. I entered them in the regression to identify whether vari-
ous academic disciplines had any unique effect on religious beliefs. Few
studies have examined this relationship, and their findings have been in-
consistent (Pascarella & Terenzini, 1991). I entered student major variables
immediately before the remaining college environment blocks, since the
academic major can characterize a student’s scholarly interests prior to col-
lege entry as well during college.
College Variables: Institutional Characteristics
Institutional characteristics, the fourth block, consisted of six dichoto-
mous measures for institutional type and national geographic location:
Catholic college, Protestant college, nonsectarian college, east region, west
LEE / Religion and College Attendance 375

region, and south region. I also used a continuous measure indicating se-
lectivity, determined by mean GPA and Standardized Assessment Test (SAT)
scores. Using this college environment block, I examined the influence of
the structural environment and institutional type on the dependent vari-
able, since Roof and Hadaway (1988) have linked geographical location to
changes in religious values. Specifically, they found a greater percentage of
the religiously disaffiliated in the West than in the South.
I also added “Mean: Attend Religious Services” to the analysis. This vari-
able is a peer measure created by aggregating the mean of student responses
by institution on how often they attended religious services. Astin (1993b)
found that “the students’ peer group is the single most potent source of
influence on growth and development during the undergraduate years” (p.
398), making its relevance obvious for this study. Astin explains that as
members, individuals conform their beliefs and values to those of the group.
Dalton (1989) adds, “The values college students develop are strongly in-
fluenced by the extent and intensity of their involvement with the college
peer culture and the values prized in that culture” (p. 2). Peer groups, by
their nature, exercise considerable influence upon their members in their
development of consensual attitudes and beliefs.
College Variables: Activities
I entered college experiences in the fifth block to assess specific college
activities that might account for a change in religious beliefs and convic-
tions. The degree to which students are involved in college, including spe-
cific involvement activities are, according to Astin (1993b), among the
primary factors that contribute to students’ development during their col-
lege years. These experience variables were measured on the same 1998 sur-
vey as the question for the dependent variable. I drew on Hadaway and
Roof (1988) for three items, to test their findings: got married, withdrew
from school, and distance from home. Previous studies also have found
that living at home decreases changes in religious preferences (Pascarella &
Terenzini, 1991). As variables representing diversity, I used five variables:
attended cultural diversity workshop, had roommate of different race/
ethnicity, enrolled in ethnic studies course, enrolled in women’s studies
course, and an additional ethnic interaction composite (socialized with a
different ethnic group, studied with a different ethnic group, and interacted
with a different ethnic group, alpha = .71). Berger (1973b) theorized that
encountering differing or pluralistic beliefs would likely lead to a weaken-
ing of personal religious belief. I also included three other experiences—
withdrew from school, participated in leadership training, and studied with
other students—on an exploratory basis.
For this block, I also created a composite of ten student-faculty interaction
measures (alpha = .86). The variables included in this composite measure
376 THE REVIEW OF HIGHER EDUCATION SUMMER 2002

were: faculty encouragement for graduate/professional school, advice about

educational program, respect, emotional support, letter of recommenda-
tion, honest feedback, intellectual challenge, opportunity to discuss home-
work, hours per week talking with faculty outside of class, and “faculty took
interest in my progress.” Previous research has demonstrated the important
effects of faculty on student development (Pascarella & Terenzini, 1991).
Intermediate Outcomes: Fourth Year
The next two blocks consisted of intermediate outcomes— measures that
reflect aspects of the college experience. However, such measures might also
be considered as college outcomes. Hence, intermediate outcomes are to be
examined only after controlling for precollege and earlier college environ-
ment measures (Astin, 1993a). While these intermediate outcomes do not
necessarily determine cause, they do serve as control measures for the
precollege variables within the second block. Moreover, including both pre-
tests and posttests shows how change during the college experience is asso-
ciated with changes in the dependent variable.
This block included college GPA, hours per week in religious services,
and the views and goals included in Block 2 and posttested in 1998. To
observe associations between the outcome variable and liberal views, I cre-
ated a composite of liberal views similar to the pretest that incorporated
three variables: having sex is okay if the couple is living together, abortion
should be legal, and marijuana should be legalized (alpha = .85).
Intermediate Outcomes: Self-Ratings
The final block consisted of self-ratings that earlier researchers have iden-
tified as being related to effects of diverse interactions: acceptance of differ-
ent races/cultures, cooperativeness, and understanding of others (Hurtado
et al., 1999). I also included the self-rating of “emotional health,” since past
studies indicated that those who drifted from their religious moorings were
more likely to experience maladjustment and unhappiness (Caplovitz &
Sherrow, 1977; Hadaway & Roof, 1988).

R ESULTS
Descriptive Analyses
Frequency distributions of students’ reports in 1998 showed that most
students experienced a change in their religious beliefs and convictions since
entering college. (See Table 1.) However, contrary to the hypothesis, over
one third of them reported a strengthening of religious convictions and
beliefs compared to 13.7% who indicated a weakening. Almost half (48%)
did not experience any change. The data suggest that the widely publicized
“secularizing” effect of college on students is overgeneralized.
LEE / Religion and College Attendance 377

TABLE 1
SELF-RATINGS OF CHANGES IN RELIGIOUS BELIEFS
AND CONVICTIONS, 1998
(N = 5,326)

Self Rating N Percentage

Faith weaker 727 13.7

No change 2,572 48.3
Faith stronger 2,019 37.9

Source: UCLA Higher Education Research Institute, Cooperative Institutional Research Program

Regression Analysis
The stepwise regression analysis identified aspects of college that would
account for changes in religious beliefs and convictions. Of the 64 precollege
and college environment variables entered into the regression, 23 initially
entered the equation in significant weights: 9 precollege variables, 5 college
environmental variables, and 9 intermediate outcomes. (See Appendix.) The
following sections reveal the entered independent variables related to the
college environment, including the intermediate outcomes. For each of these
sections, tables depict and describe betas after the entered precollege vari-
ables, college variables, and intermediate outcomes (or final betas).
Results: College Environment Characteristics
Among the twenty-five college environment variables, six entered the
regression analysis. (See Table 2.) After the list of entered variables, the next
column shows correlations among the independent variables with the de-
pendant variable. Columns 3 and 4 show that all of these predictors were
significant both at initial entry and also after controlling for the college
environment variables. Column 5, however, indicates that most of these
variables, with the exception of the peer mean of attending religious ser-
vices, became nonsignificant by the last step.
The peer mean of attending religious services accounted for half of the
variance in the dependent variable of changes in religious convictions. The
entered college environment characteristics remained throughout the re-
gression analysis, affirming prior findings on the strong influence of peer
groups (Astin 1993a; Dalton, 1989).
After controlling for all precollege and college environment variables,
getting married in college was a significant predictor in strengthening reli-
378 THE REVIEW OF HIGHER EDUCATION SUMMER 2002

TABLE 2
STANDARDIZED REGRESSION COEFFICIENTS
FOR COLLEGE ENVIRONMENT VARIABLES
(N = 4,278)

Variable Zero-Order Beta After Beta After Final Beta

Precollege College
Environment

Physical science major -.07 -.06** -.05** -.02

Peer mean: Frequently
attended services .20 .14** .16** .04*
In leadership training .11 .08** .07** .03
Faculty interactions .10 .07** .07** .00
Got married .08 .06** .06** .02

R2 = .04
* p < .01
** p < .001

gious convictions. Interestingly, its beta became nonsignificant and dropped

from .06 to .02 once attending religious services entered the equation. Based
on this beta change, it appears that married individuals are more likely to
attend religious services and that this activity contributes to the strength-
ening of religious convictions among married individuals. Conversely, pre-
vious studies have found that family tensions and marital difficulties were
associated with disaffiliation from religion (Hoge, 1988; Kotre, 1971). The
other college environment variables entered into the equation—physical
science major, interacting with faculty, and leadership training—remained
significant through most of the regression analysis, but gradually became
nonsignificant as other intermediate outcome variables entered the equation.
Unlike the precollege measures, these environmental measures occurred while
the respondents were in college, indicating that these characteristics predict
changes in religious convictions as well as in associated beliefs and values.
Results: Intermediate Outcomes
The intermediate outcomes showed the most interesting and unexpected
findings in the analysis. Nine of the 15 variables entered the regression equa-
tion. (See Table 3.) All of these qualities remained significant throughout
the regression analysis. Among them, attending religious services was most
highly associated with a strengthening of religious convictions (beta = .36).
This variable accounted for more than half of the total R-square. Similar to
LEE / Religion and College Attendance 379

TABLE 3
STANDARDIZED REGRESSION COEFFICIENTS
FOR INTERMEDIATE OUTCOMES
(N = 4,278)

Variable Zero-Order Beta After Beta After Final Beta

Precollege College
Environment

Attended religious
services (posttest) .42 .46** .43** .36**
Liberal views (posttest) -.33 -.33** -.30** -.15**
Goal: Influence social
values (posttest) .08 .16** .14** .08**
Goal: Help others in
difficulty (posttest) .19 .15** .13** .04*
View: Prohibit homosexual
relationships (posttest) .18 .16** .15** .06**
View: Individual can do
little (posttest) -.14 -.11** -.09** -.04*
Goal: Be very well off
financially (posttest) -.08 -.07** -.07** -.05**
Self rate: Acceptance of
different races/culture .17 .15** .14** .12**
Self rate: Emotional health .14 .14** .12** .07**

R2 =.17
*p < .01
**p < .001

all the variables in this block, this variable was pretested, indicating the pro-
pensity to attend church prior to college was controlled, which further shows
that religious activity during college helps to strengthen religious beliefs.
Whether these religious services are typically part of, or extraneous to the
college, remain unclear. In any case, religious activity does seem to promote
religious faith. Moreover, attending religious services while in college pre-
dicts changes in religious convictions, to an extent beyond what is predicted
by gender or precollege views. While intermediate outcomes cannot neces-
sarily determine cause, Astin (1993a) explains that some intermediate out-
comes can be considered to be environmental measures, given that the
environmental event precedes the outcome assessment. As such, one can
suppose that religious service attendance contributes to a strengthening of
religious convictions. Conversely, previous studies have found a lack of
church attendance among drop-outs (Hadaway & Roof, 1988). (See Table 3.)
380 THE REVIEW OF HIGHER EDUCATION SUMMER 2002

I also pretested intermediate outcomes for students’ goals and views, a

step demonstrating that the changes in personal goals and views during
college are associated with students’ self-ratings about religious beliefs.
Unlike the religious service attendance measure, these variables do not iden-
tify cause. However, the findings do provide a richer description of aspects
related to religious change. The second highest beta among the intermedi-
ate outcomes was liberal views (beta = -.15). Past research had found that
college attendance has a liberalizing effect on students (Astin, 1993b; Funk
& Willits, 1987). Given this tendency, it appears that this liberalizing effect
is strongly associated with a weakening of religious convictions. Another
set of associations also seems significant. The goals of improving society,
influencing social values, and helping others in difficulty are positively re-
lated to religious convictions, while the goal of becoming financially well
off is related to a weakening of religious convictions.
Two self-ratings also entered the regression equation. Although these two
particular variables were not pretested they do help to provide some expla-
nations about those who strengthen their religious beliefs. Positive self-rat-
ings regarding an acceptance of different races and cultures as well as
emotional health are positively associated with a strengthening of religious
beliefs and convictions. These characteristics seem appropriate, given the
values of acceptance and tolerance among most major religions, as well as
the emphasis on spiritual well-being.
Non-Entering Variables
Contrary to the hypothesis, involvement in diversity-related activities
had no discernible effect on religious change. In fact, none of the diversity-
related experiences entered into the regression equation and therefore none
had a direct effect on the outcome measure. Table 4 shows the betas of these
intermediate outcome variables had they entered the equation. Interest-
ingly, one of the variables—”attended cultural diversity workshop”—would
have had a significant, positive effect toward a strengthening of beliefs, after
controlling for all precollege and college environment variables (beta = .04),
though it would have gradually become insignificant after controlling for
the other independent variables. The other variables—had roommate of
different race/ethnicity, enrolled in ethnic studies course, and enrolled in women’s
studies course, and the ethnic interaction composite—did not significantly
affect the outcome after controlling for the other variables in the equation.

L IMITATIONS
One limitation of using this particular outcome variable was that there
was no matching pretest question that classified self-rated perceptions of
religious convictions. Hence, the degree of religious conviction prior to
LEE / Religion and College Attendance 381

TABLE 4
STANDARDIZED REGRESSION COEFFICIENTS FOR DIVERSITY
VARIABLES IF THEY HAD ENTERED THE REGRESSION EQUATION
(N = 4,278)

Variable Beta After Beta After Final Beta

Precollege College
Environment

Attended cultural
diversity workshop .06* .04* .02
Had roommate of
different ethnicity .01* .01 .00
Ethnic studies course .03 .03 .01
Women’s studies course .01 .01 .00
Ethnic interaction .04 .04 .00

* p < .01

college entry remains unknown. Nonetheless, the dependent variable sought

to measure the degree of change since entering college. While incoming
degrees of religiosity will likely differ by various institutional types, this
study’s purpose was to understand factors that contribute to the overall
change of religious beliefs and convictions during college. Although the self-
rating variable, measured near the end of the senior year, does not account
for the number of fluctuations in beliefs or the degree of conflict during the
college years, it does explain the change by the end of the college experi-
ence. That is, the study reports how the students perceived the strength of
their senior-year beliefs and convictions compared to their beliefs and con-
victions upon entering college. Future studies may consider measuring self-
ratings of religious beliefs at multiple points throughout college, in addition
to the senior year.
Identifying the predictors of the outcome variable was limited, as evi-
denced by an R-square of .27, most of which was attributed to a single vari-
able: attending religious services. While the study sought to test previous
findings in predicting religious change, the variables that I used accounted
for a small proportion of the variance. This limitation resulted from using a
survey that was not specifically designed to address religious change. Con-
sequently, other factors leading to religious change remain unknown and
could perhaps be better addressed in an exploratory, qualitative study.
382 THE REVIEW OF HIGHER EDUCATION SUMMER 2002

Discussion/Conclusion
The goal of this study was to understand how college influences stu-
dents’ religious beliefs and convictions. Some of the findings were unex-
pected. First, while students tend to experience changes in religious beliefs,
more students experience a strengthening of religious convictions than those
whose faith weakened. This finding challenges previous findings that col-
lege has an overall secularizing influence on students (Berger, 1973a;
Pascarella & Terenzini, 1991).
Interestingly, while this study finds an increase in religious convictions,
others claim a decrease in religious activity (Astin, 1993b; Bowen, 1997; Cox,
1988; Levine, 1980). Given that most studies on higher education and reli-
gion have observed only changes among monotheistic religions, it is pos-
sible that this finding illuminates students’ increasing acceptance of multiple
religions, beyond any single religious doctrine. For that reason, students
may engage in less church attendance and observance of the Sabbath but
may still hold strong convictions about being “spiritual” and may develop
more tolerance of multiple religious faiths compared to their beliefs when
they first entered college (Lee, 2002).
This study also demonstrated possible ways in which college attendance
could strengthen religious convictions. Beyond the findings already estab-
lished in research about college and religion, I found that attending reli-
gious services leads to stronger personal religious beliefs and convictions.
This finding is not surprising. After all, frequent religious participation and
strong religious convictions are closely intertwined. However, this study
makes the case that religious behavior (i.e., attending religious service) leads
to stronger religious convictions (i.e., stronger personal faith). This study
also found that attending an institution with peers who frequently attend
religious services strengthens one’s personal beliefs, confirming past research
on the powerful influence of peer groups. Other experiences within the in-
stitution, such as faculty interactions and leadership training, play a less-
direct but still-significant role.
Nevertheless, such explanations ought not to dismiss the sizable fraction
of students (13%) that self-reported a weakening of religious beliefs. For
these students, the college years are marked by religious questioning and
doubt. As expected from past findings, those who reported a weakening of
religious beliefs tend to be liberal and possess poor emotional health. Adopt-
ing liberal views, in particular, can result in a weakening of religious faith.
Given that college has been shown to liberalize students’ views (Astin, 1993b),
the study’s finding that liberal views are strongly related to weaker religious
beliefs makes evident that college does affect students’ religiosity. Liberal
views regarding casual sex, marijuana, and abortion arguably run contrary
to most monotheistic ideologies, and adopting such views quite logically
would lead to religious questioning.
LEE / Religion and College Attendance 383

Surprisingly, engaging in diversity-related activities does not necessarily

lead to weaker religious convictions. This finding raises questions about
Peter Berger’s theory that pluralistic interactions result in a weakening of
religious beliefs. Instead, interactions designed to increase students’ expo-
sure to diversity are not significantly associated with any religious change.
Future studies should investigate possible indirect reasons for this finding
or consider other aspects of the college experience that would further challenge
the asserted link between a pluralistic environment and religious decline.
Another unexpected finding is that institutional religious affiliation does
not affect changes in students’ religious beliefs. Although colleges certainly
vary from one to another, having a religious affiliation does not necessarily
make a difference in students’ convictions. Again, one reason might be due
to a lack of a survey item that would explain the degree of religious convic-
tion upon entering college. Students with very strong beliefs may leave col-
lege with the same high degree of convictions. Nevertheless, whether
religious schools are affecting changes in students’ religious lives while in
college remains questionable. Hence, this study has important implications
for institutional policy makers and researchers, in both sectarian and non-
sectarian schools. Nonsectarian institutions should note possible effects on
students’ religious beliefs and determine whether such effects are impor-
tant concerns for the institution. In light of this suggestion, future research
should consider exploring the effects of religious beliefs on more explicit
institutional concerns, such as leadership, societal welfare and community
service, and academic achievement. Secondly, sectarian institutions that
prioritize their religious affiliation should reconsider and reexamine their
current approaches toward developing the religious lives of their students.
Possible ways to strengthen students’ religious beliefs, based upon the find-
ings from this study, are implementing policies toward increasing student-
faculty interactions, as well as increasing support for religious student
organizations and activities, such as religious services.
In conclusion, this study has demonstrated that college does indeed impact
students’ religious beliefs and convictions. Given the importance of religion
in the lives of so many students, it is high time for institutions to consider
whether and/or how they will take further responsibility in such religious
changes. The findings presented here offer some guidance, while the challenge
to continue exploring this important domain of higher education remains.

R EFERENCES
Astin, A. W. (1993a). Assessment for excellence: The philosophy and practice of assess-
ment and evaluation in higher education. Phoenix, AZ: Oryx Press.
Astin, A. W. (1993b). What matters in college?: Four critical years revisited. San Fran-
cisco: Jossey-Bass.
384 THE REVIEW OF HIGHER EDUCATION SUMMER 2002

Berger, P. (1973a). Homeless mind. New York: Vintage Books.

Berger, P. (1973b). The social reality of religion. New York: Penguin Books.
Bowen, H. (1997). Investment in learning: The individual and social value of Ameri-
can higher education. Baltimore, MD: Johns Hopkins University Press.
Caplovitz, D., & Sherrow, F. (1977). The religious drop-outs: Apostasy among college
graduates. Beverly Hills, CA: Sage.
Condran, J., & Tamney, J. (1985). Religious nones: 1975–1982. Sociological Analy-
sis, 46(4), 415–423.
Cox, S. (1988). A longitudinal study of value preferences changes as an index of adult
development in Berea College students and alumni. Paper presented at the
meeting of the American Educational Research Association, Washington, DC.
Dalton, J. C. (1989). The influence of peer culture on college student values. Na-
tional Association of Student Personnel and Administrators Journal, 26(3), 180–
86.
Donahue, M. J., & Benson, P. L. (1995). Religion and the well-being of adolescents.
Journal of Social Issues, 51(2), 145–161.
Erikson, E. (1964). Insight and responsibility. New York: W. W. Norton.
Funk, R. B., & Willits, F. K. (1987). College attendance and attitude change: A panel
study, 1970–1981. Sociology of Education, 60(4), 224–231.
Hadaway, C., & Roof, W. (1988). Apostasy in American churches: Evidence from
national survey data. In D. Bromley (Ed.), Falling from the faith (pp. 29–46).
San Francisco, CA: Sage.
Hoge, D. (1988). Why Catholics drop out. In D. Bromley (Ed.), Falling from the
faith (pp. 81–99). Newbury Park, CA: Sage.
Hurtado S., Milem, J., Clayton-Pederson, A., & Allen, W. (1999). Enacting diverse
learning environments: Improving the climate for racial/ethnic diversity in higher
education. ASHE-ERIC Higher Education Report, Volume 26, No. 8. Wash-
ington, DC: George Washington University, Graduate School of Education
and Human Development.
Kotre, J. (1971). The view from the border. Chicago: Atherton.
Lebowitz, H. J. (1999). Resurgence of campus activism. Sojourners, 28(5), 18–23.
Lee, J. (in press). Changing worlds, changing selves: The experience of the religious
self among Catholic collegians. Journal of College Student Development.
Levine A. (1980). When dreams and heroes died. San Francisco: Jossey-Bass.
Maton, K. I., & Wells, E. A. (1995). Religion as a community resource for well-
being: Prevention, healing, and empowerment pathways. Journal of Social
Issues, 51(2), 177–194.
Pascarella, E. T., & Terenzini, P. T. (1991). How college affects students: Findings and
insights from twenty years of research. San Francisco: Jossey-Bass.
Roof, W. C., & Hadaway, C. K. (1977). Shifts in religious preference: The mid-sev-
enties. Journal for the Scientific Study of Religion, 16(3), 409–412.
Roof, W. C., & Hadaway, C. K. (1979). Denominational switching in the seventies.
Journal for the Scientific Study of Religion, 18(3), 363–379.
Roof, W. C., & McKinney, W. (1987). The new shape of American religion. New
Brunswick, NJ: Rutgers University Press.
Watanabe, T. (2000, October 18). The new gospel of academia. Los Angeles Times, p.
A-1.