Download as pdf or txt
Download as pdf or txt
You are on page 1of 36

Res High Educ (2013) 54:895929

DOI 10.1007/s11162-013-9301-z

Public Versus Private Colleges: Political Participation


of College Graduates
Joe L. Lott II. Jose Hernandez Joe P. King Tiffany Brown
Ismael Fajardo

Received: 22 June 2012 / Published online: 21 May 2013


 Springer Science+Business Media New York 2013

Abstract Using data from the Baccalaureate and Beyond Longitudinal Study (B&B:93/
03) of College Graduates, we use structural equation modeling to model the relationships
between college major, values held in college, collegiate community service participation,
and the post-college political participation of college graduates by public versus private
institutions. We use Hollands Theory of person-environment fit as lens to understand
differences in political participation across majors and institutional contexts. Over a
10-year period immediately after receiving the baccalaureate, we find that choice of major
and individual values are differentially associated with post-college political participation
for private institution graduates when compared to the counterparts at public institutions.
We relate our findings to extant literature that highlights the differences in institutional
characteristics between public and private colleges and socialization patterns of undergraduates that may inform differences in post-college political participation. Implications
for future research are also offered.
Keywords Political participation  Baccalaureate and beyond (B&B)  Structural
equation modeling (SEM)  Public/private institutional differences

J. L. Lott II. (&)  T. Brown  I. Fajardo


Department of Education Leadership and Policy Studies, College of Education, University of
Washington, Box 353600, Seattle, WA 981051, USA
e-mail: jlott1@uw.edu
T. Brown
e-mail: tjbrown1@u.washington.edu
I. Fajardo
e-mail: ifajardo@uw.edu
J. Hernandez  J. P. King
Department of Educational Psychology, University of Washington, Seattle, WA, USA
e-mail: joseh@u.washington.edu
J. P. King
e-mail: jp@joeking.com

123

896

Res High Educ (2013) 54:895929

The purpose of this paper is to understand the relationships between college major, values
held in college, community service participation during college, and the post-college
political participation of college graduates. Using data from the Baccalaureate and Beyond
Longitudinal Study (B&B) of College Graduates, which gathers information from
respondents who graduated from college during the 19921993 academic year and follows
up with them in 19931994, 1997, and 2003, this extends previous research that investigated political participation using the 1994 and 1997 waves of the B&B (Nie and Hillygus
2001; Hillygus 2005), to also include 2003 political participation10 years after
respondents graduated from college. Given the inconclusive evidence about the impact of
college experiences on political engagement outcomes in general (Pascarella and Terenzini
2005), and the underdeveloped research about the relationships between variables associated with the college experience and post-college political participation of college
graduates, we seek to understand the extent to which the relationship between 1994 and
2003 political participation differs for graduates of private higher education institutions
when compared to graduates of public higher education institutions.
We use structural equation modeling (SEM) is to understand how information collected
in 1993college major, values held in college, community service during college, and
socio-demographic controlsis associated with 1994 political participation of public and
private college graduates. Once understanding the within-institutional associations, we test
whether the relationships between 1993 predictors and 1994 political participation significantly differ for public versus private institutional modelsa between-institutional
analysis. We then investigate the relationship between 1993 predictors and 2003 political
participation. Our analytic approach allows us to understand the extent to which relationships between 1993 predictors and 1994 political participation endure in 2003. This
study makes a significant contribution to our developing understanding about the relationships between college experiences and post-college political participation of college
graduates. We conclude with implications and directions for future research to further
understand political participation.

Review of the Literature


Understanding the relationship between college experiences and post-college political
participation has become increasingly important, as many institutions and national organizations have created initiatives to address the declining interest in civic and political
participation among college students (Association of American Colleges and Universities
2012; Colby et al. 2007). Political participation is the necessary instrument to promote
democracy (Verba and Nie 1972; Verba et al. 1995); and individuals who attend college
engage in significantly more political participation than individuals who do not attend
college (Kam and Palmer 2008; Verba and Nie 1972). Studies have investigated the impact
of pre-college, academic, social, and institutional influences on the political engagement
outcomes of undergraduates (Astin 1996; Astin et al. 2006; Beaumont et al. 2006; Colby
et al. 2007; Dey 1996, 1997; Pascarella and Terenzini 2005; Sax 2004). However, there is a
gap in the literature that provides insights into the relationships between experiences and
attitudes in college and post-college political participation. Because college experiences
have the potential to mediate levels of post-college political participation, there is a
pressing need to understand which experiences may inform post-college political participation for college graduates.

123

Res High Educ (2013) 54:895929

897

A broad consensus agrees that there is a positive relationship between formal education
and political outcomes (Nie and Hillygus 2001; Pascarella and Terenzini 2005). This
positive relationship can partly be explained by the many opportunities students have to
become knowledgeable about political issues, communicate with others who have different
perspectiveswhich help them crystallize their views, values, and political standpoints
and learn how to actively engage in addressing a problem (Flanagan and Levine 2010;
Verba and Nie 1972; Colby et al. 2003). These opportunities range from a variety of
academic and co-curricular experiences, such as study abroad, class assignments that
promote critical thinking, service-learning projects, learning communities, student organization involvement and general student life activities (Pascarella and Terenzini 2005;
Colby et al. 2007). In addition, the college environment provides rich opportunities to
acquire cognitive abilities that enable comprehension of political content, and develop
civic skills and civic orientations that foster political action, and increase the opportunities
for social mobility that facilitate political participation (Kam and Palmer 2008).
There is limited research that investigates the relationships between college experiences
and post-college political participation for college graduates. Certain experiences have
been shown to increase the likelihood of post-college political engagement. For example,
involvement in large scale service-learning opportunities have been found to increase the
various attributes that are correlated with political participation, such as critical thinking,
heightened sense of civic responsibility, leadership skills, the ability to express civic
problems, and a stronger connection to civic problems, which lend themselves to participation (Eyler and Giles 1999; Astin et al. 2002; Pascarella and Terenzini 2005). Colby
et al. (2007) political engagement project (PEP) provided the most extensive insight into a
range of teaching and learning processes that inform the political development process of
college students. They investigated twenty-one courses and co-curriculum programs that
included a focus on political learning. Through interviews with faculty, program leaders,
and students across ten public 4-year colleges/universities, nine 4-year private colleges/
universities, one community college, and one association that is a collaborative of liberal
arts institutions dedicated to social justice, Colby et al. (2007) provided different examples
that show how one-semester projects, 2-year programs, summer programs, 30-day programs, and learning communities could include a diverse range of pedagogies, activities,
and collaborations that are designed to increase levels of political engagement for college
students. While these studies provide insight into the experiences during college that
inform political learning that may translate into post-college political participation, there
are few studies that expressly investigate the relationships between experiences in college
and post-college political participation for college graduates.
We extend the studies of Nie and Hillygus (2001) and Hillygus (2005), who provide the
most guidance to understand the relationship between college experiences and post-college
political participation of college graduates. In an effort to interrogate the various aspects of
the college experience that correlate with democratic citizenship outcomes, both of these
studies investigated the relationships between experiences in college and political
engagement outcomes of college graduates using the B&B. Nie and Hillygus (2001) used
data from the B&B:93/94, the first follow-up of the B&B study, to estimate ordinary least
squares (OLS) and logit regression models to explain how college major, academic performance, quality of college/university, race, and gender relate to six democratic
engagement outcomes of college graduates 1 year after receiving the baccalaureate. They
had two linear outcomes: political participation and community service participation. The
other four dichotomous outcomes included voting, political persuasion, valuing influencing
politics, and valuing financial wealth. In the full conditional model, six variables

123

898

Res High Educ (2013) 54:895929

significantly correlated with political participation. Being married, number of social science credits, and SAT verbal scores were positive correlates; and number of business
credits, number of science credits, and SAT math negatively correlated with political
participation within a year after receiving the baccalaureate. This is one of the first studies
to publish results about the relationship between college experiences and political participation of college graduates.
Whereas Nie and Hillygus (2001) investigated democratic engagement outcomes
located in the B&B:93/94, Hillygus (2005) investigated political engagement outcomes
using the B&B:93/97, 4 years after respondents received their baccalaureate degrees.
Hillygus (2005) investigated the extent to which three competing hypotheses (i.e., civic
education, social network, and political meritocracy) underlie the enduring correlation
between formal years of education and political engagement. She specifically estimated
two-stage least squares models to understand the impact of race, gender, socioeconomic
status, SAT scores, college major, college gpa, institution size, quality of college/university, marital status in 1997, graduate school enrollment, professional occupation, and
interest in politics on political participation in 1997 and voting in 1997. Hillygus found that
six variables significantly correlated with political participation in 1997: parents combined
education, SAT verbal scores, social science credits, and humanities credits had positive
associations with 1997 voting, while SAT math, and business credits had negative associations with the outcome. Seven variables had significant associations with voting in
1997. Being female, SAT verbal scores, age, social science credits, being married in 1997,
and political interest had positive associations with 1997 voting, while being Asian had the
only negative association with the outcome. Both studies provide valuable insight into the
relationship between some individual components of educationparticularly the college
environmentand political engagement outcomes. We extend these studies by investigating 1994, 1997, and 2003 political participation outcomes in the B&B:93/03.

Conceptual Framework
One of the goals of this paper is to understand how the relationships between college
experiences, values, community service, and post-college political participation differ
between public and private institutions. Given the dearth of research, a multigroup analysis
is an appropriate mechanism to begin understanding relationships. Although the weight of
empirical evidence regarding the impact of institutional control categories on political
engagement outcomes is inconclusive, research has found differences in activities that
shape students political disposition between public and private institutions (Astin 1996;
Astin et al. 2006; Kuh 1993; Dey 1996, 1997; Hanson et al. 2012; Pascarella and Terenzini
2005). This body of research has provided a foundation to further delve into differences
between educational experiences in public and private institutions and outcomesespecially post-college political participation.
Hollands Theory of person-environment fit (1966, 1973, 1985, 1997) also provides an
analytical lens to understand differences in post-college participation for private college
versus public college graduates because it considers personalities, environments, and the
interaction between personalities and environments, which allows a better understanding
about how students post-college political participation could be informed by their choice
of major and reinforced through their academic experiences (Smart et al. 2000). Hollands
(1997) typology of personality traits and environments emerged through research that
attempted to explain vocational behavior and to provide ideas for people to attain

123

Res High Educ (2013) 54:895929

899

vocational satisfaction, which proposed that individuals and environment can be classified
into six types. According to Holland (1997): (1) Realistic environments encourages
activities that entail the explicit, ordered, or systematic manipulation objects tools,
machines, and animals; (2) Investigative environments encourage intellectual activity
aimed at the creation and use of knowledge; (3) Artistic environments encourage ambiguous, free, unsystematized activities and competencies to create art form or products (4)
Social environments encourage activities that stimulate people to engage in social activities, helping others, and seeing the world in flexible ways; (5) Enterprising environments
emphasizes engaging in enterprising activities such as selling or leading others to attain
organizational or self-interest goals; and (6) Conventional environments encourage systematic manipulation of data such as keeping records, filing materials, and reproducing
materials.
Smart et al. (2000) extend Hollands (1997) work to research college faculty and
students, and particularly investigate the extent to which students self-selection into their
academic majors match their Holland type. Hollands theory assumed that each personality
type more or less aligns with environments that provide opportunities, activities, and tasks
that were congruent with the competencies and interests that parallel each personality type.
Using a sub-sample of 5,450 faculty and 4,408 students, Smart et al. found a range of
outcomes that support Hollands theory, not the least of which was that students seek out
majors that are compatible with their personality types. Smart et al. were also able to
organize student and faculty samples by specific departments according to Hollands
classification. Given that we are extending the studies of Nie and Hillygus (2001) and
Hillygus (2005), we limit our description of Smart et al. organization of the departments to
the proxies for majors used in these two studieshumanities, social science, science and
engineering, business, and education (for more extensive discussion see Smart et al.
chapter 3). Drawing from Hollands (1997) attributes of the six model environments, Smart
et al. found that students who displayed Realistic personality traits tended to major in
electrical and mechanical engineering. Students with Investigative personalities tended to
major in a range of majors including social science (i.e., anthropology, ethnic studies,
sociology), science and engineering (i.e., general biology and related biological fields,
chemistry, physical science, astronomy), and business (i.e., only finance) while Investigative faculty tended to belong to Biological/Life sciences, economics, geography,
mathematics/statistics, and physical sciences. Students with Artistic personalities tended to
major in the humanities (i.e., arts, English, language, music, and theater) while their faculty
tended to belong to Fine Arts and Foreign Languages). Students with Social personalities
tend to major in Humanities (i.e., philosophy, religion, and history), Social Science (i.e.,
political science, psychology, social work, and womens studies), and Education (i.e.,
elementary education) while their faculty tended to belong to the Social Sciences and
Humanities. Students with Enterprising personalities tended to major in Business (i.e.,
marketing, management, business administration and education) and Industrial Engineering while the faculty tend to belong to Business. Students with Conventional personalities
tend to major in accounting and no faculty were listed in this category.1
In addition to students personalities and their academic environment, the third component of Hollands theory is interaction between students personality and their academic
environment, which have implications for the socialization within each major that differentially reinforce and reward student ability and interests. Building on past studies that
1

Smart et al. (2000) noted that generic categories of Education and Engineering were unclassifiable
because of individual specialties within broad classifications that represented multiple Holland Types.

123

900

Res High Educ (2013) 54:895929

found partial support for the validity of the socialization assumption of Hollands theory,
Smart et al. (2000) investigated the differential change and stability in college students
focus on self-perceptions of their abilities and interests over a 4-year period beginning
when they were a freshman. Overall, they found that students in Investigative, Artistic,
Social, and Enterprising majors had increasing differentiation from other students in their
respective environments, providing support that these environments reinforce and reward
students personalities and respective sets of abilities. They also found that faculty create
distinctive academic environments in a manner generally consistent with the postulates of
Hollands theory (p. 98). Given past research that found differences in political participation by major, we consider Hollands theory to interpret some of the differences we may
find in post-college political participation of college graduates.
Although many colleges and universities make little formal effort to shape students
political engagement outcomes and values, the socialization processes of students across
different types of institutions are inevitably informed by organizational, interpersonal, and
intrapersonal processes (Dey 1997; Weidman 1989), which may inform political engagement
outcomes. Students self-select into a major based on a variety of factors, including fit that is
many times associated with a particular major. Faculty tend to shape the dispositions and
orientations in undergraduate education by providing a range of general educational
approaches, discipline specific types of education, educational and occupational training
(Brint et al. 2009; Pascarella and Terenzini 2005; Biglan 1973a, b). The combination of
student personality, the nature of academic disciplines, and the faculty who tend to share
characteristics ascribed to the students tend to be mutually reinforcing dimensions that in
some cases may create the conditions that promote political participation and in other cases
may create the conditions that inhibit political participation. Research has yet to understand
the relationship between college experiences and post-college political participation between
respondents from public and private institutions. We fill this void and provide evidence for
understanding the political participation of respondents who attended a public versus private
college/university in a longitudinal context.

Research Questions
Given the above, the following research questions guide this study:
(1)
(2)
(3)

Are there differences in post-college political participation for graduates of public


institutions versus graduates of private institutions?
Does the relationship between 1994 political participation and 2003 political
participation differ for private and public college graduates?
What is the relationship between college experiences and political participation in 1994?
(3a) Are there differences across college majors (i.e., social science, science,
business)?
(3b) Are there differences across values held in college (i.e., political influence,
community leadership, wealth)?
(3c) Are there differences for those who engaged in community service during
college versus those who did not?

(4)

To what extent do relationships discovered between college experiences and 1994


political participation endure over a 10-year time period (2003 political
participation)?

123

Res High Educ (2013) 54:895929

901

Hypotheses
Given research has yet to offer empirical insights into the relationships between college
experiences and post-college participation using a reliable and valid measure of political
participation, we test for several general hypotheses: (1) We expect to find graduates who
attend private colleges and universities will participate in politics to a greater degree than
graduates who attend public colleges and universities. (2) Similarly, given that the
strongest predictor of future political participation is past political participation (Verba and
Nie 1972), and given the research suggesting that private institutional educational approach
that better lends itself to political participation engagement (hypothesis 1), we believe the
relationship between 1994 and 2003 political participation will be stronger for private
institutions than public institutions. (3a) Post-college political participation will vary by
college experience. We expect majoring in the social sciences versus sciences, business,
humanities, and education will relate to differing levels of political participation; We also
expect to private institution graduates across all majors will have higher levels of political
participation than public institution graduates. (3b) Post-college political participation will
vary by values and moral orientations. Specifically, we expect political participation will
differ by values held in college: having political influence, being community leaders, and
being wealthy; and (3c) Post-college political participation will vary by community service
participation during college. (4) The relationships between 1993 predictors and 1994
political participation will be different than the relationships between 1993 predictors and
2003 political participations. We expect these sets of relationships to be stronger for
graduates from private institutions than public institutions.

Data and Measures


Data
This study uses data from the Baccalaureate & Beyond Longitudinal Study (B&B:93/03)
from the National Center for Education Statistics (NCES). Following students who completed their bachelors degree during the 19921993 academic year, the B&B study collected data from respondents across three different waves post-baccalaureate, 1994, 1997,
and 2003. Data collected from B&B:93/03 respondents include information about information about work experiences, family formation, student loans and finances, civic and
political engagement experiences, and post-baccalaureate education at the graduate level
(see methodology report: Wine et al. 2004). Over 10,000 college graduates across 1,200
institutions are represented in the B&B:93/03. This is a study using secondary analysis of
existing data. There was a large amount of missing data, particularly for the items that
comprise the 1997 and 2003 political participation factors (i.e., slightly above 40 %). In
general, we found that there were some respondents who participated in the study in 1994
but did not participate in 1997 and 2003. Also, there were some respondents who started
the study in either 1997 or 2003. There were also some respondents who participated in all
three waves but didnt respond to all of the items. Given this combination participation and
item response, our final analytic sample with complete data includes 5,143 respondents
from 501 institutions. Institutions in this study represent 246 public institutions
(n = 3,312) and 255 private institutions that includes 136 religious institutions (n = 887)
and 119 private non-religious institutions (n = 944).

123

902

Res High Educ (2013) 54:895929

The final sample is 55 % female. Respondents in the sample are mostly White (83.4 %),
while there is an adequate sample of Black respondents (5.9 %), Hispanic respondents
(4.9 %), Asian respondents (3.9 %), and Other Race respondents (1.9 %). The Other
Race category included 25 (0.5 %) Native American respondents, and 95 (1.85 %)
respondents who marked other for race. The B&B:93/03 panel weight (BNBPANL3) was
applied to approximate the population of 19921993 bachelors degree recipients to
address analytic issues associated with the use of data collected through complex sampling
designs. Problems associated with complex databases, such as B&B:93/94, are well documented in the literature (see, for example, Heck and Thomas 2009; Thomas et al. 2005;
Thomas and Heck 2001). We created a relative weight by dividing the raw panel weight by
its mean to preserve the effective sample size while still adjusting for oversampling of
some groups (Thomas and Heck 2001) thereby minimizing the influence of oversampling
on standard errors (Perna 2004).

Measures
Political Participation
Political participation is measured at three different time points in this study: 1994, 1997,
and 2003. Our political participation measures are informed by Nie and Hillygus (2001)
and Hillygus (2005) studies. Using the B&B:94, Nie and Hillygus (2001) political
participation measure is a single scale of interrelated activities that include campaign
volunteering, attending a political rally or meeting, contributing money to a political
campaign, and writing a letter to a public official. They transformed these items into a
scale using homogeneity analysis by means of alternating least squares, or Homals.
Using the B&B:97, Hillygus (2005) dichotomized political participation measure is
based on whether respondents participated in any of the following: written to a public
official, attended a political meeting, contributed money to a political candidate, or
contributed money or time to a political cause. Both studies operationalization of
political participation, although different, are consistent with extant studies political
participation measures, as traditional examinations of political participation either
aggregate an individuals total number of political actions to create a participation scale,
create a dichotomous variable indicating whether or not a respondent participated in at
least one act or not, or simply use voting to measure it (Oesterle et al. 2004; SinclairChapman et al. 2009).
Our political measures include the following (See Table 1 for descriptive information):
1994 Political ParticipationActively campaign for candidate (PolBs 1994), talked to
someone about politics (PolShw 1994), Gave money to campaign (PolMny 1994), Time or
money to political action groups (PolAct 1994), Wrote letter to public official (PolLett
1994), and attended political meeting (PolMeet1994); 1997 Political Participation
Attended political meeting (PolMeet 1997), gave money to Campaign (PolMny 1997),
Talked to someone about politics (PolSh 1997), Money to political action groups (PolAct
1997), and Wrote letter to public official (PolLett 1997); 2003 Political Participation
Attended political meetings, rallies, or dinners in past 2 years (Polit 2003), Wrote to public
official in 2003 (PolLett), E-mailed a public official in 2003 (PolEmail), and Called a
public official in 2003 (PolTelph).

123

Res High Educ (2013) 54:895929

903

Table 1 Descriptive statistics


Mean

SD

Political participation variables


1994
Political activity time or money to pol. action groups

0.07

Try to talk to someone about candidates

0.23

0.25
0.42

Gave money to campaign

0.07

0.25

Time or money to political action groups

0.15

0.36

Sent letter to public official

0.15

0.36

Attend political meeting

0.12

0.33

Attended political meetings

0.16

0.37

Gave money to campaign

0.09

0.29

Talk about politics candidates

0.26

0.44

Time or money to political action groups

0.18

0.39

Written letter to public official

0.20

0.40

Activities past 2 years attend political meetings, rallies, or dinners

0.16

0.37

Wrote to public official

0.18

0.38

Emailed public official

0.27

0.44

Called a public official

0.13

0.33

Female

0.55

0.50

Black

0.06

0.24

Hispanic

0.05

0.22

Asian

0.04

0.19

Parents education

0.00

1.81

Married in 1993

0.27

0.44

Married in 1997

0.44

0.50

Married in 2003

0.58

0.49

Kids in 1997

0.15

0.35

1997

2003

Covariates

Kids in 2003

0.51

0.49

Own home in 2003

0.60

0.48

Community leader 1993

0.40

0.49

Wealthy 1993

0.60

0.49

Political structure 1993

0.39

0.49

Humanities credits

17.35

15.57

Social science credits

22.14

18.18

Science/engineering credits

15.35

21.99

7.61

14.89

6.86

15.52

Business credits
Education credits
Age in 1993

24.35

5.953

Community service in 1993

35.9

0.46

123

904

Res High Educ (2013) 54:895929

College Curriculum
Similar to Nie and Hillygus (2001) and Hillygus (2005) we tested the relationships between
self-selection effects of the college curriculum and post-college political participation.
Academic majors have distinct socialization experiences for college students and these
academic environments shape and reinforce students personalities and dispositions
(Holland 1985; Pike 2006). Academic majors that impart the knowledge, skills, and
political familiarity help students bridge the gap between understanding their role in a
democratic system and political action (Hillygus 2005). Given that private institutions and
public institutions generally have different approaches to general education (Brint et al.
2009; Stevens 2001), which likely increases students odds of being differentially exposed
to a set of activities that promote political participation before declaring a major, the
relationship between college major and political participation may differ across institutional contexts. As a proxy for major, similar to Nie and Hillygus (2001), and Hillygus
(2005), we included the total number of credits students accrued in the humanities, social
sciences, business, and education, respectively; and these variables were treated as
continuous.
Values Held Immediately Post-Graduation
We also examine the relationships of values on post-college political participation. Values
are principles and standards that shape the fundamental aspects of our frame of reference
(Shaver and Strong 1982), and for college students these values are a point of connection
between their campus life and their personal life (Morrill 1980). A set of contrasting
values, such as hope and pessimism, materialism and idealism, and individualism and
collectivism are related to civic participation, and they are important elements in understanding political participation (Snell 2010). Nie and Hillygus (2001) considered a contrasting set of values, public versus private regard, to understand how college experiences
are associated with them. In two logit models where influencing politics (public) and
valuing being financially wealthy (private) were outcomes, they found attending graduate
school and number of social science credits taken were positively associated with influencing politics, while the number of business credits taken and SAT math scores had a
negative effect on influencing politics. They also found that Black students (compared to
White students), and the number of business credits taken during college had positive
effects on valuing being financially wealthy, while being female, undergraduate GPA, the
number of education credits, and SAT verbal scores had a negative effect on valuing being
wealthy. We will build on their findings to understand the degree to which the relationships
between these values and post-college political participation differ between public institution respondents and private institution respondents.
Three dichotomous variables measured values students held during their senior year,
which included whether they valued being a community leader, if they valued influencing
the political structure, or if they valued being financially wealthy. If this value was present,
students were coded 1, otherwise they were coded as 0. These variables come from the
National Postsecondary Student Aid Study of 19921993 (NPSAS:93)the year academic
year in which respondents graduateda study conducted by NCES to determine how
students and their families pay for postsecondary education. While the NPSAS:93 study
was designed to sample a cross-section of all students enrolled in the United States (NCES
1995), only the ones who graduated during the 19921993 academic year were included in
the B&B:93/03 study.

123

Res High Educ (2013) 54:895929

905

Community Service During College


We include community service during college as a predictor in our model. Political
engagement outcomes and civic engagement outcomes tend to be positively correlated
(Colby et al. 2007). There is a robust literature about the impact that community service
has on academic and co-curricular outcomes (Pascarella and Terenzini 2005). Research has
yet to report how community service during college may inform post-college political
participation. Also from the NPSAS:93, our community service variable is dichotomous.
More specifically, respondents were asked, between July 1, 1992 and June 30, 1993 did
you perform community service or volunteer work other than court ordered? Respondents
who did engage in service were coded 1, otherwise they were coded as 0.
Life Course
Our study uses a national longitudinal dataset, which allows us to understand how data
collected during the college years are associated with post-college political participation
10 years after respondents graduate. While college experiences may inform some of the
variance in post-college political participation, so will life course experiences of college
graduates. Stable patterns of civic and political engagement take hold once individuals
have settled into adult roles, such as steady jobs, marriage, and parenting, that build up
their stake in community affairs (Flanagan and Levine 2010, p. 160). The weight of
empirical evidence suggests that college experiences will shape the civic and political
disposition of students. However, these effects may not be realized until years after college
when their status across many social institutions and communities are more developed
(Oesterle et al. 2004). Therefore, similar to Nie and Hillygus (2001) and Hillygus (2005),
we control for some life course experiences on post-college political participation. We
included age, marital status, having children, and home ownership to control for life course
experiences. Age is a continuous variable. There were three dichotomous marriage variables; respondents who were married in 1994, 1997, or 2003 were coded 1, while nonmarried respondents were coded 0; respondents who had children in 1997 or 2003 were
coded 1, while those who did not have children were coded 0; and respondents who owned
a home in 2003 were coded 1, while those who did not own a home were coded 0.
Socio-Demographic
Several socio-demographic covariates were used. Race is comprised of four dummy
variables where a respondent was coded 1 for either being Black, Latino, Asian, or other,
and Whites were coded 0. Because of the small percentages those in the other racial
category are not shown in the final models. For gender, females were coded 1 and males
were coded 0. Parents education was used as a proxy for SES, because the parents
education and income did not yield a reliable estimate in the B&B data (Nie and Hillygus
2001) and Hillygus (2005) also used parents education.
Institutional Selectivity
We also examined the differences in the average selectivity between public and private
institutions using the Barrons Admission Competitive Index, which organizes colleges in
seven competitiveness categories from most competitive to non-competitive based on
their entrance requirements (Schmitt 2009). Institutional selectivity has been found to be

123

906

Res High Educ (2013) 54:895929

associated with a range of student outcomes (Pascarella and Terenzini 2005; Pascarella
et al. 2006; Toutkoushian and Smart 2001). If there are average selectivity differences
between public and private institutions in this study, these differences may contribute to
any varying relationships found between predictors in the study and post-college political
participation across institutional contexts. To date, there is no evidence that institutional
selectivity is related to post-college political participation.

Models
Confirmatory Factor Analysis (CFA)
The first step in understanding if political participation differs between private and public
graduates is to model the best items that measure political participation. Since the variables
hypothesized to measure political participation are measured differently across years in the
B&B, we have to understand how hypothesized items are measuring political participation.
In other words, is there congeneric measurement (Kline 2005) across these factors? To
understand if there is congeneric measurement across items that estimate political participation across three time points, confirmatory factor analysis (CFA) models were fit
using Mplus 7 (Muthen and Muthen 19982010).
Confirmatory factor analysis is employed when substantive theory is available to inform
the creation of latent constructs and prior information exists about the direction and
magnitude of the parameter relationships (Brown 2006; Loehlin 2004). In this manner,
effects on individual observed variables are used to indirectly inform unmeasured latent
variables. A CFA model determines the appropriateness of a hypothesized measurement
model by determining how well the data fit a proposed model. In statistics, a CFA model
takes on the following form (Joreskog and Sorbom 1996; Joreskog 1973; Joreskog and
Sorbom 1982; Kaplan 2000):
Let subjects be (i = 1, 2, , n) and j-items (j = 1, 2, 3, p)
yij mj ^j gi dij ;

gi  N0; w

dij  N0; H

where, yij equals the ith subjects score on the jth item and is a px1 vector of indicators
terms (m1, mp); gi is an mx1 vector of
(yi1,, yip)0 ; my, is a px1 vector of intercept
V
underlying latent variables (gi1, gim); y is a p 9 m factor loading matrix that relates yij
and gi through individual factor loading k; dij is a px1 vector of the measurement error
terms. Lastly, in order to estimate the model assumptions (2 and 3) are necessary.
32 3 2 3
2 3 2 3 2k
11 k12 . . . k1m
dyi
gi
m1
yi
6
7
6 yi 7 6 m2 7 6 k21 k22 . . . k2m 76 gi 7 6 dyi 7
6 76 76 .
76 7
4
.. 76
..
..
4 . . . 5 4 . . . 5 4 ..
. 54 . . . 5 4 . .y. 5
.
.
dp
gp
yp
mp
kp1 kp2 . . . kpm
Confirmatory factor analysis is a method designed to sort out and test the relationships
between individual observed variables and factors. Because of the categorical data being
used for this study, more robust cutoff measures are used; cutoffs C0.96 for CFI and B0.05

123

Res High Educ (2013) 54:895929

907

for the RMSEA indicate adequate model fit (Yu and Muthen 2002; Byrne 2006; Kline
2005). Figure 1 show the CFA models estimated.
If congeneric measurement is established for an overall political participation measure,
the next step is to conduct Tau equivalence tests to establish whether the individual
indicators representing the hypothesized constructs satisfy the equal factor loadings test
(Brown 2006; Graham 2006). Along with the determination of tau-equivalent models,
we conducted an individual test of reliabilitythe composite reliability for congeneric
measures models (CRCMM)for each of the three political participation constructs. The
CRCMM was first introduced by Fornell and Larcker (1981), and was later revisited by
Raykov (1997). Previous research has shown that Cronbachs coefficient, when used in
SEM, can be misleading if not problematic because it tends to underestimate reliability
(Fornell and Larcker 1981; Raykov and Widaman 1995; Raykov 1997). Using Raykov
(1997) and Fornell and Larckers (1981) the CRCMM will take the following form:
Pp
2
i1 kyi
5
qn Pp
2 Pp
i1 kyi
i1 Var di
where, kyi denotes the individual standardized factor loadings for a given construct; Var(di)
denotes the individual residual variance of each indicator; and R denotes the sum over
multiple factor loadings and error variance for the given construct.
If there is evidence of equivalent relationships, we will then assess the comparability
between public and private institutions multiple group comparison (MGC) approach
(Brown 2006; Kaplan 2000; Lubke and Muthen 2004). In an iterative process to determine
measurement invariance, we will first establish form invariance to assess the validity of
using our three factor measurement model on our groups of interest. This is done by testing
the measurement model separately for public and private institution samples, and assessing
the factor structure based on a nested covariance matrix. We will then determine if the
meaning of our political participation constructs have the same meaning in across the two
groups by testing the equality of factor loadings in a nested model for both groups by
assessing the intercepts.
We will first estimate an unconstrained model to establish a baseline model for comparisons in the multiple group framework. To establish metric invariance a model with the
factor loadings fixed equal with both groups was compared to the baseline model. Second
to assess full factorial invariance, a model with both factor loadings and term intercepts are
constrained and compared to the freely estimated baseline model. A Chi squared difference
test is used to assess invariance between models. To assess measurement invariance the
SatorraBentler Chi square difference test between the two group nested models will be
used (Satorra 1999). In short, if there is no significant difference between the constrained
model and the freely estimated model, then we have evidence for invariance (Baumgartner
and Steenkamp 1998; Brown 2006; Dimitrov 2006; Lubke and Muthen 2004; Milfont and
Fischer 2010). However, because of our large sample size, and the evidence in recent
research on the sensitivity of the Chi squared difference test in invariance testing (Cheung
and Rensvold 2002; Meade et al. 2006), aside from the traditional Chi squared difference
test, the following relative fit indices were used for establishing invariance across groups:
the change in the comparative fit index (deltaCFI: change values between 0 and 0.01);
change in the root mean squared error of approximation (delta RMESEA change values
between 0 and 0.01) and the change in the Tucker-Lewis Index (deltaTLI values between 0
and 0.01) (Baumgartner and Steenkamp 1998; Cheung and Rensvold 2002; Meade et al.

123

Fig. 1 Political participation measurement model

908

123

Res High Educ (2013) 54:895929

Res High Educ (2013) 54:895929

909

2006). In this present study change in the relative fit indices were given stronger consideration whenever the Chi squared difference test did not agree.
Once we establish measurement invariance of political participation measures between
public and private institutional models, we can now answer our primary research question.
The Wald test of parameter equality tests whether the relationships between 1994 and 2003
political participation measures are significantly different when comparing public institutions graduates versus private institution graduates. The Wald test determines whether
there is enough evidence to determine if the parameter estimate from one group is significantly different from another group, and this test is implemented in MPlus 7 via the
MODEL TEST command (Muthen and Muthen 19982010).

Structural Equation Modeling (SEM)


To answer research questions two and three we conduct MGCs using SEM so that we can
understand the relationships between college experiences and post-college political participation between public versus private institutions. SEM is the structural component of a
hypothesized CFA model (Kaplan 2000; Kline 2005; Skrondal and Rabe-Hesketh 2004).
SEM models complex dependencies of both observed and latent variables and is analogous
to running multiple regression equations simultaneously, while accounting for unique and
indirect effects of explanatory variables (Muthen 2002). Because we want to know the
predictive nature of the explanatory variables over three time points, SEM is the most
appropriate analytic technique for the study. In the case where exogenous predictors are
used, no distributional assumptions are made and their relationships with each other (if
multiple exogenous variables are used) and with the latent variables are captured and
controlled for by the model (Fox 2002; Joreskog and Sorbom 1982; Kaplan 2000; Muthen
2002). The structural model for the effects of the latent factors on the observed covariate
dependent variables is as follows:
gi a bgi CX1i fi

where, b in an m 9 m parameter matrix of slopes for regressions of latent variables on


other latent variables, and C is an m 9 q parameter matrix for regression coefficients of the
effects of the latent variables on the dependent variables X1i. Similarly as stated in the CFA
model, a is a vector of intercepts, and fi represents the unexplained components of the
model (residuals).
Similar to our CFA models, we will assess the fit of our SEM models with the comparative fit index (CFI) and the mean square error of approximation (RMSEA). Furthermore, because we had a substantial sample size in our data, we assigned less weight to the
Chi square difference test, as studies have revealed the high sensitivity of this test as
sample size increases (Chen 2007; Cheung and Rensvold 2002; Hooper et al. 2008).
We estimate three models to determine the relationships between college experience
variables and political participation across three time points. The first model estimates the
relationships between major, gender, race, parent education, values held in 1993, marital
status in 1993, age, and 1994 political participationafter controlling for all relationships
in the full model. The second model estimates the relationships between major, gender,
race, parent education, values held in 1993, marital status in 1997, have kids in 1997, age,
and 1997 political participation measuresafter controlling for all the relationships in the
full model. The third model estimates the relationships between major, gender, race, parent

123

910

Res High Educ (2013) 54:895929

Fig. 2 Full SEM model

education, values held in 1993, marital status in 2003, have kids in 2003, age, and 2003
political participation measuresafter controlling for all the relationships in the full model
(see Fig. 2 for final models so be estimated).
To understand the degree to which college experience variables are associated with
political participation measures, we estimated the parameter estimates for both models. If
the p value for the parameter estimates are below 0.05, we can conclude that there is a
statistically significant relationship between an independent variable and our political
participation measure. We observe the parameter estimates to mainly understand the within
group dynamics. In other words, after controlling for all relationships in the model, is there
a unique relationship between a covariate and political participation for within the model
of public college graduates or the model for private college graduates? To test if the same
variables path is significantly different for public institutions versus private institutions
(i.e., between group differences), similar to understanding if the relationship between 1993
and 2003 political participation significantly differ between public and private institution
graduates, we interpret results from the Wald test. There are instances where a variable

123

Res High Educ (2013) 54:895929

911

Table 2 Test of measurement invariance of public and private institutions for the three factor political
participation factors
Political participation

v2

df

Dv2

Model 0 (freely estimated


nested)

1067.67

164

Model 1 (equal factor


loadings)

1065.27

176

2.4

Model 2 (equal
factor ? intercepts)

1087.9

188

20.23

v2 TRd diff

Ddf

CFI

DCFI

RMSEA

DRMSEA

0.93
7.87245583
-55.423126***

12

0.93

0.001

0.033

24

0.93

0.001

0.032

0.001

v2 diff is nested
CFI comparative fit index, RMSEA root mean square of approximation
* p \ 0.05, ** p \ 0.01, *** p \ 0.001

(e.g., college major) may have a significant association with political participation in both
models for public and private institutions. If the Wald test is significant, it would suggest
that relationship between college major and political participation is significantly different
for one group versus the other. The higher standardized parameter estimate would give an
indication about which group has the stronger relationship between the covariate and
political participation.

Results
The following paragraphs provide the results of the study. The results of our measurement
invariant tests are shown in Table 2. Table 3 shows the results of our SEM model. We
provide the standardized estimates for total, direct, and indirect effects for each predictor.
The total effects are the sum of all direct and indirect effects of one variable on all other
variables in the model (see Kline 2005 for more information about their estimation and
decomposition). Given our goal is to understand the unique impact of variables associated
with the college experience, we only focus on and interpret the direct effect of relevant
predictor variables on post-college political participation.
Our findings show that there is congeneric measurement across the items located on
the B&B:93/03 that estimated political participation across three time points. These
factors are also shown to be reliable. This model showed an adequate fit, v2(96) = 6732,
p \ 0.05; (SRMR = 0.032, RMSEA = 0.042, CFI = 0.93). Figure 3 shows the confirmatory factor model that includes standardized path estimates. Based on the modification
indices, several pair of correlated errors (Joreskog 1973) on the same item across waves
were added due to their common relationship with each other that was measured beyond
their respective factor scores. These items include: Writing a letter to a public official in
1994 and 1997; and Contributing Money to a Campaign in 1994 and 1997.2 Based on
2

It is important to note that voting is not included in our political participation construct. As previously
stated, voting is one of the four broad modes of political participation (Verba and Nie 1972), and is often
used when created political participation measures (Oesterle et al. 2004; Sinclair-Chapman et al. 2009).
However, when we tested the voting items contribution to its respective political participation, the factor
loadings were 0.17, 0.23, and 0.24 for the 1994, 1997, and 2003 factors respectively.

123

123

0.586***

0.462***

19942003

0
0
0

0.007

-0.03

0.017

0.024

Black

Latino

Asian

Parent education
0

-0.020

-0.136***

Sci & Eng credits

Humanities credits

Social Sci. credits

0.026***

0.063

1997

0.135

Community service 93

-0.01

-0.08***

0.018

0.008***

-0.04*

0.281***

-0.07**

Age

Married

Influence political structure

Wealth desire 1993

Comm. leader in 1993

-0.04*

-0.07*

Education credits

Female

0
0

-0.09***

-0.06

Sci & Eng credits

0.392***

Business credits

0.019

0.068

Humanities credits

Social Sci. credits

1994

0.68***

19972003

-0.052*

-0.002

0.018

0.063

0.135

-0.008

0.281***

-0.042*

-0.073*

0.024

0.017

-0.029

0.007

-0.074*

-0.042*

-0.06*

-0.094***

0.068

0.019

0.07

0.586***

0.68***

-0.085**

0.139*

0.085

0.026

0.127

0.017

0.244***

-0.038

-0.034

0.082

-0.01

0.018

-0.037

-0.019

-0.084**

-0.041

-0.018***

0.207

-0.005

0.452***

0.233

0.657

Total

Direct

Total

Total indirect

Private

Public

19941997

Table 3 SEM results

-0.017

0.116***

-0.011

0.129**

Total indirect

-0.068*

-0.023

0.074*

0.026

0.127

0.017

0.244***

-0.038

-0.034

0.082

-0.01

0.018

-0.037

-0.019

-0.084**

-0.041

-0.018

0.207**

-0.005

0.323*

0.233

0.657***

Direct

0.127

2.246

1.579

0.844

0.276

0.302

3.007*

0.06

1.394

1.109

0.622

1.247

0.973

1.914

0.595

0.375

3.905*

3.855*

0.282

2.23*

2.63*

Wald

912
Res High Educ (2013) 54:895929

0.148*

-0.116***

-0.041

Sci & Eng Credits

Business credits

Education credits

-0.067***

-0.035

-0.039

Black

Latino

Asian

0.013

-0.098***

Social Sci. credits

Female

0.029

-0.050

Humanities credits

2003

Community service 93

-0.049

Kids in 1997

0.191**

0.043

Married in 97

Age

0.270***

-0.03

0.002

-0.01

0.013

-0***

-0.037

-0.024

-0.054*

0.009

-0.012

-0.042*

-0.002

-0.1***
-0.07***

-0.04

0.017

0.097**

0.167*

-0.049

0.043

0.072*

-0.006

0.024

0.066

-0.007

0.007

0.019

-0.066**

0.006

-0.054*

0.01

0.012

0.051

0.024**

0.198***

-0.04*

-0.045

Wealth desire 1993

Influence political structure

0.011**
-0.05**

0.011

-0.02

0.077

0.026

-0.018

Latino

Asian

0.009

-0.06**

0.075

0.028

Parent education

-0.126***

Female

Black

-0.05*
-0.04*

Comm. leader in 1993

-0.106***

-0.047

Business credits

-0.019

0.014

-0.044

-0.010

0.007

-0.058**

-0.046

0.021

0.057

0.043

0.036

0.111

0.060*

0.163***

-0.084*

-0.038

0.043

-0.010

0.043

0.100

-0.030

-0.105***

-0.084*

Total

Direct

Total

Total indirect

Private

Public

Education credits

Table 3 continued

-0.003

0.022

0.015

-0.017

-0.061***

-0.045*

-0.039*

0.08**

0.024

0.019

0.027

0.153***

-0.031

-0.019

0.038

-0.008

0.013

-0.022

-0.018

-0.059**

-0.039

Total indirect

-0.016

-0.008

-0.029

0.008

0.009

-0.013

0.007

-0.059

0.033

0.023

0.009

0.111

0.06*

0.01

-0.053

-0.019

-0.005

-0.002

0.03

0.078

-0.012

-0.046

-0.045

Direct

0.179

0.082

0.119

0.001

0.223

0.526

0.015

0.087

0.163

3.978*

9.204*

5.528*

0.058

1.944

0.879

1.113

2.597

0.014

0.28

0.59

1.926

1.564

0.144

Wald

Res High Educ (2013) 54:895929


913

123

123

* p \ 0.05, ** p \ 0.01, *** p \ 0.001

0.096***

0.140**

Age

Community service 93

-0.063

-0.003

Kids in 2003

Own a home 2003


0.092*

0.088**

-0.011

Married in 2003

0.185***

-0.03

0.239***

-0.056*

Wealth desire 1993

Influence political structure

0.039*
-0.02

0.045*

0.065

Parent education

0.048

0.008

-0.003

-0.063

-0.011

0.054*

-0.029

0.042

0.006

0.069

0.033

-0.043

-0.046**

0.053**

0.185**

-0.121*

-0.025

0.003

Total

Direct

Total

Total indirect

Private

Public

Comm. leader in 1993

Table 3 continued

0.023

0.035

0.12***

-0.042**

-0.03

0.029

Total indirect

0.046

-0.002

-0.043

-0.046**

0.053**

0.065*

-0.079*

0.005

-0.026

Direct

0.079

0.031

0.749

0.258

1.577

0.017

8.418**

0.889

0.362

Wald

914
Res High Educ (2013) 54:895929

Fig. 3 Political participation measurement model with estimates

Res High Educ (2013) 54:895929


915

123

916

Res High Educ (2013) 54:895929

results of z tests,3 all items had statistically significant factor loadings on their respective
political participation measure. In addition to congeneric measurement, we calculated the
reliability statistic, CRCMM, which shows that each one of our political participation
measures has a composite reliability statistic, qg = 0.99.
We also found evidence of measurement invariance of political participation between
public and private institution models. The unconstrained model which was fully estimated
between public and private institutions fit reasonably well to the data: v2(176) = 1,065,
CFI = 0.93, RMSEA = 0.033, TLI = 0.99). The constrained model shows equal factor
and intercepts, and equality of factor loadings (see Table 2). Given that we have equal
factors, we can now move forward with investigating the relationships between 1993
predictors and post-college political participation.
Hypothesis 1
The Mplus output for our multiple group CFA models provide results of the asymptotic Z test
that shows the mean differences between the political participation factors for graduates of
public versus private institutions. Mplus fixes one group mean to 0 and the mean of the other
group is the difference between the group means. Our results show that the mean difference
for the 1994 political participation factors is 0.008, suggesting that average political participation is higher for public institution graduates when compared to private institution graduates; and this difference is statistically significant (z = 2.64, p \ 0.05). However, there is a
non-significant difference in 2003, suggesting that average political participation 10 years
after respondents receive the baccalaureate is similar between the two groups.
Hypothesis 2
When examining results of the relationship between 1994 political participation and 2003
political participation, we found a statistically significant positive direct effect for private institutions (b = 0.323, p \ 0.05). For the public institution model, participation in 1994 does have a
significant effect on participation in 1997 (b = 0.68, p \ 0.001), which in turn has a significant
effect on participation in 2003 (b = 0.586, p \ 0.001); we find that while there is a significant
total effect between 1994 and 2003 political participation, most if it is explained through the
indirect effects of 19941997 and the effect of 1997 on 2003. For the private institutional model
there is a significant total effect between 1994 and 2003 political participation, and some of this
effect is explained by 19941997 political participation, and 1997 on 2003 political participation;
however, a significant direct effect remains between 1994 and 2003, which is also significantly
different than the same relationship for public institutions (Wald = 2.638, p \ 0.05). Therefore,
we are more confident that as respondents from private institutions engage in political activities
1-year after they received their baccalaureate the more they will continue and even increase their
political engagement 10 years after they attained the baccalaureate. We are less confident of this
relationship for public institution respondents in our sample.
Hypothesis 3a
Findings for the number of credits taken in an academic field, our proxy for major, show
that the number of social science credits has a statistically significant positive
3

Not shown but available upon request.

123

Res High Educ (2013) 54:895929

917

relationship with 1994 political participation for both institutional models. Therefore, the
more social science credits college graduates earned, the higher their level of political
participation in 1994. However, social science credits has a significantly larger direct
effect on 1994 political participation for private institution graduates than their public
institution counterparts (Wald = 3.855, p \ 0.05). The number of credits earned in
science and engineering has a statistically significant negative association with 1994
political participation for graduates of public institutions (b = -0.094, p \ 0.001); and
this relationship is statistically and significantly different from the association found for
graduates from private institutions (Wald = 3.905, p \ 0.05). Table 3 also shows that
the number of education credits earned had a statistically significant negative association
with 1994 political participation for graduates of private institutions (b = -0.084,
p \ 0.01).
Hypothesis 3b
For the public institution graduates, respondents who valued being a community leader
engaged in significantly less political participation versus those who did not hold this value
(b = -0.073, p \ 0.01). For both institutional categories, respondents who valued influencing the political structure engaged in significantly more political participation than
those who did not hold this value; however, this relationship was statistically and significantly stronger for public institution graduates than for private institution graduates
(Wald = 3.007, p \ 0.05).
Hypothesis 3c
Community service did not have a statistically significant association with 1994 political
participation in either of the institutional models. For the model of public institution
graduates, females engaged in significantly less political participation in 1994 than males
(b = -0.074, p \ 0.05). For the model of private institutions, parents education has a
statistically significant positive association with political participation in 1994
(b = 0.082, p \ 0.05). Age has a statistically positive association for both institutional
models.
Hypothesis 4
Major
Whereas the 1994 political model had several statistically significant direct associations
with major for both institutional models, business credits earned is the only variable that
has a statistically significant direct effect with political participation in 2003. Table 3
shows that there is a significant total effect for business credits earned for both institutional models; the total indirect effect for both models also explains a significant
amount of variance in the relationship between business and 2003 political participation.
However, only for public institutions is there a statistically significant direct effect
(b = -0.042, p \ 0.05). There is also to significant total effect for science and engineering credits in the public institutional model for 2003 political participation but most
of this effect is explained through the total indirect effectleaving a small and insignificant direct effect.

123

918

Res High Educ (2013) 54:895929

Values and Moral Orientations


Similar to the 1993 model, respondents who valued influencing the political structure in
1993 engaged in significantly more political participation in 2003. This association
endured for both institutional models. However, when interpreting the direct effects, the
associations are weaker in the 2003 model than the 1993 model. There is also a finding in
the 2003 model that was not present in the 1993 model. For graduates of private institutions, respondents who valued being wealthy in 1993 engaged in significantly less political
participation in 2003 (b = -0.079, p \ 0.05 and this relationship was significantly
stronger than the same relationship for the public institutional model (Wald = 8.418,
p \ 0.01), which has a non-significant direct effect.
Socio-Demographic Controls
For the public institutional model, Black respondents engaged in significantly less political
participation in 2003 than White respondents (b = -0.221, p \ 0.05). For both institutional models, respondents who had children in 2003 engaged in significantly less political
participation in 2003 than their counterparts who did not have children.

Limitations
There are several limitations of the study that should be considered when interpreting
results. The political participation measure of this study is limited by items located in the
B&B. Other studies might define political participation in differently, but the items in our
measure are commonly utilized. In addition, pre-college experiences have been found to
shape students civic and political dispositions (Hart et al. 2007; Kahne and Sporte 2008);
however, the B&B study did collect information about secondary experiences or other precollege experiences that may inform post-college political participation. The B&B study
also does not capture a robust set of academic and social experiences throughout
respondents college tenure to estimate socialization effects during their 4 years of college.
Policies and practices differ across institutional contexts and these varying approaches
differentially influence changes in college students outcomes over time (Weidman 1989;
Pascarella and Terenzini 2005).

Discussion
This study provides insight into academic experiences and values that inform the relationship between post-college political participation 1 year after respondents received their
baccalaureate and 10 years after receiving the baccalaureate. The first goal of the paper
was to locate a reliable and valid measure of political participation in the B&B. Based on
guidance from prior research (Verba and Nie 1972; Nie and Hillygus 2001; Hillygus 2005),
we found evidence of congeneric measurement across items that were hypothesized to
estimate political participation across three time points (1994, 1997, and 2003). This study
is the first to report CFA results of political participation measures located in the B&B.
Since we established that the three political participation factors were invariant, we were
confident in moving forward with the next phases of our investigation that examined the

123

Res High Educ (2013) 54:895929

919

relationships among the political participation measures between public institution and
private institutional respondents. Support for the hypotheses of the study was mixed.
Surprisingly, we found that graduates who attended public institutions engaged in more
political participation than graduates at private institutions (Hypothesis 1)although this
finding was present in 1994 and not in 2003. Extant research suggests that private institutions provide the academic milieu that better fosters the relationship between academic
experiences and post-college political participation, which lends support to other findings
in this study, but there are obviously other factors that need to be considered when
examining the full scope of political participation. For example, it may be the case that
average levels of political participation are different for freshman who attend public
institutions versus freshman who attend private institutions, and the academic and cocurricular experiences differentially mediate the difference between freshman year political participation and senior year political participation across private and public institutions. We know that the undergraduate socialization processthat includes pre-college
individual and family characteristics- shapes student outcomes in complex ways depending
on a number of variables, including institutional type (Pascarella and Terenzini 2005;
Weidman 1989). Our data do not allow us to examine the full range of pre-college,
academic, and co-curricular socializing functions that inform political participation of
college. Future longitudinal studies that investigate political participation of college
graduates should first understand the extent to which there are differences in the average
level of political participation of freshman across institutional contexts, and then they
should be informed by college impact models in order to adequately assess how academic
and co-curricular experiences inform the change in political participation during college
and beyond.
Hypothesis 2 was supported, as we found a significantly stronger relationship between
political participation in 1994 and political participation in 2003 for graduates from private
institutions when compared to the public institutional counterpart. Whereas findings from
hypothesis 1 shows public institution graduates engage in more political participation
1-year after college than private institution graduates, findings from hypothesis 2 shows
that the more private institutions graduates engage in political participation in 1994 the
more they will engage in political participation in 2003, and this is after controlling for
1997 political participation and all other predictors in the model; and such relationships are
not found for the public institutional model. The relationship between 1994 and 2003
political participation can be understood through the indirect effects for public institutions,
while we show that there is a significantly stronger direct effect of 1994 political participation on 2003 political participation. We found that some of the differences can be
explained by factors other than social class and institutional selectivity, as we found no
statistically significant differences in average parents education level, our proxy for SES,
or differences in the selectivity between the two institutional groups. We contribute evidence that shows institutional differences between college major, values held during
college, and service experiences improve our understanding of how college experiences
inform post-college political participation over a 10-year period for college graduates.
Major and 1994 Political Participation
We found two significant findings for major for 1994 political participation. We found the
number of social science credits positively correlates with 1994 political participation for
both institutional models, but the relationship is significantly stronger for private institutions than public institutions. We also found a statistically significant negative relationship

123

920

Res High Educ (2013) 54:895929

between science and engineering credits and 1994 political participation for public institutions, while there is no statistically significant relationship for private institutions. Our
social science findings support previous findings that found positive relationships between
social science majors and political participation (Astin et al. 2006; Nie and Hillygus 2001;
Hillygus 2005. The social sciences tend to be situated in social environments that promote
helping others and engaging in social activities (Holland 1997; Smart et al. 2000) that are
often reinforced by the types of experiences that social science faculty create for their
students, and that lend themselves to political expression, which include class discussion,
presentations, internships, volunteer work, and service-learning (Simmons and Lilly 2010).
Our science and engineering findings supports previous research that found that engineering majors have a negative relationship with political engagement, and they are less
likely to develop a personal commitment to social activism (Astin et al. 2006; Sax 2004).
These majors tend to be located in an investigative environment (Holland 1997; Smart
et al. 2000) which tends to rely more heavily than others on formal and structured
teachinglearning strategies that are strongly subject-matter centered (Smart et al. 2000,
p. 99), and their content doesnt lend themselves to the exposure of moral and civic
responsibilities as the social sciences (Colby et al. 2003). Therefore, in part our findings
support previous research and provide some new information to consider as we continue to
explore the relationship between academic environments and political participation.
The stronger relationship between social science majors and 1994 political participation
for the private institutional model and the negative relationship between science and
engineering and 1994 political participation for the public institutional model require
additional explanation; and the general education approaches between private and public
institutions may inform these differences. Many public institutions general education
approach focuses on developing basic reading, writing, and math skills to prepare student
for specializations (Stevens 2001), while many private institutions general education
approach is a blend of liberal arts, cultures, and ethics that focus on intellectual and moral
development, cultural appreciation, and the study of the human condition (Brint et al.
2009). First, the combination of smaller class sizes, high levels of faculty interactions, and
the general education requirements at private institutions may provide a stronger foundation for developing civic and social sensitivities for students at private institutions than
students at public institutions. When students get into their majorparticularly social
science and science/engineering majorsthe personality of the student, environment of the
major, and the interaction between the two, which includes faculty dispositions that
reinforce the disciplinary norms, may have an interactive effect between general education
and disciplinary norms. In the case of social science majors at private institutions, the
general education environment and the social academic environment may amplify students personalities in ways that reinforce their disposition for political participation in
ways that it does not for public institution graduates. In this case of science/engineering
majors, the combination of students personality and the general education environment
may mitigate the disciplinary norms that results in the lack of participation in the sciences
and engineering for private institution graduates, while the combination of institutional,
personal, and disciplinary norms play themselves out as expected for public institution
graduates. Future studies should attempt to further understand the impact of general
education approaches on political dispositions before students enter their majors and seek
how the interactions between the personality of the student and the environment of the
major may reinforce or mitigate the propensity to politically participate.
For both institutional models, we find a negative relationship between education credits
and 1994 political participation. We find no significant effects for education credits on

123

Res High Educ (2013) 54:895929

921

1997 or 2003 political participation; therefore, there are obviously a set of experiences for
education majors that inhibit post-college participation immediately after they graduate.
Smart et al. (2000) found little support of ascribing one of Hollands environmental traits
to different programs in the college of education. Instead, they found that many programs
in educations represent five of Hollands academic environments. Given our education
measure is an aggregate variable that combines early childhood education, elementary
education, secondary education, special education, physical education, and education:other. Hollands lens is difficult to apply. This is the first study to report such a finding,
which provides a baseline for further exploration into the relationship political participation and education. It is possible the transition to becoming a teacher and/or working in an
educational setting 1-year after graduation requires a significant amount of professional
development, investment in time, and learning the educational environment, that education
majors lack the time and opportunity to participate. Its also a possibility that given early
childhood, elementary, and secondary education is a significant part of federal and state
budget, the politics of resource allocation, and the contentious terrain about accountability,
school reform, inequities, and teacher quality may turn education students off from
political participation when they graduate. It is important that future studies attempt to
isolate the effects of the range of programs in colleges of education by collecting nuanced
data (i.e., program specific) to improve our understanding of the relationship the subdisciplines within education and post-college political participation.
Values on 1994 Political Participation
Influencing the Political Structure
Our hypothesis about the relationships between values and political participation is partially supported (Hypothesis 3.b). Valuing influencing the political structure in 1993 is
significantly and positively associated with 1994 political participation for both institutional models; and this relationship is significantly stronger for public institutions versus
private institutions. Nie and Hillygus (2001) considered this item to measure a public
regard of participation; therefore, it is reasonable to expect that the presence of this value
would be positively associated with our political participation measure, as it involves
influencing a public process and contacting public officials. They also found that the
number of social science credits taken were positively associated with influencing politics,
while the number of business credits taken negatively influenced the item (Nie and
Hillygus 2001). Students values are reinforced in their academic majors, and when they
major in areas congruent with their personalities, they tend to have greater differentiation
in their personalities from other students in their respective environments (Smart et al.
2000). We found positive bivariate relationships between the social sciences and this value
in both institutional models (rpublic = 0.09, p \ 0.001; rprivate = 0.117, p = \ 0.001); and
negative bivariate relationships between this value and business credits (rpublic = -0.08,
p \ 0.001; rprivate = -0.094, p = \ 0.001), and between science and engineering for
valuing influencing the political structure (rpublic = -0.101, p \ 0.001; rprivate = -0.094,
p = \0.001). A further examination about how students political interests are developed
within the sub-disciplines of social sciences, hard sciences, and engineering would add
considerably to our understanding of the relationships between political values and postcollege political participation. While many institutions are not formally explicit about the
values they would like their students to espouse (Hartley and Morphew 2008), this study
shows that there are certain values public institution graduates hold in college that are more

123

922

Res High Educ (2013) 54:895929

strongly related to their political activities post-college than their private institution
counterparts. Valuing influencing the political structure has been used as an outcome and
as part of civic engagement constructs in prior studies (Nie and Hillygus 2001; Astin 1993;
Pascarella et al. 1988; Antonio 2001); but its relationship with post-college political participation has yet to be understood across institutional contexts. Findings from this study
provide some baseline information to further investigate.
Community Leader
For respondents at public institutions, those who valued being a community leader during
their senior year in college engaged in significantly less political participation in 1994,
while there was a non-significant association for the private institution model in 1994. We
initially hypothesized this finding was due to our samples Generation X identitythose
born between 1965 and 1976that is partly defined by antigovernment and antipolitical
rhetoric, which resulted in less political active Gen Xers in general (Zukin et al. 2006).
Whereas previous cohorts were oriented toward the use of government to solve problems,
Gen Xers, and even the generation that succeeded them, are more likely to solve public
problems through various community-based organizations and local activities than through
political means (Zukin et al. 2006). Therefore, we inferred respondents in the public
institution model who value being a community leader during their senior year would be
more likely to be engaged in community service instead of political participation. However, a closer look at the zero-order correlations for the public institution model show a
negative relationship between valuing being a community leader and engaging in community service during their senior year (r = -0.085); but it does have a negative association with valuing influencing the political structure (r = -0.048). We do not find a
positive association with valuing being a community leader and community service in
1994, 1997, and 2003.4 Our data do not provide much more insight into this finding. It may
be the case that valuing being a community leader is a lofty notion that drives socially
desirable responses and does not equate to predictable behavior. It is important to note that
there is a statistically non-significant association between this value and political participation in 1997 and 2003. Future research would benefit from an investigation of students
perceptions of what community leaders do for those students who value this type of
leadership.
Political Participation in 2003
One of our main investigations was to understand if the relationships between political
participation 1 year after receiving the baccalaureate and political participation 10 years
after receiving the baccalaureate will differ, and we found evidence of several statistically
significant associationssome of which endured from 1994 and some of which are new.
We controlled for some life course predictors measured in 1997 (i.e., marital status, and
having children) and some measured in 2003 (i.e., marital status, having children, and
home ownership status) to better understand the unique contribution of college major,
values held during college, and community service participation on political participation
in 2003. We find unique associations with college major and values.

This is a separate analysis where community service hours was the dependent variable and valuing
community service is one of forty predictors.

123

Res High Educ (2013) 54:895929

923

Major
Examining the full model, controlling for life course experiences, credits earned in business for graduates of public institutions had a negative impact on 2003 political participation. This finding was the only academic major variable that impacted the 2003 model.
Our findings support past studies that have found negative relationships between business
majors and political participation (Astin et al. 2006; Nie and Hillygus 2001; Hillygus
2005), but the significance of this relationship only manifests itself for public institutions
and not private institutions; and this is the first study to report such findings. It may be the
case that enterprising environments, in general, do not promote political participation. The
relationship between business credits and 2003 political participation is best explained
through the combination of all other variables in our model, as the total indirect effects
explain the most of the total effects in both institutional models. Some of this variance
could be explained through life course factors such as family roles and labor force outcomes. A secondary analysis found no differences in 2003 income between business
majors who graduated from public institutions versus private institutions. Perhaps, and
similar to our previous explanation about the relationships between general education and
major in the 1994 political participation model, the relationship between political participation and business credits play themselves out as expected in public institutions, but the
combination of pre-major experiences mitigate any negative effects that are expected in
enterprising environments. As previously mentioned, some service-learning experiences
tend to promote political interest (Eyler and Giles 1999; Astin et al. 2002; Pascarella and
Terenzini 2005). As more business programs integrate service-learning components
throughout their courses (Campus Compact 2013; Zlotkowski et al. 2000) and more
guidance is provided about integrating political content into the curriculum (Colby et al.
2007), more research is needed to understand the extent to which these experiences are
integrated across institutional contexts, and the ways in which they inform post-college
political participation for all major, especially business majors. We provide results based
on longitudinal data for further study of this underdeveloped area of research.
Influencing the Political Structure
The positive association between valuing influencing the political structure and political
participation endured during the 10-year period immediately after respondents received
their baccalaureate. It is important to note that these 2003 findings exist in the presence of
life course considerations, which highlight the unique impact of college experiences. This
is the only finding that shows the similar relationships between 1993 and 2003 variables for
both institutional models. This is the first study to report enduring relationships between
values and political participation for college graduates. College students values are
changed and maintained through the various socialization processes associated with the
college experience (Weidman 1989). Current research has found that a range of academic
and co-curricular experiences have impacted students values in general, including but not
limited to completion of an ethnic studies class, womans studies class, diversity classes,
participating in study abroad, volunteering, study abroad, student government, and a range
of peer learning activities and faculty/staff interactions (Antonio 2001; Astin 1993; Colby
et al. 2007, 2003; Pascarella and Terenzini 2005; Eyler and Giles 1999; Boyte and Kari
2000; Lott 2013). However, there is less information about how academic environments
contribute to the development of individuals attitudes and values (Pike 2006). More finergrained categories within academic fields and disciplines would allow a more nuanced

123

924

Res High Educ (2013) 54:895929

analysis of the values that are embedded within individual majors as previous research did
with six majors within engineering (Lattuca et al. 2010). A deeper investigation into the
extent to which individual academic environments shape values of students would not only
extend Hollands theory, it may also improve our understanding about the relationship
between college experiences and post-college political participation.
Valuing being Wealthy
Surprisingly, we found a negative association between valuing being wealthy and 2003
political participation for the private institutional model when we did not find a significant
effect in the 1994 model. This is the first study to report such a finding. Nie and Hillygus
(2001) view valuing being wealthy as an individualistic characteristic ascribed to an individual who is more interested in private matters than public matters. There are obviously a set
of post-baccalaureate experiences for graduates from private institutions who value being
wealthy that results in greater differentiation in 2003 political participation than their public
institutional counterparts who do not hold such a value. We conducted t tests for valuing
being wealthy and 2003 income for both the private and public institution graduates to see if
we could better understand our findings through a relationship between institution type,
income, and valuing being wealthy. We found statistically significant mean differences in
both models,5 suggesting that those who value being wealthy have significant higher 2003
incomes than those who do not hold such value. However, the effect sizes were similar for
both institutional models, meaning that the difference in 2003 income between those who
value being wealthy and those who do not were similar for private and public institution
graduates. Similar to our suggestions for the political interest value, a deeper investigation
into the ways in which college experiencesparticularly within majorsdevelop values
will provide much needed insight into our findings. Follow-up studies should attempt to
control for as many life course experiences as possible so we can get a better understanding of
how college experiences may inform this longitudinal relationship.

Future Research
There are several ways in which future research could expand on our study. An important
part of understanding any unique effect of the college experience is understanding how
pre-college experiences may inform the relationship between college experiences and postcollege political participation. Some of our findings, particularly the ones associated with
values, may be a function of students pre-college levels of political interest. Some of these
pre-college interests and values are informed by students involvement in politically salient
youth organizations (Settle et al. 2011; McFarland and Thomas 2006), high school community service (Hart et al. 2007), classroom-based experiences that target civic goals
(Kahne and Sporte 2008), and various aspects related to parental socialization and their
non-college reference groups (Weidman 1989). In addition, these pre-college factors may
also affect where a student chooses to attend college. A complex set of economic and
sociological variables impacts individuals and their families decision about where to
attend college. The college choice literature is robust and studies have investigated how a
combination of policies (i.e., federal, state, and institutional), academic preparation, and/or
SES influences individuals decisions to attend a range of higher education institutions,
5

Not shown but available upon request.

123

Res High Educ (2013) 54:895929

925

including public and private (see, for example, Kim 2012; Paulsen and St John 2002; Hu
and Hossler 2000; Klugman 2012; Perna and Titus 2004; Wiese and Townsend 1991;
Perna et al. 2005). Some of our findings may be explained by differences in the disposition
to participate in politics between individuals who attend public versus private schools.
Therefore, future studies should be informed by the college choice literature.
Future studies should also consider replicating this study with more recent cohorts of
students and find ways to have a more expansive political participation measure. The most
recent generation of the American populationsthe DotNets who are born after 1976are
more likely to engage in political activities such as boycotting against corporations and
buycotting to support a company, and are less likely to engage in voting, contacting public
officials, etc., because they believe that the private sector of business has greater influence
on our lives than does government (Zukin et al. 2006). Finally, future survey design
approaches should attempt to have identical items across political participation measures
so that latent growth models and other robust models can be estimated. The B&B provides
more than adequate data but the inconsistent nature of political nature across the waves
limited our modeling approach.

Conclusion
Political participation of college graduates will be necessary to continue promoting
democracy and the democratic process. The research about the relationships between
college experiences and post-college political participation is developing in complex ways.
This research applied SEM to investigate some college experiences and post-college
political participation for public institution and private institution graduates. Hollands
theory of vocational personalities provided a useful analytical lens to understand how some
of our findings are related to students personalities and their academic environments
across public and private institution. This study is the first to document the relationship
between college major, values, and post-college political participation 10 years after students received the baccalaureate.
Our research shows differences in post-college political participation across institutional
contexts. Practically, our findings may be useful for those seeking to strengthen or even
integrate political content in their curricular and co-curricular environments. Colby et al.
(2007) found that in order to accomplish deep and enduring political learning, students need
to simultaneously engage in intellectually, emotionally, socially, and personal learning. This
is many times done through connecting students with ideas and people who are able to
deepen their political engagement through mentors, speakers, and staff at service-learning/
placement sites. They argue, and we agree, that the challenges and problems that confront
teaching for political development point to the value of a more cumulative, institutionally
integrated approach (Colby et al. 2007, p. 293). Through assessing the curriculum across all
majors with a political learning lens, and being intentional about creating opportunities and
experiences for students and faculty, institutions may get to a point where they are able to
provide context-specific guidance and encouragement across campus units about promoting
political development across all majors and environments.
Acknowledgments This research was supported by a grant from the American Educational Research
Association which receives funds for its AERA Grants Program from the National Science Foundation
under #DRL-0634035. Opinions reflect those of the author and do not necessarily reflect those of the
granting agencies.

123

926

Res High Educ (2013) 54:895929

References
Antonio, A. L. (2001). Diversity and the influence of friendship groups in college. Review of Higher
Education, 25(1), 6389.
Association of American Colleges and Universities. (2012). Core commitments: Educating students for personal and social responsibility. Retrieved June 1, 2012, from http://www.aacu.org/core_commitments/
index.cfm.
Astin, A. W. (1993). What matters in college: Four critical years revisited. San Francisco: Jossey-Bass.
Astin, A. W. (1996). Involvement in learning revisited: Lessons we have learned. Journal of College Student
Development, 36, 123134.
Astin, A., Vogelgesang, L. J., Ikeda, E. K., & Yee, J. A. (2002). How service learning affects students:
Executive summary. Los Angeles: Service Learning Clearing Project, Higher Education Research
Institute.
Astin, A. W., Vogelgesang, L. J., Misa, K., Anderson, J., Denson, N., Jayakumar, U., et al. (2006).
Understanding the effects of service-learning: A study of students and faculty. Los Angeles: Higher
Education Research Institute, UCLA.
Baumgartner, H., & Steenkamp, J.-B. E. M. (1998). Multi-group latent variable models for varying numbers
of items and factors with cross-national and longitudinal applications. Marketing Letters, 9(1), 2135.
Beaumont, E., Colby, A., Ehrlich, T., & Torney-Purta, J. (2006). Promoting political competence and
engagement in college students: An empirical study. Journal of Political Science Education, 2(3),
249270.
Biglan, A. (1973a). The characteristics of subject matter in different academic areas. Journal of Applied
Psychology, 57, 195203.
Biglan, A. (1973b). Relationships between subject matter characteristics and the structure and output of
university departments. Journal of Applied Psychology, 57, 204213.
Boyte, H., & Kari, N. (2000). Renewing the democratic spirit in American Colleges and Universities: Higher
education as public work. In T. Ehrlic (Ed.), Civic responsibility and higher education (pp. 3759).
Phoenix: Oryx Press.
Brint, S., Proctor, K., Murphy, S. P., Turk-Bicakci, L., & Hanneman, R. A. (2009). General education
models: Continuity and change in the U.S. undergraduate curriculum, 19752000. Journal of Higher
Education, 80(6), 605642.
Brown, T. A. (2006). Confirmatory factor analysis for applied research. New York: Guilford Press.
Byrne, B. M. (2006). Structural equation modeling with EQS: Basic concepts, applications, and programming. Mahwah, NJ: Lawrence Erlbaum Associates.
Campus Compact. (2013). Syllabi: Business/management. Campus Compact. Retreived January 18, 2003,
from http://www.compact.org/category/syllabi/.
Chen, F. F. (2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural
Equation Modeling, 14(3), 464504.
Cheung, G., & Rensvold, R. (2002). Evaluating goodness-of-fit indexes for testing measurement invariance.
Structural Equation Modeling: A Multidisciplinary Journal, 9(2), 233255.
Colby, A., Beaumont, E., Ehrlich, T., & Corngold, J. (2007). Educating for democracy: Preparing
undergraduates for responsible political engagement. San Francisco, CA.: Jossey-Bass.
Colby, A., Ehrlich, T., Beaumont, E., & Stephens, J. (2003). Educating citizens: Preparing Americas
undergraduates for lives of moral and civic responsibility. San Francisco, CA: Jossey-Bass.
Dey, E. (1996). Undergraduate political attitudes: An examination of peer, faculty, and social influences.
Research in Higher Education, 37, 535554.
Dey, E. (1997). Undergraduate political attitudes: Peer influence in changing social contexts. Journal of
Higher Education, 68, 398413.
Dimitrov, D. M. (2006). Comparing groups on latent variables: A structural equation modeling approach.
Work, 26(4), 429436.
Eyler, J., & Giles, D. (1999). Wheres the learning in service-learning?. San Francisco: Jossey-Bass.
Flanagan, C., & Levine, P. (2010). Civic engagement and the transition to adulthood. The Future of
Children, 20(1), 159179.
Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and
measurement error. Journal of Marketing Research, 48, 3950.
Fox, J. (2002). An R and S-Plus companion to applied regression. Thousand Oaks, CA: Sage Publications.
Graham, J. M. (2006). Congeneric and (essentially) tau-equivalent estimates of score reliability: What they
are and how to use them. Educational and Psychological Measurement Psychological, 66(6), 930944.
Hanson, J. M., Weeden, D. D., Pascarella, E. T., & Blaich, C. (2012). Do liberal arts colleges make students
more liberal? Some initial evidence. Higher Education, 64(3), 355369.

123

Res High Educ (2013) 54:895929

927

Hart, D., Donnelly, T. M., Youniss, J., & Atkins, R. (2007). High school community service as a predictor of
adult voting and volunteering. American Educational Research Journal, 44(1), 197219.
Hartley, M., & Morphew, C. C. (2008). Whats being sold and to what end? A content analysis of college
viewbooks. Journal of Higher Education, 79(6), 671691.
Heck, R. H., & Thomas, S. L. (2009). An introduction to multilevel modeling techniques (2nd ed.). New
York, NY: Routledge.
Hillygus, S. D. (2005). The missing link: Exploring the relationship between higher education and political
engagement. Political Behavior, 27(1), 2547.
Holland, J. L. (1966). The psychology of vocational choice. Waltham, MA: Blaisdell.
Holland, J. L. (1973). Making vocational choices: A theory of careers. Englewood Cliffs, NJ: Prentice-Hall.
Holland, J. L. (1985). Making vocational choices: A theory of vocational personalities and work environments (2nd ed.). Englewood Cliffs, NJ: Prentice-Hall.
Holland, J. L. (1997). Making vocational choices: A theory of vocational personalities and work environments (3rd ed.). Odessa, FL: Psychological Assessment Resources.
Hooper, D., Coughlan, J., & Mullen, M. R. (2008). Structural equation modelling: Guidelines for determining model fit. Electronic Journal of Business Research Methods, 6(1), 5360.
Hu, S., & Hossler, D. (2000). Willingness to pay and preference for private institutions. Research in Higher
Education, 41(6), 685701.
Joreskog, K. (1973). A general method for estimating a linear structural equation system. In A. S. Goldberger & O. D. Duncan (Eds.), Structural equation models in the social sciences (pp. 85112). New
York: Academic Press.
Joreskog, K. G., & Sorbom, D. (1996). LISREL 8 users reference guide. Chicago, IL: Scientific Software
International.
Joreskog, K. G., & Sorbom, D. (1982). Recent developments in structural equation modeling. Journal of
Marketing Research, 19(4), 404416. doi:10.2307/3151714.
Kahne, J. E., & Sporte, S. E. (2008). Developing citizens: The impact of civic learning opportunities on
students commitment to civic participation. American Educational Research Journal, 45(3), 738766.
Kam, C. D., & Palmer, C. L. (2008). Reconsidering the effects of education on political participation. The
Journal of Politics, 70(3), 612631.
Kaplan, D. (2000). Structural equation modeling: Foundations and extensions. Thousand Oaks, CA: Sage
Publications.
Kim, J. (2012). Exploring the relationship between state financial aid policy and postsecondary enrollment
choices: A focus on income and race differences. Research in Higher Education, 53, 123151.
Kline, R. B. (2005). Principles and practice of structural equation modeling. New York: Guilford Press.
Klugman, J. (2012). How resources inequalities among high school reproduce class advantages in college
destination. Research in Higher Education, 53, 803830.
Kuh, G. (1993). In their own words: What students learn outside the classroom. American Educational
Research Journal, 30, 277304.
Lattuca, L. R., Terenzini, P. T., Harper, B. J., & Yin, A. C. (2010). Academic environments in detail:
Hollands theory at the subdiscipline level. Research in Higher Education, 51, 2139.
Loehlin, J. C. (2004). Latent variable models: An introduction to factor, path, and structural equation
analysis. Mahwah, NJ: L. Erlbaum Associates.
Lott, J. (2013). Predictors of civic values: Understanding student-level and institutional-level effects.
Journal of College Student Development, 54(1), 116.
Lubke, G., & Muthen, B. (2004). Applying multigroup confirmatory factor models for continuous outcomes
to Likert scale data complicates meaningful group comparisons. Structural Equation Modeling: A
Multidisciplinary Journal, 11(4), 514534.
McFarland, D. A., & Thomas, R. J. (2006). Bowling young: How youth voluntary associations influence
adult political participation. American Sociological Review, 71(3), 401425.
Meade, A. W., Johnson, E. C., & Braddy, P. W. (2006). The utility of alternative fit indices in tests of
measurement invariance. Academy of Management Annual Meeting Proceedings, 8(1), B1B6.
Milfont, T., & Fischer, R. (2010). Testing measurement invariance across groups: Applications in crosscultural research. International Journal of psychological research, 3(1), 111121.
Morrill, R. L. (1980). Teaching values in college. San Francisco, CA: Jossey-Bass.
Muthen, B. (2002). Beyond SEM: General latent variable modeling. Behaviormetrika, 29(1), 81117.
Muthen, L. K., & Muthen, B. O. (19982010). Mplus users guide. (6th ed.). Los Angeles, CA: Muthen &
Muthen.
National Center for Education Statistics (NCES). (1995). Methodology Report for the National Postsecondary Student Aid Study, 19921993 (NPSAS:93). Washington, DC: Department of Education.

123

928

Res High Educ (2013) 54:895929

Nie, N., & Hillygus, D. (2001). Education and democratic citizenship. In D. Ravitch & J. Viteritti (Eds.),
Making good citizens: Education and civil society (pp. 3057). New Haven: Yale University Press.
Oesterle, S., Johnson, M. K., & Mortimer, J. T. (2004). Volunteerism during the transition to adulthood: A
Life course perspective. Social Forces, 82(3), 11231149.
Pascarella, E. T., Cruce, T., Umbach, P. D., Wolniak, G. C., Kuh, G. D., Carini, R. M., et al. (2006).
Institutional selectivity and good practices in undergraduate education: How strong is the link? The
Journal of Higher Education, 77(2), 251285.
Pascarella, E. T., & Terenzini, P. T. (2005). How college affects students: A third decade of research (Vol.
2). San Francisco: Jossey-Bass.
Paulsen, M. B., & St John, E. P. (2002). Social class and college costs: Examining the financial nexus
between college choice and persistence. The Journal of Higher Education, 73(2), 189236.
Perna, L. (2004). Understanding the decision to enroll in graduate school: Sex and racial/ethnic group
differences. The Journal of Higher Education, 75(5), 487527.
Perna, L. W., Steel, P., Woda, S., & Hibbert, T. (2005). State public policies and the racial/ethnic stratification of college access and choice in the state of Maryland. The Review of Higher Education, 28(2),
245272.
Perna, L. W., & Titus, M. A. (2004). Understanding differences in the choice of college attended: The role
of state public policies. The Review of Higher Education, 27(4), 501525.
Pike, G. R. (2006). Students personality types, intended majors, and college expectations: Further evidence
concerning psychological and sociological interpretations of Hollands theory. Research in Higher
Education, 47(7), 801822.
Raykov, T. (1997). Multivariate structural modeling of plasticity in fluid intelligence of aged adults.
Multivariate Behavioral Research, 30(2), 255287.
Raykov, T., & Widaman, K. E. (1995). Issues in applied structural equation modeling research. Structural
Equation Modeling, 2, 289318.
Satorra, A. (1999). Scaled and adjusted restricted tests in multi-sample analysis of moment structures.
Barcelona: Department of Economics and Business, Universitat Pompeu Fabra. http://ideas.repec.org/
p/upf/upfgen/395.html. Accessed 16 Nov 2012.
Sax, L. (2004). Citizenship development and the American college student. New Directions for Institutional
Research, 2004(122), 6579.
Schmitt, C. M. (2009). Documentation for the restricted-use NCES-Barrons admissions competitive index
data files: 1972, 1982, 1992, 2004, and 2008 (NCES 2010-330). Washington, DC: National Center for
Education Statistics, Institute of Education Sciences, U.S. Department of Education.
Settle, J. E., Bond, R., & Levitt, J. (2011). The social origins of adult political behavior. American Politics
Research, 39(2), 239263.
Shaver, J. P., & Strong, W. (1982). Facing value decisions: Rationale-building for teachers. New York:
Teachers College Press.
Simmons, J. R., & Lilly, B. (2010). The university and student political engagement. PS: Political Science &
Politics, 43(2), 347350.
Sinclair-Chapman, V., Walker, R. W., & Gillion, D. Q. (2009). Unpacking civic participation: Analyzing
trends in black [and white] participation over time. Electoral Studies, 28(4), 550561.
Skrondal, A., & Rabe-Hesketh, S. (2004). Generalized latent variable modeling: Multilevel, longitudinal,
and structural equation models. Boca Raton: Chapman & Hall/CRC.
Smart, J. C., Feldman, K. A., & Ethington, C. A. (2000). Academic disciplines: Hollands theory and the
study of college students and faculty. Nashville, TN: Vanderbilt University Press.
Snell, P. (2010). Emerging adult civic and political disengagement: A longitudinal analysis of lack of
involvement with politics. Journal of Adolescent Research, 25(2), 258287.
Stevens, A. H. (2001). The philosophy of general education and its contradictions: The influence of
Hutchins. The Journal of General Education, 50(3), 165191.
Thomas, S. L., & Heck, R. H. (2001). Analysis of large-scale secondary data in higher education research:
Potential perils associated with complex sampling designs. Research in Higher Education, 42(5),
517540.
Thomas, S. L., Heck, R. H., & Bauer, K. W. (2005). Weighting and adjusting for design effects in secondary
data analyses. In P. D. Umbach (Ed.), Surveying research: Emerging issues (pp. 5172). San Francisco,
CA: Jossey-Basss.
Toutkoushian, R. K., & Smart, J. C. (2001). Do institutional characteristics affect student gains from
college? Review of Higher Education, 25(1), 3961.
Verba, S., & Nie, N. H. (1972). Participation in America: Political democracy and social equality. Chicago,
IL: University of Chicago Press.

123

Res High Educ (2013) 54:895929

929

Verba, S., Schlozman, K. L., & Brady, H. E. (1995). Voice and equality: Civic voluntarism in American
politics. Cambridge, MA: Harvard University Press.
Weidman, J. (1989). Undergraduate socialization. In J. C. Smart (Ed.), Higher education handbook of theory
and research (Vol. 5, pp. 289322). New York: Agathon.
Wiese, M. D., & Townsend, B. K. (1991). College choice in the church-affiliated sector of higher education:
The influence of the congregational cultural orientation. Research in Higher Education, 332(2),
107122.
Wine, J., Cominole, M., Carwile, S., Franklin, J., Carley-Baxter, L., and Wheeless, S. (2004). 1993/03
Baccalaureate and Beyond Longitudinal Study (B&B:93/03) Field Test Methodology Report (NCES
200402). U.S. Department of Education. Washington, DC: National Center for Education Statistics.
Yu, C.-Y., & Muthen, B. (2002). Evaluation of model fit indices for latent variable models with categorical
and continuous outcomes. Paper presented at the annual meeting of the American Educational
Research Association, New Orleans, LA.
Zlotkowski, E., Godfrey, P. C., & Grasso, E. T. (2000). Working for the common good: Concepts and
models for service learning in management. American Association for Higher Education Series on
Service-Learning in Disciplines. Retrieved December 29, 2011, from http://www.eric.ed.gov/PDFS/
ED449733.pdf.
Zukin, C., Keeter, S., Andolina, M., Jenkins, K., & Delli Carpini, M. X. (2006). A new engagement: political
participation, civic life, and the changing American citizen. New York, NY: Oxford University Press.

123

Copyright of Research in Higher Education is the property of Springer Science & Business Media
B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

You might also like