Changes in Politics and Religiosity Among Students at a Protestant University
[Correction added on 29 November 2023, after first online publication: Sarah A. Schnitker's ORCID was added.]
Abstract
College is a setting and time of profound change in the lives of emerging adults. This change can include shifts in identity related to politics and religion. Given widespread attention to the alignment of religious people with conservative politics and less religious people with liberal politics (i.e., the “God Gap”), we ask: do college students who become politically liberal lose their religion in the process? Using longitudinal panel data, this study examines changes in political identity and religiosity among students at a Protestant university. Findings reveal changes in students’ politics align with changes in public and private religious behaviors, certainty in belief, agreement with core tenets of the Christian faith, faith maturity, and closeness to God. Whereas students who become more politically conservative increase their religiosity, the inverse is true for those whose politics become more liberal in college.
Introduction
For the majority of the United States' history, political parties drew comparable support from Americans of differing religious traditions. Since the 1970s and 1980s, however, this relationship has changed. Democratic and Republican parties adopted divergent positions on moral issues, and religious groups organized politically to influence public policy in ways that aligned these groups with a singular political party (Hartman 2015; Layman 2001; Putnam and Campbell 2012). This shift resulted in a significant social cleavage, often referred to as the “God Gap,” with highly religious Americans clustering in the Republican Party and less religious Americans in the Democratic Party. This divergence in political commitments by religiosity has been discussed at length in the academic community (Abramowitz 2010; Layman 2001) and is an association widely known to the general public.
The narrative of religious voters coalescing over time in support of the Republican Party remains a persuasive explanation for the creation of the God Gap (Hout and Fischer 2002; Putnam and Campbell 2012), but certain scholars are now advancing another narrative that contends religiosity is no longer as determinative of political affiliation as political affiliation is of religiosity (Campbell et al. 2018; Egan 2020; Margolis 2018b). Today, this theory suggests, a growing number of American political partisans have internalized the God Gap and correspondingly shifted their religious beliefs and behaviors to align with their political identity. At the same time, political partisans have become more affectively polarized, leading to more salient “battle” lines drawn between the parties (Iyengar, Sood, and Lelkes 2012).
Our study analyzes concurrent changes in politics and religiosity among students at a Protestant university in the southwest United States. Numerous studies based in the United States point to the self-reinforcing generational or cohort-based nature of religious decline in recent decades (Brauer 2018; Chaves 2017; Voas and Chaves 2016), encouraging the study of recent generations’ religious lives. Prior research shows a marked decline in religiosity, most notably religious participation, for emerging adults during their college years (Astin, Astin, and Lindholm 2010; Uecker, Regnerus, and Vaaler 2007). Applying theoretical work on affective political polarization and social sorting, we ask: Is this trend a function of, or accelerated for, those whose politics shift toward the political left while in college? In other words, do college students who become politically liberal lose their religion in the process?
Political Polarization and Expressive Political Identity
People take on numerous identities—religious, political, gender, racial, regional, and so forth—that define them and help locate them with similar others in social groups (Stryker and Burke 2000). Once people internalize membership within a social group, they experience high levels of in-group identification, intergroup differentiation, and in-group bias (Roccas and Brewer 2002). Political identity operates as a social identity with the power to shape individuals’ policy preferences toward political and moral issues, voting behavior, and evaluations of members who do and do not share their political identity (Green, Palmquist, and Schickler 2004). In the last four decades, the United States has experienced increasing affective political polarization, in which people employ their political identity as a primary script for deciding everyday judgments toward others (Iyengar and Westwood 2015; Iyengar, Sood, and Lelkes 2012).
In identifying with their political party, partisans take on negative evaluations of the opposing party, thereby strengthening in-group ties and distancing themselves from outsiders. Iyengar and colleagues (2012) argued that polarization has resulted from individuals emphasizing the “social distance” between adherents of the two parties. Their data demonstrated that, while showing consistently favorable ratings of the in-group, ratings of the out-party among Republicans and Democrats plummeted from 1978 to 2008. Partisan affect, not ideology, seemed to drive political polarization and the authors concluded there is sufficient animosity for political affiliation to be relevant in everyday interpersonal relations. Affective polarization created social distance between parties by causing Americans to perceive those who belong to an opposing political party in stereotypical ways (Ahler and Sood 2018). Moreover, affective polarization created hostility toward opposing party members outside the realm of politics to a degree sometimes exceeding race-based discrimination (Iyengar and Westwood 2015).
Political identity is consequential in other ways in the United States. It can operate in an expressive way that determines positions on political and moral issues with the effect varying according to the salience of political identity and awareness of issue-based party differences (Carsey and Layman 2006). For example, although most Americans have moderate stances on the situational morality of abortion, people who change party identification from Republican to Democrat take on the position toward abortion characteristic of the Democratic Party (Killian and Wilcox 2008). Other studies document how priming political identity motivates change in participants’ voting behavior. After sending a treatment group a mailer reminding them of the need to register with a political party and vote, Gerber and colleagues (2010) found that the treatment group reported a higher level of party identification and alignment with their party on partisan political issues when surveyed 4 months later.
Expressive partisanship reifies and is reified by “social sorting,” as American political identities have come to be more closely associated with other social identities, be they religious, racial, or another group identity (Mason 2018). This phenomenon creates an increasingly homogeneous set of parties whose cultures are seen and internalized by members as direct counterparts. Layman (2001) traced the growing linkage between religious and political identity as an instrumental relationship in which religious people are guided by their religious identity and religious values to ally with a particular political party's platform, but more recent scholars have found affective polarization and social sorting made political identity more deterministic of individuals’ religious identity (Campbell et al. 2018; Margolis 2018b; Patrikios 2008).
Shifting Politics, Shifting Religion
Religious identity is influenced by political attitudes. U.S. panel data demonstrate shifts in people's religious affiliation and behavior in response to political convictions (Campbell et al. 2018; Hout and Fischer, 2002, 2014; Margolis 2018; Patrikios 2008; Putnam and Campbell 2012b). For example, Margolis (2018a) found that respondents updated their previously reported religious identity to better align with their political identity following priming of their political identity. Furthermore, the salience of political identities exceeded that of religious identities for many Americans (Margolis 2018b). Other scholars have found that conservative Republicans were more likely than liberal Democrats to be born-again Christians and liberal Democrats were more likely than conservative Republicans to become less actively religious over time or disaffiliate entirely (Egan 2020). Hout and Fischer (2002) and Putnam and Campbell (2012) analyzed the doubling of religious “nones” in the 1990s. They hypothesized that the rise was best explained as disaffiliation from religion in rejection of traditional religion's association with conservative politics and the Republican Party. Later analysis clarified this shift to be a phenomenon of religious disaffiliation and not religious disbelief, with people who became more politically liberal typically reporting declines in religious participation or disaffiliation altogether but not substantial change in religious beliefs (Hout and Fischer 2014; Rhodes 2010).
In contrast, the perception that Republicans are religious makes political conservatives less inclined to disaffiliate from religious groups than either moderates or liberals (Loveland 2003). Marginal congregants are more likely to be Democrat than Republican (Djupe, Neiheisel, and Sokhey 2018). Democrats who are religious tend to experience dissonance between their political identity and religious identity, feel out of place, and may leave their religion (Campbell et al. 2018; Patrikios 2008). Furthermore, the population of those who are both religious and politically liberal has steadily decreased in size and level of political involvement over the last two decades (Baker and Martí 2020). Altogether, these studies provide evidence for a growing bifurcation of religion and politics, with Republicans increasingly composed of religious conservatives and Democrats of religious liberals or irreligious people. Recently, Braunstein (2022) analyzed whether the broad rise in religious disaffiliation stems from political backlash against the religious right, finding support for narrow backlash against the religious right and counter backlash in the form of purification and radicalization within the religious right. Notably, cohort succession explained 60 percent of the variance in religious beliefs in Hout and Fischer's (2014) study. The contemporary association between religion and politics, according to some researchers (Braunstein 2022; Hout and Fischer 2014; Margolis 2018b), works hand in hand with consistent trends of cohort- and period-based religious decline among recent generations (Brauer 2018; Chaves 2017; Voas and Chaves 2016).
Leaving the Faith?
The separation of religious and nonreligious people by political identity is augmented by religious and political identity formation in the life course. Adolescence and early adulthood are times of religious transition for many. As teenagers move into their early twenties, decreases typically occur in religious service attendance (Astin, Astin, and Lindholm 2010; Smith and Snell 2009; Uecker, Regnerus, and Vaaler 2007), prayer (Smith and Snell 2009), and the importance of religious identity (Smith and Snell 2009).
A popular belief has been that college students leave organized religion as a result of their involvement in higher education (Wuthnow and Glock 1973). The assumption is that postsecondary education encourages students to question their religious beliefs, making colleges and universities “the breeding grounds for apostasy” (Caplovitz and Sherrow 1977:109). Yet, contemporary research contests the common wisdom that religiosity declines sharply during emerging adulthood. Findings from the National Study of Youth and Religion demonstrated that one half or more of all young adults retain their religious tradition from their teenage years to adulthood (Smith and Snell 2009). Similarly, Astin and coauthors (2010) found that though college students attend religious services less often, their levels of religious commitment (measured by a scale of religious attitude and belief items) change very little during college and college students grow in measures of spirituality and spiritual seeking.
The trend of religious decline seems to apply primarily to religious attendance. The most significant impediment to college students’ religious lives is time. Students’ decrease in religious service attendance is often simply explained by conflict between their prior religious lives and the pressures attached to accumulation of academic knowledge and geographic relocation (Astin, Astin, and Lindholm 2010; Hill 2009). In fact, much contemporary research points to a weak relationship between higher education and religiosity. A drop in religious service attendance among emerging adults is no more likely for college students, and sometimes less likely, when comparing college attendees to those who do not attend college (Mayrl and Uecker 2011; Uecker, Regnerus, and Vaaler 2007), buttressing Astin et al.’s argument that attending college can positively impact students’ religious lives by cultivating their spirituality. The weakening, and sometimes disappearing, effect of higher education on religiosity has been explained by intergenerational growth in higher education (Schwadel 2014) and more educationally diverse religious organizations (Schwadel 2009). Whatever the cause, higher education is no longer associated with a categorical loss of religion.
Religious Engagement by College Type
Religious participation and commitment vary greatly by type of college or university. Astin, Astin, and Lindholm (2010:97–98) found that the percentage of incoming students who frequently attended religious services in the year prior to enrollment was far lower for public universities (39 percent) and private nonreligious universities (44 percent) than at religiously affiliated universities. Of those attending religiously affiliated universities, it was 90 percent for evangelical universities, 49 percent for Catholic universities, and 61 percent for other Christian universities. Religious universities provide a more comfortable, nurturing community for those who are religious, leading to a stark separation in the levels of religious engagement and belief compared to those at secular universities (Hill 2011).
In addition, the rate at which students’ religious participation decreases while at college varies according to college type. Hill (2009) found those at evangelical Protestant colleges tended to decrease religious service attendance less than do students at Catholic and mainline Protestant schools, with Catholic and mainline universities students’ religiosity often declining more than public schools. Moreover, measures of religiosity have been consistently highest for students at evangelical Protestant universities (Astin, Astin, and Lindholm 2010; Hill, 2009, 2011). Hill (2009) attributed this finding in large part to selection as well as to the ability of evangelical Protestant universities to create and maintain a moral community rooted in religious values. The social networks of students at these universities are more religiously homogeneous, contributing to the legitimation of religious belief and practice. A religious university devoted to sustaining the faith of its students thus serves as a rigorous test of political influence on religious commitment.
The Present Study
The extant literature clearly demonstrates that political polarization has occurred in the United States and that increased salience of politics increasingly seems to influence religious behavior and identity. What is missing from existing research is attention to when and how in a person's life course political changes and concomitant religious changes occur. Emerging adulthood is a segment of life when individuals’ partisan political identity is highly influenced (Margolis 2018a; Smith and Snell 2009). A growing body of research finds the same is true for religious identity (Astin, Astin, and Lindholm 2010; Hall, Edwards, and Wang 2016; Hill, 2009, 2011; Smith and Snell 2009). Yet, to the best of our knowledge, no previous study has examined change in political identity and its relation to religious belief, spirituality, and behavior at this key period in the life course using longitudinal data.
- Hypothesis 1: Becoming more politically liberal in the college years will be associated with decreases in public and private religious activities.
- Hypothesis 2: Becoming more politically liberal will be associated with increasing doubts or conflicts in religious/spiritual beliefs among students.
- Hypothesis 3: Becoming more politically liberal will be associated with feeling increasingly distant in relation to the divine.
- Hypothesis 4: Becoming more politically liberal will be associated with decreasing Vertical Faith Maturity.
- Hypothesis 5: Becoming more politically liberal will be associated with increasing Horizontal Faith Maturity.
Data, Measures, and Methods
This study uses two waves of longitudinal panel data from a research project exploring undergraduate students’ spiritual formation and character development at a large Protestant university in the South (Dougherty et al. 2022). Though it is limited to one faith-based university, these data are well suited for our purposes of analyzing the changing association between political identity and religiosity due to the university's position as a leading Christian university. As such, it is a bellwether for other Christian colleges and universities, where the effects of religious moral community are most robust.
The first wave of data were collected in August 2018, on the first day of chapel, a weekly religious service students are required to attend for two semesters at the university. Students received an email from the Office of Institutional Research and Testing with a link to the online survey instrument. Of 3718 new undergraduate students entering the university in the fall of 2018, 3369 completed an online religion survey for a 91 percent response rate. In the summer of 2020, the same survey was distributed to 2799 students entering their third year at the university who had participated in the initial survey. A total of 561 students completed the survey for a response rate of 20 percent. After exclusion of cases with missing data for key variables, the analytic sample was 400 students.
Dependent Variables
Religious activities. Change in students’ public and private religious activities is analyzed using three measures: religious service attendance, prayer or meditation, and reading the Bible. To measure students’ frequency of attending religious services (a reliable test of public religiosity and involvement in moral community), they were asked “How often do you attend religious services at a church, mosque, synagogue, or other place of worship?” Available answers ranged from 0 = “never” to 7 = “several times a week.” Frequency of private prayer or meditation was included among a set of activities beneath the question “During the past year, how much time do you spend during a typical week doing the following activities?” Response categories ranged from 0 = “never” to 7 = “over 20 hours.” Bible reading was assessed using the item “How often do you read the Bible?” with response options ranging from 0 = “never” to 7 = “daily.” Change in the frequency in each religious activity was assessed by subtraction of the Year 1 value from the value at Year 3 so that positive scores represent increased attendance, prayer, or Bible reading.
-
Jesus was the divine Son of God.
-
The Bible may be an important book of moral teachings, but it was no more inspired by God than were many other such books in human history. (Reverse coded)
-
The concept of God is an old superstition that is no longer needed to explain things in the modern era. (Reverse coded)
-
Through the life, death, and resurrection of Jesus, God provided a way for the forgiveness of people's sins.
-
Despite what many people believe, there is no such thing as a God who is aware of our actions. (Reverse coded)
-
Jesus was crucified, died, and was buried but on the third day, He arose from the dead.
Scores on the Christian Orthodoxy Scale range from one to seven, with higher values indicating more orthodox beliefs. Cronbach's alphas for the scale were .83 in Year 1 and .94 in Year 3. Others have used this scale to test religious beliefs among alumni at two Christian colleges (Cook et al. 2014) and students at a public university (Cummings et al. 2017). A change score was created by subtracting Year 1 Christian Orthodoxy from Year 3 Christian Orthodoxy for each respondent. Positive Christian Orthodoxy change scores correspond to increased acceptance of traditional Christian beliefs. Negative change scores signify reduced orthodox Christian beliefs over time.
In the past year, how often have you had struggles related to religion or spirituality? Struggles can include: feelings of confusion or doubt about your religious/spiritual beliefs; feelings as though your life had no deeper meaning; conflicts with other people about religious/spiritual matters; feeling hurt/mistreated by religious people; or feeling attacked by evil forces.
Response options ranged from 1 = “never/not at all” to 5 = “very often.” For both variables, Year 1 scores were subtracted from Year 3 scores to create variables depicting increasing doubt in or conscious wrestling with religion.
Closeness to God and faith maturity. To measure different aspects of faith, we employ three variables. The first variable is a participant's perceived closeness to God from the question “How close do you feel to God?” Answers available to students ranged from 1 = “not at all close” to 5 = “extremely close.” Year 1 scores were subtracted from those at Year 3 to capture change in perceived intimacy with God.
Finally, we assess Vertical Faith Maturity and Horizontal Faith Maturity using subscales from the Faith Maturity Scale developed by Benson and colleagues (1993). The Faith Maturity Scale contains indicators of faith as lived out in everyday life rather than general statements of belief. It must be noted that although these measures refer to spirituality in terms of “maturity,” they are not judgments of the development of individuals’ faith; they are measurements of orientations toward the divine, society, or other people. We use the names of these variables in keeping with prior research. The Vertical Faith Maturity subscale has seven items that measure the strength of a respondent's relationship with the metaphysical, attempts to find God, and personal experiences with the divine. Sample items from the vertical subscale include “I have a real sense that God is guiding me” and “I seek opportunities to help me grow spiritually.” Vertical Faith Maturity scores range from one to seven, with higher numbers indicating more active prioritization of the relationship between the believer and the divine. Cronbach's alpha for the vertical subscale was .91 in Year 1 and .93 in Year 3. The Horizontal Faith Maturity subscale is a five-item measure of the degree that faith motivates people toward prosocial or communal action, with Likert scale scores ranging from one to seven. Sample items include “I feel a deep sense of responsibility for reducing pain and suffering in the world” and “I help others with their religious questions and struggles.” Cronbach's alpha for the horizontal subscale was .73 in Year 1 and .72 in Year 3. For both variables, Year 1 scores were subtracted from Year 3 scores to measure change in a student's faith maturity over time.
Independent Variable: Political Liberalization
The key independent variable measures change in a student's political identity from Year 1 to Year 3 at the university, which we refer to as “political liberalization.” The survey asked, “How would you describe yourself politically?” Response options were coded 1 = “extremely conservative,” 2 = “conservative,” 3 = “leaning conservative,” 4 = “moderate,” 5 = “leaning liberal,” 6 = “liberal,” and 7 = “extremely liberal.” Responding to this question, students indicate both their political orientation (conservative or liberal) as well as their degree of commitment to this orientation (from “leaning” to “extremely”). We calculated a change score by subtracting political identity at Year 1 from Year 3 political identity. Values less than zero indicate movement toward political conservatism over time. A value of zero indicates no change in reported political identity from Year 1 to Year 3. Values greater than zero indicate movement toward political liberalism. It is important to clarify that the change score does not measure a categorical change in identity from political conservative to political liberal, or vice versa. Instead, it measures the direction of movement from a student's initial political leanings on the seven-point ordinal scale.
Control Variables
Controls were included for students’ gender, socioeconomic status (SES), race, or ethnicity, and their scores on each of the dependent variables at Year 1. Gender was coded as a dummy variable where 0 = male and 1 = female. The survey did not include other response options for gender. SES was assessed using a survey question that asked, “Which social class would you consider yourself to belong to?” We created dummy variables for the SES categories of “lower class,” “working class,” “middle class,” and “upper class.” Lower class was the reference group. Similarly, we use a set of dummy variables for race and ethnicity to distinguish “White,” “Black,” “Hispanic,” “Asian or Pacific Islander,” and “Other race.” White was the reference group.
Analytic Approach
First, descriptive statistics for the sample are reported in Table 1 to provide a general overview of the difference in the variables of interest for the students’ first and third years at the university. Next, we show the percentages for decline and growth in each measure of religion by change in political identity, gender, race or ethnicity, and SES in Table 2. This allows a simplified picture of how change in these measures of religiosity is associated with change in politics. In particular, Table 2 focuses on how change in religiosity compares by three different paths of change in politics (shifting toward the political right, shifting toward the left, or no reported change in political alignment) across survey waves. Finally, we build upon these descriptive findings with results from nine ordinary least squares (OLS) regression models in Tables 3, 4, and 5. Each model employs the regressor variable approach, positioning one of the measures of change in religion as the dependent variable and political liberalization as the key independent variable, controlling for gender, race and ethnicity, and SES. Following Allison (1990), we also control for each religiosity variable at Year 1, thereby reducing the likelihood of spuriousness.
| M | SD | Min. | Max. | |
|---|---|---|---|---|
| Student classification | ||||
| Year 1 | ||||
| Freshman | 0.96 | 0 | 1 | |
| Transfer | 0.04 | 0 | 1 | |
| Year 3 | ||||
| Sophomore | 0.01 | 0 | 1 | |
| Junior | 0.86 | 0 | 1 | |
| Senior | 0.13 | 0 | 1 | |
| Fifth-year undergraduate | 0.01 | 0 | 1 | |
| Graduate student | 0.00 | 0 | 1 | |
| Dependent variables | ||||
| Religious service attendance (Year 1) | 5.34 | 2.44 | 0 | 7 |
| Religious service attendance (Year 3) | 4.93 | 2.51 | 0 | 7 |
| Prayer (Year 1) | 1.73 | 1.25 | 0 | 7 |
| Prayer (Year 3) | 1.67 | 1.28 | 0 | 7 |
| Bible reading (Year 1) | 3.74 | 2.20 | 1 | 7 |
| Bible reading (Year 3) | 3.53 | 2.49 | 1 | 7 |
| Christian Orthodoxy (Year 1) | 6.13 | 1.24 | 1 | 7 |
| Christian Orthodoxy (Year 3) | 6.08 | 1.47 | 1 | 7 |
| Questioning beliefs (Year 1) | 1.83 | 0.71 | 1 | 3 |
| Questioning beliefs (Year 3) | 1.79 | 0.71 | 1 | 3 |
| Religious struggles (Year 1) | 2.94 | 1.06 | 1 | 5 |
| Religious struggles (Year 3) | 2.94 | 1.09 | 1 | 5 |
| Closeness to God (Year 1) | 3.00 | 1.13 | 1 | 5 |
| Closeness to God (Year 3) | 3.01 | 1.26 | 1 | 5 |
| Vertical Faith Maturity (Year 1) | 4.96 | 1.37 | 1 | 7 |
| Vertical Faith Maturity (Year 3) | 5.03 | 1.53 | 1 | 7 |
| Horizontal Faith Maturity (Year 1) | 4.69 | 1.23 | 1 | 7 |
| Horizontal Faith Maturity (Year 3) | 4.82 | 1.13 | 1 | 7 |
| Change scores | ||||
| Religious service attendance | −0.41 | 1.94 | −7 | 7 |
| Prayer | −0.07 | 1.17 | −4 | 5 |
| Bible reading | −0.22 | 2.05 | −6 | 5 |
| Christian Orthodoxy | −0.05 | 0.97 | −6 | 3 |
| Questioning beliefs | −0.05 | 0.78 | −2 | 2 |
| Religious struggles | 0.01 | 1.16 | −4 | 4 |
| Closeness to God | 0.01 | 1.03 | −3 | 4 |
| Vertical Faith Maturity | 0.07 | 1.13 | −5 | 4 |
| Horizontal Faith Maturity | 0.13 | 1.26 | −4 | 4 |
| Independent variables | ||||
| Political liberalization | 0.43 | 1.05 | −3 | 4 |
| Political identity | ||||
| Conservative (Year 1) | 0.55 | 0 | 1 | |
| Conservative (Year 3) | 0.46 | 0 | 1 | |
| Moderate (Year 1) | 0.24 | 0 | 1 | |
| Moderate (Year 3) | 0.23 | 0 | 1 | |
| Liberal (Year 1) | 0.20 | 0 | 1 | |
| Liberal (Year 3) | 0.31 | 0 | 1 | |
| Gender | ||||
| Male | 0.30 | 0 | 1 | |
| Female | 0.70 | 0 | 1 | |
| Race or Ethnicity | ||||
| White | 0.70 | 0 | 1 | |
| Black | 0.03 | 0 | 1 | |
| Hispanic | 0.12 | 0 | 1 | |
| Asian or Pacific Islander | 0.11 | 0 | 1 | |
| Other race | 0.04 | 0 | 1 | |
| Socioeconomic status | ||||
| Lower class | 0.01 | 0 | 1 | |
| Working class | 0.09 | 0 | 1 | |
| Middle class | 0.69 | 0 | 1 | |
| Upper class | 0.21 | 0 | 1 |
| Religious Service Attendance | Prayer | Bible Reading | Christian Orthodoxy | Questioning Beliefs | Religious Struggles | Closeness to God | Vertical Faith Maturity | Horizontal Faith Maturity | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| − | + | − | + | − | + | − | + | − | + | − | + | − | + | − | + | − | + | |
| All students | 40.3 | 24.3 | 31.8 | 25.3 | 36.0 | 25.0 | 32.3 | 32.0 | 25.5 | 20.5 | 28.0 | 30.3 | 25.3 | 27.3 | 45.3 | 52.8 | 44.0 | 52.0 |
| Political identity | ||||||||||||||||||
| No change | 39.2 | 21.6 | 30.9 | 29.8 | 35.4 | 23.2 | 28.2 | 32.0 | 28.7 | 13.3 | 26.0 | 27.6 | 22.1 | 29.3 | 40.3 | 56.9 | 45.3 | 51.4 |
| Shift toward conservatism | 28.9 | 38.5a | 34.6 | 26.9 | 30.8 | 32.7 | 21.2 | 57.7a | 26.9 | 3.1 | 36.5 | 23.1 | 17.3 | 38.5 | 36.5 | 61.5 | 40.4 | 55.8 |
| Shift toward liberalism | 44.9b | 22.8b | 31.7 | 19.8a | 38.3 | 24.6 | 40.1ab | 24.0b | 21.6 | 27.5a | 27.5 | 35.3b | 31.1ab | 21.6b | 53.3ab | 45.5ab | 43.7 | 51.5 |
| N | 161 | 97 | 127 | 101 | 144 | 100 | 129 | 128 | 102 | 82 | 112 | 121 | 101 | 109 | 181 | 211 | 176 | 208 |
- Note: “-” refers to decreased scores across survey waves, and “+” refers to increased scores on the respective measure of religion.
- a Significantly different from no change at p < .05.
- bSignificantly different from shift toward conservatism at p < .05.
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Variable | (Religious Service Attendance) | (Prayer) | (Bible Reading) | |||
| b | SE | b | SE | b | SE | |
| (β) | (β) | (β) | ||||
| Political liberalization | −.321*** | .087 | −.071 | .051 | −.277** | .094 |
| (−.174) | (−.064) | (−.142) | ||||
| Year 1 control | −.275*** | .037 | −.419*** | .043 | −.302*** | .044 |
| (−.346) | (−.447) | (−.325) | ||||
| Femalea | .191 | .196 | .034 | .115 | −.073 | .044 |
| (.045) | (.013) | (−.016) | ||||
| Race/ethnicityb | ||||||
| Black | −.031 | .695 | −0.081 | .406 | .324 | .752 |
| (−.002) | (−.009) | (.021) | ||||
| Hispanic | .161 | .286 | .260 | .167 | −.409 | .310 |
| (.027) | (.072) | (−.064) | ||||
| Asian or Pacific Islander | −.187 | .313 | .425* | .183 | .124 | .338 |
| (−.028) | (.107) | (.018) | ||||
| Other race | −349 | .441 | −0.241 | .258 | −.103 | .477 |
| (−.037) | (−.043) | (−.010) | ||||
| Socioeconomic statusc | ||||||
| Working class | −.075 | .801 | .068 | .466 | −.545 | .864 |
| (−.011) | (.017) | (−.077) | ||||
| Middle class | .357 | .751 | .367 | .438 | −.057 | .810 |
| (.081) | (.137) | (−.012) | ||||
| Upper class | .276 | .782 | .451 | .454 | .302 | .842 |
| (.051) | (.139) | (.053) | ||||
| Constant | 0.783 | 0.770 | 0.261 | 0.445 | 1.168 | 0.833 |
| R2 | .169 | .227 | .132 | |||
- Note: Standardized coefficients appear in parentheses below the unstandardized coefficients.
- a Reference category = male.
- b Reference category = white.
- c Reference category = lower class
- †p< .10; *p <.05; **p < .01; ***p < .001.
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Variable | (Christian Orthodoxy) | (Questioning Religious Beliefs) | (Religious Struggles) | |||
| b | SE | b | SE | b | SE | |
| (β) | (β) | (β) | ||||
| Political liberalization | −.189*** | .046 | .107*** | .032 | .114* | .048 |
| (−.203) | (.143) | (.103) | ||||
| Year 1 control | −.103** | .039 | −.601*** | .046 | −.569*** | .047 |
| (−.131) | (−.548) | (−.520) | ||||
| Femalea | .196† | .103 | −.008 | .071 | .286** | .108 |
| (.092) | (−.005) | (.113) | ||||
| Race/ethnicityb | ||||||
| Black | −.148 | .365 | −.161 | .253 | −.269 | .383 |
| (−.020) | (−.027) | (−.031) | ||||
| Hispanic | .051 | .150 | −.025 | .104 | −.055 | .157 |
| (.017) | (v.010) | (−.015) | ||||
| Asian or Pacific Islander | −.136 | .165 | .035 | .114 | −.015 | .173 |
| (−.041) | (.013) | (−.004) | ||||
| Other race | −.616** | .232 | .076 | .161 | −.115 | .243 |
| (−.131) | (.020) | (−.021) | ||||
| Socioeconomic statusc | ||||||
| Working class | −.012 | .423 | −.136 | .291 | −.503 | .440 |
| (−.003) | (−.050) | (−.126) | ||||
| Middle class | .009 | .400 | −.193 | .273 | −.282 | .413 |
| (.004) | (−.109) | (−.107) | ||||
| Upper class | −.025 | .415 | −.234 | .284 | −.259 | .428 |
| (−.009) | (−.108) | (−.081) | ||||
| Constant | 0.563 | 0.432 | 1.204*** | 0.286 | 1.740*** | 0.434 |
| R2 | .093 | .321 | .295 | |||
- Note: Standardized coefficients appear in parentheses below the unstandardized coefficients.
- a Reference category = male.
- b Reference category = white.
- c Reference category = Lower class.
- †p< .10; *p <.05; **p < .01; ***p < .001
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Variable | (Closeness to God) | (Vertical Faith Maturity) | (Horizontal Faith Maturity) | |||
| b | SE | b | SE | b | SE | |
| (β) | (β) | (β) | ||||
| Political liberalization | −.161*** | .047 | −.163** | .053 | .001 | .050 |
| (−.164) | (−.152) | (.000) | ||||
| Year 1 control | −.302*** | .043 | −.215*** | .040 | −.606*** | .042 |
| (−.332) | (−.261) | (−.594) | ||||
| Femalea | (.113) | .105 | .064 | .119 | .163 | .113 |
| (.068) | (.026) | (.059) | ||||
| Race/ethnicityb | ||||||
| Black | .501 | .374 | .233 | .421 | −.340 | .397 |
| (.064) | (.027) | (−.035) | ||||
| Hispanic | .064 | .154 | .127 | .173 | −.045 | .163 |
| (.020) | (.037) | (−.012) | ||||
| Asian or Pacific Islander | −.299† | .168 | −.187 | .189 | −.169 | .179 |
| (−.085) | (−.049) | (−.039) | ||||
| Other race | −.026 | .238 | −.161 | .267 | −.032 | .253 |
| (−.005) | (−.030) | (−.005) | ||||
| Socioeconomic statusc | ||||||
| Working class | .024 | .430 | −.263 | .484 | .208 | .457 |
| (.007) | (−.068) | (.048) | ||||
| Middle class | .223 | .403 | −.211 | .454 | .174 | .428 |
| (.095) | (−.082) | (.061) | ||||
| Upper class | .302 | .418 | −.217 | .472 | .336 | .445 |
| (.106) | (−.069) | (.096) | ||||
| Constant | 0.679 | 0.424 | 1.385** | 0.485 | 0.485 | 0.485 |
| R2 | .151 | .099 | .360 | |||
- Note: Standardized coefficients appear in parentheses below the unstandardized coefficients.
- a Reference category = male.
- b Reference category = white.
- c Reference category = lower class.
- †p< .10; *p <.05; **p < .01; ***p < .001.
Results
Table 1 shows each measure of religion and the percentage distribution of students by political identity at Years 1 and 3. Among first-year students, 55 percent were conservatives, 24 percent were moderates, and 20 percent were liberals. In Year 3, fewer students were politically conservative (46 percent), more were liberal (31 percent), and a similar percentage (23 percent) identified as moderates. In terms of within-individual change, 13 percent of students shifted right politically, 46 percent stayed the same, and 41 percent shifted leftward (results not shown).
Table 2 shows the percent of students whose religious activities, beliefs, and faith maturity increased or decreased for path of change in politics. Results are clearest for religious activities, where more students reported a decrease in attendance, prayer, and Bible reading at Year 3. The largest percentage drop was in public religious involvement. Forty percent of students attended religious services less often, compared to nearly a third or more who decreased frequency of prayer and Bible reading. Roughly a quarter of students attended religious services, prayed, and read the Bible more often. No clear trends appeared for overall change in Christian Orthodoxy, questioning religious beliefs, religious struggles, or perceived closeness to God. Slightly more than half of students scored higher on Vertical Faith Maturity (53 percent) and Horizontal Faith Maturity (52 percent).
Change in the religious lives of students varied in expected ways by whether their political identity shifted right or left during their first 2 years of college. Whereas fewer than 30 percent of those who became more conservative attended religious services less often, this was true for 45 percent of those who became more politically liberal. A significantly higher percentage of those who became more conservative increased their attendance compared to those whose politics did not change (p = .01) or shifted leftward (p = .02). Those who shifted left prayed more at significantly lower rates (p = .03) than those whose political identity remained constant. Those with a stable political identity were no different than those who shifted right on changes in prayer.
These trends held for each measure of religious belief and doubt. Differences by political shift were most stark in the rates of students reporting lower and higher scores on the Christian Orthodoxy Scale. Forty percent of students who liberalized politically also liberalized in terms of Christian Orthodoxy scores. Conversely, nearly 60 percent of students who shifted rightward politically had higher Orthodoxy scores in Year 3. Students liberalizing politically reported questioning their religious beliefs more often in Year 3 at a proportion twice the size of those whose politics did not change and almost 10 times the proportion of students who became more politically conservative. Higher proportions of liberalizing students also struggled with their religion or spirituality than did other students.
Students’ perceived closeness to God and Vertical Faith Maturity changed along with changes in political identity. A significantly higher proportion of liberalizing students reported feeling less close to God at Year 3 than in their first year compared to those whose politics did not change (p = .05) or shifted rightward (p = .05). They also reported feeling closer to God than in Year 1 at a proportion significantly lower (p = .02) than for those who shifted rightward politically. Over half of liberalizing students’ Vertical Faith Maturity scores dropped across survey waves, indicating less active investment in spiritual endeavors and a less intimate relationship with God. Those whose politics did not change (p = .03) or shifted rightward (p = .04) scored higher on the Vertical Faith Maturity subscale compared to liberalizing students. There were no statistically significant differences in Horizontal Faith Maturity scores from Year 1 to Year 3 by change in political identity.
Table 3 reports results from three regression models estimating the effect of change in political liberalization from the first to third year at the university in relation to changes in (1) religious service attendance, (2) prayer or mediation, and (3) reading the Bible, controlling for gender, race or ethnicity, and SES. The models for change in religious service attendance and Bible reading support the hypothesis that students would be less publicly and privately active in their faith if their politics move leftward in college. A one standard deviation increase in political liberalization is significantly associated with a 0.174 standard deviation lower attendance at religious services in Model 1, and a 0.142 standard deviation decrease in Bible reading in Model 3. The effect of liberalization predicting change in prayer frequency in Model 2 is not significant.
In Table 4, we find consistent support for Hypothesis 2. Each measure of change in religious belief shows a statistically significant inverse relationship between liberalizing politically from Years 1 to 3 and religious belief or certainty in belief. A one standard deviation increase in political liberalization is associated with a 0.203 standard deviation decrease in scores on the Christian Orthodoxy Scale. Liberalizing students report more questioning of their religious beliefs and general religious struggles after 2 years at the university, as well. A one standard deviation increase in liberalization is associated with a 0.143 standard deviation increase in the frequency of questioning of beliefs and a 0.103 standard deviation increase in religious struggles.
Table 5 provides support for Hypothesis 3. In Model 1, liberalization in political identity is related to feeling more personally distant from God as students experience dissonance between normative expectations of religiosity and political identity within their social networks. The effect of liberalizing politically is significantly related to change in perceived closeness to God. Models 2 and 3 examine the effect of liberalization on students’ orientation toward God. The coefficient for liberalization in Model 2 is significant and negative, indicating that a one standard deviation shift leftward in political identity across survey waves is associated with a 0.152 standard deviation reduction in Vertical Faith Maturity scores. This aligns with Hypothesis 4 predicting less active spiritual engagement behaviors and a less intimate relationship with God. The coefficient for liberalization in Model 3 of Table 5 is not significant, failing to support Hypothesis 5. Political liberalization within the sample does not explain gains or losses in Horizontal Faith Maturity across waves.
Discussion and Conclusions
Our findings from this study align with research from the past two decades showing a robust association between political identity and normative expectations of religious life (Hout and Fischer, 2002, 2014; Margolis 2018; Rhodes 2010b). Politically conservative Republicans are popularly perceived to be religious, but liberal Democrats are not (Hout and Fischer, 2002, 2014). In this study, students who became more politically liberal across 2 years of college exhibited less frequent displays of public and private religiosity, less certainty in and reduced religious belief, and a less intimate relationship with God. Conversely, those who became more politically conservative typically became more religious.
Our findings also align with other research suggesting that the popular association between political conservatism and religiosity creates dissonance for those on the political left, encouraging those on the left to become passively and actively secular (Campbell et al. 2018). Most of the students in our data are members of a generation whose life has been defined by the previously discussed trends of affective polarization and social sorting in political life. As the inheritors of a social reality defined by these broad trends, students in our study seem to be negotiating perceived misalignment between their religious and political identities and becoming less traditionally religious as they become more politically liberal over their college years.
On average, students at this Protestant university reported engaging in religious activities less over time. This is most apparent for change in religious service attendance; however, these results should be interpreted in light of potential changes in services due to the COVID-19 pandemic. Results in terms of belief and faith maturity are more varied. Although the percentage of students attending religious services less often almost doubled that of those who increased their attendance, the percentages of students who experienced growth and decline in Christian orthodoxy was nearly identical. Relative to the other measures of religiosity, the inverse relationship between political liberalization and religiousness is most apparent for Christian orthodoxy, religious service attendance, and closeness to God.
These data complement findings from others showing a drop off in college students’ religious activities that could in large part be explained by students’ struggle to adjust their lives to competing interests (Astin, Astin, and Lindholm 2010; Berger 2014; Hill 2009; Mayrl and Uecker 2011; Uecker, Regnerus, and Vaaler 2007). Geographic relocation from childhood religious communities, academic pursuits, available substitutes for religious service attendance provided by faith communities (such as involvement in Bible studies or campus parachurch organizations), and nonreligious demands such as working for pay may work together to discourage formal religious participation in the college years. The argument by Astin and colleagues (2010) that the college years have an overall positive impact on the spiritual lives of emerging adults is not supported here. Students’ initially high scores and relative stability in Vertical Faith Maturity show that they were, on average, negotiating the complexities of religious life and their relationship to God at both points in time at the university. Though students’ engagement in religious practices decreased, measures associated with belief and spirituality remained relatively stable from Years 1 to 3 at the university.
The institutional contexts in which students navigate their identities can serve to either abate or exacerbate the effects of general social trends of skepticism toward metaphysical explanations and declining religious participation in the college years (Hill 2011). Our findings offer mixed support for the idea that Christian colleges or universities act as “moral communities” by legitimating and reproducing religious beliefs, behaviors, and values. Protestant colleges and universities work to foster an environment that encourages religiosity and promotes the development of students’ religion and spirituality by providing scheduled time for and places of worship; incorporating religious symbology and rhetoric into their missions, recruitment, orientation, and traditions; and hosting student activities that encourage religious and spiritual exploration and commitment. Despite slight declines in religious activities compared with Year 1, students within the sample maintained high Christian orthodoxy and Vertical and Horizontal Faith Maturity scores at Year 3. Nearly as many students increased as decreased in the measures associated with religious belief and its certainty, closeness to God, and Faith Maturity. The story in our data is not one of growth in religiosity and spirituality across the college years in conservative Christian moral community, but one of maintenance. These students reported high religiosity in their first year at the university as well as their third year. This may be due to the university's efforts to produce a community that legitimizes religiousness.
Change in political identity over the college years may moderate the overall effect of embeddedness within a religious moral community, as students whose politics shifted leftward also evinced declines in religiosity at higher rates than the overall student population. Norms associated with religiosity permeate everyday life in these communities. In an age of social sorting (Mason 2018), politically liberalizing students might become more sensitized to reciprocal norms of conservative politics and religiosity by their status as a political minority in orthodox Protestant universities. It is also possible that as students become less religious in this social context, they eschew the conservative political views as they adopt a different religious minority identity. In future research, researchers might test whether political or religious minority status among religious students at a Protestant school pushes these students to experience parallel changes in religious or political attitudes, respectively, than at Catholic, more nominally Protestant, or religiously unaffiliated universities.
Limitations and Future Directions
There are limitations to this study. The particular university setting for this study provides a unique lens through which to interpret and generalize the findings. As is common at many Protestant institutions, this university requires that students attend at least two semesters of chapel and two semesters of Christian religion courses. One of the numerous ways faith and spirituality permeate life at the university is the high proportion of religiously affiliated clubs and organizations. Further, employment for faculty is conditional upon membership with either a Jewish or Christian faith tradition. The relationship between political change and religious change needs to be examined at secular universities and other types of religious universities as well. But even with respect to Protestant institutions, there are several issues that may affect the findings’ generalizability.
As a large university, this setting may be unique among Protestant universities in the diversity of options available for those seeking community among like-minded others. Students experiencing cognitive dissonance resulting from misaligned political and religious identities are likely to be able to find others at the university who can offer support in integration of liberal political identity with religious life or in further dissociation from their previously held religious identity. Choices for community are more constrained at smaller evangelical Protestant colleges and universities. Politically liberal and liberalizing students at smaller Protestant institutions may perceive social pressure to engage in typical religious behaviors at the university, which could keep them in religious services even when their disbelief makes them feel like outsiders within the religious community.
Thus, we would expect that replicability of our findings in other Protestant colleges and universities will depend on the size and diversity of the student body as well as the dynamic interplay between social pressure and belonging. Toward this end, further research should seek a more thorough understanding of the motivating factors for students’ public displays of religiosity in various types of universities.
Beyond constraints on the generalizability of findings, it is important to note that a definitive causal relationship between political identity change and concurrent change in religious beliefs and behavior cannot be established with these data. This article's focus is a test of the reversal of the traditional “God Gap” causal arrow posited by recent studies (Campbell et al. 2020; Hout and Fischer 2014; Margolis 2018b) among college students, but it is worth noting that the God Gap explanation is also possible. We do not know whether political or religious change between the study waves came first. One of our study's novel contributions is its longitudinal tracking of students’ religion and politics across two relevant time points in college, but our findings would be improved with more frequent administration of the survey during students’ tenure at the university. Longitudinal data with four or more time points would allow better assessment of within-person change as well as evaluation of cross-lagged effects that test directionality.
Another limitation concerns the study's administration. The Year 1 survey that took place during a chapel service could invite bias with students feeling the need to respond in ways that align with the university's religious norms. To combat this potential, researchers administering the survey gave verbal instructions to students to answer honestly, reminding them that the survey is anonymous and confidential. Nevertheless, the setting may have influenced student responses and thus we may overestimate the amount of religious change across waves.
Attrition between survey waves and missing cases among variables of interest are other limitations of these data. Only 16.7 percent of the 3369 students who participated in the first survey wave completed the second, and, after listwise deletion for missing cases on variables of interest, the final sample represented 11.9 percent of the Wave 1 sample. The low response rate of 20 percent (561 of the available 2799 students entering their third year) at Wave 2 can be partially attributed to the survey's distribution in the summer of 2020, mere months after the onset of the COVID-19 pandemic.
Those who participated in the Wave 2 survey differed from those who did not in several respects. Mean initial scores on some measures of religiosity—religious service attendance, Bible reading frequency, and Christian Orthodoxy—were significantly lower for participants who attrited. Those who dropped out were also more likely to have been male and a person of color (most often Black or Hispanic), indicating that participants who attrited were not missing at random.
Our sample of 400 obtained by listwise deletion had significantly higher mean religious service attendance and Bible reading frequency, as well as a higher proportion of white students compared to the full Wave 2 sample of 561 students. Multiple strategies were used to make up for these between-sample differences, including full information maximum likelihood and multiple imputation by chained equations (MICE), but these techniques did not significantly alter the results. Additionally, correlations between political orientation and our measures of religion at Year 1 were consistently similar for those who attrited between survey waves and students in our Year 3 sample, however, showing that the relationship between religion and politics did not differ substantially at Year 1 between the samples.1
Survey attrition is the most obvious impact of the COVID-19 pandemic on this study, but the pandemic has other implications for our findings. Political acrimony was growing over differences in opinion over COVID mitigation strategies, including mitigations that restricted in-person religious services. We do not believe that students in our sample were shifting political allegiances due to the pandemic, but we do not know this for certain. Religious beliefs, however, seemed to remain largely stable, even though congregations were closed for weeks or months for some religious people (Pew Research Center 2021). These closures influenced our attendance variable. However, given the number and variety of religion variables we test, we are confident that our findings reveal relationships that transcend the global COVID-19 pandemic.
Despite these limitations, the study's findings meaningfully demonstrate changes in navigating religious and political life among college students in conservative Christian moral community, where the norms associated with alignment in religiosity and conservative politics are more salient than in other universities. This is a profound illustration of the association between affective political polarization and religiousness that likely reflects social sorting in the identity formation of today's younger generations. The popular conflation of religious involvement and conservative politics makes it difficult for students to retain their religious beliefs and practices when their politics shift in a more liberal direction or, perhaps in some circumstances, to retain their political beliefs when their religiousness shifts. Future research on change in political identity and religiosity over the college years should be done in other university settings to allow for a comparison of this meaning-making process by context, be it religious, social, or geographic.
Acknowledgments
We thank the editor and anonymous reviewers for their helpful feedback on earlier drafts of this article.
References
- 1 These correlations are available online in a separate file of supplemental tables.
