Abstract
Drawing data from a longitudinal survey of college students from 514 institutions of higher education, we add to the discussion on the education–religion puzzle by providing information on specifically which college students experience the most religiosity change, investigating multiple change measures (conviction strength, service attendance, and religious identity), and estimating which programs of study and collegiate experiences cause the most change. We also provide an analysis of students who seek or initially sought an occupation within the clergy. Among our findings, 56% of students report changes in the strength of their religious convictions during college, while 45% report changes in religious service attendance frequency. Of those who matriculate as religious, about 9% lose their religion by graduation. Of those who matriculate with no religious identity, an impressive 33% graduate with one. Choice of institution, major of study, academic success, and many other collegiate experiences are shown to be determinants of these changes.
1 Religion and Education
Religious belief and tradition have long served in how humans interpret their place in the cosmos. As individuals age, they are exposed to ideas and experiences shaping these beliefs. Religious convictions are significant from a societal perspective due to both religious service attendance and belief being associated with important individual-level outcomes such as income (Gruber 2005), mortality (Hummer et al. 1999), marriage and divorce rates (Gruber 2005; Lehrer and Chiswick 1993), and, on a macro level, with GDP growth (McCleary and Barro 2006). Prior research on religion and education is mixed with some studies finding positive relationships (Iannaccone 1998; Gruber 2005) and others negative relationships (Hungerman 2014; Mocan and Pogorelova 2014; Hout and Fischer 2002).
Glaeser and Sacerdote (2008) address the education–religion puzzle, noting service attendance increases with education on an individual level, but attendance decreases as education increases at the denomination level. This same pattern holds true as they turn to higher levels of aggregation, ultimately arguing this can be explained if education increases the return to social aspects of service attendance while reducing amount of belief, and if belief differs by denomination. Akerlof and Kranton (2000) relate identity and college attendance. As individuals attend college and begin to identify themselves with peers, they will view appropriateness of behavior and social phenomena based on the perspective of their associated group. Gaviria and Raphael (2001) find strong peer effects at the school level during secondary education on church attendance. Once students leave the direct supervision of parents, peers they interact with will likely have an even greater effect on religious preferences. Eswaran (2011) argues the incentives religious organizations face differ by the level of competition. Thus, there are expectations of differential impacts according to the makeup of the student body and religious affiliation for private universities. While these studies consider what might impact religious belief or service attendance, they do not focus on what specific mechanisms in the undergraduate college environment drive changes in religiosity.
Hungerman (2014) incorporates changes in Canadian compulsory schooling laws and finds for each additional year of education there is a 4 percentage point decline in the likelihood someone identifies with a religious tradition. Mocan and Pogorelova (2014) use individual-level data combined with changes in compulsory schooling laws in 11 European countries finding additional compulsory schooling has large negative effects on attending religious services and on prayer frequency. These studies offer compelling evidence for the realms of compulsory and secondary education, although non-compulsory post-secondary education is arguably geared more toward greater introspection and individual development leading to differing impacts on religiosity.
Schwadel (2011) challenges the negative education–religion relationship. He finds a positive relationship between education and changes in affiliation toward mainline protestant denominations, changing beliefs in religion’s role in society, increases in religious participation, and decreasing belief in biblical literalism. Schwadel also finds individuals with greater levels of education are less likely to hold literal interpretations of the bible and affirmations of biblical literalism are less likely when more educated members dominate a congregation.
Kimball et al. (2009) discuss whether the difference in the college experience across majors impacts religiosity, arguing the scientific worldview has negative effects. Their results show a decline in religiosity for those majoring in the social sciences and humanities, but increases for students in business and education. Gervais and Norenzayan (2012) find analytical thinking promotes disbelief in laboratory settings, although over a short-term time horizon. Thus, it could be students enrolled in more analytical fields are more apt to become less religious.
Lee (2002) uses similar data to our own to investigate the religiosity effects of college experiences. Our study differs by use of nonlinear estimation, multiple outcomes, a significantly larger and more diverse sample, and controlling for pre-college religiosity, among other differences. Another related study is Mooney (2010), who uses data on students at elite colleges and universities and finds those who attend religious services more regularly spend time on academic pursuits, have higher grades, and exhibit higher levels of satisfaction with their college even after controlling for overall time spent on religious activities.
We add to the discussion of the education–religion puzzle by providing information on specifically which college students experience the most religiosity change, investigating multiple change measures (conviction strength, religious service attendance, and religious identity), and estimating which programs of study and collegiate experiences cause the most change. In addition, since we appear to be the first to have quality data on the subject, we provide a brief analysis of students who seek or initially sought occupations within the clergy.
2 Religiosity Change during College
2.1 College Student Surveys
Data used in this study were collected as part of the Cooperative Institutional Research Program (CIRP), conducted by the Higher Education Research Institute (HERI), which is housed at the University of California – Los Angeles. HERI conducts numerous collegiate surveys. We make use of their The Freshman Survey (TFS) and College Senior Survey (CSS). The TFS is administered at or very near a student’s matriculation, while the CSS is administered at or very near graduation. These data are longitudinal in that we are able to match TFS and CSS responses. All freely available data on students who took both surveys at the time of writing have been combined with two additional years of data (2003–2004) which we requested and received from HERI. This merged data set contains 141,082 respondents who graduated between 1994 and 2004.
Although this sample is large and comes from 514 different institutions, these data are not necessarily nationally representative. Private school students are over-represented, as these institutions more often participate in the CIRP. By its nature, the sample includes only students who both finish their undergraduate program and do so at their original institution or another institution which participates in the CIRP. Thus, we can only investigate students who graduate and do so at a participating institution. We advise the reader to keep these two data limitations in mind. These surveys include multiple measures of religiosity not previously analyzed in detail by others, leading us to believe there is value in both descriptive and multivariate analyses. We begin with the former.
2.2 Changes in Conviction Strength
On the CSS, students were asked to relate how their religious beliefs and convictions changed over their college tenure. Students could respond with [They are] “much weaker,” “weaker,” “no change,” “stronger,” or “much stronger.” Table 1 presents the responses to this question for the full sample and across several subsamples. These include gender, race, institutional type, geographic region, student distance from home, religious identity at matriculation, and aggregate graduating major. Values are the percent of students who chose each response, within each subsample.
Sample | Much weaker | Weaker | No change | Stronger | Much stronger | N |
Full | 2.9 | 8.8 | 44.1 | 28.6 | 15.5 | 141,082 |
Male | 4.2 | 10.0 | 43.7 | 27.3 | 14.6 | 54,087 |
Female | 2.1 | 8.1 | 44.4 | 29.4 | 16.1 | 86,995 |
White | 2.9 | 9.2 | 44.3 | 28.7 | 14.9 | 116,332 |
African-American | 1.2 | 4.1 | 39.6 | 32.2 | 22.8 | 6,026 |
Hispanic | 3.3 | 7.2 | 43.8 | 28.6 | 17.2 | 4,568 |
Asian | 2.5 | 6.8 | 48.7 | 26.6 | 15.4 | 5,534 |
American Indian | 4.5 | 9.5 | 46.2 | 23.5 | 16.3 | 231 |
Other race | 3.9 | 8.1 | 41.2 | 29.1 | 17.8 | 2,324 |
Multiracial | 4.0 | 9.2 | 43.2 | 26.2 | 17.4 | 6,067 |
Secular college/univ. | 3.2 | 9.5 | 48.5 | 26.3 | 12.4 | 82,261 |
Religious college/univ. | 2.8 | 8.4 | 41.3 | 30.0 | 17.5 | 58,821 |
College/university region | ||||||
East | 3.3 | 9.0 | 50.5 | 25.7 | 11.5 | 54,168 |
Midwest | 3.0 | 9.2 | 39.6 | 31.4 | 16.7 | 42,846 |
South | 2.4 | 8.6 | 43.7 | 28.5 | 16.8 | 21,671 |
West | 2.4 | 7.8 | 37.7 | 30.3 | 21.8 | 22,397 |
Distance between college/university and home | ||||||
10 miles or less | 2.3 | 6.0 | 46.8 | 29.1 | 15.6 | 11,719 |
11–50 miles | 2.3 | 6.8 | 46.8 | 29.4 | 14.6 | 24,033 |
51–100 miles | 2.7 | 8.4 | 45.8 | 28.2 | 14.9 | 21,594 |
101–500 miles | 3.0 | 9.8 | 43.5 | 28.1 | 15.5 | 54,234 |
More than 500 miles | 3.6 | 10.1 | 40.6 | 29.0 | 16.8 | 29,502 |
Religious identity at matriculation | ||||||
Catholic | 3.3 | 10.9 | 43.9 | 29.0 | 12.9 | 43,682 |
Protestant | 2.2 | 8.2 | 40.0 | 29.7 | 19.9 | 79,634 |
Jewish | 3.3 | 9.7 | 57.3 | 21.0 | 8.8 | 1,987 |
Other | 3.0 | 6.8 | 44.8 | 28.8 | 16.5 | 4,711 |
None | 4.9 | 6.3 | 65.4 | 16.1 | 7.3 | 11,068 |
Aggregate graduating college major | ||||||
Agriculture | 2.3 | 5.1 | 56.3 | 23.8 | 12.4 | 782 |
Biological sciences | 3.0 | 9.3 | 44.6 | 28.1 | 15.0 | 11,184 |
Business | 1.9 | 8.1 | 48.2 | 29.9 | 11.9 | 20,837 |
Education | 1.1 | 5.1 | 38.2 | 34.6 | 21.0 | 9,479 |
Engineering | 3.1 | 10.0 | 50.7 | 25.8 | 10.5 | 6,054 |
English | 5.1 | 12.6 | 38.4 | 27.4 | 16.5 | 6,122 |
Fine arts | 3.7 | 8.6 | 43.1 | 26.9 | 17.6 | 4,911 |
Health professional | 1.3 | 5.7 | 45.3 | 32.8 | 14.9 | 6,078 |
History or poli. sci. | 4.0 | 11.3 | 43.3 | 26.2 | 15.1 | 9,622 |
Humanities | 4.3 | 9.6 | 33.3 | 27.6 | 25.1 | 7,241 |
Mathematics or stats | 3.1 | 10.5 | 40.4 | 28.3 | 17.7 | 2,110 |
Other non-technical | 2.4 | 7.7 | 44.6 | 29.4 | 16.0 | 8,319 |
Other technical | 3.5 | 9.4 | 49.2 | 26.4 | 11.5 | 2,870 |
Physical sciences | 4.2 | 10.6 | 45.3 | 26.5 | 13.4 | 3,384 |
Social sciences | 3.2 | 10.0 | 43.9 | 27.7 | 15.2 | 17,761 |
Approximately 56% of college students report a change in the strength of their beliefs. Across all samples, there is a general strengthening in convictions, that is, there is no sample investigated where more students report weaker than stronger beliefs. Females report more strengthening than males, while African-American students are the most likely to do so among the seven racial groups. Unsurprisingly, so are students attending a religious-oriented school when compared to those at a secular institution. What may be surprising is that these two groups of students are not very different in this regard.
There appear to be clear, positive relationships between students’ distance from their parental home and the probabilities their convictions experience change and/or weaken. Protestant Christians report the largest increase in convictions of the different matriculating religious identities, while, not surprisingly, those stating no religious identity upon arrival have the lowest. Education, health professional, and humanities majors seem to show the largest increases in convictions, with English and physical sciences majors the largest percentage of weakening convictions.
Perhaps the most noteworthy finding here is the consistency of stronger self-reported convictions across student groupings. Exactly as many students (44.1%) report strengthened convictions as do no change, while only 11.7% relate weakened beliefs. The religiosity-education paradox seems to be exemplified here. While others have shown that society at large becomes less religious as the average education level increases, individual-level data show convictions more often increase over college tenure. We note, however, that comparing conviction strength for students across time is, of course, not the same as comparing those with to those without a higher education.
2.3 Changes in Religious Service Attendance
Another measure of religiosity change is the change in the frequency one attends religious services. On the TFS and CSS, respectively, students were asked about their religious service attendance in the year immediately prior to matriculation and in their last year of college. Table 2 presents the percent of college students who report increased, decreased, or equal attendance during these two separate years across all the same samples as before.[1]
Sample | Less frequent attendance | Equally frequent attendance | More frequent attendance | N |
Full | 38.8 | 54.7 | 6.5 | 141,082 |
Male | 40.8 | 53.0 | 6.2 | 54,087 |
Female | 37.5 | 55.8 | 6.7 | 86,995 |
White | 39.2 | 54.9 | 5.9 | 116,332 |
African-American | 35.7 | 54.9 | 9.4 | 6,026 |
Hispanic | 36.4 | 53.5 | 10.1 | 4,568 |
Asian | 35.8 | 54.2 | 10.1 | 5,534 |
American Indian | 38.6 | 53.8 | 7.6 | 231 |
Other race | 37.2 | 53.2 | 9.6 | 2,324 |
Multiracial | 38.0 | 54.5 | 7.5 | 6,067 |
Secular college/univ. | 40.2 | 52.9 | 6.8 | 82,261 |
Religious college/univ. | 37.9 | 55.9 | 6.3 | 58,821 |
College/university region | ||||
East | 42.7 | 50.7 | 6.7 | 54,168 |
Midwest | 37.8 | 56.6 | 5.7 | 42,846 |
South | 39.1 | 54.2 | 6.7 | 21,671 |
West | 31.0 | 61.5 | 7.6 | 22,397 |
Distance between college/university and home | ||||
10 miles or less | 32.1 | 59.9 | 8.0 | 11,719 |
11–50 miles | 36.3 | 56.5 | 7.2 | 24,033 |
51–100 miles | 40.9 | 53.0 | 6.1 | 21,594 |
101–500 miles | 40.9 | 53.3 | 5.8 | 54,234 |
More than 500 miles | 37.9 | 55.3 | 6.8 | 29,502 |
Religious identity at matriculation | ||||
Catholic | 41.9 | 52.3 | 5.8 | 43,682 |
Protestant | 38.2 | 56.6 | 5.2 | 79,634 |
Jewish | 32.0 | 58.8 | 9.1 | 1,987 |
Other | 38.0 | 53.5 | 8.5 | 4,711 |
None | 30.4 | 57.2 | 12.4 | 11,068 |
Aggregate graduating college major | ||||
Agriculture | 40.8 | 53.2 | 6.0 | 782 |
Biological sciences | 38.0 | 55.0 | 7.0 | 11,184 |
Business | 39.4 | 54.2 | 6.4 | 20,837 |
Education | 30.6 | 63.3 | 6.2 | 9,479 |
Engineering | 39.6 | 54.2 | 6.2 | 6,054 |
English | 43.2 | 51.8 | 5.1 | 6,122 |
Fine arts | 39.6 | 54.1 | 6.3 | 4,911 |
Health professional | 33.7 | 59.4 | 6.8 | 6,078 |
History or poli. sci. | 41.0 | 52.7 | 6.3 | 9,622 |
Humanities | 36.5 | 57.6 | 6.0 | 7,241 |
Mathematics or stats | 35.3 | 57.5 | 7.1 | 2,110 |
Other non-technical | 38.2 | 55.3 | 6.5 | 8,319 |
Other technical | 39.5 | 54.4 | 6.1 | 2,870 |
Physical sciences | 38.9 | 54.4 | 6.7 | 3,384 |
Social sciences | 41.3 | 52.2 | 6.5 | 17,761 |
Whether we should expect a general increase or decrease in attendance is not clear. For students attending school far from home, we might anticipate a decrease in attendance, at least near the beginning of tenure, as they are now living away from their congregation. However, the CSS variable refers to the last year of tenure, giving students ample time to find new services to attend if they choose to do so. Also, college campuses offer many opportunities to join new congregations. In addition to the opportunities/requirements offered by religious institutions themselves, student religious clubs and organizations abound on the typical American college campus (including at public colleges and universities).
Approximately 55% of students report no change in attendance frequency, while a large minority of about 39% report decreased attendance. Less than 7% of college students relate increased religious service attendance. Males, Caucasians, students at east coast colleges, those further from home, and students matriculating as Catholics decrease their service attendance more than their peer counterparts. English, history, and political science majors are the most likely to decrease attendance among the aggregate major categories, though differences across major are relatively small.
Perhaps most interestingly, Table 2 shows students who choose to attend religious colleges are not very dissimilar in terms of attendance change from those who choose a secular college. Overall, Table 2 demonstrates a significant decrease in attendance. While beliefs may strengthen, as shown in Table 1, religious service attendance appears to become much less frequent. Beliefs are also shown to change more often than service attendance.
2.4 Changes in Religious Identity
Yet a third measure of religiosity change, and perhaps the most important, is a change in religious identity. On both college surveys, students were asked to report their religious identity, that is, what specific religion and denomination they consider themselves to be. One choice option was “none,” which refers to either agnosticism or atheism. Using responses from these two variables, we are able to create a categorical variable with four designations. Students who report the same identity on both surveys (including those who chose “none” on both surveys) are designated “kept identity.” Students who report different identities at matriculation and graduation are separated into the three remaining categories: one for those who report “none” on the CSS but not TFS, one for those who do the opposite, and one for those who report two different identities where neither is “none.” These categories are named “lost identity,” “gained identity,” and “changed identity,” respectively. Table 3 presents this categorical variable across our investigated subsamples.
Sample | Kept identity | Changed identity | Lost identity | Gained identity | N |
Full | 85.3 | 3.6 | 8.5 | 2.6 | 141,082 |
Male | 84.0 | 3.4 | 9.7 | 2.9 | 54,087 |
Female | 86.1 | 3.7 | 7.8 | 2.4 | 86,995 |
White | 86.1 | 3.3 | 8.4 | 2.2 | 116,332 |
African-American | 83.6 | 7.8 | 5.8 | 2.8 | 6,026 |
Hispanic | 84.0 | 6.2 | 7.2 | 2.6 | 4,568 |
Asian | 82.6 | 2.2 | 10.6 | 4.6 | 5,534 |
American Indian | 74.3 | 8.8 | 11.7 | 5.2 | 231 |
Other race | 77.3 | 5.8 | 12.3 | 4.6 | 2,324 |
Multiracial | 78.9 | 5.2 | 11.9 | 4.0 | 6,067 |
Secular college/univ. | 85.5 | 3.2 | 8.2 | 3.1 | 82,261 |
Religious college/univ. | 85.2 | 3.9 | 8.7 | 2.2 | 58,821 |
College/university region | |||||
East | 85.1 | 2.9 | 9.3 | 2.7 | 54,168 |
Midwest | 86.0 | 4.3 | 7.7 | 2.0 | 42,846 |
South | 86.1 | 4.3 | 6.9 | 2.7 | 21,671 |
West | 83.7 | 3.4 | 9.6 | 3.3 | 22,397 |
Distance between college/university and home | |||||
10 miles or less | 84.9 | 5.2 | 7.1 | 2.8 | 11,719 |
11–50 miles | 85.9 | 4.2 | 7.5 | 2.4 | 24,033 |
51–100 miles | 86.4 | 3.7 | 7.7 | 2.2 | 21,594 |
101–500 miles | 85.6 | 3.4 | 8.5 | 2.5 | 54,234 |
More than 500 miles | 83.8 | 2.7 | 10.4 | 3.1 | 29,502 |
Religious identity at matriculation | |||||
Catholic | 88.5 | 5.4 | 6.1 | N/A | 43,682 |
Protestant | 89.5 | 4.0 | 6.5 | N/A | 79,634 |
Jewish | 89.1 | 5.2 | 5.7 | N/A | 1,987 |
Other | 39.7 | 46.8 | 13.5 | N/A | 4,711 |
None | 67.1 | N/A | N/A | 32.9 | 11,068 |
Aggregate graduating college major | |||||
Agriculture | 90.0 | 4.4 | 4.2 | 1.4 | 782 |
Biological sciences | 85.3 | 3.0 | 9.1 | 2.6 | 11,184 |
Business | 88.5 | 3.6 | 6.0 | 1.9 | 20,837 |
Education | 89.1 | 5.5 | 3.8 | 1.6 | 9,479 |
Engineering | 86.2 | 2.2 | 9.0 | 2.6 | 6,054 |
English | 80.5 | 3.6 | 13.2 | 2.7 | 6,122 |
Fine arts | 81.0 | 3.1 | 12.5 | 3.4 | 4,911 |
Health professional | 89.8 | 4.3 | 3.9 | 2.0 | 6,078 |
History or poli. Sci. | 84.8 | 2.1 | 10.5 | 2.6 | 9,622 |
Humanities | 81.4 | 3.3 | 12.2 | 3.1 | 7,241 |
Mathematics or stats | 86.2 | 3.4 | 8.2 | 2.2 | 2,110 |
Other non-technical | 86.1 | 3.9 | 7.8 | 2.2 | 8,319 |
Other technical | 83.8 | 2.9 | 10.6 | 2.7 | 2,870 |
Physical sciences | 82.7 | 2.5 | 11.5 | 3.3 | 3,384 |
Social sciences | 83.6 | 3.7 | 10.1 | 2.6 | 17,761 |
A change in religious identity is undoubtedly a bigger, more important shift than changes in conviction strength or service attendance. Approximately 15% of students report such a shift over college tenure. The majority of these (8.5% of the total student pool, or 8.7% of students who matriculate with an identity other than “none”) lost their identity, that is, they became atheist or agnostic during college tenure. At the same time, an impressive 32.9% of students who begin college as atheist or agnostic (or 2.6% of the total student pool) leave college with a religion. Males appear more likely to change than females. Whites were the most likely to hold onto their identity, while American Indians the least among the racial groups. Perhaps surprisingly, differences across institutional secular status are again very small.
Students attending college in the western region report the greatest changes in identity, though regional differences are relatively small. Students attending a college greater than 500 miles from home report having lost their identity at the greatest rate (at 10.4%). Decreased parental proximity might allow for the possibility of greater independent thought leading to the development of different personal religious preferences. Also, students who would be willing to attend college a greater distance from home may already be more independent and self-reliant, and thus be predisposed toward religious identity shifts. English, fine arts, and humanities majors are shown to be the least likely to hold onto their identity through tenure. Students majoring in agriculture, health, and education are shown to be the most likely.
These statistics point toward more movement in religious identity during college than they may have been anticipated given prior research. There do appear to be sizable movements in belief which differ by major, region, and especially for those who matriculate with no identity.
2.5 College Success and Religiosity Change
If there are indeed aspects of the college educational experience that alter one’s religiosity, then we would expect these impacts to differ across the level at which students are “attached to” or “engaged in” their collegiate studies. Students’ grades are perhaps the only objective measure of this level of attachment, at least the only measure we observe here. Table 4 presents our three religiosity change variables across students’ graduating grade point averages (GPAs).
Changes in religious convictions | |||||
Graduating GPA | Much weaker | Weaker | No change | Stronger | Much stronger |
3.75–4.00 | 3.1 | 9.8 | 37.2 | 30.9 | 19.0 |
3.25–3.74 | 3.0 | 9.3 | 42.6 | 28.8 | 16.2 |
2.75–3.24 | 2.8 | 8.3 | 46.2 | 28.1 | 14.6 |
2.25–2.74 | 2.8 | 7.6 | 49.6 | 27.5 | 12.6 |
1.75–2.24 | 2.9 | 7.7 | 54.8 | 23.8 | 10.7 |
Below 1.75 | 6.3 | 6.7 | 60.0 | 16.0 | 11.3 |
Changes in religious service attendance frequency | |||||
Decreased attendance | Equal attendance | Increased attendance | |||
3.75–4.00 | 32.0 | 62.2 | 5.7 | ||
3.25–3.74 | 37.5 | 56.1 | 6.4 | ||
2.75–3.24 | 41.2 | 52.1 | 6.6 | ||
2.25–2.74 | 43.2 | 49.8 | 7.1 | ||
1.75–2.24 | 44.4 | 48.0 | 7.6 | ||
Below 1.75 | 43.8 | 44.8 | 11.4 | ||
Changes in religious identity | |||||
Kept identity | Changed identity | Lost identity | Gained identity | ||
3.75–4.00 | 86.1 | 2.6 | 9.3 | 2.0 | |
3.25–3.74 | 85.1 | 3.2 | 9.2 | 2.5 | |
2.75–3.24 | 85.4 | 4.0 | 8.0 | 2.6 | |
2.25–2.74 | 85.1 | 4.8 | 7.2 | 2.9 | |
1.75–2.24 | 84.1 | 5.2 | 7.0 | 3.7 | |
Below 1.75 | 80.9 | 5.9 | 6.4 | 6.8 |
We turn first to changes in conviction strength. We see a strong, positive relationship between GPA and the probability there was a change in belief strength. Students in the highest GPA category (3.75–4.00) are disclosed to be 22.8 percentage points more likely to experience a conviction strength change than those in the lowest category (below 1.75). Students with the highest grades were the most likely to feel their beliefs became stronger, much stronger, but also weaker when compared to other students. Students of all grade levels are more likely to experience strengthened than weakened convictions.
There is also a clear pattern between academic performance and service attendance change. The better the student’s performance, the less likely they are to change their service attendance frequency. Similarly, the lower a student’s performance, the more likely they are to either decrease or increase attendance. Specifically, students with the lowest grades are 17.4 percentage points more likely to change attendance frequency than the highest academic performance.
Students in the highest GPA category are additionally the most likely to keep the same religious identity over tenure, but also the most likely to switch to agnosticism or atheism from some other identity. They are thus the least likely to fall within the categories of a “changed” or “gained” an identity. For all four religious identity categories, the bottom panel of Table 4 presents monotonic relationships across GPA.
It seems collegiate success is related to some important religiosity shifts, with it being positively related to changing belief strength, negatively related to service attendance change, and also having some clear relationships with identity changes. It could be that those students with the best academic performance also experienced the most inner development as well. As part of being engaged in learning course material, it is possible they also learned more about themselves and their religion. Overall, like others have shown for education level and religion, we conclude there exists a clear yet complex relationship between academic success and religiosity.
2.6 Students Seeking Positions in the Clergy
Students aiming for post-collegiate religious occupations are an interesting subsample in their own right with regards to any education–religion relationships. Tables 5 and 6 present summary statistics on students who on the TFS and/or CSS indicated a position in the clergy was their desired occupation. Our large student sample contains 1,076 students who sought a clergy position at matriculation, with 40.4% specifically seeking to be a minister or priest and the rest some “other clergy” occupation. While we are unable to match respondents to their respective institutions, we know four specific schools represented in our merged sample are seminary schools, explaining the large subsample here.
Statistic | Number | Percent of subsample |
Matriculating students seeking a job in the clergy | 1,076 | |
Specifically minister/priest | 435 | 40.43 |
Specifically “other” clergy | 641 | 59.57 |
Desire maintained by graduation | 478 | 44.42 |
Majored in theology | 160 | 33.47 |
Religious college/university | 441 | 92.26 |
Have parent in clergy | 66 | 13.81 |
Male | 368 | 76.99 |
Mean graduating college GPA (std. dev.) | 3.40 | (0.42) |
Desire lost by graduation | 598 | 55.58 |
Majored in theology | 72 | 12.04 |
Religious college/university | 502 | 83.95 |
Have parent in clergy | 55 | 9.20 |
Male | 368 | 61.54 |
Mean graduating college GPA (std. dev.) | 3.30 | (0.60) |
Graduating students seeking a job in the clergy | 1,423 | |
Specifically minister/priest | 899 | 63.18 |
Specifically “other” clergy | 524 | 36.82 |
Desire formed during college tenure | 945 | 66.41 |
Majored in theology | 159 | 16.83 |
Religious college/university | 791 | 83.70 |
Have parent in clergy | 98 | 10.37 |
Male | 573 | 60.63 |
Mean graduating college GPA (std. dev.) | 3.32 | (0.48) |
Changes in religious convictions | |||||
Much weaker | Weaker | No change | Stronger | Much stronger | |
Clergy desire maintained | 0.2 | 1.9 | 4.9 | 30.6 | 62.4 |
Clergy desire lost | 2.2 | 7.3 | 14.3 | 36.2 | 39.9 |
Clergy desire gained | 0.3 | 1.2 | 4.7 | 25.5 | 68.3 |
Changes in religious service attendance frequency | |||||
Decreased attendance | Equal attendance | Increased attendance | |||
Clergy desire maintained | 2.5 | 96.9 | 0.6 | ||
Clergy desire lost | 19.3 | 77.9 | 2.9 | ||
Clergy desire gained | 4.9 | 89.1 | 6.0 | ||
Changes in religious identity | |||||
Kept identity | Changed identity | Lost identity | Gained identity | ||
Clergy desire maintained | 95.8 | 3.0 | 0.6 | 0.6 | |
Clergy desire lost | 90.3 | 5.1 | 2.3 | 2.3 | |
Clergy desire gained | 91.3 | 6.5 | 1.1 | 1.1 |
Of these students, only 44.4% still maintained this occupational goal at graduation. Of the subsample who maintained this goal, they were more often theology majors, more often attending non-secular institutions, more often have a parent in the clergy, male, and earned higher grades. Though these relationships are all expected, differences here are mostly small. More interestingly, more than 66% of students who desire a clergy occupation at graduation had another preferred occupation in mind at matriculation. These students are also about 23 percentage points more likely to minister/preach than those who had wanted a clergy job at matriculation.
Table 6 shows our three religiosity change variables for the three clergy subsamples. These students much more often increased their convictions during college. Even among respondents stating they lost the desire to join the clergy during college, 76.1% responded to having stronger religious beliefs at graduation when compared to matriculation. They were also much more likely to keep their identities and not change their service attendance frequency.
3 Models of Student Religiosity Change
With the results of a descriptive analysis in hand, we turn to multivariate techniques to better understand the relationship between religiosity change and the overall college experience, with a focus on students’ majors of study. We are interested in which majors of study, and other college experiences, appear to have significant effects on student religiosity change after controlling students’ religious characteristics at matriculation among many other relevant factors. Thus, our models take the form
where
We specify and estimate three versions of this model. The first specification uses the self-reported ordinal change in religious conviction over college tenure variable (presented in Table 1) as the dependent variable (
The third specification makes use of categorical religious identity change (presented in Table 3) as the dependent variable and is estimated with multinomial logistic regression. Here, the marginal effects calculated from
Control vector
The top panel in Table 7 displays the college experiences captured by these surveys. These variables include the student’s graduating GPA and nine college event indicators. The events captured are: joining a social Greek letter organization (a fraternity or sorority); living in the college dormitories; having a roommate of a different race or ethnicity; an indicator for having frequently socialized with such students outside of the dormitories; working full-time; becoming married; an indicator for those reporting they frequently felt overwhelmed by their college studies; and two indicators for students whose political leaning has changed over college tenure, one for these views becoming more liberal, and the other more conservative. These last two indicators were created using two categorical political leaning variables, one from each survey, relating the students’ current leaning.[2] The other eight event indicators were reported directly by students on the CSS. There are undoubtedly other collegiate events that would be interesting in this context, but we must work within the confines of what is captured and made available to us by HERI.
Variable | Mean | Std. dev. | Min | Max |
Collegiate experiences | ||||
Graduating GPA | 3.209 | 0.458 | 1 | 4 |
Joined fraternity/sorority | 0.192 | 0.393 | 0 | 1 |
Lived in dorms | 0.871 | 0.335 | 0 | 1 |
Roommate of different race/ethnicity | 0.311 | 0.463 | 0 | 1 |
Socialized with other races/ethnicities | 0.464 | 0.499 | 0 | 1 |
Worked full-time while in school | 0.082 | 0.274 | 0 | 1 |
Became married during college | 0.027 | 0.163 | 0 | 1 |
Felt overwhelmed by college | 0.328 | 0.470 | 0 | 1 |
College changed political views: more liberal | 0.239 | 0.427 | 0 | 1 |
College changed political views: more conservative | 0.175 | 0.380 | 0 | 1 |
Religion-oriented controls | ||||
College’s religious affiliation: very important | 0.164 | 0.370 | 0 | 1 |
College’s religious affiliation: somewhat important | 0.244 | 0.429 | 0 | 1 |
Different religious identity from mother | 0.094 | 0.292 | 0 | 1 |
Different religious identity from father | 0.129 | 0.335 | 0 | 1 |
Plans to enter the clergy | 0.008 | 0.087 | 0 | 1 |
Mother in the clergy | 0.003 | 0.054 | 0 | 1 |
Father in the clergy | 0.016 | 0.126 | 0 | 1 |
Self-rating: emotional health | 3.751 | 0.885 | 1 | 5 |
Essential goal: raising a family | 0.440 | 0.496 | 0 | 1 |
Essential goal: developing a life philosophy | 0.252 | 0.434 | 0 | 1 |
Experienced depression | 0.078 | 0.268 | 0 | 1 |
Frequently lonely/homesick during college | 0.057 | 0.231 | 0 | 1 |
Religious service attendance: frequent | 0.555 | 0.497 | 0 | 1 |
Religious service attendance: occasional | 0.328 | 0.469 | 0 | 1 |
Considers oneself “born-again Christian” | 0.151 | 0.358 | 0 | 1 |
Demographics and other controls | ||||
Male | 0.383 | 0.485 | 0 | 1 |
Age at matriculation | 18.280 | 0.597 | 17 | 21 |
Age at graduation | 22.600 | 1.053 | 21 | 28 |
Part-time student | 0.048 | 0.213 | 0 | 1 |
International student | 0.023 | 0.151 | 0 | 1 |
Native language is English | 0.925 | 0.263 | 0 | 1 |
Family income ($0,000) | 64.350 | 54.898 | 0 | 200 |
Student loans ($0,000) | 14.900 | 22.023 | 0 | 100 |
Mother’s education (years of schooling) | 14.880 | 2.481 | 8 | 20 |
Father’s education (years of schooling) | 15.450 | 2.678 | 8 | 20 |
First-generation college student | 0.171 | 0.377 | 0 | 1 |
Parents divorced | 0.170 | 0.376 | 0 | 1 |
One or more parents deceased | 0.033 | 0.180 | 0 | 1 |
Distance between college and family home (miles) | 160.300 | 177.450 | 5 | 500 |
Attending first choice college | 0.752 | 0.432 | 0 | 1 |
Attending second choice college | 0.181 | 0.385 | 0 | 1 |
Non-traditional student (not straight from HS) | 0.035 | 0.184 | 0 | 1 |
High school GPA | 3.368 | 0.377 | 1 | 4 |
The middle and bottom panels of Table 7 collectively display our set of individual-level control variables. These variables were split into two panels for ease of reference. The middle panel contains the set of controls one would hope to obtain and use in a study of religiosity change. This subset of controls includes 14 indicators: separate indicators for students reporting the college’s religious affiliation was very important or somewhat important in its choosing (omitted category “not important”); separate indicators for students who have different religious identities from their parents at matriculation; one for students who plan to enter the clergy; separate indicators for students whose parents are in the clergy; separate indicators for students reporting at matriculation the goals of raising a family and developing a life philosophy were “essential”; one for students reporting they experienced depression during college tenure; one for students who frequently felt lonely and/or homesick during college; separate indicators for frequent and occasional religious service attendance in the year prior to matriculation (omitted category “not at all”); and one for students reporting they are “born-again Christians.”
The bottom panel of Table 7 contains the demographic and other controls. These include a gender (male) dummy; age at matriculation; age at graduation; annual family income in the year before matriculation; student loan amount; mother’s and father’s total years of schooling; distance (in miles) between the college/university and the student’s previous home; high school GPA; and separate indicators for part-time students, international students, those whose first language is English, first-generation students, parental marital status, those who have one or more deceased parents, non-traditional students (those not straight from high school), students attending their first choice institution, and those attending their second choice (omitted category third or lower choice).
4 Results
Before discussing the results of our multivariate models, we remind the reader that religion is a complex subject. Religiosity can and is in fact meant to influence many aspects of an individual’s life, especially for those with the strongest convictions. The opposite is also true, that is, life experiences can affect religiosity. Discussed in the previous section, we have implemented a large level of control including 19 variables pertaining to religious belief at matriculation alongside school and time fixed effects, and 24 other control variables. Even so, given the complexity of religiosity, there may simply be endogeneity concerns for which there are no clear empirical solutions. However, given our high level of control, and that our outcome variables are all religiosity changes as opposed to religious characteristics at a particular time, these concerns are (perhaps greatly) lessened when compared to many other existing empirical studies of religiosity.
Table 8 displays the odds ratios and marginal effects from the major of study indicators (
Aggregate major | Model | |||
(1) Change in convictions | (2) Change in attendance | (3) Change in identity | ||
Changed identity | Lost identity | |||
Agriculture | 1.008 | 1.100 | –0.008*** | –0.029*** |
(0.071) | (0.073) | (0.002) | (0.005) | |
Biological sciences | 0.940*** | 0.932*** | –0.003*** | 0.010*** |
(0.020) | (0.018) | (0.001) | (0.003) | |
Business | 0.914*** | 1.112*** | –0.006*** | –0.016*** |
(0.016) | (0.020) | (0.001) | (0.002) | |
Education | 1.559*** | 1.744*** | –0.005*** | –0.026*** |
(0.035) | (0.043) | (0.001) | (0.002) | |
Engineering | 0.736*** | 0.949* | 0.004*** | 0.037*** |
(0.020) | (0.027) | (0.001) | (0.004) | |
English | 0.875*** | 0.861*** | 0.001 | 0.039*** |
(0.024) | (0.024) | (0.001) | (0.004) | |
Fine arts | 1.013 | 1.070 | –0.001 | –0.001 |
(0.030) | (0.043) | (0.001) | (0.003) | |
Health professional | 1.157*** | 1.643*** | –0.003*** | –0.022*** |
(0.030) | (0.044) | (0.001) | (0.002) | |
History or poli. Sci. | 0.848*** | 0.977 | 0.001 | 0.013*** |
(0.019) | (0.022) | (0.001) | (0.003) | |
Humanities | 1.383*** | 1.282*** | –0.003** | –0.027*** |
(0.036) | (0.033) | (0.001) | (0.002) | |
Mathematics or stats | 1.031 | 1.012 | –0.001 | –0.001 |
(0.044) | (0.064) | (0.001) | (0.005) | |
Other technical | 0.777*** | 0.841*** | –0 001 | 0.014*** |
(0.028) | (0.031) | (0.001) | (0.003) | |
Physical sciences | 0.804*** | 0.940*** | –0.001 | 0.012** |
(0.028) | (0.020) | (0.001) | (0.005) | |
Social sciences | 0.922*** | 0.946*** | 0.003*** | 0.022*** |
(0.017) | (0.017) | (0.001) | (0.005) | |
Pseudo- |
0.285 | 0.533 | 0.361 | |
N | 141,082 | 141,082 | 130,014 |
The remaining 10 aggregate major categories show statistically significant results in all three models. Biological science, engineering, English, other technical, physical science, and social science majors are all shown to have both reduced convictions and service attendance relative to the average college graduate. The opposite can be said of education, health professional, and humanities majors. Business majors are unique in that, while their convictions weakened, they increased their service attendance. It may be the case that business majors most benefit from the networking effects (or “social-pressure” effects, as proposed by Azzi and Ehrenberg (1975)) of service attendance, though this is speculation on our part. Our finding business majors generally had declining religiosity is in opposition to Kimball et al. (2009) finding they exhibited increasing belief. Though smaller in magnitude, the identity change model results seem to mirror those of these two ordered logits. At an increased probability of approximately 4 percentage points, English majors are shown to be the most likely to lose their identity over college tenure.
Table 9 presents odds ratios and marginal effects from the college experience variables (
College experience | Model | |||
(1) Change in convictions | (2) Change in attendance | (3) Change in identity | ||
Changed identity | Lost identity | |||
Graduating GPA | 1.191*** | 1.561*** | –0.051*** | 0.005*** |
(0.015) | (0.020) | (0.003) | (0.001) | |
Joined fraternity/sorority | 0.773*** | 0.764*** | 0.012*** | –0.001 |
(0.011) | (0.011) | (0.003) | (0.001) | |
Lived in dorms | 1.084*** | 1.135*** | –0.021*** | 0.008*** |
(0.019) | (0.020) | (0.004) | (0.001) | |
Roommate of different race/ethnicity | 1.057*** | 0.941*** | 0.013*** | 0.005*** |
(0.014) | (0.012) | (0.003) | (0.001) | |
Frequently socialized with other races/ethnicities | 1.047*** | 0.980*** | 0.007** | 0.003*** |
(0.013) | (0.012) | (0.003) | (0.001) | |
Worked full-time while in school | 0.952** | 0.872*** | 0.010** | 0.010*** |
(0.021) | (0.019) | (0.005) | (0.002) | |
Became married during college | 1.833*** | 1.365*** | 0.108*** | –0.006** |
(0.065) | (0.090) | (0.007) | (0.003) | |
Felt overwhelmed by college | 1.022* | 1.106*** | –0.010*** | 0.007*** |
(0.013) | (0.014) | (0.003) | (0.001) | |
College changed political views: more liberal | 0.881*** | 0.859*** | 0.010*** | 0.040*** |
(0.011) | (0.011) | (0.002) | (0.002) | |
College changed political views: more conservative | 1.134*** | 1.029** | 0.012*** | –0.008*** |
(0.016) | (0.014) | (0.002) | (0.002) | |
Pseudo- |
0.285 | 0.533 | 0.361 | |
N | 141,082 | 141,082 | 130,014 |
Interaction with those of a different race and/or ethnicity exposes students with new views and ideas, including those pertaining to religion. Results here show that the prevalence of these interactions, whether as a roommate or otherwise, is associated with slightly stronger personal convictions, slightly less frequent service attendance, and very small increases in the probabilities of changing or losing one’s religious identity.
5 Conclusions
Undergraduate education is a time of great change for students as they are greatly increasing their knowledge sets and are often without direct parental supervision for the first time in their lives. This study of undergraduate education confirms our preconceptions that religiosity would be among those things to change. Over half the students sampled report having changes in the strength of their religious convictions, with a non-trivial portion of students (about 15%) fully changing their religious identity. Our descriptive results are in line with the side of the education–religion debate that argues individual-level religious beliefs strengthen during college, on average. However, we find there are many programs of study and college experiences that decrease religious convictions, again on average.
Glaeser and Sacerdote (2008) claim education has a negative effect on religious convictions and that is due to the emphasis that secular education places on concepts that are contrary to many traditional religious views. These effects are expected to be more prevalent for students of disciplines like the physical sciences, history, and perhaps some social sciences. We find this to be true. We also find not all courses of study have the same impacts on religiosity; thus, it may be more appropriate to discuss disciplines individually when confronting the education–religion puzzle than thinking of education as an aggregate, homogeneous variable. Perhaps the most interesting of our findings were the relationships between collegiate academic success and religiosity changes. Success was associated with increases in convictions, increases in religious service attendance, and less movement in terms of religious identity.
Our sample only includes those that both attended and completed college. Thus, we are unable (and it was not our goal) to claim results here are in favor of an overall positive or negative education–religion relationship. We sought to add to the discussion by discovering which college students experience the most change and to demonstrate the complexity of the higher education–religiosity relationship. While detailed religiosity data is notoriously difficult to find, future avenues for research would preferably contain both college students and those who did not opt to enter post-secondary education to better isolate the effects of college and more directly address the education–religion puzzle.
Appendix
Year before college | Last year of college | |||
Not at all | Occasionally | Frequently | ||
Secular college or university (N = 82,261) | ||||
Not at all | 74.3 | 22.1 | 3.6 | 12.2 |
Occasionally | 39.2 | 50.2 | 10.6 | 35.0 |
Frequently | 9.4 | 40.7 | 49.8 | 52.8 |
27.8 | 41.8 | 30.5 | 100.0 | |
Religious college or university (N = 58,821) | ||||
Not at all | 71.8 | 24.7 | 3.6 | 10.3 |
Occasionally | 38.4 | 51.0 | 10.6 | 31.7 |
Frequently | 8.3 | 36.1 | 55.7 | 58.0 |
24.4 | 39.6 | 36.0 | 100.0 |
Variable | Model | |||
(1) Change in convictions | (2) Change in attendance | (3) Change in identity | ||
Changed identity | Lost identity | |||
College’s religious affiliation: very important | 1.294*** | 1.149*** | –0.016*** | –0.026*** |
(0.035) | (0.021) | (0.002) | (0.001) | |
College’s religious affiliation: somewhat important | 1.359*** | 1.621*** | –0.011*** | –0.013*** |
(0.017) | (0.021) | (0.002) | (0.001) | |
Different identity from mother | 0.923*** | 0.744*** | 0.007** | –0.005*** |
(0.019) | (0.017) | (0.003) | (0.002) | |
Different identity from father | 0.862*** | 0.834*** | 0.007*** | 0.015*** |
(0.015) | (0.016) | (0.002) | (0.002) | |
Plans to enter the clergy | 1.727*** | 1.686*** | –0.019*** | –0.024*** |
(0.168) | (0.039) | (0.007) | (0.006) | |
Mother in clergy | 1.133 | 0.933 | 0.019 | 0.005 |
(0.108) | (0.100) | (0.015) | (0.012) | |
Father in clergy | 1.450*** | 1.821*** | –0.002 | –0.005 |
(0.059) | (0.094) | (0.006) | (0.005) | |
Self-rating: emotional health | 1.219*** | 1.157*** | 0.001 | –0.006*** |
(0.008) | (0.008) | (0.001) | (0.001) | |
Essential goal: raising a family | 1.356*** | 1.517*** | –0.010*** | –0.029*** |
(0.014) | (0.017) | (0.001) | (0.001) | |
Essential goal: developing a life philosophy | 1.524*** | 1.055*** | 0.020*** | 0.029*** |
(0.019) | (0.014) | (0.002) | (0.001) | |
Experienced depression | 0.858*** | 0.864*** | 0.008** | 0.018*** |
(0.019) | (0.021) | (0.003) | (0.003) | |
Frequently lonely/homesick during college | 1.108*** | 1.114*** | 0.004 | –0.005** |
(0.027) | (0.030) | (0.003) | (0.002) | |
Religious service attendance: frequent | 1.550*** | 1.758*** | –0.028*** | –0.047*** |
(0.028) | (0.415) | (0.003) | (0.002) | |
Religious service attendance: occasional | 1.070*** | 1.393*** | 0.003 | –0.003** |
(0.019) | (0.071) | (0.002) | (0.001) | |
Considers oneself “born-again Christian” | 1.750*** | 1.233*** | 0.032*** | –0.016*** |
(0.026) | (0.038) | (0.002) | (0.002) | |
Catholic at matriculation | 0.657*** | 0.919*** | –0.040*** | –0.075*** |
(0.007) | (0.010) | (0.001) | (0.001) | |
Jewish at matriculation | 0.450*** | 0.274*** | –0.032*** | –0.039*** |
(0.017) | (0.010) | (0.003) | (0.001) | |
Other non-Christian religion at matriculation | 0.701*** | 0.196*** | 0.270*** | –0.019*** |
(0.019) | (0.006) | (0.007) | (0.002) | |
No religious identity at matriculation | 0.251*** | 0.060*** | N/A | N/A |
(0.004) | (0.001) | N/A | N/A |
Variable | Model | |||
(1) Change in convictions | (2) Change in attendance | (3) Change in identity | ||
Changed identity | Lost identity | |||
Male | 0.803*** | 0.786*** | 0.001 | 0.011*** |
(0.009) | (0.009) | (0.001) | (0.001) | |
African-American | 1.681*** | 1.288*** | –0.050*** | –0.020*** |
(0.041) | (0.031) | (0.004) | (0.003) | |
Hispanic | 1.122*** | 0.896*** | –0.024*** | –0.004 |
(0.031) | (0.025) | (0.005) | (0.004) | |
Asian | 1.018 | 0.562*** | 0.011*** | 0.005 |
(0.026) | (0.015) | (0.004) | (0.004) | |
American Indian | 0.888 | 0.696*** | 0.081*** | 0.025 |
(0.112) | (0.085) | (0.025) | (0.019) | |
Other race | 1.131*** | 0.719*** | 0.064*** | 0.028*** |
(0.044) | (0.028) | (0.008) | (0.006) | |
Multiracial | 1.015 | 0.807*** | 0.040*** | 0.025*** |
(0.025) | (0.020) | (0.004) | (0.004) | |
Age at matriculation | 0.999 | 1.003 | –0.006*** | –0.009*** |
(0.011) | (0.013) | (0.002) | (0.001) | |
Age at graduation | 1.073*** | 1.030*** | 0.013*** | 0.008*** |
(0.006) | (0.006) | (0.001) | (0.001) | |
Part-time student | 1.007 | 1.102*** | 0.003 | 0.001 |
(0.025) | (0.029) | (0.003) | (0.003) | |
International student | 0.927* | 0.814*** | –0.024*** | 0.006 |
(0.036) | (0.035) | (0.004) | (0.004) | |
Native language is English | 0.924*** | 1.024 | –0.044*** | 0.004* |
(0.021) | (0.025) | (0.004) | (0.002) | |
Family income ($0,000) | 0.998*** | 0.998*** | –0.001*** | 0.001 |
(0.000) | (0.000) | (0.000) | (0.001) | |
Student loans ($0,000) | 1.001*** | 0.998*** | 0.001** | 0.001*** |
(0.000) | (0.000) | (0.000) | (0.000) | |
Mother’s education (years) | 0.994** | 0.981*** | 0.001* | 0.001** |
(0.003) | (0.003) | (0.000) | (0.000) | |
Father’s education (years) | 0.989*** | 0.992** | –0.001*** | –0.001 |
(0.003) | (0.003) | (0.000) | (0.001) | |
First-generation college student | 0.962** | 0.952** | –0.001 | –0.006*** |
(0.018) | (0.020) | (0.002) | (0.002) | |
Parents divorced | 1.027* | 0.864*** | 0.013*** | 0.008*** |
(0.015) | (0.013) | (0.002) | (0.002) | |
One or more parents deceased | 0.997 | 0.965 | 0.009** | 0.002 |
(0.029) | (0.031) | (0.004) | (0.003) | |
Distance: college to home | 1.000 | 0.999** | 0.001** | 0.001*** |
(0.003) | (0.000) | (0.000) | (0.000) | |
Attending first choice college | 1.079*** | 1.125*** | –0.009*** | –0.004** |
(0.023) | (0.026) | (0.003) | (0.002) | |
Attending second choice college | 0.988 | 1.042 | –0.008*** | –0.002 |
(0.023) | (0.026) | (0.003) | (0.002) | |
Non-traditional student | 0.899*** | 1.133*** | –0.015*** | –0.001 |
(0.030) | (0.042) | (0.004) | (0.003) | |
High school GPA | 0.906*** | 1.279*** | –0.015*** | 0.012*** |
(0.013) | (0.020) | (0.002) | (0.002) |
References
Akerlof, G. A. , and R. E.Kranton. 2000. “Economics and Identity.” The Quarterly Journal of Economics115 (3):715–53.10.1162/003355300554881Search in Google Scholar
Azzi, C. , and R.Ehrenberg. 1975. “Household Allocation of Time and Church Attendance.” The Journal of Political Economy83 (1):27–56.10.1086/260305Search in Google Scholar
Eswaran, M. 2011. “Competition and Performance in the Marketplace for Religion: A Theoretical Perspective.” The BE Journal of Economic Analysis & Policy11 (1). http://www.degruyter.com/view/j/bejeap.2011.11.issue-1/bejeap.2011.11.1.2723/bejeap.2011.11.1.2723.xmlSearch in Google Scholar
Gaviria, A. , and S.Raphael. 2001. “School-Based Peer Effects and Juvenile Behavior.” Review of Economics and Statistics83 (2):257–68.10.1162/00346530151143798Search in Google Scholar
Gervais, W. M. , and A.Norenzayan. 2012. “Analytic Thinking Promotes Religious Disbelief.” Science336 (6080):493–6.10.1126/science.1215647Search in Google Scholar
Glaeser, E. L. , and B. I.Sacerdote. 2008. “Education and Religion.” Journal of Human Capital2 (2):188–215.10.1086/590413Search in Google Scholar
Gruber, J. H . 2005. “Religious Market Structure, Religious Participation, and Outcomes: Is Religion Good for You?” Advances in Economic Analysis & Policy5 (1). http://www.degruyter.com/view/j/bejeap.2005.5.issue-1/bejeap.2005.5.1.1454/bejeap.2005.5.1.1454.xml10.2202/1538-0637.1454Search in Google Scholar
Hout, M. , and C. S.Fischer. 2002. “Why More Americans Have No Religious Preference: Politics and Generations.” American Sociological Review67 (2):165–90.10.2307/3088891Search in Google Scholar
Hummer, R. A. , R. G.Rogers, C. B.Nam, and C. G.Ellison. 1999. “Religious Involvement and US Adult Mortality.” Demography36 (2):273–85.10.2307/2648114Search in Google Scholar
Hungerman, D. M. 2014. “The Effect of Education on Religion: Evidence from Compulsory Schooling Laws.” Journal of Economic Behavior & Organization104:52–63.10.1016/j.jebo.2013.09.004Search in Google Scholar
Iannaccone, L. R. 1998. “Introduction to the Economics of Religion.” The Journal of Economic Literature36 (3):1465–96.Search in Google Scholar
Kimball, M. S. , C. M.Mitchell, A. D.Thornton, and L. C.Young-Demarco. 2009. “Empirics on the Origins of Preferences: The Case of College Major and Religiosity.” National Bureau of Economic Research Working Paper No. w15182.10.3386/w15182Search in Google Scholar
Lee, J. J. 2002. “Religion and College Attendance: Change among Students.” The Review of Higher Education25 (4):369–84.10.1353/rhe.2002.0020Search in Google Scholar
Lehrer, E. L. , and C. U.Chiswick. 1993. “Religion as a Determinant of Marital Stability.” Demography30 (3):385–404.10.2307/2061647Search in Google Scholar
McCleary, R. M. , and R. J.Barro. 2006. “Religion and Economy.” The Journal of Economic Perspectives20 (2):49–72.10.1257/jep.20.2.49Search in Google Scholar
Mocan, N. , and L.Pogorelova.2014. “Compulsory Schooling Laws and Formation of Beliefs: Education, Religion and Superstition.” National Bureau of Economic Research Working Paper No. w20557.Search in Google Scholar
Mooney, M. 2010. “Religion, College Grades, and Satisfaction among Students at Elite Colleges and Universities.” Sociology of Religion71 (2):197–215.10.1093/socrel/srq035Search in Google Scholar
Schwadel, P. 2011. “The Effects of Education on Americans’ Religious Practices, Beliefs, and Affiliations.” Review of Religious Research53 (2):161–82.10.1007/s13644-011-0007-4Search in Google Scholar
©2015 by De Gruyter