ARTICLE IN PRESS

Social Science & Medicine 57 (2003) 20492054

Public and private domains of religiosity and adolescent health risk behaviors: evidence from the National Longitudinal Study of Adolescent Health
James M. Nonnemakera,*, Clea A. McNeelyb, Robert Wm. Blumb
a

Research Triangle Institute International, Health, Social, and Economic Research, 3040 Cornwallis Road, P.O. Box 12194, Research Triangle Park, NC 22709-2194, USA b Division of General Pediatrics and Adolescent Health, University of Minnesota, USA

Abstract The purpose of this study was to examine the association of public and private domains of religiosity and adolescent health-related outcomes using data from the National Longitudinal Study of Adolescent Health (Add Health), a nationally representative sample of American adolescents in grades 712. The public religiosity variable combines two items measuring frequency of attendance at religious services and frequency of participation in religious youth group activities. The private religiosity variable combines two items measuring frequency of prayer and importance of religion. Our results support previous evidence that religiosity is protective for a number of adolescent health-related outcomes. In general, both public and private religiosity was protective against cigarettes, alcohol, and marijuana use. On closer examination it appeared that private religiosity was more protective against experimental substance use, while public religiosity had a larger association with regular use, and in particular with regular cigarette use. Both public and private religiosity was associated with a lower probability of having ever had sexual intercourse. Only public religiosity had a signicant effect on effective birth control at rst sexual intercourse and, for females, for having ever been pregnant. However, neither dimension of religiosity was associated with birth control use at rst or most recent sex. Public religiosity was associated with lower emotional distress while private religiosity was not. Only private religiosity was signicantly associated with a lower probability of having had suicidal thoughts or having attempted suicide. Both public and private religiosity was associated with a lower probability of having engaged in violence in the last year. Our results suggest that further work is warranted to explore the causal mechanisms by which religiosity is protective for adolescents. Needed is both theoretical work that identies mechanisms that could explain the different patterns of empirical results and surveys that collect data specic to the hypothesized mechanisms. r 2003 Elsevier Science Ltd. All rights reserved.
Keywords: USA; Religiosity; Adolescents; Risk behaviors; Substance use; Sexual behavior

Introduction In recent years there has been growing interest in the role religion may play in reducing harm (e.g., Resnick et al., 1997) and contributing to the resilience of young people (e.g., Werner & Smith, 1992). Evidence from
*Corresponding author. Tel.: +1-919-541-7064; fax. +1919-541-6683. E-mail address: jnonnemaker@rti.org (J.M. Nonnemaker).

national survey data for the United States also has consistently indicated that religion is important to American adolescents (Bridges & Moore, 2002; Donahue & Benson, 1995; Johnston, Bachman, & OMalley, 1999; Regnerus, 2003). Recently, faith-based initiatives for addressing adolescent issues have received greater attention because of the Bush administrations policy to fund interventions for adolescents through faith-based communities (Bridges & Moore, 2002; Glazer, 2001; Miller & Gur, 2002).

0277-9536/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0277-9536(03)00096-0

ARTICLE IN PRESS
2050 J.M. Nonnemaker et al. / Social Science & Medicine 57 (2003) 20492054

Relationship of religiosity to adolescent health-related outcomes Religiosity has been found to be protective against participation in adolescent health risk behaviors. Higher levels of religiosity have been associated with lower levels of cigarette use (Amey, Albrecht, & Miller, 1996; Wallace Jr. & Forman, 1998), alcohol use (Amey et al., 1996; Cochran & Akers, 1989; Miller, Davies, & Greenwald, 2000; Wallace & Forman, 1998), marijuana use (Amey et al. 1996, Cochran & Akers, 1989; Miller et al., 2000; Wallace & Forman, 1998), and illicit drug use (Amey et al., 1996; Miller et al., 2000). A few studies have found that religiosity is related to lower levels of suicidal thoughts or attempts (Donahue & Benson, 1995; Zhang & Jin, 1996). A number of studies have also documented an association between greater religious involvement and lower levels of depression (see Koenig, McCullough, & Larson, 2001, for a summary). Evidence of the relationship between religiosity and violence is sparse, though available evidence would suggest that religiosity is protective (Wallace & Forman, 1998). Regarding sexual behavior, there is some evidence that while religiosity may delay the initiation of sexual intercourse (for a review, see Wilcox, Rostosky, Randall, & Wright, 2000), it might also be associated with less contraceptive use (Cooksey, Rindfuss, & Guilkey, 1996). Miller and Gur (2002) found that two indicators of religiosity, personal devotion and frequent religious service attendance, were associated with greater sexual responsibility, including responsible and planned birth control use. Conversely, an indicator of personal conservatism was associated with unprotected sex. The purpose of this paper is to explore the relationship between two dimensions of religiosity, public and private religiosity, and various measures of adolescent health and health risk behaviors.

Table 1 presents descriptions of the religiosity variables and the health-related outcomes as well as the descriptive statistics for these variables. Analytic strategy To examine the relationship between public and private religiosity and the health-related outcomes, we used OLS and logistic regression, controlling for age, gender, race, ethnicity (black, Latino, and other race/ ethnicity), family structure (single-parent biological family, step family, and other family structure), household size, household income, whether respondent is 2 years older than the average age in their grade, whether respondent is 2 years younger than the average age in their grade, whether respondent appears older or younger than his or her peers, modied PPVT score, and religious denomination.1 Analyses were done in Stata 6.0 using weights and adjusting for the complex sample design. We also statistically tested for the difference between the public and private religiosity coefcients within each model using a Wald test adjusted for sample design. The substance use categorizations and modeling strategies are motivated by models of the smoking uptake process (Choi, Pierce, Gilpin, Farkas, & Berry, 1997; Flay, 1993; Leventhal & Cleary, 1980; Mowery, Brick, & Farrelly, 2000). For cigarette and alcohol use, we modeled the effect of religiosity on the probability that an adolescent is an experimental or occasional substance user compared to being a non-user. Then, we examined the effect of religiosity on the probability of regular or problem use compared to experimental or occasional use. For marijuana use, we estimated the effect of religiosity on the probability of any use and then the effect on problem use conditional on any use.
1 We do not consider differences in the effects of our public and private religiosity variables on outcomes by religious afliation mostly because of concerns over the poor measurement of religious afliation in our data. Given the categorization of afliations in our data it is likely that there is signicant heterogeneity within religious afliations that would complicate any interpretation of results by afliation. That is, the variation within religious denomination could be greater than the variation between religious denominations on characteristics of the denomination that are related to outcomes (e.g., the proscriptiveness of denomination regarding specic behaviors). Matthews et al. (1999) discuss this issue in more detail. To the extent that public and private religiosity varies for adolescents by religious afliation, omitting an interaction between religious afliation and our religiosity variables or not stratifying by religious afliation is a limitation of this study. Note that we do control for different levels of the outcomes by afliation.

Methods Data and measures Data for this study were drawn from the National Longitudinal Study of Adolescent Health (Add Health), a nationally representative sample of American adolescents in grades 712 (Bearman, Jones, & Udry, 1997). For the present analysis we used data from the Wave 1 in-home sample with sampling weights (n 18; 924). Those adolescents who reported no religious afliation (n 2260) were excluded from the analysis because they were not asked to answer the religiosity questions. An additional 358 respondents who did not answer the religiosity questions were excluded. The sample size is therefore 16,306.

ARTICLE IN PRESS
J.M. Nonnemaker et al. / Social Science & Medicine 57 (2003) 20492054 Table 1 Weighted descriptive statistics for health-related outcomes, National Longitudinal Study of Adolescent Health Measures Religiositya Public religiosity Private religiosity Health-related outcomes Cigarette use in last 30 days Any Experimental (smoked 119 days) Regular (smoked 2030 days) Alcohol use in past 12 months Any Ever got drunk or very, very high on alcohol Occasional Got drunk or very, very high on alcohol 12 days in past 12 months up to once a month or less (regular users excluded) Regular Got drunk or very, very high on alcohol 23 days a month or more Marijuana use in past 30 days Any Regular (used more than 3 times) Sexual behaviors Ever had sex Birth control at 1st intercourse viz condom, birth control pills, Depo Provera, or diaphragm Birth control at most recent intercourse Ever pregnant (females only) Mental health Emotional distressb Suicide attempt Ever attempted suicide in the past 12 months Suicidal ideation Ever seriously thought about committing suicide in the past 12 months (among those who have not attempted suicide) Weapon-related violence in last 12 monthsc
a

2051

Mean/ proportion 2.59 3.16

SD

Minimum

Maximum

Unweighted N

1.01 0.92

1 0

4 4

16,300 16,301

0.26 0.14 0.13

0 0 0

1 1 1

16,197 16,197 16,197

0.28 0.20

0 0

1 1

16,146 14,576

0.10

16,146

0.13 0.06

0 0

1 1

16,061 16,061

0.36 0.63

0 0

1 1

16,174 6275

0.65 0.19

0 0

1 1

6275 3080

8.65 0.04 0.09

6.83

0 0 0

51 1 1

16,298 16,306 15,614

0.22

16,242

Our measurement strategy was based on the available items for measuring religiosity in Add Health. Public religiosity was the mean response to 2 questions about the past 12 months: How often did you attend religious services? and, Many churches, synagogues, and other places of worship have special activities for teenssuch as youth groups, Bible classes, or choir. How often did you attend such religious activities? Responses range from once a week or more (4) to never (1). Private religiosity was the mean response to two questions: How important is religion to you? (responses range from very important (4) to not important at all (1)) and How often do you pray? (responses range from at least once a day (5) to never (1)). b Mean of responses to 17 questions asking how often in the past week or year the adolescent felt depressed, lonely, sad, fearful, moody, had cried or had a poor appetite. Responses range from never or rarely (1) to most or all of the time (4) (a 0:87). c In last year committed at least one of the following acts: threatened to use a weapon to get something from someone, pulled a knife or gun on someone, shot or stabbed someone, used a weapon in a ght, or hurt someone badly enough to need bandages or medical care.

ARTICLE IN PRESS
2052 J.M. Nonnemaker et al. / Social Science & Medicine 57 (2003) 20492054

Results Substance use The results from the models examining the relationship between public and private religiosity and substance use outcomes are summarized in Table 2. When the outcome was dened as any use, i.e., any cigarette use, any alcohol use, or any marijuana use, both public and private religiosity was equally protective. Higher levels of public and private religiosity were associated with lower levels of any use and the coefcients for public and private religiosity were not statistically different. For experimental use (compared to no use) of cigarettes and
Table 2 Weighted logit coefcients (SE) for the effect of public and private religiosity on substance use, National Longitudinal Study of Adolescent Health (A) Cigarette use Public religiosity Private religiosity N (B) Alcohol use Public religiosity Private religiosity N Any use 0.19*** (0.03) 0.21*** (0.04) 14,719 Any use 0.08* (0.04) 0.17 (0.04) 8320 Experimental use Regular use 0.07 (0.04) 0.19*** (0.04) 13022 Occasional use 0.06 (0.04) 0.15** (0.04) 6868 0.28***a (0.06) 0.07 (0.06) 3681 Problem use 0.11* (0.06) 0.06 (0.06) 4105 Problem use 0.14 (0.07) 0.01 (0.07) 1970

alcohol, private religiosity was protective but public religiosity was not, though the coefcients were not statistically signicant. When we examined regular substance use, conditional on being a user, public religiosity was associated with less use, whereas private religiosity is not; the difference was signicant for regular cigarette use. Sexual behaviors Table 3 shows that both public and private religiosity were signicantly associated with lower levels of ever having had sex, public religiosity having a signicantly stronger relation. Only public religiosity had a signicant effect on effective birth control use at rst sexual intercourse. Neither public nor private religiosity was associated with birth control use at most recent intercourse. Among females, public religiosity was protective for having ever been pregnant, but private religiosity had no signicant effect. Mental health and violence Table 4 presents the results for the association between public and private religiosity and the three mental health outcomes. Higher values of public religiosity are associated with lower levels of emotional distress, whereas private religiosity had no signicant effect. However but the difference between coefcients was not signicant. Private religiosity, but not public religiosity, was signicantly associated with the suicidal ideation and suicide attempt variables. Private but not public religiosity was associated with involvement in weapon-related violence in the last year, but the difference was not signicant.

(C) Marijuana use Any use Public religiosity Private religiosity N 0.24*** (0.04) 0.23*** (0.04) 14,643

Discussion The purpose of this study was to examine the association of religiosity and adolescent health-related outcomes. In general, both public and private religiosity

po0:05; po0:01; po0:001: a Difference of religiosity coefcients po0:05:

Table 3 Weighted logit coefcients (SE) for the effect of public and private religiosity on sexual behaviors, National Longitudinal Study of Adolescent Health Ever had sex Public religiosity Private religiosity N 0.27***a (0.03) 0.15*** (0.04) 16,163 Effective birth control at rst sex 0.03 (0.04) 0.11*a (0.05) 5753 Effective birth control at last sex 0.05 (0.04) 0.03 (0.05) 5753 Ever pregnant (females only) 0.22**a (0.07) 0.00 (0.08) 2860

po0:05; po0:01; po0:001: a Difference of religiosity coefcients, po0:05:

ARTICLE IN PRESS
J.M. Nonnemaker et al. / Social Science & Medicine 57 (2003) 20492054 Table 4 Weighted coefcients (SE) for the effect of public and private religiosity on mental health and weapon-related violence Mental Health Emotional distress (OLS coefcients) Public religiosity Private religiosity R2 N 0.26** (0.08) 0.12 (0.12) 0.08 14,805 Suicidal ideation (logit coefcients) 0.08 (0.04) 0.14** (0.04) 14,262 Suicide attempt (logit coefcients) 0.07 (0.08) 0.26**a (0.08) 14,809 Weapon-related violence Weapon-related violence in last year (logit coefcients) 0.06 (0.03) 0.12** (0.03) 14,801 2053

*po0:05; po0:01; po0:001: a Difference of religiosity coefcients, po0:05:

is protective against any use of cigarettes, alcohol, and marijuana. These results are consistent with the general consensus in the literature that higher levels of religiosity are associated with lower levels of substance use (Amey et al., 1996; Wallace & Forman, 1998; Cochran & Akers, 1989; Miller et al., 2000). On closer examination it appears that private religiosity is more protective against initiating substance use (assuming that the probability of being in the experimental category is related to initiation), whereas public religiosity has a larger association with regular use, and in particular with regular cigarette use. The different pattern of results for public and private religiosity on experimental and regular smoking outcomes suggests the need for more work, both theoretical and empirical, to understand this process. Public religiosity and to a lesser extent private religiosity were associated with a lower probability of having ever had sex. This is consistent with prior research (Wilcox et al., 2000). Public, but not private, religiosity is also associated with a lower probability of ever having been pregnant, conditional on having had sex. Private religiosity was modestly associated with use of an effective birth control method at rst sexual intercourse, but neither religiosity variable was associated with effective birth control at most recent sex. Thus, it would appear that religiosity has a protective effect on the probability an adolescent has had sex, but once the decision is made to have sex there is little association with birth control use. Given this result, it is somewhat surprising that public religiosity appears to be protective against pregnancy. This may be due to less than truthful responses from those who are highly religious, perhaps because they do not want to disclose abortions. Alternatively, adolescents who frequently attend religious services or youth group activities could have sex less frequently. Frequency of sexual intercourse was not measured in Add Health. A third explanation that cannot be ruled out by the cross-sectional analysis presented here is that adolescents attend religious

services or youth groups more frequently after becoming pregnant. It is not surprising to nd little relation between religiosity and birth control use. Whereas most religious denominations have consistent values about sex, specically, that it is better to delay rst sexual intercourse, there are not consistent norms for contraception across religions. Cooksey et al. (1996) found that birth control use varies by denomination, perhaps reecting the lack of a consistent norm. We found that public religiosity was associated with lower emotional distress but, somewhat surprisingly, private religiosity was not. For suicidal ideation and suicide attempts, in contrast, public religiosity was not protective but private religiosity was (see Koenig et al., 2001, Chapters 8 and 15; Stark, Doyle, & Rushing, 1983). While public religiosity increases opportunities for social support, private religiosity may increase selfesteem or a sense of self-efcacy, as well as openness to social support. Consistent with previous research (Wallace & Forman, 1998), and current theories of social control and social learning, we found that both public and private religiosity were protective for violence. In general, our results support previous evidence that religiosity is protective for a number of adolescent health risk behaviors and outcomes. We found evidence for independent effects for both public and private religiosity for some health-related outcomes. But for other outcomes only one of the two dimensions of religiosity had a signicant protective effect. Without a clear theoretical model that makes testable empirical predictions it is impossible to suggest specic causal mechanisms. Similarly, the cross-sectional design limits our ability to make any causal conclusions.

Acknowledgements This study was supported by a grant (No. 1999-00218) of the Charles Stuart Mott foundation.

ARTICLE IN PRESS
2054 J.M. Nonnemaker et al. / Social Science & Medicine 57 (2003) 20492054 havior risk modication. Psychological Bulletin, 88, 370405. Matthews, D. A., McCollough, M. E., Swyers, J. P., Milano, M. G., Larson, D. B., & Koenig, H. G. (1999). Religious commitment and health studies. Archives of Family Medicine, 8, 76. Miller, L., Davies, M., & Greenwald, S. (2000). Religiosity and substance use and abuse among adolescents in the National Comorbidity Survey. Journal of the American Academy of Child and Adolescent Psychiatry, 39, 11901197. Miller, L., & Gur, M. (2002). Religiousness and sexual responsibility in adolescent girls. Journal of Adolescent Health, 31, 401406. Mowery, P. D., Brick, P. D., & Farrelly, M. C. (2000). Legacy rst look report 3. Pathways to established smoking: Results from the 1999 National Youth Tobacco Survey. Washington, DC: American Legacy Foundation. Regnerus, M. D. (2003). Religion and positive adolescent outcomes: A review of research and theory. Review of Religious Research, 44(4) in press. Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman, K. E., Harris, K. M., & Jones, J., et al. (1997). Protecting adolescents from harm: Findings from the National Longitudinal Study of Adolescent Health. Journal of the American Medical Association, 278(10), 823832. Stark, R., Doyle, D., & Rushing, J. (1983). Beyond Durkheim: Religion and suicide. Journal for the Scientic Study of Religion, 22, 120131. Wallace Jr., J. M., & Forman, T. A. (1998). Religions role in promoting health and reducing risk among American youth. Health Education & Behavior, 25, 721741. Werner, E., & Smith, R. (1992). Overcoming the odds: High-risk children from birth to adolescence. Ithaca, NY: Cornell University Press. Wilcox, B. L., Rostosky, S. S., Randall, B., & Wright, M. L. C. (2000). Adolescent religiosity and sexual behavior: A research review. National Campaign to Prevent Teen Pregnancy: Washington, DC (unpublished manuscript.). Zhang, J., & Jin, S. (1996). Determinants of suicide ideation: A comparison of Chinese and American college students. Adolescence, 31, 451467.

References
Amey, C. H., Albrecht, S. L., & Miller, M. K. (1996). Racial differences in adolescent drug use: The impact of religion. Substance Use and Misuse, 31, 13111332. Bearman, P. S., Jones, J., & Udry, J. R. (1997). The National Longitudinal Study of Adolescent Health: Research design. Available at http://www.cpc.unc.edu/addhlth. Bridges, L. J., & Moore K. A. (2002). Religious involvement and childrens well-being: What research tells us (and what it doesnt). Child trends research brief. Child Trends: Washington, DC. Available at http://www.childtrends.org. Choi, W. S., Pierce, J. P., Gilpin, E. A., Farkas, A. J., & Berry, C. C. (1997). Which adolescent experimenters progress to established smoking in the United States. American Journal of Preventive Medicine, 13, 385391. Cochran, J. K., & Akers, R. L. (1989). Beyond hellre: An exploration of the variable effects of religiosity on adolescent marijuana and alcohol use. Journal of Research in Crime and Delinquency, 26, 198225. Cooksey, E. C., Rindfuss, R. R., & Guilkey D, K. (1996). The initiation of adolescent sexual and contraceptive behavior during changing times. Journal of Health and Social Behavior, 37(1), 5974. Donahue, M. J., & Benson, P. L. (1995). Religion and the well-being of adolescents. Journal of Social Issues, 51, 145160. Flay, B. R. (1993). Youth tobacco use: Risks, patterns, and control. In J. Slade, & C. T. Orleans (Eds.), Nicotine addiction: Principles and management. New York: Oxford University Press. Glazer, S. (2001). Faith based initiatives. Congressional Quarterly Researcher, 11, 377400. Johnston, L. D., Bachman, J. G., & OMalley, P. M. (1999). Monitoring the future: Questionnaire responses from the nations high school seniors. Ann Arbor: Institute for Social Research. Koenig, H. G., McCullough, M. E., & Larson, D. B. (2001). Handbook of religion and health. Oxford: Oxford University Press. Leventhal, H., & Cleary, P. D. (1980). The smoking problem: A review of the research and theory in be-