Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Mar 1.
Published in final edited form as: Addict Behav. 2020 Nov 24;114:106739. doi: 10.1016/j.addbeh.2020.106739

Patterns of Bi-directional Relations Across Alcohol use, Religiosity, and Self-Control in Adolescent Girls

Meredith H Palm a,*, Shawn J Latendresse a, Tammy Chung b, Alison E Hipwell c, Carolyn E Sartor d
PMCID: PMC8341452  NIHMSID: NIHMS1657009  PMID: 33307407

Abstract

Examining predictors of alcohol use among adolescent girls is increasingly important to enhance prevention efforts, given the gender gap in alcohol use is steadily closing. While both religiosity and self-control have been independently associated with decreased alcohol use, little research has explored 1) whether religiosity and self-control are reciprocally related and 2) whether the reciprocal association between these constructs may indicate different patterns in the development of alcohol use. As such, this study examined whether there are multiple patterns of reciprocal relationships across religiosity, self-control, and alcohol use among adolescent girls. Latent variable mixture modeling was combined with an autoregressive cross-lagged panel model to identify discrete, prototypical patterns of longitudinal associations (i.e., subgroups) across religiosity, self-control, and alcohol use among 2,122 girls ages 13–17. Psychosocial covariates (e.g., conduct problems) were examined as predictors of each subgroup. Two subgroups were identified. Self-control was associated with reduced alcohol use in both the majority (87.56% of the sample) and minority (12.44% of the sample) subgroups, but only the majority subgroup also demonstrated associations between religiosity, self-control and alcohol use. Religiosity may predict lower alcohol use in a majority of adolescent girls but this association may not be present among all girls, suggesting that there is a qualitative difference in how religiosity is associated with self-control and alcohol use between subgroups. Results also suggest that higher levels of conduct problems may predict which girls are more likely to demonstrate associations between only self-control and alcohol use, and demonstrate no significant associations with religiosity.

Keywords: alcohol, girls, adolescence, religiosity, self-control, mixture modeling

1. Introduction

Adolescence is an important developmental period in which to study alcohol use, as alcohol use is typically initiated during this time and even relatively low levels of use are associated with an increased risk for physical and social consequences (Marshall, 2014; Schulenberg et al., 2014). Alcohol is the most common substance used by adolescents, but studies of alcohol use specifically in girls have been less prevalent (Johnston et al., 2017). While boys have typically consumed higher levels of alcohol than girls, adolescent girls now have equivalent or higher levels of alcohol use, including binge drinking, compared to their same-age male counterparts, indicating that special attention to alcohol use among adolescent girls is necessary (Johnston et al., 2017).

Religiosity and self-control have been identified as protective factors that may reduce one’s likelihood of using alcohol (McCullough & Willoughby, 2009). A review of the literature on the relationship between religiosity, “formal, institutional, and outward expression of the sacred” (Cotton et al., 2006, p. 472) and adolescent health suggested that overall religiousness was related to lower frequency of alcohol use (Cotton et al., 2006). Religiosity has also been related to lower adolescent alcohol use across multiple definitions of both alcohol use (e.g., quantity/frequency, binge drinking) and religiousness, including religious service attendance and importance of religion (Jankowski et al., 2013; Salas-Wright et al., 2014). Religiosity may be an especially important protective factor to address in girls given their higher levels of religiosity compared to boys’ (Miller & Stark, 2002; Wink et al., 2007).

In addition, self-control, the ability to inhibit inappropriate responses or adapt one’s behavior to the demands of a situation, has been linked to lower alcohol use (McCullough & Willoughby, 2009). Using behavioral and emotional self-regulatory strategies specifically in social situations (e.g., avoiding situations where alcohol may be available, having an excuse to turn down a drink) has been associated with lower levels of alcohol use across a variety of alcohol outcomes (D’Lima et al., 2012; Glassman et al., 2007). As such, both religiosity and self-control have independently been associated with lower levels of alcohol use.

Not only are religiosity and self-control associated with alcohol use, but prior research suggests religiosity and self-control are associated with each other. McCullough and Willoughby (2009) have provided an extensive review of the relationships between these constructs. Experiments that subliminally exposed participants to religious passages or words suggest that religiosity is associated with higher levels of self-control (Fishbach et al., 2003; Rounding et al., 2012; Watterson & Giesler, 2012). Even though this experimental research implies that religiosity may lead to higher levels of self-control, additional research suggests that one’s traits (including self-control) may predict later religiosity (McCullough & Carter, 2013; McCullough & Willoughby, 2009). Personality factors (e.g., conscientiousness, agreeableness) indicative of self-control predict adult religiosity even after controlling for adolescent religiosity, suggesting that individuals with higher levels of self-control may be more likely to self-select into religious participation (McCrae & Costa, 1987; McCullough et al., 2003; McCullough & Willoughby, 2009; Wink et al., 2007). The collective results of these studies suggest bidirectional influences between religiosity and self-control. However, longitudinal research examining reciprocal relations between religiosity and self-control in adolescents is scarce. To the best of our knowledge, extant research is limited to a single study, with a selective sample of primarily (86.4%) male justice system-involved adolescents, that found higher levels of religiosity were associated with lower levels of criminal behavior, and that this relationship was partially mediated by self-control (Pirutinsky, 2014). Examining the simultaneous influences of religiosity and self-control may provide insight into whether religiosity and self-control have unique or distinct roles in predicting alcohol use. If bidirectional relationships are present between religiosity and self-control, and both are associated with lower levels of alcohol use, it may suggest that both constructs are areas for prevention and intervention strategies.

However, there may be unique developmental differences in the extent to which these constructs are related. Developmental systems theories suggest that reciprocal interactions across multiple constructs within an individual and her environment occur over time; that is, the bidirectional association between religiosity and self-control may be indicative of unique developmental processes (Ford & Lerner, 1992). By examining the person as the unit of analysis, we may find evidence for distinctive patterns of associations across religiosity, self-control, and alcohol use that adequately characterize some individuals, but not others. For instance, prior research by our group suggests that there are multiple patterns of bidirectional associations between parental monitoring and alcohol use in adolescent girls, providing initial evidence that a single pattern of cross-construct associations may not adequately model developmental relationships equally well for all individuals within a sample (Latendresse et al., 2017). Further, prior research has suggested that unique patterns of religiosity demonstrate differential effects on alcohol use; that is, the effect of religiosity on alcohol use within a population may not be uniform and religiosity may not be protective against alcohol use for some subgroups of adolescents (Hoyland et al., 2017). While we expect religiosity to be positively associated with self-control and negatively associated with alcohol use among most girls, it may be that there are subgroups of individuals for whom religiosity is not associated with self-control or alcohol use.

Further, individuals who are likely to demonstrate unique patterns of associations across religiosity, self-control, and alcohol use may be differentiated by other psychosocial characteristics. Conduct problems are correlated with lower levels of religiosity and higher levels of alcohol use (King & Boyatzis, 2015; Kuperman et al., 2013), while Black families tend to be more religious than White families (Taylor et al., 1996). It may be that for individuals who already demonstrate conduct problems, religiosity is not a sufficient protective factor for alcohol use, but higher levels of self-control may provide a protective effect. However, race may indicate a variety of cultural practices that predict patterns of behavior characterized by lower levels of alcohol use and higher levels of religiosity. These factors, combined with other predictors of alcohol use including poor parental supervision (Latendresse et al., 2017), early pubertal development (Biehl et al., 2007), and low socioeconomic status (Green et al., 2013) may differentiate individuals likely to demonstrate different patterns of associations across religiosity, self-control, and alcohol use. The goal of the present study is to identify these patterns of cross-construct associations and the psychosocial characteristics that predict likelihood of demonstrating these patterns of associations.

2. Method

2.1. Sample

Data were drawn from the Pittsburgh Girls Study (PGS), a multi-cohort community sample of adolescent girls, assessed annually in an accelerated longitudinal design (Hipwell et al., 2002; Keenan et al., 2010). The PGS consists of 2,450 girls and their parents, initially sampled in 1999–2000. At enrollment, girls were between the ages of 5–8, and represented four age cohorts. Families living in low-income neighborhoods were over sampled (Hipwell et al., 2002; Keenan et al., 2010) and 85% of families who were contacted during recruitment consented to participation (Keenan et al., 2010). Participant retention has been high, on average 88.5% through ages 13–17. In order to facilitate interpretation of possible racial differences in substance use, as has been done in previous studies using PGS data (Latendresse et al., 2017), we excluded 145 girls whose primary caregiver identified them as members of a multi-racial or an “other” race group, only retaining participants who identified as Black or White. We excluded another 183 participants who were missing data on religiosity, self-control, and alcohol use variables at all time points between ages 13–17. The final analytic sample therefore was comprised of 2,122 girls (57.4% Black, 42.6% White).

Approval for all study procedures was obtained from the University of Pittsburgh Institutional Review Board. Written informed consent from the caregiver and verbal assent from the child were obtained prior to data collection. Interviews were conducted separately for the parent and child in the home by trained interviewers using a laptop computer.

2.2. Measures

2.2.1. Religiosity.

Religiosity was assessed annually across ages 13–17 using four items. Girls reported their frequency of prayer (0 = “Never” to 3 = “About once a day”) and frequency of attendance at religious services (0 = “Never” to 3 = “More than one time a week”) in the past year. Girls also indicated how important religion was to them (0 = “Not important” to 2 = “Very important”) and how frequently they engaged in other religious activities such as youth group or choir (recoded to 0 = “Never” to 5 = “Just about every day”). Due to different response scales across the four items, we created a composite indicator of religiosity by standardizing each item and then summing the standardized items (99.9% of the sample responded to all religiosity items within age). Internal consistency of the standardized items was acceptable at all ages (α = .73–.78). The means of the standardized indicators ranged from 0.00 to 0.12, standard deviations ranged from 2.96 to 3.10, and the range of the indicators were from −6.41 to 7.62 across ages 13–17 (see Supplementary Table S1 for details on this indicator). While an indicator of girls’ religious affiliation was unavailable in this data, primary caregivers reported their religious affiliation which is commonly used as an indicator of adolescent religiosity given the correlation between adolescent and parents’ religiosity (Boyatzis, 2009).

2.2.2. Self-control.

Self-control was assessed at ages 13–17 by self-report on the Social Skills Rating Scale (SSRS; Gresham & Elliott, 1990). Responses to 8 items (e.g., “I control my temper when people are angry with me”) reported on 3-point Likert scales (0 = “Often” to 2 = “Never” scale) were summed. Internal consistency was acceptable at each age (α = .69–.73). The SSRS was removed from the PGS battery in annual wave 11 and so was not administered to cohort 5 girls (aged 5 in PGS wave 1) at ages 16 and 17 (23.6% of full sample), or to cohort 6 girls at age 17 (25.7% of full sample). Thus, all cohorts provided self-control data at ages 13–15, only cohorts 6–8 provided self-control data at age16, and only cohorts 7–8 provided self-control data at age 17. One-way analysis of variance suggested that there was no difference in self-control at ages 13–15 (for which data were available for all cohorts) among girls who were ages 5–8 at study inception (see Supplementary Table S2).

2.2.3. Alcohol.

Alcohol use was assessed annually across ages 13–17 via the Nicotine, Alcohol, and Drug Substance Use self-report measure (Pandina et al., 1984). Girls indicated how frequently they consumed beer, wine, and liquor, including sips or tastes, over the past year on a scale from 0 = “I did not within the past year” to 7 = “Every day or more than once a day” and the reason they drank. We examined any level of past-year alcohol use in order to capture alcohol in any context at any level of use. These items were summarized into a single item representing alcohol use in the past year with 0 indicating no alcohol use and 1 indicating any use. As participants were not directly asked whether they had ever consumed a full drink of alcohol, we created a proxy for full drink, defined as consuming alcohol in any quantity 5 or more times in the past year or consuming 1–2 drinks per occasion at any frequency in the past year to compare relative frequencies of any use compared to a full drink of alcohol.

2.2.4. Psychosocial Characteristics.

All psychosocial covariates were assessed at age 13. Primary caregivers’ reports of receipt of public assistance (0 = “No” or 1 = “Yes”), their highest level of education (0 = more than 12 years or 1 = less than 12 years), and single parent-headed household status (0 = “No” or 1 = “Yes”) were used as indicators of low socioeconomic status (Loeber et al., 1998). Early puberty (i.e., menarche by age 11) was assessed by girl’s self-report using a single item from the Pubertal Development Scale (0 = “No” or 1 = “Yes”; Petersen et al., 1988). An indicator of conduct problem severity was created from the Adolescent Symptom Inventory-4 (ASI-4; Gadow & Sprafkin, 1999). Adolescents rated their frequency of 15 DSM-IV conduct disorder symptoms on a scale from 0 = “Never” to 3 = “Very often” and responses were summed to create a severity score (α = .77). An indicator of poor parental supervision was created from the sum of child responses to four items on the Supervision Involvement Scale (SIS) reported on a 3 = “Almost Never” to 1 = “Almost Always” scale (α = .60), with higher scores indicating lower supervision (Loeber et al., 1998).

2.3. Analytic Plan

Analyses were conducted using Mplus version 8.2 (Muthén & Muthén, 1998). We followed the procedure outlined by Latendresse et al. (2017) to combine an autoregressive cross-lagged path model with latent variable mixture modeling (LVMM). LVMM assumes that parameters for a specific process may vary within the population such that the data may be best represented by discrete patterns, each with a unique set of parameter estimates. We conducted LVMM to iteratively assess the fit of k versus k-1 discrete patterns (e.g., whether two distinct sets of parameters describe the longitudinal relationships among religiosity, self-control, and alcohol use better than a single set of parameters, and if so, whether three sets are better than two, etc.), using standard model fit statistics – Bayesian Information Criteria (BIC; Schwartz, 1978), entropy (Pastor et al., 2007), and the Vuong-Lo-Mendell-Rubin Likelihood Ratio test (VLMR; Lo et al., 2001; Vuong, 1989) – to assess whether extracting an additional pattern resulted in a significant decrement in model fit. Once the appropriate number of patterns was identified, Wald Tests were used to evaluate whether certain sets of within-pattern autoregressive and cross-lagged paths could be constrained to equality. Lastly, posterior probabilities of the resulting patterns were regressed onto a set of relevant covariates to determine whether they differentially characterized the discrete patterns. Although maximum likelihood estimation allows for missingness on endogenous variables (i.e., those used in the LVMM), it excludes all cases with missing data on exogenous covariates. As such, we conducted multiple imputation on the covariates to impute missing values on 94 cases, resulting in 10 complete datasets on which the final analyses were run.

3. Results

3.1. Preliminary Analyses

Comparisons in demographic characteristics between those excluded and retained in the analytic sample are available in Table 1. As a preliminary model-building step, we fit a model that estimated all autoregressive and cross-lagged paths among assessments of religiosity, self-control, and alcohol use from ages 13–17. Details on this initial model are available in supplementary material. This initial model (BIC = 97703.06; see Supplementary Figure S1) indicated small but positive cross-lagged associations between religiosity and self-control, negative associations between religiosity and alcohol use, and a negative effect of self-control on alcohol use, but not the reverse.

Table 1.

Comparisons Between the Analytic Sample and Excluded Cases on Demographic Variables

Variable Analytic Sample (n = 2,122)
% or Mean (SD)		 Excluded Cases (n = 328)
% or Mean (SD)		 t(df) or
χ2(df)
Race (Black) 57% 42% 16.17(1) ***
Low Caregiver Education 14% 10% 0.39(1)
Public Assistance 37% 10% 8.89(1) **
Single Parent Household 44% 28% 3.16(1)
Early puberty 29% 85% 236.75(1) ***
Conduct Problem Severity 0.97 (1.76) 0.44 (0.65) −3.91(28.44) **
Poor Parental Supervision 4.62 (1.10) 4.63 (1.46) 0.02(2,058)
Religious Denomination
 Roman Catholic 24.7% 21.6% 32.85(5) ***
 Protestant 46.0% 38.1%
 No preference 15.9% 10.8%
 No religious belief 4.3% 11.5%
 Jewish 3.9% 5.0%
 Other 5.2% 12.9%
Open in a new tab
**

p < .01,

***

p < .001. % indicates percentage of the sample for frequency variables. SD = Standard Deviation, df = degrees of freedom.

3.2. Identifying Discrete Patterns of Associations

Next, we incorporated LVMM to assess whether there were multiple patterns of cross-lagged and autoregressive associations across the three constructs. A two-pattern model best fit the data (BIC = 97,422.72, entropy = .907, VLMR = −48,647.85, p = .004). A three-pattern model was unidentified. We conducted Wald Tests on the two-pattern model, which indicated that a model in which all autoregressive and cross-lagged paths were set to equality did not fit the data (χ2(54) = 635.94, p < .001), but fit substantially improved when autoregressive paths for religiosity and alcohol use were freed in both patterns (χ2(42) = 42.29, p = .459). As such, we reran the two-pattern model, fixing all cross-lagged paths and the autoregressive self-control paths to equality across ages 13–17 (BIC = 96,271.27, entropy = .975, VLMR = −48,655.72, p < .001).

Within this restricted model, the first pattern represented associations within a minority of the sample (n = 264, 12.44%) and demonstrated few significant cross-construct relationships across religiosity, self-control, and alcohol use after controlling for each construct at all prior ages. However, self-control and likelihood of alcohol use were reciprocally related, although self-control was negatively associated with subsequent likelihood of alcohol use and alcohol use associated with higher subsequent self-control. Figure 1 depicts the autoregressive and cross-construct associations for both patterns. Although alcohol use was coded dichotomously and thus should not be interpreted as a continuous variable, we present linear coefficients to maintain a consistent metric across all relationships in this model.

Figure 1.

Figure 1.

Open in a new tab

Unstandardized coefficients resulting from the best fitting-cross-lagged panel mixture model.

The first panel depicts the set of parameter estimates that best characterize a minority of the sample (12.44%), while the second panel depicts the set of parameter estimates that characterize a majority of the sample (87.56%). Significant cross-lagged paths are bolded. Standard errors are in parentheses. *p ≤ .05, **p ≤ .01, ***p < .001.

The second pattern best characterized the majority of the sample (n = 1,858, 87.56%). All autoregressive and cross-construct relationships were significant even after controlling for prior levels of all constructs, except that alcohol use did not predict subsequent self-control. Religiosity was negatively associated with alcohol use and positively associated with self-control, while self-control had a positive effect on religiosity and a negative effect on alcohol use. Alcohol use was associated with lower religiosity but was unrelated to subsequent self-control.

Mean levels/rates of endorsement for all study variables are presented separately for individuals best characterized by majority and minority patterns in Table 2. Relative to girls characterized by the majority pattern, those represented by the minority pattern had lower average self-control and were more likely to report alcohol use starting at age 14 but did not differ on religiosity. The autoregressive coefficients of alcohol use across ages 13–15 in the minority pattern are unusual due to a lack of variance in alcohol use at age 14 in this pattern. Only one individual endorsed alcohol use at age 13 while all individuals endorsed age 14 alcohol use, resulting in a nearly perfect correlation between alcohol use at ages 13 and 14 and no variance at age 14 with which to predict age 15 alcohol use. It is possible that age 14 alcohol use influenced subgroup membership but is not the only distinguishing feature as 11% of girls in the majority subgroup also endorsed alcohol use at age 14.

Table 2.

Demographic Characteristics and Variables of Interest by Most Likely Subgroup Membership

Variable Full Sample
% or Mean (SD)		 Majority Subgroup
% or Mean (SD)		 Minority Subgroup
% or Mean (SD)		 t(df)
χ2 (df)
n (percentage of sample) 2,122 (100%) 1,858 (87.56%) 264 (12.44%)
Covariates (Age 13)
 Race (Black) 57% 60% 43% 25.10 (1) ***
 Low Caregiver Education 45% 45% 47% 0.71 (1)
 Public Assistance 37% 38% 34% 1.21 (1)
 Single Parent Household 44% 45% 37% 6.23 (1) **
 Early puberty 29% 30% 24% 3.11 (1)
 Conduct Problem Severity 0.97 (1.76) 0.95 (1.77) 1.14 (1.68) 1.63 (2071)
 Poor Parental Supervision 7.11 (1.29) 7.11 (1.29) 7.10 (1.28) −0.10 (2064)
Religious Denomination
 Roman Catholic 24.7% 33.2% 23.5% 22.51 (5) ***
 Protestant 46.0% 34.4% 47.7%
 No preference 15.9% 15.6% 16.0%
 No beliefs 4.3% 4.2% 4.3%
 Jewish 3.9% 6.1% 3.5%
 Other 5.2% 6.5% 5.0%
Any Alcohol Use
 Age 13 16.0% 18.0% 0.4% 54.75 (1) ***
 Age 14 22.1% 11.0% 100% 1061.25 (1) ***
 Age 15 28.0% 25.0% 52% 84.99 (1) ***
 Age 16 33.2% 30.0% 55% 56.14 (1) ***
 Age 17 40.2% 37.0% 61% 50.41 (1) ***
Full Drink of Alcohol
 Age 13 4.8% 5.5% 0.4% 12.94 (1) ***
 Age 14 10.6% 6.2% 40.1% 277.94 (1)***
 Age 15 15.6% 13.2% 32.1% 60.14 (1)***
 Age 16 22.0% 19.5% 39.6% 49.31(1)***
 Age 17 33.4% 30.6% 53.8% 49.79 (1) ***
Religiosity Composite
 Age 13 0.01(2.97) 0.03 (2.94) −0.14 (3.16) −0.82 (331.52)
 Age 14 0.00(2.98) 0.02 (2.94) −0.10 (3.20) −0.58 (329.94)
 Age 15 0.01(3.04) 0.03 (3.01) −0.17 (3.21) −0.98 (1997)
 Age 16 0.00(3.06) 0.02 (3.03) −0.10 (3.18) −0.59 (1951)
 Age 17 0.00(3.10) 0.04 (3.12) −0.27 (3.00) −1.43 (1936)
Self-Control
 Age 13 12.63(3.56) 12.71 (3.57) 12.07 (3.42) −2.71 (2065) **
 Age 14 12.32(3.37) 12.45 (3.36) 11.48 (3.28) −4.34 (2019) ***
 Age 15 12.47(3.44) 12.55 (3.45) 11.85 (3.27) −3.16 (336.95) **
 Age 16 12.79(3.31) 12.85 (3.32) 12.44 (3.18) −1.60 (1489)
 Age 17 13.04(3.42) 13.19 (3.44) 12.16 (3.22) −3.46 (195.21)**
Open in a new tab
**

p ≤ .01,

***

p < .001. % indicates percentage of the sample for frequency variables. Religiosity is measured as the sum of standardized items, where M = 0. AA = African American, EA = European American, SD = Standard Deviation, df = degrees of freedom.

3.3. Psychosocial Covariates Predicting Discrete Patterns

Lastly, we assessed whether psychosocial covariates were associated with likelihood of membership in the minority pattern. We found that relative to White girls, Black girls were less likely to manifest the minority pattern (b = −0.081, SE = 0.018, p < .001). Likewise, higher levels of conduct disorder symptoms were associated with an increased likelihood of demonstrating the minority pattern (b = 0.018, SE = 0.004, p < .001). In contrast, neither early puberty (b = −0.015, SE = 0.015, p = .335) nor poor parental supervision (b = −0.001, SE = 0.006, p < .892) were associated with minority/majority patterns. Similarly, socioeconomic status indicators did not predict probability of manifesting the minority pattern: receipt of public assistance (b = 0.003, SE = 0.016, p = .831), living in a single parent-headed household (b = −0.012, SE = 0.016, p = .470), and low caregiver education (b = 0.016, SE = 0.015, p = .296).

3.4. Post-Hoc Analyses

We also assessed possible alternative reasons for alcohol use within the entire sample to rule out the direct cause of religious involvement in alcohol use (see Supplementary Table S3). Only a small proportion of girls indicated drinking alcohol as part of a religious ceremony, indicating that the alcohol use demonstrated in this sample was likely not driven by religious participation. The majority of alcohol use in early adolescence was in the context of family or holiday celebrations.

4. Discussion and Conclusions

The present study aimed to assess whether reciprocal associations among religiosity, self-control, and alcohol use were best characterized by one or more discrete patterns within a large community sample of adolescent girls. Person-centered analyses probabilistically identified two subgroups of girls likely to exist within this population, each characterized by a unique set of longitudinal, cross-construct associations among religiosity, self-control, and alcohol use; these results support systems theories suggesting that individuals experience reciprocal relationships between religiosity and self-control resulting in unique developmental patterns (Ford & Lerner, 1992). One subgroup represents the vast majority of the sample, for whom religiosity and self-control were positively associated with each other and negatively associated with alcohol use. The second subgroup represented a minority of the sample, for whom only self-control and alcohol use were reciprocally related across adolescence. The presence of two patterns of associations differentiated by the relevance of religiosity to the pattern of cross-construct associations suggests that for the majority of adolescent girls, religiosity and self-control are independently associated with lower alcohol use (Dickens et al., 2018; Dvorak et al., 2011; Wallace et al., 2016) but are also bidirectionally associated with each other across time.

These results suggest that for a majority of adolescent girls, religiosity predicts higher levels of self-control, aligning with prior literature that suggests religion replenishes and bolsters self-control (Fishbach et al., 2003; Rounding et al., 2012; Watterson & Giesler, 2012). These results also suggest that self-control predicts higher levels of religiosity. Prior research, rooted in personality theory, has suggested that individuals with higher levels of personality traits related to self-control (e.g., conscientiousness, agreeableness) may be more likely to self-select into religious participation, indicating that self-control may be predictive of religiousness (McCrae & Costa, 1987; McCullough et al., 2003; McCullough & Willoughby, 2009). The results of our study extend these findings to suggest that the predictive nature of personality on religiosity may exist in adolescence, complementing research that documents these associations in late adulthood (Wink et al., 2007). Collectively, these reciprocal associations associations may predict lower alcohol use. As such, continuing to reinforce and support religiosity and self-control may further reduce the likelihood of alcohol use for most adolescent girls.

The absence of significant associations between religiosity and alcohol use in the minority subgroup is inconsistent with prior research that did not consider the role of self-control in predicting adolescent alcohol use (Dickens et al., 2018; Wallace et al., 2016). Notably, the minority subgroup did not differ from the majority subgroup in the extent to which they endorsed religiosity, suggesting that there is a qualitative difference in how religiosity is associated with self-control and alcohol use between subgroups of girls, rather than a difference in mean levels of religiosity driving these relationships. Continuing to promote self-control may reduce the likelihood of alcohol use among all individuals, which aligns with public health research on the importance of self-control in promoting healthy behaviors (Kang & You, 2018), while religiosity as a means of reducing alcohol use may be effective for only some girls.

The patterns we identified are probabilistic, meaning that girls are not definitively classified into a subgroup; rather, girls have a higher probability of demonstrating one pattern. We also found that the probability of demonstrating each pattern was predicted by different psychosocial covariates. We found that higher levels of conduct problems were associated with an increased likelihood of demonstrating the minority pattern, which was characterized by cross-construct associations between only self-control and alcohol use. Conduct problems and religiosity are typically inversely related (King & Boyatzis, 2015; Kuperman et al., 2013); it may be that girls who have higher levels of conduct problems are likely to be less religious to an extent that religiosity is not an effective protective factor. Rather, higher self-control is associated with lower levels of alcohol use even among girls with higher levels of conduct problems; this subgroup of girls may be indicative of one who is especially likely to benefit from efforts towards improving self-control with implications for both health promotion (e.g., reducing alcohol use) and reducing conduct problems (Kang & You, 2018).

Black girls were more likely to reflect the majority subgroup pattern, comprised of inverse relationships between alcohol use and both self-control and religiosity. This is somewhat surprising, given that religious involvement has been found to be more protective against alcohol use in White than Black youth (Agrawal et al., 2017; Wallace et al., 2007). Rather, these results may be indicative of cultural practices that predict patterns of behavior characterized by lower levels of alcohol use and higher levels of religiosity. For instance, Black families are more religious on average than White families (Taylor et al., 1996); this may increase the likelihood that girls’ religiosity is reciprocally associated with both self-control and alcohol use. It may be that this subgroup, which represents the majority of adolescent girls, is indicative of a general cultural background that promotes factors associated with lower levels of alcohol use.

4.1. Limitations

The alcohol use outcome was dichotomized within the present study due to the relatively limited level of alcohol involvement within our sample and our interest in associations with any level of use, and we were unable to include a zero-inflated component to this model due to the complexity of the panel mixture model; as such, conclusions cannot be drawn about associations with quantity or frequency of alcohol use, but should be studied in future research. Further, self-control data were missing at ages 16 and 17 for a substantial number of participants due to the study design (see Section 2.2.2.). Although robust maximum likelihood is the most appropriate way to account for incomplete data, and our results suggested that self-control did not differ across the cohorts at earlier ages when full data was available, the presence of missing data may indicate that our study underestimates the role of self-control in the religiosity-alcohol use relationship. Lastly, while we retained the use of this model due to the complex nature of combining two modeling approaches, recent research has criticized the cross-lagged panel model approach, indicating additional research with a more sophisticated cross-lagged modeling approach may be needed to fully understand the nature of reciprocal relationships across religiosity, self-control, and alcohol use (Hamaker et al., 2015). Including a more advanced cross-lagged modeling approach may also improve our ability to capture developmental dimensions of alcohol use and religiosity across adolescence, which is an additional limitation of our current study.

4.2. Future Directions

Future directions include using other longitudinal approaches to examine the mechanistic role of self-control in the religiosity-alcohol use relationship (Desmond et al., 2013; DeWall et al., 2014). The purpose of the present investigation was to identify discrete patterns of associations that may describe individuals with unique developmental experiences with religiosity, alcohol use, and self-control, but future research should investigate the mechanisms through which conduct problems influence these discrete patterns of associations. Given the multidimensional nature of religiosity (Rew & Wong, 2006), research may consider examining differences in these relationships across religions (particularly those that prohibit alcohol use) and various exhibitions of religiosity (e.g., public vs. private). Future research should also examine the extent to which self-control of social behaviors in an alcohol use setting may translate to prevention related skills, such as refusal self-efficacy and confidence in one’s ability to avoid drinking, and the extent to which religiosity may be related to the application of these skills in drinking situations (Lee & Oei, 1993). Finally, studies of heterogeneity in patterns of associations across religiosity, self-control, and alcohol use should be extended to include boys and individuals of other racial backgrounds.

Supplementary Material

1
2
3
4

Highlights.

  • Two patterns of relationships across religiosity, self-control, and alcohol use.

  • Minority subgroup shows relationships between self-control and alcohol use.

  • Majority subgroup demonstrates relationships between all three constructs.

  • Girls with higher levels of conduct disorder symptoms demonstrate minority pattern.

Acknowledgements:

Special thanks go to the families of the Pittsburgh Girls Study for their participation in this research, and to our dedicated research team for their continued efforts.

Funding: This project was supported by grants funded through the National Institute of Mental Health (MH056630), the National Institute of Drug Abuse (DA012237), and the National Institute on Alcohol Abuse and Alcoholism (AA023549), the FISA Foundation, and the Falk Fund. These institutions had no role in the study design, collection, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Author Agreement

All authors have seen and approved the final version of the manuscript. The article is the authors’ original work. Portions of preliminary analyses directly stemming from this manuscript were previously presented as posters at the Society for Research in Child Development biennial meeting and the Society for Research on Adolescence biennial meeting. Further, the Pittsburgh Girls Study, the source of data for the present manuscript, is a large-scale longitudinal study that was initiated in 1999 and thus has been the source of data for numerous published studies over the last two decades. The assessments covered a wide range of constructs, many of which are unrelated to religious involvement or substance use. As such, the majority of Pittsburgh Girls Study based publications do not overlap in content area with the current manuscript. Seventeen publications based on Pittsburgh Girls Study data are related to alcohol use, religiosity, or selfcontrol.

Although these studies made use of some of the same data analyzed in the current study, none addressed religiosity, self-control, and alcohol use in a single manuscript. Despite these related prior publications, none of the original material contained within this manuscript has or will be submitted for review/publication in another journal while under consideration by Addictive Behaviors.

Conflict of Interest

All other authors declare that they have no conflicts of interest.

Declaration of interests: none.

References

  1. Agrawal A, Grant JD, Haber JR, Madden PAF, Heath AC, Bucholz KK, & Sartor CE (2017). Differences between White and Black young women in the relationship between religious service attendance and alcohol involvement. The American Journal on Addictions, 26(5), 437–445. 10.1111/ajad.12462 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Biehl MC, Natsuaki MN, & Ge X (2007). The influence of pubertal timing on alcohol use and heavy drinking trajectories. Journal of Youth and Adolescence, 36(2), 153–167. 10.1007/s10964-006-9120-z [DOI] [Google Scholar]
  3. Boyatzis CJ (2009). Examining Religious and Spiritual Development During Childhood and Adolescence. In Souza M. de, Francis LJ, O’Higgins-Norman J, & Scott D (Eds.), International Handbook of Education for Spirituality, Care and Wellbeing (pp. 51–67). Springer; Netherlands. 10.1007/978-1-4020-9018-9_4 [DOI] [Google Scholar]
  4. Cotton S, Zebracki K, Rosenthal SL, Tsevat J, & Drotar D (2006). Religion/spirituality and adolescent health outcomes: A review. Journal of Adolescent Health, 38(4), 472–480. 10.1016/j.jadohealth.2005.10.005 [DOI] [PubMed] [Google Scholar]
  5. Desmond SA, Ulmer JT, & Bader CD (2013). Religion, self-control, and substance use. Deviant Behavior, 34(5), 384–406. 10.1080/01639625.2012.726170 [DOI] [Google Scholar]
  6. DeWall CN, Pond RS, Carter EC, McCullough ME, Lambert NM, Fincham FD, & Nezlek JB (2014). Explaining the relationship between religiousness and substance use: Self-control matters. Journal of Personality and Social Psychology, 107(2), 339–351. 10.1037/a0036853 [DOI] [PubMed] [Google Scholar]
  7. Dickens DD, Jackman DM, Stanley LR, Swaim RC, & Chavez EL (2018). Alcohol consumption among rural African American and White adolescents: The role of religion, parents, and peers. Journal of Ethnicity in Substance Abuse, 17(3), 273–290. 10.1080/15332640.2016.1179155 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. D’Lima GM, Pearson MR, & Kelley ML (2012). Protective behavioral strategies as a mediator and moderator of the relationship between self-regulation and alcohol-related consequences in first-year college students. Psychology of Addictive Behaviors, 26(2), 330–337. 10.1037/a0026942 [DOI] [PubMed] [Google Scholar]
  9. Dvorak RD, Simons JS, & Wray TB (2011). Alcohol use and problem severity: Associations with dual systems of self-control. Journal of Studies on Alcohol and Drugs, 72(4), 678–684. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Fishbach A, Friedman RS, & Kruglanski AW (2003). Leading us not into temptation: Momentary allurements elicit overriding goal activation. Journal of Personality and Social Psychology, 84(2), 296–309. 10.1037/0022-3514.84.2.296 [DOI] [PubMed] [Google Scholar]
  11. Ford DH, & Lerner RM (1992). Developmental systems theory: An integrative approach (pp. xi, 259). Sage Publications, Inc. [Google Scholar]
  12. Gadow KD, & Sprafkin J (1999). Adolescent symptom inventory-4 (ASI-4). Checkmate Plus. [Google Scholar]
  13. Glassman T, Werch C. (Chad), & Jobli E (2007). Alcohol self-control behaviors of adolescents. Addictive Behaviors, 32(3), 590–597. 10.1016/j.addbeh.2006.06.003 [DOI] [PubMed] [Google Scholar]
  14. Green MJ, Leyland AH, Sweeting H, & Benzeval M (2013). Socioeconomic position and adolescent trajectories in smoking, drinking, and psychiatric distress. Journal of Adolescent Health, 53(2), 202–208.e2. 10.1016/j.jadohealth.2013.02.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gresham FM, & Elliott SN (1990). Social skills rating system: Manual. American Guidance Service. [Google Scholar]
  16. Hamaker EL, Kuiper RM, & Grasman RP (2015). A critique of the cross-lagged panel model. Psychological Methods, 20(1), 102. [DOI] [PubMed] [Google Scholar]
  17. Hipwell AE, Loeber R, Stouthamer-Loeber M, Keenan K, White HR, & Kroneman L (2002). Characteristics of girls with early onset disruptive and antisocial behaviour. Criminal Behaviour and Mental Health: CBMH, 12(1), 99–118. [DOI] [PubMed] [Google Scholar]
  18. Hoyland MA, Rowatt WC, & Latendresse SJ (2017). Prior delinquency and depression differentially predict conditional associations between discrete patterns of adolescent religiosity and adult alcohol use patterns. Substance Abuse: Research and Treatment, 2017(10), 0–0. 10.4137/SART.S34948 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Jankowski PJ, Hardy SA, Zamboanga BL, & Ham LS (2013). Religiousness and hazardous alcohol use: A conditional indirect effects model. Journal of Adolescence, 36(4), 747–758. 10.1016/j.adolescence.2013.06.001 [DOI] [PubMed] [Google Scholar]
  20. Johnston LD, O’Malley PM, Miech RA, Bachman JG, & Schulenberg JE (2017). Monitoring the Future national survey results on drug use, 1975–2016: Overview, key findings on adolescent drug use. Institute for Social Research, The University of Michigan. [Google Scholar]
  21. Kang N-G, & You M-A (2018). Association of perceived stress and self-control with health-promoting behaviors in adolescents. Medicine, 97(34). 10.1097/MD.0000000000011880 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Keenan K, Hipwell A, Chung T, Stepp S, Stouthamer-Loeber M, Loeber R, & McTigue K (2010). The Pittsburgh Girls Studies: Overview and initial findings. Journal of Clinical Child and Adolescent Psychology, 39(4), 506–521. 10.1080/15374416.2010.486320 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. King PE, & Boyatzis CJ (2015). Religious and spiritual development. In Handbook of Child Psychology and Developmental Science. John Wiley & Sons, Inc. http://onlinelibrary.wiley.com/doi/10.1002/9781118963418.childpsy323/abstract [Google Scholar]
  24. Kuperman S, Chan G, Kramer JR, Wetherill L, Bucholz KK, Dick D, Hesselbrock V, Porjesz B, Rangaswamy M, & Schuckit M (2013). A model to determine the likely age of an adolescent’s first drink of alcohol. Pediatrics, 131(2), 242–248. 10.1542/peds.2012-0880 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Latendresse SJ, Ye F, Chung T, Hipwell A, & Sartor CE (2017). Parental monitoring and alcohol use across adolescence in black and white girls: A cross-lagged panel mixture model. Alcoholism, Clinical and Experimental Research, 41(6), 1144–1153. 10.1111/acer.13386 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lee NK, & Oei TPS (1993). The importance of alcohol expectancies and drinking refusal self-efficacy in the quantity and frequency of alcohol consumption. Journal of Substance Abuse, 5(4), 379–390. 10.1016/0899-3289(93)90006-W [DOI] [PubMed] [Google Scholar]
  27. Lo Y, Mendell NR, & Rubin DB (2001). Testing the number of components in a normal mixture. Biometrika, 88(3), 767–778. 10.1093/biomet/88.3.767 [DOI] [Google Scholar]
  28. Loeber R, Farrington DP, Stouthamer-Loeber M, Kammen WBV, Farrington DP, touthamer-Loeber M, & Kammen WBV (1998). Antisocial Behavior and Mental Health Problems: Explanatory Factors in Childhood and Adolescence. Psychology Press. 10.4324/9781410602930 [DOI] [Google Scholar]
  29. Marshall EJ (2014). Adolescent alcohol use: Risks and consequences. Alcohol and Alcoholism, 49(2), 160–164. 10.1093/alcalc/agt180 [DOI] [PubMed] [Google Scholar]
  30. McCrae RR, & Costa PT (1987). Validation of the five-factor model of personality across instruments and observers. Journal of Personality and Social Psychology, 52(1), 81–90. 10.1037/0022-3514.52.1.81 [DOI] [PubMed] [Google Scholar]
  31. McCullough ME, & Carter EC (2013). Religion, self-control, and self-regulation: How and why are they related? In Pargament KI, Exline JJ, & Jones JW (Eds.), APA Handbook of Psychology, Religion, and Spirituality, Vol. 1: Context, Theory, and Research (pp. 123–138). American Psychological Association. [Google Scholar]
  32. McCullough ME, Tsang J, & Brion S (2003). Personality traits in adolescence as predictors of religiousness in early adulthood: Findings from the Terman Longitudinal Study. Personality and Social Psychology Bulletin, 29(8), 980–991. 10.1177/0146167203253210 [DOI] [PubMed] [Google Scholar]
  33. McCullough ME, & Willoughby BLB (2009). Religion, self-regulation, and self-control: Associations, explanations, and implications. Psychological Bulletin, 135(1), 69–93. 10.1037/a0014213 [DOI] [PubMed] [Google Scholar]
  34. Miller AS, & Stark R (2002). Gender and religiousness: Can socialization explanations be saved? American Journal of Sociology, 107(6), 1399–1423. 10.1086/342557 [DOI] [Google Scholar]
  35. Muthén LK, & Muthén BO (1998). Mplus User’s Guide. (8.2). Muthén & Muthén. [Google Scholar]
  36. Pandina RJ, Labouvie EW, & White HR (1984). Potential contributions of the life span developmental approach to the study of adolescent alcohol and drug use: The Rutgers Health and Human Development Project, a working model. Journal of Drug Issues, 14(2), 253–268. 10.1177/002204268401400206 [DOI] [Google Scholar]
  37. Pastor DA, Barron KE, Miller BJ, & Davis SL (2007). A latent profile analysis of college students’ achievement goal orientation. Contemporary Educational Psychology, 32(1), 8–47. 10.1016/j.cedpsych.2006.10.003 [DOI] [Google Scholar]
  38. Petersen AC, Crockett L, Richards M, & Boxer A (1988). A self-report measure of pubertal status: Reliability, validity, and initial norms. Journal of Youth and Adolescence, 17(2), 117–133. 10.1007/BF01537962 [DOI] [PubMed] [Google Scholar]
  39. Pirutinsky S (2014). Does religiousness increase self-control and reduce criminal behavior? A longitudinal analysis of adolescent offenders. Criminal Justice and Behavior, 41(11), 1290–1307. 10.1177/0093854814531962 [DOI] [Google Scholar]
  40. Rew L, & Wong YJ (2006). A systematic review of associations among religiosity/spirituality and adolescent health attitudes and behaviors. Journal of Adolescent Health, 38(4), 433–442. 10.1016/j.jadohealth.2005.02.004 [DOI] [PubMed] [Google Scholar]
  41. Rounding K, Lee A, Jacobson JA, & Ji L-J (2012). Religion replenishes self-control. Psychological Science, 23(6), 635–642. 10.1177/0956797611431987 [DOI] [PubMed] [Google Scholar]
  42. Salas-Wright CP, Vaughn MG, Maynard BR, Clark TT, & Snyder S (2014). Public or private religiosity: Which is protective for adolescent substance use and by what pathways? Youth & Society, 82–93. 10.1177/0044118X14531603 [DOI] [Google Scholar]
  43. Schulenberg J, Patrick ME, Maslowsky J, & Maggs JL (2014). The epidemiology and etiology of adolescent substance use in developmental perspective. In Handbook of Developmental Psychopathology (pp. 601–620). Springer, Boston, MA. 10.1007/978-1-4614-9608-3_30 [DOI] [Google Scholar]
  44. Schwartz G (1978). Estimating the dimensions of a model. The Annals of Statistics, 6(2), 461–464. [Google Scholar]
  45. Taylor RJ, Chatters LM, Jayakody R, & Levin JS (1996). Black and white differences in religious participation: A multisample comparison. Journal for the Scientific Study of Religion, 35(4), 403–410. 10.2307/1386415 [DOI] [Google Scholar]
  46. Vuong QH (1989). Likelihood ratio tests for model selection and non-nested hypotheses. Econometrica, 57(2), 307–333. 10.2307/1912557 [DOI] [Google Scholar]
  47. Wallace JM, Delva J, O’Malley PM, Bachman JG, Schulenberg JE, Johnston LD, & Stewart C (2007). Race/ethnicity, religiosity and adolescent alcohol, cigarette and marijuana use. Social Work in Public Health, 23(2–3), 193–213. 10.1080/19371910802152059 [DOI] [PubMed] [Google Scholar]
  48. Wallace JM, O’Malley PM, Bachman JG, Schulenberg JE, & Johnston LD (2016). Race/Ethnicity, Religiosity and Differences and Similarities in American Adolescents’ Substance Use. In Thomas YF & Price LN (Eds.), Drug Use Trajectories Among Minority Youth (pp. 105–121). Springer; Netherlands. 10.1007/978-94-017-7491-8_7 [DOI] [Google Scholar]
  49. Watterson K, & Giesler RB (2012). Religiosity and self-control: When the going gets tough, the religious get self-regulating. Psychology of Religion and Spirituality, 4(3), 193–205. 10.1037/a0027644 [DOI] [Google Scholar]
  50. Wink P, Ciciolla L, Dillon M, & Tracy A (2007). Religiousness, spiritual seeking, and personality: Findings from a longitudinal study. Journal of Personality, 75(5), 1051–1070. 10.1111/j.1467-6494.2007.00466.x [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1657009-supplement-1.docx (111.7KB, docx)
2
NIHMS1657009-supplement-2.docx (13.1KB, docx)
3
NIHMS1657009-supplement-3.docx (13.1KB, docx)
4
NIHMS1657009-supplement-4.docx (12.7KB, docx)

RESOURCES