Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Nov 4.
Published in final edited form as: J Res Adolesc. 2019 Mar 25;30(Suppl 2):499–515. doi: 10.1111/jora.12494

Adolescent social norms and alcohol use: Separating between- and within-person associations to test reciprocal determinism

Samuel N Meisel 1, Craig R Colder 1
PMCID: PMC7610152  NIHMSID: NIHMS1640415  PMID: 30908811

Abstract

Despite perceived drinking norms being robust predictors of adolescent alcohol use, few studies have assessed the development of perceived norms across adolescence and processes accounting for the strong associations between perceived norms and drinking. Using reciprocal determinism as a theoretical basis for understanding the development of adolescent drinking norms, the current study examined reciprocal associations across nine waves of data spanning early to late adolescence. Bivariate latent curve models with structured residuals demonstrated consistent within-person reciprocal associations between descriptive and injunctive norms and alcohol use after accounting for growth in norms and alcohol use. Results suggest the need for developmentally informed intervention efforts targeting perceived drinking norms during early and middle adolescence.

Keywords: adolescent alcohol use, descriptive norms, injunctive norms, reciprocal determinism, latent growth curve modeling with structured residual

Introduction

Adolescent alcohol use predominately occurs in the context of peers (Brechwald & Prinstein, 2011). Accordingly, social influence figures predominantly in developmental models of adolescent alcohol use (Chassin, Colder, Hussong, & Sher, 2016). Perceived drinking norms are one mechanism of social influence, and they refer to perceptions of peer drinking and approval of drinking (Borsari & Carey, 2001; Rimal & Real, 2005). Perceived drinking norms are among the strongest predictors of alcohol use throughout adolescence (Pedersen et al., 2017; Neighbors et al., 2008).

Although the literature supports a robust relationship between perceived drinking norms and adolescent alcohol use, few studies have examined their development or the mechanisms that may account for the strong relationship between norms and drinking (Geber, Buamann, & Kilmmt, 2017; Mollen, Rimal, & Lapinski, 2010). The current study uses Bandura’s (1978) theory of reciprocal determinism to examine growth in perceived drinking norms and alcohol use across adolescence and reciprocal relationships between descriptive (perceptions of peer drinking) and injunctive norms (perceptions of peer approval of drinking) and alcohol use. Moreover, we use Latent Curve Models with Structured Residuals (LCM-SR) to distinguish between- and within-person relationships, which we argue is a statistical approach theoretically consistent with Bandura’s notion of reciprocal determinism.

Social Norms and Adolescent Drinking

The Focus Theory of Normative Conduct distinguishes between two distinct, yet related social norms – descriptive and injunctive norms (Cialdini, Reno, & Kallgren, 1990). Descriptive norms, as applied to alcohol use, refer to perceptions of how much others drink or how many individuals drink and they serve as a quick heuristic to guide behavior because they inform what peers do in a particular context (Jacobson, Mortensen, & Cialdini, 2011). Injunctive norms refer to perceptions of others’ approval of drinking, and are thought to guide behavior after reflecting on what constitutes acceptable behavior in a given context (Jacobson et al., 2011). According to The Focus Theory of Normative Conduct, individuals align their behavior with descriptive and injunctive norms because they inform what constitutes socially acceptable behavior (Cialdini et al., 1990).

Perceived drinking norms may be particularly impactful during adolescence (Pedersen et al., 2017). Peer admiration, group acceptance, and social status are highly rewarding during adolescence (Crone & Dahl, 2012; Yeager, Dahl, & Dweck, 2018), and complying with social norms can help achieve these social rewards (Meisel & Colder, 2015). Indeed, adolescents perceive drinking to be prototypical of adult status, popularity, and gregariousness (Balsa, Homer, French & Norton, 2011; Gerrard et al., 2002). Further, adolescents who fail to adopt what is perceived to be normative drinking may place themselves at risk for friendship dissolution (Rimal & Real, 2005).

The Development of Social Norms

To date, few studies have assessed the development of perceived drinking norms during adolescence (Duan et al., 2009; Weaver et al., 2011). Duan and colleagues (2009) found that descriptive norms for close friends and peers increased from middle school through high school with the largest increase in norms occurring during the transition to high school. Weaver et al. (2011) observed a similar developmental pattern for White, but not Black adolescents. To our knowledge, no studies have assessed the development of adolescent injunctive drinking norms.

Despite the importance of drinking norms in the etiology of adolescent alcohol use, mechanisms accounting for the development of drinking norms and their relationship with alcohol use remain unstudied. Several reviews have concluded that there is a need for studies to explicate the development of both descriptive and injunctive norms (Mollen et al., 2010). Reciprocal determinism is a tenet of Social Learning Theory (Bandura, 1977, 1978) that provides a theoretical basis for understanding the development of adolescent drinking norms. According to Social Learning Theory, alcohol use is a learned behavior and the social environment plays a central role in shaping drinking behaviors (Bandura, 1969). The peer environment is a critical social context during adolescence (Collins & Steinberg, 2006). Adolescents look to the actual as well as perceived behaviors of their peers to guide their own behaviors, such as drinking, and they are motivated to align their behaviors with those of their peers to build and maintain close relationships (Bandura, 1969, Maisto, Carey, & Bradizza, 1999; Meisel & Colder, 2015). Reciprocal determinism thus provides an account of how an adolescent’s alcohol use and perceived drinking norms reinforce each other across time (Bandura, 1978). Alcohol-related cognitions, such as descriptive and injunctive norms, and alcohol use should reciprocally influence one another through a dynamic learning process. Alcohol use shapes social norms, and social norms predict alcohol use, thus reinforcing each other over time.

To date, studies assessing reciprocal determinism for social norms and alcohol use have predominantly used college samples (e.g., Lee, Geisner, Patrick, & Neighbors, 2010; Lewis, Litt, & Neighbors, 2015), despite the evidence that norms are particularly dynamic during adolescence. Only one study, to our knowledge, has examined reciprocal effects between adolescent social norms and alcohol use. Marks and colleagues (1992) used two time points to examine reciprocal associations between descriptive norms and alcohol, and findings supported bidirectional associations. However, this study was limited by using only two assessments, precluding the examination of growth in social drinking norms. Moreover, injunctive norms were not assessed, and this is problematic given empirical evidence for the importance of injunctive norms in the etiology of adolescent drinking (Pedersen et al., 2017; Meisel, Colder, & Read, 2016). The current study addresses these gaps by examining reciprocal determinism across early through late adolescence and examining both descriptive and injunctive social norms.

Disaggregating Between- and Within-Person Effects

To date, studies examining reciprocal determinism with respect to alcohol use and drinking norms have tested between-person cross-lagged associations (e.g., Lee et al., 2010; Lewis et al., 2015). That is, these studies examined whether a person’s relative standing on a sample distribution (e.g., on social norms) is related to relative standing on another distribution (e.g., on alcohol use). A focus on between-person associations has both conceptual and statistical limitations. On a conceptual level, reciprocal determinism is a theory of individual change because it argues that an individual’s cognition (e.g., perceived norm) can influence his/her behavior (e.g., alcohol use) and vice versa. For example, when an individual experiences an increase in the perceived normativeness of drinking (relative to his/her typical perception) this should predict subsequent increases in drinking (relative to his/her typical level of drinking). We argue that testing reciprocal determinism according to a social learning perspective seeks to explain how an individual’s perceived norms and alcohol use develop, rather than examining between person differences, and this has not been done in prior work.

A statistical limitation of assessing reciprocal determinism only on the between-person level is that such analyses can lead to inaccurate estimation of reciprocal associations if within-person associations are present (Curran, Howard, Bainter, Lane, & McGinley, 2014). The failure of prior work to distinguish between- and within-person associations may account for why some studies have found support for reciprocal determinism (Lee et al., 2010; Lewis et al., 2015; Litt, Lewis, Rhew, Hodge, & Kaysen, 2015; Marks et al., 1992; Neighbors, Dillard, Lewis, Bergstrom, & Neil, 2006; Wardell & Read, 2013) and others have not (Farrell, 1994; Read, Wood, & Capone, 2005).

The Current Study

Few studies have examined developmental mechanisms that might account for the strong relationship between perceived norms and drinking during adolescence. Reciprocal determinism provides a social learning mechanism that can account for the strong relationship between norms and alcohol use across adolescence. Based on theory and prior research, we hypothesized:

  1. Descriptive and injunctive norms, and alcohol use would demonstrate significant growth from early through late adolescence.

  2. At the between-person level, growth in descriptive and injunctive norms would be associated with growth in alcohol use.

  3. At the within-person level, descriptive and injunctive norms would be reciprocally and prospectively associated with alcohol use.

Methods

Participants

Participants were from a longitudinal study of risk and protective factors for adolescent substance use. The community sample of 387 families (1 child, 1 caregiver) was recruited between 2007 and 2009 and was evenly split on gender (N = 213 female adolescents, 55%). The majority were non-Hispanic White (83.1%), and the remaining adolescents were African American (9.1%), Hispanic (2.1%), and Asian (1.0%) or of mixed ethnicity (4.7%). Median family income was $70,000 (range $1,500 to $500,000), and 6.2% of the families received public income assistance. Sample demographics are similar to those of Erie County from whence the sample came (see Trucco, Colder, Wieczorek, Lengua, & Hawk, 2014 for more information on sample characteristics and recruitment).

This study utilized data from Waves 1 (W1) through 9 (W9) of the longitudinal project. Participants had an average age (in years) of 12.1, 13.1, 14.1, 15.1, 16.1, 17.1, 17.9, 18.9, and 19.9 (SD range=0.59–0.73) at W1 to W9, respectively.

Procedures

At W1-W3 and W7-W9, adolescents and their caregivers were interviewed annually in university research offices. After assent/consent procedures, parents and adolescents were escorted into separate interview rooms. Assessments took approximately 2.5 to 3 hours. At W1 to W3, families were compensated $75, $85, and $125, respectively. Adolescents were compensated $125, $135, and $145, and caregivers were compensated $40, $45, and $50, at W7 to W9, respectively.

W4, W5, and W6 consisted of a brief telephone-based audio-Computer-Assisted Self-Interview (CASI) survey of substance use that took 10–15 min to complete. Parents provided consent over the phone and were given a phone number and PIN for their adolescent to use. Assent from the adolescent was obtained at the initiation of the audio-CASI survey. Adolescents were compensated, $15, $15, and $20 at W4 to W6, respectively. All procedures were approved by the University’s Institutional Review Board (Study title: Internalizing problems, motivation, peers, & development of adolescent drug use; MODCR00000706).

Measures

Alcohol use (W1-W9).

Items from the National Youth Survey (Elliot & Huizinga, 1983) were used to assess past year alcohol use at W1 to W6. Adolescents reported the number of times in the past year they used alcohol with a fill-in-the-blank question. At W7, W8, and W9, participants reported past year alcohol frequency using an 8-point response scale (1=not at all, 2=one or twice, 3=one day per month, 4=2–3 days/month, 5=1 day per week, 6=two to three days per week, 7=four or five days per week, 8=everyday). This scale was converted to represent the number of drinking days in the past year to be consistent with our W1 to W6 measurement. To reduce the influence of outliers, extreme values were recoded to three standard deviations above the mean at each wave (Tabachnick & Fidell, 2007).

Social Norms (W1-W3, W7-W9).

Descriptive and injunctive norms were assessed at W1 to W3 and W7 to W9 with three items from the Monitoring the Future Study (Johnston et al., 2003). The descriptive norms items asked participants to report how many of their friends drink occasionally, drink regularly, and consume more than 5 drinks during a single drinking occasion using six response options (1 = none to 6 = all). Injunctive norms were assessed with 3 items that asked participants to rate how their three close friends would feel about them doing each of the following behaviors drinking alcohol occasionally, drinking alcohol regularly, and having five or more drinks of alcohol at 1 time using a 5-point response scale (1 = strongly disapprove to 5 = strongly approve). Items were averaged to form descriptive and injunctive scale scores at each assessment. Cronbach’s alpha for descriptive norms was 0.85, 0.90 and 0.87 at W1, W2, and W3, respectively, and 0.86, 0.89 and 0.91 for injunctive norms.

Missing Data and Analytic Strategy

When data are either missing completely at random (MCAR) or Missing at Random (MAR), full information maximum likelihood estimation (FIML) provides accurate and unbiased parameter estimates (Enders, 2013; Schafer & Graham, 2002). To assess for MCAR, Little’s test of missing completely at random (MCAR) was conducted in SAS 9.4 on a data set using raw scores for descriptive norms, injunctive norms, and alcohol use at W1-W9 as well as minority status, gender, and age. Results of Little’s MCAR test were significant (χ2=1212.26(881), p<.001) suggesting our data were not MCAR. Although our data were not MCAR, retention was excellent across the nine waves of the study (W2=96%, W3=96%, W4=96%, W5=94%, W6=91%, W7=91% W8= =91%, W9=91%). Further, comparison of those with and without missing data on W1 variables suggested no significant differences (ps > 0.05) for age, gender, ethnicity, descriptive norms, injunctive norms, and alcohol use. The low attrition rate and lack of differences suggest that missing data met requirements for MAR and did not have a substantial impact on the findings of the current study. Further, we used full-information maximum likelihood estimation, which permits the inclusion of cases with missing data.

Bivariate Latent Curve Modeling with Structured Residuals (LCM-SR; Curran et al., 2014) was used to test reciprocal associations between social norms and alcohol use. LCM-SR allows for modeling growth (which is important when considering perceived drinking norms and alcohol use in adolescence), but also disaggregates within- and between-person associations. This modeling framework imposes structure on time-specific residuals of observed measures of norms and alcohol use, which represents individual variability after taking into account growth. When cross-lags between the structured residuals are introduced, this allowed us to examine whether increases in norms (relative to an individual’s typical level of perceived norms) was associated with increases in subsequent alcohol use (relative to an individual’s typical level of drinking) and vice versa. As with parallel process growth curves, covariances between growth factors can be estimated, and these represent between-person associations.

LCM-SR models were estimated in Mplus 8.0 with Maximum Likelihood Robust estimation (MLR, Muthén & Muthén, 1998–201) which adjusts fit indices and standard errors for non-normality. Prior to estimating the LCM-SR, measurement invariance for descriptive and injunctive norms was assessed from W1-W9. We used a sequential model building strategy as suggested by Curran and colleagues (Curran et al., 2014) starting with univariate growth curves. Multiple forms of growth (e.g., linear, piecewise) were specified for social norms and alcohol use and the Satorra-Bentler chi-square difference test (Satorra & Bentler, 2001) was used to determine the best fitting growth models. Next structured residuals were specified and chi-square difference tests were used to determine whether the addition of autoregressive paths for the structured residuals were supported, and whether these paths could be constrained to be equal across time. Finally, bivariate LCM-SR models were estimated separately for each social norm and alcohol use, and nested model tests were conducted to determine whether the bivariate models supported the inclusion of covariances across intercepts and slopes of social norms and alcohol use, within-time residual covariances, and reciprocal effects between social norms and alcohol use. Gender, minority status, and age were included as covariates in the final bivariate models by regressing the growth factors for norms and alcohol use on these covariates.

Model fit was assessed using the Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), Root Mean Square Error Approximation (RMSEA), and Standardized Root-Mean-Square Residual (SRMR). Specific cut-offs for assessing “good” fit cannot be generalized across all models (Marsh, Hau, & Wen, 2004), therefore, ranges were used to determine model fit acceptability (for CFI and TLI, <.90 is poor, .90 to .94 is acceptable, and >.95 is excellent; for RMSEA, .08 is poor, .05 to .07 is acceptable, and <.05 is excellent; and for SRMR, .09 is poor, .06 to .09 is acceptable, and <.06 is excellent).

Results

Descriptive Statistics

Table 1 provides descriptive statistics for norms and alcohol use. The means suggest increases in all three variables from early through late adolescence. There was a large increase in past-year alcohol frequency from W6 (M=4.17) to W7 (M=20.79) which corresponded with the change in response format for the assessment of alcohol frequency. This increase was likely attributable to change in age and context as this interval corresponds to when much of our sample transitioned out of high school, but it is also possible that this increase was a function of the change in response format.

Table 1.

Zero-Order Correlations

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1. AU W1 ---
2. AU W2 0.34 ---
3. AU W3 0.27 0.39 ---
4. AU W4 0.26 0.38 0.55 ---
5. AU W5 0.18 0.28 0.46 0.65 ---
6. AU W6 0.11 0.25 0.45 0.55 0.70 ---
7. AU W7 0.08 0.23 0.30 0.33 0.44 0.60 ---
8. AU W8 0.11 0.14 0.25 0.26 0.32 0.48 0.64 ---
9. AU W9 0.07 0.08 0.16 0.18 0.23 0.42 0.56 0.73 ---
10. IN W1 0.33 0.24 0.25 0.15 0.08 −0.01 0.01 0.07 0.04 ---
11. IN W2 0.17 0.40 0.45 0.41 0.33 0.27 0.30 0.20 0.16 0.34 ---
12. IN W3 0.19 0.32 0.54 0.52 0.46 0.41 0.32 0.22 0.12 0.23 0.55 ---
13. IN W7 0.07 0.14 0.21 0.26 0.31 0.40 0.58 0.55 0.45 −0.02 0.25 0.28 ---
14. IN W8 0.12 0.08 0.13 0.19 0.22 0.29 0.49 0.57 0.44 −0.04 0.08 0.21 0.63 ---
15. IN W9 0.06 −0.04 0.00 −0.01 0.13 0.24 0.35 0.43 0.44 −0.08 −0.03 0.05 0.51 0.62 ---
16. DN W1 0.33 0.27 0.36 0.20 0.10 0.13 0.12 0.10 0.09 0.39 0.28 0.22 0.12 0.11 0.03 ---
17. DN W2 0.21 0.44 0.45 0.38 0.36 0.28 0.23 0.17 0.16 0.28 0.60 0.37 0.23 0.15 0.05 0.42 ---
18. DN W3 0.17 0.36 0.61 0.52 0.45 0.38 0.31 0.26 0.17 0.19 0.47 0.61 0.31 0.23 0.08 0.34 0.53 ---
19. DN W7 0.05 0.14 0.18 0.22 0.31 0.43 0.59 0.51 0.47 −0.10 0.17 0.20 0.69 0.55 0.45 0.06 0.26 0.28 ---
20. DN W8 0.07 0.07 0.18 0.16 0.25 0.36 0.46 0.59 0.52 −0.05 0.10 0.13 0.52 0.64 0.51 0.10 0.20 0.26 0.62 ---
21. DN W9 0.14 0.02 0.07 0.09 0.15 0.30 0.32 0.49 0.53 0.01 0.03 0.04 0.44 0.51 0.66 0.04 0.10 0.15 0.46 0.61 ---
22. Age 0.01 0.16 0.21 0.25 0.20 0.23 0.20 0.12 0.12 0.03 0.21 0.22 0.13 0.14 0.05 0.08 0.20 0.27 0.15 0.04 0.05 ---
23. Minority −0.02 0.09 0.02 0.04 0.05 −0.06 −0.11 −0.14 −0.14 0.07 0.06 0.08 −0.10 −0.22 −0.17 −0.01 −0.01 −0.04 −0.11 −0.14 −0.21 −0.01 ---
24. Gender 0.08 −0.02 −0.03 −0.02 −0.06 −0.02 0.02 0.06 −0.01 0.10 0.05 0.08 0.09 0.07 0.07 0.07 −0.06 −0.05 0.00 0.08 0.00 −0.04 −0.02 ---
Mean 0.34 0.65 1.18 1.27 2.30 4.17 20.79 44.80 58.63 1.24 1.41 1.62 2.78 2.90 3.06 1.12 1.33 1.54 2.85 3.14 3.33 12.09 0.17 0.45
SD 1.00 1.76 2.50 2.86 4.74 8.15 21.69 56.11 66.96 0.53 0.74 0.85 0.99 0.96 0.91 0.39 0.76 0.86 1.30 1.27 1.25 0.59 0.37 0.50
Skew 3.73 3.57 2.80 2.78 2.72 2.76 0.48 1.47 1.41 3.36 2.03 1.43 −0.04 −0.28 −0.21 5.69 3.32 1.84 0.32 0.15 −0.04 −0.16 1.78 0.20
Kurtosis 14.70 12.95 7.70 7.39 6.90 7.29 −1.49 1.02 1.14 15.31 3.85 1.33 −0.37 −0.21 0.06 40.38 12.31 2.96 −0.79 −0.70 −0.74 −1.12 1.17 −1.97
Open in a new tab

Note. AU=alcohol use, IN=injunctive norms, DN=descriptive norms. Minority=minority status where 0=non-Hispanic White and 1=other. Correlations significant at p<.05 are in bold.

To investigate these possibilities, alcohol frequency was compared to data from Monitoring the Future (Johnston, O’Malley, Bachman, & Schulenberg 2010), and we broke these data down by educational status. Past-year alcohol frequency rates in the current study were very similar to Monitoring the Future. When comparing Monitoring the Future to the current study, past year rates of abstaining in 8th, 10th, and 12th grade were 71% vs. 73%, 48% vs. 63%, and 34% vs. 37%, and past year rates of use on 1–5 occasions were 21% vs. 23%, 30% vs. 27%, and 31% vs. 29%. Of note is that our rates of use in 12th grade are comparable to those of MTF even though 32% of our participants in 12th grade were assessed at W7, when our measurement of alcohol use frequency changed.

W7 was the first year of our study where a significant portion of our participants graduated high school and entered college. As of W7, 13.6% of our participants were in a 2-year college, 41.5% were in a 4-year college, 1.7% were in technical school, 32.4% were in high school, and 9.7% were not in school. Consistent with the broader literature that finds greater alcohol frequency for adolescents who graduated high school relative to those still in high school (White, Labouvie, & Papadaratsakis, 2005; White et al., 2006), rates of use differed in our sample for participants in versus out of high school. Broken down by educational status, the average past year alcohol frequency for high school students at W7 was 13.26 (SD=18.6), 2-year college students was 24.46 (SD=22.32), 4-year college students was 24.27 (SD=22.1), technical school students was 30.78 (SD=23.76), and not currently attending school was 25.50 (SD=23.00). These means highlight that the large change in past year frequency was likely largely driven by participants who had graduated high school. We ran an ANOVA with a planned contrast comparing high school students to students in 2 and 4 year colleges, and participants not currently in school, to determine whether these differences were statistically significant. Indeed, there was a significant difference across the groups (F(3, 337)=7.19, p=.001) and the planned contrast indicated that the high school students had significantly lower past year alcohol frequency relative to 2 and 4 year college students, and non-school attending participants (F(1)=19.46, p<.001; η2=.05). In sum, although we cannot rule out that some bias resulting from change in our measurement may have influenced reportage of frequency (Gmel & Lokosha, 2000), the data strongly suggest that change in age and context likely account for the increases we observe across W6 and W7.

Zero-order spearman correlations are presented in Table 1 because of the non-normality of alcohol use. Within waves, alcohol use was correlated with descriptive norms and injunctive norms. Age was also positively correlated with alcohol use, injunctive norms, and descriptive norms at most waves. Gender was not associated with alcohol use or either norms variable. As seen in Table 1, social norms and alcohol use were non-normally distributed, and therefore, these variables were log-transformed. The log transformation reduced the degree of non-normality of descriptive norms (skew<4.12, kurtosis<21.32), injunctive norms (skew<2.36, kurtosis<6.05), and alcohol frequency (skew<2.60, kurtosis<6.04).

Longitudinal Measurement Invariance

Measurement invariance across time was assessed for descriptive and injunctive norms by creating latent descriptive and injunctive norms variables at each wave norms were assessed (W1-W3 and W7-W9). The three items each assessing descriptive and injunctive norms from the Monitoring the Future Study were specified as indicators of latent descriptive norms and injunctive norms, respectively. After establishing configural invariance for descriptive norms, nested model tests indicated that constraining factor loadings to be equal across time did not result in a significant decrement in model fit (χ2=8.99 (10), p=.53) supporting full metric invariance (χ2=191.63(124), p=.001, CFI=.97, TLI=.96, RMSEA=.03, SRMR=.05). For injunctive norms, constraining all factor loadings to be equal across time resulted in a significant decrement in model fit (χ2=57.58(10), p<.001). Partial metric invariance was supported across W1-W9 (χ2=9.87(6), p=.12) after freeing the factors loadings for the item assessing the perceived acceptability of drinking occasionally at W1, W2, and W3, and freeing the item loading assessing the acceptability of drinking regularly at W1 (χ2=221.86(122), p<.001, CFI=.97, TLI=.96, RMSEA=.04, SRMR=.04). These factor loadings were all statistically significant (p<.05) and substantial (standardized loadings >.72), suggesting the differences in item loadings do not substantially change the interpretation of the factors.

Univariate LCM-SR Models for Perceived Norms and Alcohol Use

Alcohol Use.

Considering the low rates of use in our sample at W1 and W2, the intercept for alcohol use was set at W3. A piecewise growth model provided the best fit for growth in alcohol use (see Appendix A Figure A1). The first growth segment represented change from W1 to W6 with freely estimated slope factor loadings at W4 and W6 (slope loadings of −2, −1, 0, 0.26, 2, 3.62, for W1-W6, respectively) and the second growth segment represented change from W6 to W9 with freely estimated slope factor loadings at W7 and W8 (slope loadings of 0, 1.73, 2.44, 3 for W6-W9, respectively). The slope factor means for W1 to W6 (M=0.13, p<.001, 95% CI [.11, .15]) and W6 to W9 (M=0.77, p<.001, 95% [.71, .83]) indicated significant increases in both segments. There was significant variability in the intercept of alcohol use at W3 (σ2=0.16, p<.001, 95% [.12, .20]), both slopes (W1 to W6, σ2=0.03, p<.001, 95% CI [.02, .03]; W6 to W9, σ2=0.12, p<.001, 95% CI [.07, .17]). Covariances between the intercept and slope from W1 to W6 was significant (covariance=0.05, p<.001, 95% CI [.03, .06]), such that higher initial levels of alcohol use were related to more rapid increases in alcohol use from W1 to W6. The intercept was not related to growth from W6 to W9 and the slope factors were not significantly associated.

Next, structured residuals were specified and nested model tests supported including autoregressive pathways for the residuals (χ2=42.70(8), p<.001) and equality of the autoregressive paths (χ2=5.84(7), p=.56). Non-significant covariances between the intercept of alcohol use and growth in alcohol use from W6 to W9 and between the growth terms were constrained to zero (χ2=3.81(2), p=.14) to reduce model complexity. The final model provided an excellent fit to the data (χ2=41.92(34), p=.16, CFI=.99, TLI=.99, RMSEA=.02, SRMR=.04).

Descriptive Norms.

To be consistent with the alcohol use LCM-SR model, the intercept for descriptive norms was set at W3. Non-linear slope factors provided the best fit for descriptive norms with freely estimated loadings for the last three waves (slope loadings of −2, −1, 0, 5.39, 6.39, 7.02, for W1-W3 and W7-W9, respectively). Descriptive norms increased from early through late adolescence (M=0.08, p<.001, 95% CI [.06, .09]). There was significant variability in the intercept (σ2=0.01, p<.001, 95% CI [.01, .02]) and slope (σ2=0.001, p=.004, 95% CI [.000, ,001]) suggesting that there were significant individual differences in levels of descriptive norms at W3 and significant variability in growth in descriptive norms from early through late adolescence. The intercept and slope of descriptive norms were unrelated and this covariance was constrained to be zero (see Appendix A Figure A2). Next, structured residuals were specified and autoregressive paths for the structured residuals were estimated, and the model supported constraining autoregressive paths to be equal across time (Δχ2=6.19(3), p=.10). The final model provided a strong fit to the data (χ2=20.73(13), p=.07, CFI=.98, TLI=.98, RMSEA=.03, SRMR=.06).

Injunctive Norms.

The intercept for injunctive norms was set at W3. Non-linear slope factors provided the best fit for injunctive norms with the last three loadings freely estimated (slope loadings of −2, −1, 0, 5.88, 6.37, 7.07, for W1-W3 and W7-W9, respectively). Injunctive norms significantly increased from early through late adolescence (M=0.06, p<.001, 95% CI [.05, ,08]). There was significant variability in the slope (σ2=0.000, p=.001, 95% CI [.000, .001]), but not the intercept (σ2=0.003, p=.33, 95% CI [.00, .00]). The variance of the intercept was constrained to be zero in subsequent models to reduce model complexity (χ2=0.61(1), p=.43). Next, structured residuals were specified and nested model supported the inclusion of autoregressive pathway for the residuals (χ2=96.96(5), p<.001) and constraining them to be equal with the exception of W2 predicting W3 and W3 predicting W7 (χ2=5.95(3), p=.11) (see Appendix A Figure A3). Modification indices suggested adding a residual covariance between W2 injunctive norms and W7 injunctive norms. Including this covariance resulted in a significant increment in model fit (χ2=19.95(1), p<.001) and the final model including this covariance provided an excellent fit to the data (χ2=11.69(11), p=.38, CFI=.99, TLI=.99, RMSEA=.01, SRMR=.04).

Bivariate LCM-SR Models of Perceived Norms and Alcohol Use

Descriptive Norms and Alcohol Use.

On the between-person level, nested model tests supported the inclusion of covariances between the intercepts and growth factors for descriptive norms and alcohol use (χ2=161.96(6), p<.001). High levels of W3 descriptive norms were associated with high levels of alcohol use at W3 (covariance=0.02, p<.001, 95% CI [.01, .03]), with rapid growth in alcohol use from W1 to W6 (covariance=0.003, p=.01, 95% CI [.001, .005]) and from W6 to W9 (covariance=0.01, p=.04, 95% CI [.003, .01]). The latter covariances support prospective effects of descriptive norms on growth in alcohol use. High levels of W3 alcohol use were associated with rapid growth in descriptive norms (covariance=.002, p=.02, 95% CI [.001, .003]), and this supports prospective between-person effects of alcohol on growth in descriptive norms. Growth in descriptive norms was also associated with growth in alcohol use from W1 to W6 (covariance=0.002, p<.001, 95% CI [.001, .003]) and from W6 to W9 (covariance=.003, p=.002, 95% CI [.001, .005]), suggesting that descriptive norms and alcohol use travel together during adolescent development.

Considering the non-significant variability in the structured residual of descriptive norms at W1 (σ2=0.01, p=.10), the within-time covariance between the structured residuals of alcohol use and descriptive norms at W1 was constrained to zero. Estimating the remaining within-time covariances improved the model fit (χ2=131.98(5), p<.001). Constraining the residual within-time covariances to be equal across time led to a significant decrement in fit (χ2=28.20(4), p<.001). However, constraining the within-time covariances between the structured residuals at W2 and W3 to be equal and at W8 and W9 to be equal (χ2=5.71(2), p=.05) was supported. The significant within-person covariances between the structured residuals for descriptive norms and alcohol use indicated that when a person experiences more than their typical or average level of descriptive norms, he/she is likely to use alcohol use more frequently than his/her typical level of drinking.

Adding within-person cross-lagged paths from descriptive norms to alcohol use (χ2=21.171(5), p<.001) and alcohol use to descriptive norms (χ2=12.71(4), p=.02) was supported as were equality constraints for these cross-lags (from descriptive norms to alcohol use, χ2=0.70(4), p=.95; and from alcohol use to descriptive norms, χ2=8.92(4), p=.06). When an adolescent’s descriptive norms were higher than his/her average or typical normative perception, levels of alcohol increased one year later (B=0.69, p<.001, 95% [.49, .84]). When an adolescent’s alcohol use was higher than his/her average or typical use, levels of descriptive norms increased one year later (B=0.04, p=.001, 95% CI [.02, .05]). Modification indices suggested adding pathways from W3 descriptive norms to alcohol use at W5, W6, and W7, and adding these paths improved model fit (χ2=18.62(3), p<.001). Higher than average descriptive norms at W3 were significantly associated with subsequent increases in W5, W6, and W7 levels of alcohol use (B=0.60, p<.001, 95% CI [.33, .86]). The final bivariate LCM-SR model for descriptive norms and alcohol use, depicted in Figure 1, provided a good fit to the data (χ2=219.24(119), p<.001, CFI=.96, TLI=.95, RMSEA=.04, SRMR=.06).

Figure 1.

Figure 1.

Open in a new tab

Bivariate LCM-SR for alcohol use and descriptive norms. AU=alcohol use, DN= descriptive norms, ε=structured residual, β=latent slope, α=latent intercept. Standardized regression coefficients are presented in the figure and superscripts indicate which parameters were constrained to be equal based on nested model tests. All presented coefficients are significant at p<.05.

Injunctive Norms and Alcohol Use.

Between-person covariances among growth factors for injunctive norms and alcohol use were supported by nested model tests (χ2=128.27(3), p<.001). The slope of injunctive norms was associated with high levels of alcohol use at W3 (covariance=0.003, p=.001, 95% CI [.001, .004]), and more rapid growth in alcohol use from W1 to W6 (covariance=0.001, p=.001, 95% CI [.000, .002],) and from W6 to W9 (covariance=0.004, p<.001, 95% CI [.002, .006]). These between-person covariances suggest that alcohol use prospectively effects the development of injunctive norms and that injunctive norms and alcohol use travel together during adolescence. Recall that there was little variability in the injunctive norms intercept, precluding testing the prospective association between injunctive norms at W3 and growth in alcohol use.

On the within-person level, nested model tests supported the inclusion of within-time covariances between the structured residual for injunctive norms and alcohol use (χ2=122.61(6), p<.001). Constraining them to be equal across time led to a significant decrement in fit (χ2=35.53(4), p<.001). Constraining the within-time covariances between the structured residuals at W1, W2, W3, and W9 to be equal and at W7 and W8 to be equal led to a non-significant decrement in fit (χ2=6.44(4), p=.16). High levels of injunctive norms were associated with high levels of alcohol use (W1, W2, W3, and W9 covariance=0.04, p<.001, 95% CI [.03, .04]; W7 and W8 covariance=.09, p<.001, 95% CI [.07, .12]). This positive within-person association suggests that when a person experiences more than their typical or average level of injunctive norms, he/she is likely to use alcohol use more frequently.

Adding within-person cross-lagged paths from injunctive norms to alcohol use (χ2=67.84(5), p<.001) and alcohol use to injunctive norms (χ2=13.76(5), p=.02) was supported by nested model tests as were across time equality constraints for these cross-lagged paths (injunctive norms to alcohol use, χ2=7.53(4), p=.11; alcohol use to injunctive norms, χ2=9.06(4), p=.05). When an adolescent’s injunctive norms were higher than his/her average or typical normative perception, levels of alcohol increased one year later (B=0.81, p<.001, 95% CI [.65, .97]). When an adolescent’s alcohol use was higher than his/her average or typical use, levels of injunctive norms increased one year later (B=0.02, p=.01, 95% CI [.01, .04]). Modification indices suggested adding pathways from W3 injunctive norms to alcohol use at W5, W6, and W7. Nested model tests supported including these three pathways (χ2=65.87(3), p<.001). Higher than average injunctive norms at W3 were significantly associated with increases in alcohol use at W5, W6, and W7 (B=0.90, p<.001, 95% CI [.67, 1.12]). The bivariate LCM-SR model for injunctive norms and alcohol use, depicted in Figure 2, provided a good fit to the data (χ2=225.236(124), p<.001, CFI=.96, TLI=.95, RMSEA=.04, SRMR=.07).

Figure 2.

Figure 2.

Open in a new tab

Bivariate LCM-SR for alcohol use and injunctive norms. AU=alcohol use, IN= injunctive norms, ε=structured residual, β=latent slope, α=latent intercept. Standardized regression coefficients are presented in the figure and superscripts indicate which parameters were constrained to be equal based on nested model tests. All presented coefficients are significant at p<.05.

Sensitivity Analyses

Cross-lagged panel models for descriptive norms and alcohol use and injunctive norms and alcohol use were estimated in Mplus to determine whether these models replicated the cross-lagged effects found in our LCM-SR models. The cross-lagged panel models for descriptive norms and alcohol use (χ2=221.55(131), p<.001, CFI=.96, TLI=.95, RMSEA=.04, SRMR=.06) and injunctive norms and alcohol use (χ2=214.48(129), p<.001, CFI=.96, TLI=.95, RMSEA=.04, SRMR=.06) both provided a good fit to the data. Models replicated the LCM-SR models such that descriptive and injunctive norms were consistently associated with alcohol use at the next time point and descriptive and injunctive norms at W3 were associated with alcohol use at W5, W6, and W7. Alcohol use significantly predicted descriptive and injunctive norms across-waves.

A benefit of LCM-SR models is that they are thought to provide more accurate as well as less biased parameter estimates of cross-lagged effects (Curran et al., 2014). Table 2 compares cross-lagged parameter estimates from norms to alcohol use and alcohol use to norms. Comparison of the cross-lagged reciprocal pathways indicated that 19 out of the 26 reciprocal pathways were stronger in the LCM-SR models relative to the cross-lagged panel models, and this is expected given that failure to account for growth (as is done in conventional cross-lagged panel models) can lead to bias. These findings highlight the utility of disentangling between- and within-person effects for finding stronger reciprocal pathways when studying constructs with underlying growth processes.

Table 2.

Comparison of Standardized Parameter Estimates for LCM-SR and Cross-Lagged Panel Models

Alcohol Use and Descriptive Norms Alcohol Use and Injunctive Norms
Descriptive Norms Predicting Alcohol Use Alcohol Use Predicting Descriptive Norms Injunctive Norms Predicting Alcohol Use Alcohol Use Predicting Descriptive Norms
Path LCM-SR CLPM LCM-SR CLPM LCM-SR CLPM LCM-SR CLPM
W1 Predicting W2 0.11 0.14 0.05 0.05 0.28 0.19 0.03 .10
W2 Predicting W3 0.25 0.21 0.08 0.22 0.31 0.21 0.04 .14
W3 Predicting W4 0.28 0.22 - - 0.34 0.21 - -
W3 Predicting W5 0.23 0.12 - - 0.34 0.15 - -
W3 Predicting W6 0.23 0.11 - - 0.34 0.13 - -
W3 Predicting W7 0.13 0.08 - - 0.21 0.10 - -
W6 Predicting W7 - - 0.08 0.33 - - 0.07 0.25
W7 Predicting W8 0.16 0.13 0.15 0.13 0.15 0.28 0.12 0.10
W8 Predicting W9 0.18 0.13 0.17 0.13 0.16 0.10 0.14 0.12
Open in a new tab

Note. LCM-SR=latent curve model with structured residuals, CLPM =cross-lagged panel model, - = non-estimated path. All reported pathways are significant at p<.05.

Discussion

Perceived drinking norms are a prominent feature of theories of alcohol use and are robust predictors of alcohol use, however, few studies have examined theoretically informed models of how social norms develop and influence alcohol use (Mollen et al., 2010). Reciprocal determinism suggests that perceived drinking norms and alcohol use develop through a reciprocal learning process whereby these constructs both shape and are shaped by one another (Bandura, 1978). Expanding on prior work that assessed reciprocal determinism primarily with college student samples and only examined between-person associations, the current study assessed reciprocal associations between descriptive and injunctive social norms and alcohol use on between- and within-person levels of analysis from early through late adolescence. Results supported reciprocal determinism for both descriptive and injunctive norms with alcohol use and provide further evidence for the important role of perceived drinking norms in the development of adolescent alcohol use.

Development of Social Norms and Alcohol Use

In support of our first hypothesis, both descriptive and injunctive norms increased from early through late adolescence. Evidence for growth in descriptive norms is consistent with the two prior studies that have demonstrated significant increases in descriptive norms during adolescence (Duan et al., 2009; Weaver et al., 2011). To our knowledge, the current study is the first to demonstrate significant growth in injunctive norms during adolescence, suggesting that adolescents perceived approval of drinking as well as perceived normativeness of drinking increases with age. Additionally, in line with the large body of work, we found significant growth in alcohol use from early through late adolescence. These findings highlight that adolescence is a dynamic period for studying drinking norms (Miech et al., 2017). Furthermore, that growth in norms and alcohol use suggests that traditional cross-lagged panels are inappropriate for testing reciprocal determinism and this point is discussed in more detail below.

Our findings provided strong support for between-person associations between descriptive norms and alcohol use and injunctive norms and alcohol use in the form of growth factor covariances (hypothesis 2). Of note is that perceiving drinking as normative in early adolescence (W3) was associated with rapid escalation in alcohol use (W1 to W6 and W6 to W9), and that high levels of drinking in early adolescence (W3) was associated with rapid escalation in perceived descriptive norms. Similarly, growth in perceptions of peer approval of drinking was associated with high levels of alcohol use at W3 and more rapid growth in alcohol use (from W1 to W6 and W6 to W9). Non-significant variability in W3 injunctive norms precluded assessing its relationship with growth in alcohol use. These findings are in line with prior work demonstrating that growth in descriptive norms and alcohol use during adolescence are related (Duan et al., 2009; Weaver et al., 2011) and is the first study, to our knowledge, to demonstrate that growth in injunctive norms and alcohol are related during adolescence. Scalco and colleagues (2016) found adolescents who misperceived their peers’ substance use were more likely to select into groups where peers were engaged in substance use. Moreover, alcohol use was found to strongly influence whether an adolescent perceived their peers to be engaging in substance use. Taken together, one explanation for the associations between growth in descriptive and injunctive social norms and alcohol use is that both perceiving alcohol as normative and acceptable and drinking alcohol may lead adolescents to select peers who are more accepting of and regularly engage in drinking.

At the within-person level, cross-lagged associations provided strong support for reciprocal determinism (hypothesis 3). Specifically, descriptive and injunctive norms consistently predicted alcohol use such that when norms were elevated at a given time behind an adolescent’s typical normative perception, his/her drinking increased in the following year. Conversely, during years when an adolescent’s alcohol use was higher than usual, perceived descriptive and injunctive norms increased the following year. These findings are the first to demonstrate reciprocal associations between descriptive and injunctive norms and alcohol use on the within-person level of analysis and are consistent with prior studies assessing reciprocal effects on the between-person level of analysis (e.g., Lewis et al., 2015; Marks et al., 1992; Wardell & Read, 2013). Moreover, support for reciprocal associations between perceived drinking norms and alcohol use from early through late adolescence supports reciprocal determinism in that social norms and alcohol use both shape and are shaped by one another.

Interestingly, findings suggest that the strength of the association between norms and alcohol use and alcohol use and norms does not change across adolescence despite considerable differences in the rates of alcohol use, alcohol accessibility, social contexts of alcohol use, and acceptability of drinking from early to late adolescence (Schulenberg & Maggs, 2002). Some have proposed early and middle adolescence (Meisel & Colder, 2015) and late adolescence (Neighbors, Lee, Lewis, Fossos, & Larimer, 2007) as critical developmental periods for social drinking norms. Reciprocal pathways of equivalent magnitude suggest that the entire developmental period of adolescence may be a critical developmental window for the impact of social drinking norms on behavior.

Utility of Distinguishing Between and Within-Person Associations

The use of LCM-SR allowed the current study to build on prior work by examining reciprocal effects on the within-person level of analyses, but also examining between-person associations. Studies to date have failed to account for growth in social norms and alcohol use when testing reciprocal associations, and this likely has led to faulty conclusions regarding reciprocal determinism and to the mixed evidence for this theory. Curran and colleagues (2014) noted that parameter estimates for cross-lagged associations can be biased when models fail to account for underlying growth processes in the constructs being studied. Indeed, compared to the cross-lagged panel models estimated in our sensitivity analyses, the LCM-SR models provided larger standardized regression coefficients of reciprocal effects. Findings from the present study point to the importance of future work accounting for underlying growth processes in descriptive and injunctive social norms when assessing reciprocal associations.

Clinical Implications and Limitations

Despite increasing evidence that norms develop rapidly during adolescence and are robustly associated with alcohol use in early and middle adolescence, norms interventions are primarily used with college student and young adult populations (Perkins, 2002). Considering the strength of within-person prospective associations from descriptive and injunctive norms to alcohol use did not change from early through late adolescence, normative feedback interventions may help reduce drinking behaviors if applied to early and middle adolescence in a developmentally appropriate manner. Indeed, developmentally informed social norms interventions framed to advance the developmental goals of autonomy, competence, and acceptance, have shown promise in altering junk food behaviors (Bryan et al., 2016; Yeager et al., 2018). For example, in an intervention targeting junk food consumption, Bryan et al. (2016) had adolescents read an exposé highlighting industry practices aimed at shaping adolescent junk food behaviors and then read quotes from high status peers stating that they would not eat junk food out of protest. This intervention targeted a highly salient referent group during adolescence for manipulating social norms, high status peers, and facilitated autonomy development by framing industry practices as trying to undermine an adolescent’s independence and freedom to decide whether they wish to consume junk food. Similarly designed social norms interventions that highlight industry practices to promote adolescent alcohol use and use high status peers to alter descriptive and injunctive norms may help reduce the impact of social norms on adolescent drinking behaviors.

The current study contained a number of limitations. First, social norms researchers have noted the importance of assessing the unique effects of descriptive and injunctive norms because these normative influences are often moderately correlated (Meisel et al., 2016). Despite efforts to estimate a trivariate LCM-SR including descriptive norms, injunctive norms, and alcohol use, these models could not be estimated due to convergence issues. The models were likely too complex to be estimated with our sample. Second, the reference group used in measures of perceived norms impacts the strength of the association between norms and alcohol use, and keying the norms measures to different targets may yield different findings than we observed (students in your grade) (Neighbors et al., 2008). Third, our measures were all self-report, which may have inflated some of the associations. Although one has to rely on self-report for perceived norms, other strategies for assessment of alcohol use could be used to reduce this concern (e.g., verified reports from peers, biological measures). Fourth, although the ethnic breakdown of our sample was consistent with the county from whence the sample came, combining individuals across ethnicities who were not non-Hispanic White may have obfuscated differences in social norms and alcohol use for adolescents across multiple ethnic groups (Weaver et al., 2011). Lastly, our measurement of alcohol frequency changed from W6 to W7. Our observed longitudinal trends are similar to those found in other studies, and are consistent with a research suggesting a steep increment in alcohol use after high school, however, we cannot eliminate the possibility that some of the increases in alcohol use we observed are partially due to change in response format.

Conclusion

Perceived norms play a critical role in many etiological models of alcohol use, yet few studies have examined the development of these norms during adolescence. The current study provided strong evidence for reciprocal determinism, a tenet of Bandura’s (1978) Social Learning Theory, which provides a theoretical explanation for the development of descriptive and injunctive norms. Specifically, after accounting for between-person associations between norms and alcohol use, descriptive and injunctive norms both prospectively predicted and were predicted by alcohol use. The current study demonstrates the importance of disaggregating between- and within-person effects when examining reciprocal effects between social norms and alcohol use in adolescence and highlights social norms as an important target for alcohol use interventions.

Acknowledgments:

This research was supported by a grant from the National Institute on Drug Abuse (R01DA019631) awarded to Craig R. Colder and a grant from the National Institute of Alcohol Abuse and Alcoholism (F31AA025521) awarded to Samuel N. Meisel.

Appendix A

Figure A1.

Figure A1.

Open in a new tab

Model implied means for univariate LCM-SR for alcohol use. Panel A depicts model implied means for log-transformed past year alcohol frequency and Panel B depicts model implied means using non-transformed past year alcohol frequency.

Figure A2.

Figure A2.

Open in a new tab

Model implied means for univariate LCM-SR for descriptive norms. Panel A depicts model implied means for log-transformed descriptive norms and Panel B depicts model implied means using the original metric for descriptive norms (1 = none to 6 = all).

Figure A3.

Figure A3.

Open in a new tab

Model implied means for univariate LCM-SR for injunctive norms. Panel A depicts model implied means for log-transformed injunctive norms and Panel B depicts model implied means using the original metric for injunctive norms (1 = strongly disapprove to 5 = strongly approve).

References

  1. Balsa AI, Homer JF, French MT, & Norton EC (2011). Alcohol use and popularity: Social payoffs from conforming to peers’ behavior. Journal of Research on Adolescence, 21(3), 559–568. 10.1111/j.15327795.2010.00704.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bandura A (1969). Principles of Behavior Modification. New York: Holt; Rinehart & Winston. [Google Scholar]
  3. Bandura A (1977). Social Learning Theory. Englewood Cliffs, NJ: Prentice Hall Publishing. [Google Scholar]
  4. Bandura A (1978). The self system in reciprocal determinism. American Psychologist, 33(4), 344 10.1037/0003-066X.33.4.344 [DOI] [Google Scholar]
  5. Borsari B, & Carey KB (2001). Peer influences on college drinking: A review of the research. Journal of Substance Abuse, 13(4), 391–424. 10.1016/S08993289(01)00098-0 [DOI] [PubMed] [Google Scholar]
  6. Brechwald WA, & Prinstein MJ (2011). Beyond homophily: A decade of advances in understanding peer influence processes. Journal of Research on Adolescence, 21(1), 166–179. 10.1111/j.15327795.2010.00721.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bryan CJ, Yeager DS, Hinojosa CP, Chabot AM,Bergen H, Kawamura M, & Steubing F (2016). Harnessing adolescent values to reduce unhealthy snacking. Proceedings of the National Academy of Sciences, USA, 113, 10830–10835. doi: 10.1073/pnas.1604586113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chassin L, Colder CR, Hussong A, & Sher KJ (2016). Substance use and substance use disorders In Cicchetti D (Ed.), Developmental psychopathology: Maladaptation and psychopathology (pp. 833–897). Hoboken, NJ, US: John Wiley & Sons Inc. [Google Scholar]
  9. Cialdini RB, Reno RR, & Kallgren CA (1990). A focus theory of normative conduct: Recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58(6), 1015 DOI: 10.1037/00223514.58.6.1015-1026. [DOI] [Google Scholar]
  10. Collins WA, & Steinberg L (2006). Adolescent development in interpersonal context In Damon W & Lerner RM (Eds.), Handbook of child psychology (pp. 551–578). Hoboken, NJ: Wiley. [Google Scholar]
  11. Crone EA, & Dahl RE (2012). Understanding adolescence as a period of social affective engagement and goal flexibility. Nature Reviews Neuroscience, 13(9), 636–650. doi: 10.1038/nrn3313 [DOI] [PubMed] [Google Scholar]
  12. Curran PJ, Howard AL, Bainter SA, Lane ST, & McGinley JS (2014). The separation of between-person and within-person components of individual change over time: A latent curve model with structured residuals. Journal of Consulting and Clinical Psychology, 82(5), 879–894. DOI: 10.1037/a0035297 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Duan L, Chou CP, Andreeva VA, & Pentz MA (2009). Trajectories of peer social influences as long-term predictors of drug use from early through late adolescence. Journal of Youth and Adolescence, 38(3), 454–465. DOI 10.1007/s10964-008-9310-y [DOI] [PubMed] [Google Scholar]
  14. Elliott DS, & Huizinga D (1983). Social class and delinquent behavior in a national youth panel:1976–1980. Criminology, 21:149–177 doi: 10.1111/j.17459125.1983.tb00256.x. [DOI] [Google Scholar]
  15. Enders CK (2013). Dealing with missing data in developmental research. Child Development Perspectives, 7(1), 27–31. 10.1111/cdep.12008 [DOI] [Google Scholar]
  16. Farrell AD (1994). Structural equation modeling with longitudinal data: Strategies for examining group differences and reciprocal relationships. Journal of Consulting and Clinical Psychology, 62, 477–487. doi: 10.1037/0022-006X.62.3.477 [DOI] [PubMed] [Google Scholar]
  17. Geber S, Baumann E, & Klimmt C (2017). Where do norms come from? Peer communication as a factor in normative social influences on risk behavior. Communication Research, 0093650217718656 10.1177/0093650217718656 [DOI] [Google Scholar]
  18. Gerrard M, Gibbons FX, Reis-Bergan M, Trudeau L, Vande Lune LS, & Buunk B (2002). Inhibitory effects of drinker and nondrinker prototypes on adolescent alcohol consumption. Health Psychology, 21(6), 601–609. DOI: 10.1037//02786133.21.6.601. [DOI] [PubMed] [Google Scholar]
  19. Gmel G, & Lokosha O (2000). Self-reported frequency of drinking assessed with a closed-or open-ended question format: a split-sample study in Switzerland. Journal of Studies on Alcohol, 61(3), 450–454. 10.15288/jsa.2000.61.450 [DOI] [PubMed] [Google Scholar]
  20. Jacobson RP, Mortensen CR, & Cialdini RB (2011). Bodies obliged and unbound: Differentiated response tendencies for injunctive and descriptive social norms. Journal of Personality and Social Psychology, 100(3), 433–448. DOI: 10.1037/a0021470. [DOI] [PubMed] [Google Scholar]
  21. Johnston LD, O’Malley PM, & Bachman JG (2003). Monitoring the Future: National Survey Results on Drug Use, 1975–2002 Volume I: Secondary School Students (NIH Publication No. 03–5375). National Institute on Drug Abuse, Bethesda,MD. [Google Scholar]
  22. Johnston LD, O’Malley PM, Bachman JG, & Schulenberg JE (2010). Monitoring the Future national survey results on drug use, 1975–2009. Volume I: Secondary school students (NIH Publication No. 10–7584). Bethesda, MD: National Institute on Drug Abuse, 734 pp. [Google Scholar]
  23. Lee CM, Geisner IM, Patrick ME, & Neighbors C (2010). The social norms of alcohol related negative consequences. Psychology of Addictive Behaviors, 24(2), 342–348. doi: 10.1037/a0018020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lewis MA, Litt DM, & Neighbors C (2015). The chicken or the egg: Examining temporal precedence among attitudes, injunctive norms, and college student drinking. Journal of Studies on Alcohol and Drugs, 76(4), 594–601. 10.15288/jsad.2015.76.594 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Litt DM, Lewis MA, Rhew IC, Hodge KA, & Kaysen DL (2015). Reciprocal relationships over time between descriptive norms and alcohol use in young adult sexual minority women. Psychology of Addictive Behaviors, 29(4), 885–893. 10.1037/adb0000122 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Neighbors C, Dillard AJ, Lewis MA, Bergstrom RL, & Neil TA (2006). Normative misperceptions and temporal precedence of perceived norms and drinking. Journal of Studies on Alcohol, 67(2), 290–299. 10.15288/jsa.2006.67.290 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Neighbors C, Lee CM, Lewis MA, Fossos N, & Larimer ME (2007). Are social norms the best predictor of outcomes among heavy-drinking college students?. Journal of Studies on Alcohol and Drugs, 68(4), 556–565. 10.15288/jsad.2007.68.556 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Neighbors C, O’connor RM, Lewis MA, Chawla N, Lee CM, & Fossos N (2008). The relative impact of injunctive norms on college student drinking: The role of reference group. Psychology of Addictive Behaviors, 22(4), 576–581. DOI: 10.1037/a0013043. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Maisto SA, Carey KB, & Bradizza CM (1999). Social learning theory In: Leonard KE, & Blane HT (Eds.), Psychological theories of drinking and alcoholism (2nd ed.). New York: Guilford Press. [Google Scholar]
  30. Marks G, Graham JW, & Hansen WB (1992). Social projection and social conformity in adolescent alcohol use: A longitudinal analysis. Personality and Social Psychology Bulletin, 18(1), 96–101. 10.1177/0146167292181014 [DOI] [Google Scholar]
  31. Marsh HW, Hau KT, & Wen Z (2004). In search of golden rules: Comment on hypothesis testing approaches to setting cutoff values for fit indexes and dangers in overgeneralizing Hu and Bentler’s (1999) findings. Structural Equation Modeling, 11(3), 320–341. 10.1207/s15328007sem1103_2 [DOI] [Google Scholar]
  32. Miech RA, Johnston LD, O’Malley PM, Bachman JG, & Schulenberg JE, & Patrick ME (2017). Monitoring the Future national survey results on drug use, 1975–2016: Volume I, secondary school students. Ann Arbor: Institute for Social Research, The University of Michigan, 683 pp. [Google Scholar]
  33. Meisel SN, & Colder CR (2015). Social goals and grade as moderators of social normative influences on adolescent alcohol use. Alcoholism: Clinical and Experimental Research, 39(12), 2455–2462. 10.1111/acer.12906 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Meisel SN, Colder CR, & Read JP (2016). Addressing Inconsistencies in the Social Norms Drinking Literature: Development of the Injunctive Norms Drinking and Abstaining Behaviors Questionnaire. Alcoholism: Clinical and Experimental Research, 40(10), 2218–2228. 10.1111/acer.13202 [DOI] [PubMed] [Google Scholar]
  35. Mollen S, Rimal RN, & Lapinski MK (2010). What is normative in health communication research on norms? A review and recommendations for future scholarship. Health Communication, 25(6–7), 544–547. 10.1080/10410236.2010.496704 [DOI] [PubMed] [Google Scholar]
  36. Muthén LK, & Muthén BO (1998–2018). Mplus User’s Guide. Eight Edition Los Angeles, CA: Muthén & Muthén. [Google Scholar]
  37. Pedersen ER, Osilla KC, Miles JN, Tucker JS, Ewing BA, Shih RA, & D’Amico EJ (2017). The role of perceived injunctive alcohol norms in adolescent drinking behavior. Addictive Behaviors, 67, 1–7. 10.1016/j.addbeh.2016.11.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Perkins HW (2002). Social norms and the prevention of alcohol misuse in collegiate contexts. Journal of Studies on Alcohol, supplement, (14), 164–172. 10.15288/jsas.2002.s14.164 [DOI] [PubMed] [Google Scholar]
  39. Read JP, Wood MD, & Capone C (2005). A prospective investigation of relations between social influences and alcohol involvement during the transition into college. Journal of Studies on Alcohol, 66(1), 23–34. 10.15288/jsa.2005.66.23 [DOI] [PubMed] [Google Scholar]
  40. Rimal RN, & Real K (2005). How behaviors are influenced by perceived norms: A test of the theory of normative social behavior. Communication Research, 32(3), 389–414. DOI: 10.1177/0093650205275385 [DOI] [Google Scholar]
  41. Satorra A, & Bentler PM (2001). A scaled difference chi-square test statistic for moment structure analysis. Psychometrika, 66(4), 507–514. 10.1007/BF02296192 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Scalco MD, Meisel SN, & Colder CR (2016). Misperception and accurate perception of close friend substance use in early adolescence: Developmental and intervention implications. Psychology of Addictive Behaviors, 30(3), 300–311. 10.1037/adb0000175 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Schafer JL, & Graham JW (2002). Missing data: our view of the state of the art. Psychological Methods, 7(2), 147–177. DOI: 10.1037//1082-989X.7.2.147. [DOI] [PubMed] [Google Scholar]
  44. Schulenberg JE, & Maggs JL (2002). A developmental perspective on alcohol use and heavy drinking during adolescence and the transition to young adulthood. Journal of Studies on Alcohol, Supplement, (14), 54–70. 10.15288/jsas.2002.s14.54 [DOI] [PubMed] [Google Scholar]
  45. Tabachnick BG, & Fidell LS (2007). Using multivariate statistics. Allyn & Bacon/Pearson Education. [Google Scholar]
  46. Trucco EM, Colder CR, Wieczorek WF, Lengua LJ & Hawk LW (2014). Early adolescent alcohol use in context: How neighborhoods, parents, and peers impact youth. Development and Psychopathology, 26, 425–436. 10.1017/S0954579414000042 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Wardell JD, & Read JP (2013). Alcohol expectancies, perceived norms, and drinking behavior among college students: examining the reciprocal determinism hypothesis. Psychology of Addictive Behaviors, 27(1), 191–196. DOI: 10.1037/a0030653. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Weaver SR, Cheong J, MacKinnon DP, & Pentz MA (2011). Investigating ethnic differences in adolescent alcohol use and peer norms using semi continuous latent growth models. Alcohol and Alcoholism, 46(5), 620–626. 10.1093/alcalc/agr034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. White HR, Labouvie EW, & Papadaratsakis V (2005). Changes in substance use during the transition to adulthood: A comparison of college students and their noncollege age peers. Journal of Drug Issues, 35(2), 281–306. 10.1177/002204260503500204 [DOI] [Google Scholar]
  50. White HR, McMorris BJ, Catalano RF, Fleming CB, Haggerty KP, & Abbott RD (2006). Increases in alcohol and marijuana use during the transition out of high school into emerging adulthood: The effects of leaving home, going to college, and high school protective factors. Journal of Studies on Alcohol, 67(6), 810–822. 10.15288/jsa.2006.67.810 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Yeager DS, Dahl RE, & Dweck CS (2018). Why interventions to influence adolescent behavior often fail but could succeed. Perspectives on Psychological Science, 13(1), 101–122. 10.1177/1745691617722620 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES