Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2012 May 1.
Published in final edited form as: Addict Behav. 2010 Dec 10;36(5):448–455. doi: 10.1016/j.addbeh.2010.12.011

Early Adolescent Cognitions as Predictors of Heavy Alcohol Use in High School

Judy A Andrews 1, Sarah Hampson 1, Missy Peterson 1
PMCID: PMC3102557  NIHMSID: NIHMS258216  PMID: 21195554

Abstract

The present study predicts heavy alcohol use across the high school years (age 14 through 18) from cognitions regarding the use of alcohol assessed in middle school. Using Latent Growth Modeling, we examined a structural model using data from 1011 participants in the Oregon Youth Substance Use Project. In this model, social images and descriptive norms regarding alcohol use in grade 7 were related to willingness and intention to drink alcohol in grade 8 and these variables were subsequently related to the intercept and slope of extent of heavy drinking across the high school years (grades 9 through 12). Across the sample, both descriptive norms and social images influenced the intercept of heavy drinking (in the 9th grade) through willingness to drink alcohol. Multiple sample analyses showed that social images also were directly related to the intercept of heavy drinking, for girls only. Results suggest that cognitions regarding alcohol use in middle school predict subsequent heavy drinking in high school. These findings emphasize the need for prevention programs targeting changing students’ social images and encouraging a more accurate perception of peers’ use when students are in middle school.

Keywords: cognitions, heavy alcohol use, adolescence, predictors

Introduction

The identification of modifiable early risk factors is central to the design of prevention programs targeting behaviors that have deleterious effects on the health of adolescents. Heavy or binge drinking is one such health behavior. Deleterious health effects occur both as a direct consequence of heavy drinking and as a consequence of behaviors associated with heavy drinking. Heavy drinking during adolescence has more serious neurological effects than heavy drinking later in adulthood as a result of the number of structural and functional changes that occur in the brain during adolescence, (White, Ghia, Levin & Schwartwelder, 2000). For example, preliminary evidence suggests that heavy drinking during this developmental period results in a reduction in the volume of the hippocampus (De Bellis, Clark, Beers, Soloff, Boring et al., 2000) and lower coherence of white matter fibers in several regions of the brain, which could compromise decision making and emotional functioning (McQueeny et al., 2009). Several surveys including the Harvard School of Public Health College Alcohol Study and the Southern Illinois University’s Core Institute study, suggest that adolescents who engage in heavy drinking are likely to experience a number of problems beyond that of drunkenness (Jackson, Sher & Park, 2006). Problems include driving while drunk or being driven by someone who has been drinking extensively, unwanted or unintended sexual intercourse, physical and sexual assault, lowered academic achievement, and increased risk for homicide and suicide (Jackson et al., 2006). Thus, heavy drinking directly and indirectly contributes to significant problems in mid to late adolescence, including a substantial proportion of deaths (Chrikritzhs, Jonas, Stockwell, Heale & Dietz, 2001).

Adolescents in the United States tend to initiate heavy drinking when they reach high school age. According to the most recent Monitoring the Future Study (Johnston, O’Malley, Bachman, & Schulenberg, 2009), the annual prevalence of heavy alcohol use, as defined by “being drunk”, among 8th graders is 12.7% increasing to 30.0% among 10th graders and 45.6% among 12th graders. Thus, heavy drinking increases as a function of grade, and almost half of all students engage in heavy drinking in the last year by the time they are in the 12th grade, at age 18.

In this paper we examined the effect of health behavior cognitions, specifically descriptive norms and social images in early adolescence, on the initial level and growth in heavy drinking across the high school years, encompassing ages 14 to 18. Health behavior cognitions have been shown to be modifiable (Andrews, Gordon, Hampson, Christiansen, Gunn, Slovic & Severson, 2010; Andrews, Hampson & Gordon, 2009; Sussman, 1989) and thus are important etiological mechanisms to target in prevention programs. However previous research has not linked health behavior cognitions regarding alcohol use in early adolescence to heavy drinking in mid adolescence, when youth are aged 14 to 18.

The Prototype/Willingness Model of Gibbons and Gerrard (Gibbons, Gerrard, Blanton & Russell, 1998; Gibbons & Gerrard, 1997; Gibbons, Gerrard, & Lane, 2003; Gibbons, Houlihan, & Gerrard, 2009) and the Theory of Planned Behavior (Ajzen, 1988) guided our selection of health behavior cognitions as risk factors. The Prototype/Willingness model is based on the assumption that much of adolescent risky behavior, such as alcohol use and heavy drinking, is the result of responding to fortuitous risk-taking opportunities. In other words, behavior is the result of social reaction rather than intentional action. The path to reactive behavior is through the adolescent’s willingness to engage in that behavior. In contrast, within the Theory of Planned Behavior, adolescent risky behavior is planful and rational, and the path to intentional behavior is through the adolescent’s deliberate intention to engage in the behavior.

In the proposed model, shown in Figure 1, both intentions and willingness predict growth in heavy drinking during the high school years. Consistent with the Prototype/Willingness model, prototypes influence the adolescent’s willingness to engage in the behavior. Prototypes are the social images adolescents have of those who engage in risky behaviors, such as their social images of a typical young person who uses alcohol. Adolescents believe that if they engage in the behavior, others will view them as having the attributes of the prototype. The more favorable the social image (prototype), the more likely adolescents are to be willing to engage in the behavior (Gibbons & Gerrard, 1997; Gibbons, Gerrard, & Lane, 2003). Thus, in the model, social images are hypothesized to influence heavy drinking through willingness.

Figure 1.

Figure 1

Open in a new tab

The reactive and intentional pathways to heavy drinking in high school.

According to the Theory of Planned Behavior, attitudes and normative beliefs influence behavioral intentions. Since social images are evaluative, they are also considered attitudes. Normative beliefs include both descriptive norms, which are operationalized as beliefs about the extent to which other people are perceived to be engaging in the activity, and prescriptive norms or social approval of use. We included descriptive norms in the model which we measured by participants’ perceptions of how many of their peers were drinking alcohol. In the model, consistent with the Theory of Planned Behavior, both social images (comparable to attitudes) and descriptive norms influence heavy drinking through intentions, the rationale or planned pathway. The Prototype/Willingness model also incorporates descriptive norms as antecedent to behavior. However, the Prototype/ Willingness model suggests that the influence of descriptive norms on behavior is mediated by willingness rather than intentions. Thus, consistent with the Prototype/Willingness model, these two variables influence behavior through willingness, the social reaction pathway. Our etiological research (Andrews, Hampson, Barckley, Gerrard & Gibbons, 2008; Hampson, Andrews, Barckley, & Severson, 2006) and that of others (Marcoux & Shope, 1997; Webb, Baer, & McKelvey, 1995) has shown that descriptive norms influence early adolescents’ willingness and intentions to drink alcohol, and eventual use. However, whether descriptive norms regarding alcohol use among middle school students influence the extent of heavy drinking in high school is an empirical question.

Although behavioral intentions and behavioral willingness are moderately correlated, they are expected to have independent effects on behavior (Andrews et al., 2008; Gibbons, et al., 1998). Young people may not intend to try alcohol but, under risk-conducive circumstances, they may be willing to try. Therefore, it is important to address factors influencing the development of both intentions and willingness in alcohol prevention programs. In summary, drawing on these theoretical frameworks, the purpose of the present study is to examine a model relating social images and descriptive norms regarding alcohol use in early adolescence to the growth of their subsequent heavy use of alcohol in mid adolescence (across high school). We hypothesize that both social images and descriptive norms, at Grade 7, will influence subsequent behavior through both adolescents’ willingness and intentions, assessed at Grade 8. Alcohol use, assessed at Grade 6, was included in the model as a control variable.

Gender differences in parameter estimates were also explored. There is some suggestion that, compared to girls, more boys engage in heavy drinking (DHHS, SAMHSA, 2000), binge drink more frequently (Johnston, O’Malley & Bachman, 2002) and engage in daily drinking more often (Gilvarry, McCarthy, McArdle, 1995). Therefore, we hypothesized a higher initial level and greater growth in heavy drinking for boys than girls. There is also some suggestion of gender differences in the etiological process associated with heavy drinking (See Andrews, 2005, for a review). Both social images and descriptive norms can be conceptualized as cognitions that are affected by perceptions of peers. Results from adolescent studies investigating gender differences on the effects of peer influences on substance use have been mixed. While a majority of studies do not find gender differences (e.g., Andrews, Tildesley, Hops & Li, 2002; Schulenberg et al., 1999, Tucker, Martinez, Ellickson & Edelen, 2008, when gender differences were found, the effect of peer influence was stronger for girls than boys (Berndt & Keefe, 1995; Brooks, Stewig & Croy, 1998; Duncan, Duncan, & Hops, 1994; Kung & Farrell, 2001). Thus, we hypothesized a stronger relation between social images and descriptive norms and subsequent willingness and intentions, and ultimate heavy use, for girls than boys.

Data are from a community sample of youth, participating in an ongoing cohort-sequential longitudinal study the Oregon Youth Substance Use Project (OYSUP). Our previous work with this data set (Andrews et al., 2008; Hampson et al., 2006; Hampson, Andrews, & Barckley, 2007), and that of others (Gibbons & Gerrard, 1997) suggests that adolescent’s cognitions are reliable and valid predictors of subsequent behavior.

Method

Overview of Design

OYSUP is an ongoing cohort-sequential longitudinal project (Schaie, 1965; 1970), funded by the National Institute of Drug Abuse (DA10767). OYSUP began in the 1997– school year wherein students in five grade cohorts (defined by grade at T1), were in the first through fifth grade. These five grade cohorts were assessed annually or biannually for 10 assessments, across 11 years until the 2007 – 2008 school year. Funding necessitated skipping one assessment, between T4 and T5. For this paper, data are from assessments conducted with adolescents who were in the 6th through the 12th grade. Thus data from four cohorts were used in Grades 6, 7, 8, and 9. Data from all five cohorts was used in Grade 10, and since the youngest cohorts had not yet reached 11th and 12th grade during the 2007 – 2008 school year, data from four cohorts was used in Grade 11 and data from three cohorts was used in Grade 12. To analyze data by grade, it was necessary to combine data from students across cohorts.

Participants

Participants were recruited from a single school district in a working class community in Western Oregon. Using a stratified random sample, parents of 2,127 students in 15 elementary schools were sent a letter followed by a phone call describing the project and soliciting participation. Prior to T1, we obtained parental consent for 1075 students (50.7%) to participate in assessments for the first four years of the study. An average of 215 students per grade (1st through 5th) participated in the study at T1 with an even distribution by gender (50.3% female, N = 538). As reviewed in Andrews, Tildesley, Hops, Duncan & Severson (2003), the T1 participants were comparable to elementary students in the district on race/ethnicity and participation in the free-lunch program. However, they had significantly higher (albeit a small difference) on academic achievement tests in both reading and math. Students in the study were comparable to students in Oregon on 30 day prevalence of use of all substances in the 6th grade (DHS, State of Oregon, 2000), with the exception of inhalants. The prevalence of inhalant use was slightly higher in the Oregon sample than in the OYSUP sample.

For this paper, we included data from those 1011 participants (199 in cohort 1, 203 in cohort 2, 300 in cohort 3, 207 in cohort 4 and 200 in cohort 5) who participated in at least one of the grade 6 through grade 12 assessments. Half of the sample were female (50.0%), 85.5% were Caucasian, and 40.3% of the sample were of low income as indicated by their eligibility for free and/or reduced lunch. At grade 6, they were an average age of 13.41. Most (93.6%) of their mothers had graduated from high school and 70.9% had further education, with 14.3% graduating from college. Similarly, the majority of fathers had graduated from high school (88.2%), with 77.8% having further education and 19% graduating from college.

Attrition

The 1011 participants in the data set for this paper were compared to the 64 who did not participate in the study in grade 6 through grade 12. The two samples were comparable on demographic variables, including grade, gender, race/ethnicity, father’s education and mother’s education, and income (as measured by eligibility for free lunch).

Assessment Procedures

Adolescents completed a questionnaire annually. If they continued to attend school in the study school district and were present for assessment days, they were assessed at school (71.0% of 6th through 8th graders; 50.8% of 9th through 12th graders). If they were absent for assessment days or if they lived outside of the district but within driving range of the Oregon Research Institute, they were assessed at the institute (16.0% of 6th through 8th graders; 21.9% of 9th through 12th graders). If they did not live within driving range, sixth through eighth graders were assessed via the telephone (13.0% of sixth through eighth graders) and ninth through twelve graders completed mailed questionnaires (21.9%). If ninth through twelve graders did not return the questionnaire, they were assessed via the telephone (5.5%).

Measures

Alcohol use in the last year at grade 6

Alcohol use in the last year, was included as a control variable. It was assessed by the single item, “During the last 12 months did you drink alcohol”. Using a six point scale, responses ranged from “never” (0) to “some each day” (5).

Social images at grade 7

Characteristics of substance users for the assessment of social images were selected from a list of attributes of smokers examined by Dinh and colleagues (1995) in a prospective study of 5th and 7th graders. Attributes selected for the present study were “exciting”, “cool or neat”, and “popular” (See Andrews & Peterson, 2006, for more details regarding attribute selection). To assess social images, a predictor, students were asked if they thought that “kids who drink alcohol” were each of these attributes. A three-point response format was used for each item, with “Yes” coded as 2, “No” as 0, and “Maybe” as 1. Responses were averaged across the three items. Cronbach’s alpha for this measure in grade 7 was .78.

Descriptive norms at grade 7

To assess peer-based descriptive norms, a predictor, 7th graders were asked “How many of the kids at school or in the neighborhood have tried a drink of alcohol (beer, wine, or hard liquor)?” and “How many of your friends have tried a drink of alcohol?” (Responses were coded “None” = 0, “Some”, “Most” or “All” = 1 to both questions). Responses were summed across the two items. The phi correlation between these two items was .49. Hampson et al. (2006) examined the convergent and discriminate correlations among the two items assessing norms and the three items assessing social images and found that the items assessing the same construct were consistently higher (mean convergent r for prototype items = .40, mean convergent r for norm items = .34) than the correlations between items assessing divergent constructs (mean divergent r = .14).

Behavioral intentions

To assess intentions in the 8th grade, a hypothesized mediator, students were asked the following two items, “Do you think you would drink alcohol when you are an adult?” and “when you are in high school?”. Responses to each item were “No” (coded as 0), “Maybe” (coded as 1) and “Yes” (coded as 2). Responses were summed across the two items. For 8th graders, the correlation between the two items was .61.

Behavioral willingness

To assess willingness in the 8th grade, a hypothesized mediator, students were given the following scenario, “Suppose you were with a group of friends and there was some alcohol there that you could have if you wanted. How willing you would be to…”. Three items assessing willingness followed this statement. Items were “drink one drink”, “have more than one drink”, and “get drunk”. Students indicated their willingness to engage in each behavior on a 5-point Likert type scale, ranging from “not at all willing” (1) to “very willing” (5). Responses were averaged across the three items. For 8th graders, the internal consistency across the three items, as measured by Cronbach’s alpha was .90.

Heavy alcohol use

To assess heavy drinking in grade 9 through grade 12, which was used to construct the criterion, growth in heavy alcohol use, we used two items, to increase the reliability of this construct. The first item was “How many times in the last month have you had 5 or more drinks of beer, wine, or hard liquor?” Responses were: “zero” (0), “once” (1), “twice” (2), “3 to 5” (3), “6 to 9” (4), and “10 or more times” (5). The second item was “How many times in the last 12 months have you gotten really drunk from drinking too much alcohol?” Responses on the second item were recoded to range from 0 to 5, with “never” (0), “1 to 2 times” (1.67), “3 or 4 times” (3.34) and “more than 5 times” (5), to give both items equal weighting. The responses (0 to 5) on these two items were summed to form the construct, heavy alcohol use, with a scale from 0 to 10. The average correlation between these two items across the four grades was .57, with a range from .55 to .60.

Overview of Analyses

We used the M-Plus program, version 3.0 (Muthen & Muthen, 1994–2004) to test the fit of the model to the data. Within this model, missing data are estimated using maximum likelihood estimates. The fit of the model was evaluated with three criteria: The chi-square statistic, the root mean square error of approximation (RMSEA) and the comparative fit index (CFI). The chi-square statistic is the traditional method of assessing model fit and it assesses the fit of the model to the data, specifically the covariance matrix. Using this criterion, a model with a good fit has a non-significant chi-square (it does not deviate significantly from the proposed model). However, since the significance of the chi-square statistic is a function of the sample size, the chi-square statistic is often significant with large samples (Joreskog & Sorbom, 1993). Therefore, other fit criteria are often used in addition to the chi-square. The RMSEA is another indicator of how the model fits the covariance matrix of the population. A smaller RMSEA indicates a better fit. Although cut-offs vary recommended by statisticians vary, a cut-off of .06 is uniformly thought to be represent a good fit (Hu and Bentler, 1999) with a stringent upper limit of .07 (Steiger, 2007). The CFI compares the chi-square value of the model to a null model (where all covariance are set to zero), and takes into account sample size. A CFI greater than .95 is indicative of a good fit (Hu & Bentler, 1999).

Prior to testing the complete model, development in heavy drinking across grades 9 through 12 was modeled using a latent growth model (LGM). Within LGM, measures of variables across grade were used to estimate the intercept (level at grade 9) and the slope (rate of change over time) of heavy drinking. Within the final model, the significance of all indirect paths was tested using the Sobel test (Sobel, 1982). To test for gender differences, we used multiple sample analysis to compare the fit of two models, one with the respective parameter fixed to be equal between the two genders, and the other model with the same parameter freed. A chi-square difference test with one degree of freedom was used to evaluate the difference in fit. Each parameter was tested sequentially while the other parameters in the model remained fixed to be equal.

Results

Cohort differences

Within this cohort sequential design, we collapsed across cohorts to examine and predict growth in heavy alcohol use as a function of grade in school. Predictors included social images and descriptive norms at grade 7 and intention and willingness at grade 8. There were no significant differences as a function of cohort, as defined by grade at T1, on social images of alcohol users at grade 7, descriptive norms regarding alcohol use at grade 7, and willingness and intentions at grade 8. There were also no cohort differences in heavy drinking at any of the grades in high school (grades 9 through 12).

Means and standard deviations in variables by gender across grades

The means and standard deviations by gender for social images and descriptive norms for grade 7 and for willingness and intentions at grade 8 are shown in Table 1. Although girls tended to have higher intentions and willingness, and more favorable social images, gender differences were not significant. The means and standard deviations of heavy drinking from grade 9 through 12 are also shown in Table 1. Again, gender differences were not significant. Extent of heavy drinking increased between grades 9 and 12. In grade 9, 20.9% reported heavy drinking in the last year (either drinking five or more drinks in a row or getting drunk), with the prevalence increasing to 28.1% in the 10th grade and 34.5% in the 11th grade. By the time the students reach 12th grade, 45.3% of the students reported heavy drinking in the last year.

Table 1.

Means and Standard Deviations of Social Images, Descriptive Norms, Intention, Willingness, and Heavy Drinking for the Sample (Collapsing across Cohorts) by Relevant Grades and Gender

Males Females
Variable and Grade Scale
range		 n Mean SD n Mean SD
Social Images 0 – 3
  Grade 7 400 .55 1.09 390 .88 1.35
Descriptive Norms 0 – 2
  Grade 7 397 1.13 .82 393 1.19 .85
Willingness 1 – 5
  Grade 8 190 1.34 .76 207 1.58 .95
Intention 0 – 4
  Grade 8 380 1.31 1.23 401 1.55 1.35
Heavy Drinking 0 – 5
  Grade 9 378 .60 1.65 394 .71 1.66
  Grade 10 452 .87 1.90 470 1.05 2.00
  Grade 11 364 1.04 1.90 371 1.31 2.15
  Grade 12 262 1.59 2.31 271 1.64 2.30
Open in a new tab

Note: Data was not imputed for this table. An assessment was skipped, resulting in data from four cohorts for Grades 7 – 9; The measurement of willingness did not begin until T5, thus willingness was measured in two cohorts; The youngest cohorts were not assessed in grades 11 or 12.

Latent Growth Model for heavy alcohol use

A latent growth model using measures of heavy alcohol use across high school (grades 9 through 12) fit the data moderately well (X2 (5, n = 962) = 26.58, p<.0001; CFI = .972, RMSEA =.067, 90% C. I. =.043, .093). The intercept (.661) and slope (.292) were both positive and significant (p<.001) suggesting that they both were greater than zero (p<.001). The variances of both the intercept (1.495) and slope (.232) were also both significant (p<.001). The slope was not significantly correlated with the intercept (r = .062, ns). Multiple sample analysis by gender showed no significant differences in any parameter, including mean and variance of the intercept and slope. Thus, there were no gender differences in level or growth of heavy drinking across time.

Model relating cognitions in early adolescence to heavy alcohol use

We tested the fit of a model wherein social images and descriptive norms regarding alcohol use at grade 7 were related to willingness and intentions at grade 8. Willingness and intentions at grade 8 were subsequently related to the intercept and slope of heavy drinking across the four years of high school. Alcohol use during the past year at grade 6 was included in the model as a control variable. Thus, the paths from 6th grade alcohol use to prototype and descriptive norms at grade 7, willingness and intentions at grade 8, and the intercept and slope of heavy drinking were included in the model. The final model fit the data well (X2 (14, n = 1011) = 34.37, p<.01; CFI = .987, RMSEA =.038, 90% C. I. =.022, .054). The final model is shown in Figure 2. All paths are standardized. As shown, the direct paths from alcohol use in the last year at grade 6 to all variables with the exception of the intercept (initial level) and slope of heavy drinking were significant. The intercept of heavy drinking was predicted by social images both directly and indirectly through willingness and by descriptive norms indirectly through willingness. A Sobel test (Sobel, 1982) of these indirect effects was significant for both descriptive norms through willingness (Sobel = 3.28, p<.01) and social images through willingness (Sobel = 3.20, p<.01) to the intercept of heavy drinking.

Figure 2.

Figure 2

Open in a new tab

The indirect and direct effects of subjective norms and social images on the initial level and growth of heavy drinking in grades 9 through 12.

As shown in Table 2, multiple sample analyses by gender showed a significant difference on four parameter estimates. For three of these, the effects were stronger for girls than for boys. The path from the willingness to the slope of alcohol was in different directions for boys than girls, and for both genders, this effect was not significant. Social images significantly predicted the intercept of heavy drinking only for girls. All other significant paths in the model in Figure 1, were significant for both genders.

Table 2.

Gender Differences in Standardized Path Coefficients

Gender
Path χ2 Difference test Boys Girls
Slope of heavy drinking on willingness 5.24 .25 −.12
Intercept of heavy drinking on social images 3.80 .04 .20***
Social images on grade 6 alcohol use 5.38 .16** .25***
Social images with subjective norms 7.04 .28*** .39***
Open in a new tab

Note: all χ2‘ significant at p<.05

***

path significant at p<.001

**

path significant at p<.01

Discussion

Despite controlling for the alcohol use in the 6th grade, this study showed that both adolescents’ social images regarding alcohol use and their descriptive norms assessing perception of alcohol use among their peers when they are in 7th grade predict the initial level of heavy use in the 9th grade. This finding is particularly important since it emphasizes the influence of children’s cognitions in early adolescence, affecting their behavior in mid adolescence. It is also of interest that social images and descriptive norms were not regarding adolescents who drink heavily, but were regarding youth, their own age, who drink alcohol, without a reference to extent. Yet, these cognitions influenced a severe form of the behavior, heavy drinking, which included getting drunk and drinking more than five drinks in a row.

Findings were supportive of the Prototype/Willingness Model set forth by Gibbons and Gerrard (Gibbons et al., 1998; Gibbons & Gerrard, 1997; Gibbons et al., 2003). In this model, social images (prototypes) and descriptive norms influence willingness and subsequently health behavior. Consistent with this model, these cognitive variables influenced the intercept of initial level of heavy drinking in 9th grade through willingness, but not through intentions. Thus the path through which cognitions influenced behavior was a socially reactive path, rather than a reasoned path.

In contrast to the theories of Ajzen and Fishbein (Ajzen & Fishbein, 1980; Ajzen, 1988) intentions to drink alcohol in the future were not related to subsequent heavy drinking in mid adolescence. This latter finding is particularly thought provoking, since in previous research conducted with this data set both intentions and willingness predicted extent of alcohol use in the 10th grade (Andrews et al., 2008). There are two possible explanations for this finding. First, the measure assessing behavioral intentions assesses intentions to drink, but does not assess intentions to engage in heavy drinking. In contrast, the measure of willingness included an item assessing willingness to engage in heavy drinking. Thus there is a stronger methodological link between willingness and heavy drinking than between intentions and heavy drinking. However, another intriguing possibility could also explain this finding. For some adolescents, behavioral disinhibition increases as a function of alcohol consumption (Fillmore, Ostling, Martin, & Kelly, 2009). When an adolescent drinks, they are less likely to inhibit their behavior and terminate a behavior (Fillmore, 2003), including alcohol use. Thus, one drink can easily lead to another, resulting in heavy drinking. Perhaps individuals who engage in heavy drinking are not intending to do so, but once they start drinking, it is more difficult to inhibit their behavior. Thus, willingness is a stronger predictor, since if they are more willing to drink and to drink more than one drink, they may be less willing to conscientiously inhibit their behavior. Willingness and disinhibition may work together to lead to heavy drinking.

While health-related cognitions and subsequent willingness in early adolescence predicted initial level, and thus the level of use throughout high school, they did not predict growth or increase in heavy use across the high school years. However, the variance in growth of heavy use was significant and level of initial use and growth were not correlated. This suggests that the predictors of these two constructs, initial level and slope, are independent. Perhaps while early cognitions prospectively influence initial use, more proximal factors, such as associating with peers who use, influence growth in use across the high school years. Further research regarding the identification of etiological factors predictive of this growth is needed.

Gender Differences

Although we predicted gender differences in initial level and growth in heavy alcohol use, we did not find them. This lack of gender differences could be due to secular trends. Johnson and Gerstein (1998), based on retrospective reports from the National Household Survey, noted that the magnitude in gender differences decreased as a function of time, with gender differences progressively narrowing. The most recent report of the MTF Study (Johnston, O’Malley, Bachman & Schulenberg, 2009) did not report differences as a function of gender, but the authors noted that whereas in general males have traditionally had higher rates, by 2005, in 10th grade, girls caught up to boys and have remained equivalent since. Hence the lack of a gender difference found here is reflected in current population based studies.

We found a direct effect of social images on the intercept of heavy drinking that was not mediated through willingness or intentions, for girls only. This finding is supportive of our hypothesis suggesting a stronger effect of peer influence, as assessed by social images, for girls than boys. Since this effect was direct, this finding suggests the potential for other mediators not specified in the model. Further research is needed to identify these meditational mechanisms. A possible mechanism may be related to girls’ relatively greater social awareness and more of an ability to empathize with others as compared to boys (Zahn-Waxler, Race & Duggal, 2005).

Strengths and Limitations

This study has both strengths and limitations. The strengths including the large longitudinal data set providing the ability to predict initial level and growth in heavy drinking across the high school years from health-related cognitions measured in early adolescence. Despite this strength, there are several limitations. The participants in the OYSUP study were from one small working class community, which is primarily Caucasian, in Western United States. Thus the generalizeability of these results to other regions of United States and other countries as well as non-Caucasian racial/ethnic groups is limited.

However, the similarity of prevalence in this sample to data from the most recent Monitoring the Future study (MTF; Johnston et al., 2009) that assessed 46,000 students across 386 secondary schools nationwide provides some support the generalizeability of our findings. Although the MTF study used a different measure of heavy drinking, “feeling drunk”, it reported an annual prevalence of 30.0% in 10th grade, and we found an annual prevalence of 28.1% in 10th grade. Similarly, the MTF study reported an annual prevalence of 45% in 12th grade, and we found an annual prevalence of and 45.3%.

Our measure of heavy drinking has two limitations. First, it is based on self report. Although, self reports of alcohol use among adolescents have good test-retest reliability (Dollinger & Malmquist, 2009; Gruenewald & Johnson, 2006) and we attempted to increase the reliability of the construct by including two items, heavy drinking, like any other self report variable, is subject to respondent bias and social desirability. Second, one of the indicators of heavy drinking, the measure of binge drinking used in this study, i.e., drinking five or more drinks in a row, may underestimate binge drinking for women (Wechsler, Dowdall, Davenport & Rimm, 1995). These authors found that women who drunk four drinks in a row had the same likelihood of experiencing drinking related problems as men who drunk five drinks in a row.

A limitation of structural equation modeling is that a specific model is tested and the fit of the model to the data is evaluated. Alternative models could potentially fit the data as well as the model that is being evaluated. Thus, while the model that we evaluated in this paper fit the data well, other models could have fit as equally well. However, a strength of this study was that the model was derived from two theoretical perspectives and supported by previous work. Further, an additional statistical technique, beyond the fit of the model, was used to assess the significance of the indirect pathways.

Implications for Prevention

The finding that social images and descriptive norms in 7th grade predicted the intercept of heavy drinking in high school suggests the need for prevention programs targeting heavy drinking prior to high school, during the middle or junior high school years. This finding further suggests that programs target social influence variables, such as social images and descriptive norms. These programs must counteract the current cultural norm of heavy drinking among adolescents residing in North American and Western Europe. A major source of influence on alcohol use (Sargent, Wills, Stoolmiller, Gibson, and Gibbons, 2006; Wills, Sargent, Gibbons, Gerrard & Stoolmiller, 2009) is the media. For example Wills and colleagues showed that exposure to alcohol cues in movies predicted level of alcohol use three years later, including binge drinking, in a national sample of ten to fourteen year olds. Further, Cin and colleagues (2009), using data from the same study as Wills showed that the effect of exposure to alcohol use in movies on the extent of alcohol use four years later, was mediated by the adolescent’s social images. These studies further emphasize the need for interventions to counteract the cultural norms promoted by the media.

A prevention program targeting social images and descriptive norms is not without precedence. Previous effective school-based tobacco prevention programs (Sussman, Dent, Burton, Stacy, & Flay, 1995; Andrews et al., 2010) have successfully targeted social influence variables, such as descriptive norms and prototypes, and at least one alcohol prevention program is based on changing prototypes or social images (Gerrard et al., 2006). For example, the Click City©: Tobacco program (Andrews and colleagues, 2010) changed 5th graders social images of cigarette smokers, by showing the student that someone just like them doesn’t think smoking is cool, and illustrating that smokers are uncool and not at all exciting. Further, the Click City©: Tobacco program changed early adolescent’s descriptive norms by giving students feedback regarding what behaviors their classmates actually engaged in, and at the same time, teaching the student about the tendency of people to overestimate the prevalence of behaviors. A program targeting heavy drinking designed for students in early adolescence which includes changing social images and descriptive norms could be potentially effective at counteracting the cultural norms of today.

Research Highlights

  • Prospective study relating cognitions in early adolescence to heavy drinking

  • Norms and social images in early adolescence predict subsequent heavy drinking

  • The pathway to heavy drinking is reactive, through willingness

  • Social images are directly related to subsequent heavy drinking for girls only

Acknowledgements

This research was supported by Grant DA10767 from the National Institute of Drug Abuse. We gratefully acknowledge the assistance of Niraja Lorenz, Martha Hardwick, and Erika Westling for managing the collection of data, the assessment staff for helping with data collection, and Christine Lorenz for manuscript preparation.

Role of Funding Sources

NIDA 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 citable 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.

Contributors

Judy Andrews and Sarah Hampson designed the study and wrote the protocol. Judy Andrews conducted literature searches and provided summaries of previous research studies. Missy Peterson conducted the statistical analysis. Judy Andrews wrote the first draft of the manuscript and all authors contributed to and have approved the final manuscript.

Conflict of Interest

All authors declare that they have no conflicts of interest.

Contributor Information

Judy A. Andrews, Email: judy@ori.org.

Sarah Hampson, Email: sarah@ori.org.

Missy Peterson, Email: missyP@ori.org.

References

  1. Ajzen I. Attitudes, personality and behavior. New York: Open University Press; 1988. [Google Scholar]
  2. Ajzen I, Fishbein M. Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall; 1980. [Google Scholar]
  3. Andrews JA. Substance abuse in girls. In: Bell-Dolan D, Foster S, Mash E, editors. Handbook of behavioral and emotional problems in girls. New York: Kluwer Academic Press/Plenum Publishers; 2005. pp. 181–209. [Google Scholar]
  4. Andrews JA, Gordon J, Hampson SE, Christiansen SM, Gunn B, Slovic P, Severson HH. Short-term efficacy of Click City©: Tobacco: Changing etiological mechanisms related to the onset of tobacco use. 2010 doi: 10.1007/s11121-010-0192-3. Manuscript submitted for publication. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Andrews JA, Hampson SH, Gordon J. Development and efficacy of a school-based tobacco prevention program: A component evaluation approach to changing mediating mechanisms; Paper presented at the annual meeting of the Society for Prevention Research; Washington, D. C.. 2009. [Google Scholar]
  6. Andrews JA, Hampson SE, Barckley M, Gerrard M, Gibbons FX. The effect of early cognitions on cigarette and alcohol use in adolescence. Psychology of Addictive Behaviors. 2008;22:96–106. doi: 10.1037/0893-164X.22.1.96. PMCID: PMC18298235. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Andrews JA, Peterson M. The development of social images of substance users in children: A Guttman unidimensional scaling approach. Journal of Substance Use. 2006;11(5):305–321. doi: 10.1080/14659890500419774. PMCID: PMC2443056. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Andrews JA, Tildesley HH, Hops H, Duncan SC, Severson HH. Elementary school age children’s future intentions and use of substances. Journal of Clinical Child and Adolescent Psychology. 2003;32(4):556–567. doi: 10.1207/S15374424JCCP3204_8. PMCID: PMC1764642. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Andrews JA, Tildesley E, Hops H, Li F. The influence of peers on young adult substance use. Health Psychology. 2002;21:349–357. doi: 10.1037//0278-6133.21.4.349. [DOI] [PubMed] [Google Scholar]
  10. Berndt TJ, Keefe K. Friends' influence on adolescents' adjustment to school. Child Development. 1995;66(5):1312–1329. [PubMed] [Google Scholar]
  11. Brooks A, Stuewig J, LeCroy CW. A family based model of Hispanic adolescent substance use. Journal of Drug Education. 1998;28:65–86. doi: 10.2190/NQRC-Q208-2MR7-85RX. [DOI] [PubMed] [Google Scholar]
  12. Chikritzhs TN, Jonas HA, Stockwell TR, Heale PF, Dietze PM. Mortality and life-years lost due to alcohol: A comparison of acute and chronic causes. Medical Journal of Australia. 2001;174:281–284. doi: 10.5694/j.1326-5377.2001.tb143269.x. [DOI] [PubMed] [Google Scholar]
  13. Cillessen A, Rose A. Understanding Popularity in the Peer System. Current Directions in Psychological Science. 2005;14(2):102–105. [Google Scholar]
  14. Cin SD, Worht KA, Gerrard M, Gibbons FX, Stoolmiller M, Wills TA, Sargent JD. Watching and drinking: Expectancies, prototypes, and peer affiliations mediate the exposure to alcohol use in movies on adolescent drinking. Health Psychology. 2009;28:473–483. doi: 10.1037/a0014777. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. DeBellis MD, Clark DB, Beers SR, Soloff PH, Boring AM, et al. Hippocampal volume in adolescent-onset alcohol use disorders. American Journal of Psychiatry. 2000;157:737–744. doi: 10.1176/appi.ajp.157.5.737. [DOI] [PubMed] [Google Scholar]
  16. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration. Rockville, MD: Author; National Household Survey on Drug Abuse: Main Findings, 1988. 2000
  17. Dinh KT, Sarason IG, Peterson AV, Onstad LE. Children’s perceptions of smokers and nonsmokers: A longitudinal study. Health Psychology. 1995;14(1) doi: 10.1037//0278-6133.14.1.32. 32-30. [DOI] [PubMed] [Google Scholar]
  18. Dollinger SJ, Malmquist D. Reliability and validity of single-item self reports: With special relevance to college students’ alcohol sue, religiosity, study and social life. The Journal of General Psychology. 2009;136:231–241. doi: 10.3200/GENP.136.3.231-242. [DOI] [PubMed] [Google Scholar]
  19. Duncan TE, Duncan SC, Hops H. The effect of family cohesiveness and peer encouragement on the development of adolescent alcohol use: A cohort-sequential approach to the analysis of longitudinal data. Journal of Studies on Alcohol. 1994;55:588–599. doi: 10.15288/jsa.1994.55.588. [DOI] [PubMed] [Google Scholar]
  20. Fillmore MT. Drug abuse as a problem of impaired control: Current approaches and findings. Behavioral Cognitions in Neuroscience, Rev. 2003;3:179–197. doi: 10.1177/1534582303257007. [DOI] [PubMed] [Google Scholar]
  21. Fillmore MT, Ostling EW, Martin CA, Kelly TH. Acute effects of alcohol on inhibitory control and information processing in high and low sensation-seekers. Drug and Alcohol Dependence. 2009;100:91–99. doi: 10.1016/j.drugalcdep.2008.09.007. PMID: 19004578. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Gerrard M, Gibbons FX, Brody GH, Murry VM, Cleveland MJ, Wills TA. A theory-based dual focus alcohol intervention for pre-adolescents: Social cognitions in the Strong African American Families Program. Psychology of Addictive Behavior. 2006;20:185–195. doi: 10.1037/0893-164X.20.2.185. [DOI] [PubMed] [Google Scholar]
  23. Gibbons FX, Gerrard M. Health images and their effects on health behavior. In: Buunk BP, Gibbons FX, editors. Health, coping, and well-being: Perspectives from social comparison theory. Mahwah, NJ: Erlbaum; 1997. pp. 63–94. [Google Scholar]
  24. Gibbons FX, Gerrard M, Blanton H, Russell DW. Reasoned action and social reaction: Willingness and intention as independent predictors of health risk. Journal of Personality and Social Psychology. 1998;74:1164–1181. doi: 10.1037//0022-3514.74.5.1164. [DOI] [PubMed] [Google Scholar]
  25. Gibbons FX, Gerrard M, Lane DJ. A social reaction model of adolescent health. In: Suls JM, Wallston K, editors. The handbook of social-health psychology. Oxford, England: Blackwell; 2003. pp. 107–136. [Google Scholar]
  26. Gibbons FX, Houlihan AE, Gerrard M. Reason and reaction: the utility of a dual-focus, dual-processing perspective on promotion and prevention of adolescent health risk behavior. British Journal of Health Psychology. 2009;14:231–248. doi: 10.1348/135910708X376640. [DOI] [PubMed] [Google Scholar]
  27. Gilvarry E, McCarthy S, McArdle P. Substance use among school children in the north of England. Drug and Alcohol Dependence. 1995;37:255–259. doi: 10.1016/0376-8716(94)01073-t. [DOI] [PubMed] [Google Scholar]
  28. Gruenewald PJ, Johnson FW. The stability and reliability of self-reported drinking measures. Journal of Studies on Alcohol. 2006;67:738–745. doi: 10.15288/jsa.2006.67.738. [DOI] [PubMed] [Google Scholar]
  29. Hampson SE, Andrews JA, Barckley M. Predictors of the development of elementary-school children’s intentions to smoke cigarettes: Hostility, prototypes, and subjective norms. Nicotine and Tobacco Research. 2007;7:751–760. doi: 10.1080/14622200701397908. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Hampson SE, Andrews JA, Barckley M, Severson HH. Personality predictors of the development of elementary-school children’s intentions to drink alcohol: The mediating effects of attitudes and subjective norms. Psychology of Addictive Behaviors. 2006;20:288–297. doi: 10.1037/0893-164X.20.3.288. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling. 1999;6:1–55. [Google Scholar]
  32. Jackson K, Sher K, Park A. Drinking among college students: Consumption and consequences. In: Galanter M, editor. Alcohol problems in adolescents and young adults: Epidemiology, neurobiology, prevention, and treatment. NY: Springer Science; 2006. pp. 85–117. [Google Scholar]
  33. Johnson RA, Gerstein DR. Initiation of use of alcohol, cigarette, marijuana, cocaine, and other substances in US birth cohorts since 1919. American Journal of Public Health. 1998;88:27. doi: 10.2105/ajph.88.1.27. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Johnston LD, O’Malley PM, Bachman JG. Bethesda, MD: National Institute on Drug Abuse; Monitoring the Future: National Survey Results on Drug Use, 1975–2001. 2002
  35. Johnston LD, O’Malley PM, Bachman JG, Schulenberg JE. Bethesda, MD: National Institute on Drug Abuse; Monitoring the Future national results on adolescent drug use: Overview of key findings, 2008 (NIH Publication No. 09-7401) 2009
  36. Joreskog K, Sorbom D. Lisrel 8: Structural Equation Modeling with the SIMPLIS Command Language. Chicago, IL: Scientific Software International, Inc.; 1993. [Google Scholar]
  37. Kung EM, Farrell AD. The role of parents and peers in early adolescent substance use: An examination of mediating and moderating effects. Journal of Child and Family Studies. 2001;9:509–528. [Google Scholar]
  38. Marcoux BC, Shope JT. Application of the Theory of Planned Behavior to adolescent use and misuse of alcohol. Health Education Research. 1997;12:323–331. [Google Scholar]
  39. McQueeny T, Schweinsburg B, Schweinsburg A, Jacobus J, Bava S, Frank L, et al. Altered white matter integrity in adolescent binge drinkers. Alcoholism: Clinical and Experimental Research. 2009;33(7):1278–1285. doi: 10.1111/j.1530-0277.2009.00953.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Muthén LK, Muthén BO. Mplus user’s guide. 3rd ed. Los Angeles, CA: Muthén & Muthén; 1994–2004. [Google Scholar]
  41. Sargent J, Wills T, Stoolmiller M, Gibson J, Gibbons F. Alcohol use in motion pictures and its relation with early-onset teen drinking. Journal of Studies on Alcohol. 2006;67(1):54–65. doi: 10.15288/jsa.2006.67.54. [DOI] [PubMed] [Google Scholar]
  42. Schaie KW. A general model for the study of developmental problems. Psychological Bulletin. 1965;64:92–107. doi: 10.1037/h0022371. [DOI] [PubMed] [Google Scholar]
  43. Schaie KW. A re-interpretation of age-related changes in cognitive structure and functioning. In: Goulet LR, Baltes PB, editors. Life -Span developmental psychology: Research and theory. San Diego, CA: Academic Press; 1970. pp. 485–507. [Google Scholar]
  44. Schulenberg J, Maggs J, Dielman T, Leech S, Kloska D, Shope J, et al. On peer influences to get drunk: A panel study of young adolescents. Merrill-Palmer Quarterly: Journal of Developmental Psychology. 1999;45(1):108–142. [Google Scholar]
  45. Sobel ME. Asymptotic confidence intervals for indirect effects in structural equation models. In: Leinhardt S, editor. Sociological methodology, 1982. Washington, DC: American Sociological Association; 1982. pp. 290–312. [Google Scholar]
  46. Steiger JH. Understanding the limitations of global fit assessment in structural equation modeling. Personality and Individual Differences. 2007;42:893–898. [Google Scholar]
  47. Sussman S. Two social influence perspectives of tobacco use development and prevention. Health Education Research: Theory and Practice. 1989;4:213–223. [Google Scholar]
  48. Sussman S, Dent CW, Burton D, Stacy AW, Flay BR. Developing school-based tobacco use prevention and cessation programs. Thousand Oaks, CA: Sage Publications; 1995. [Google Scholar]
  49. Tucker JS, Martinez JF, Ellickson PL, Edelen MO. Temporal associations of cigarette smoking with social influences, academic performance, and delinquency: A four-wave longitudinal study from ages 13 to 23. Psychology of Addictive Behaviors. 2008;22:1–11. doi: 10.1037/0893-164X.22.1.1. [DOI] [PubMed] [Google Scholar]
  50. Webb JA, Baer PE, McKelvey RS. Development of a risk profile for intentions to use alcohol among fifth and sixth graders. Journal of the American Academy of Child and Adolescent Psychiatry. 1995;34:1772–1778. doi: 10.1097/00004583-199506000-00018. [DOI] [PubMed] [Google Scholar]
  51. Wechsler H, Dowdall GW, Davenport A, Rimm EG. A gender-specific measure of binge drinking among college students. American Journal of Public Health. 1995;85:982–985. doi: 10.2105/ajph.85.7.982. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. White AM, Ghia AJ, Levin ED, Schwartwelder HS. Binge pattern alcohol exposure: Differential impact on subsequent responsiveness to alcohol. Alcoholism: Clinical and Experimental Research. 2000;24:1251–1256. [PubMed] [Google Scholar]
  53. Wills TA, Sargent JD, Gibbons FX, Gerrard M, Stoolmiller M. Movie exposure to alcohol cues and adolescent alcohol problems; A longitudinal analysis in a national sample. Psychology of Addictive Behaviors. 2009;23:23–25. doi: 10.1037/a0014137. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Zahn-Waxler C, Race E, Duggal S. Mood disorders and symptoms in girls. In: Bell DJ, Foster SL, Mash EJ, editors. Handbook of Behavioral and Emotional Problems in Girls. New York: Kluwer Academic; 2005. [Google Scholar]

RESOURCES