Addictive
Pergamon
Behaviors, Vol. 21. No. 5, pp. 633444. 1996 Copyright 0 1996 Elsevier Science Ltd Printed in the USA. All rights reserved
0306-4603/96 $15.00 + IllI
SSDI 03064603(95)00087-9
THE RELATIONSHIP OF SOCIAL CONTEXT OF DRINKING, PERCEIVED SOCIAL NORMS, AND PARENTAL INFLUENCE TO VARIOUS DRINKING PATTERNS OF ADOLESCENTS KENNETH
H. BECK and KATHERINE University
A. TREIMAN
of Maryland
Abstract - An anonymous questionnaire was administered to a sample of over 890 adolescent drinkers. The questionnaire contained measures of their social context of drinking, perceived social norms, and perceived parental behaviors regarding drinking. These measures were related to a variety of alcohol problem behaviors including binge drinking, driving while alcohol-impaired, riding with an alcohol-impaired driver. and experiencing one or more negative consequences because of drinking. Results indicated that drinking frequently in a social context of social facilitation, stress control or school defiance. as well as having close friends engage in these behaviors, tended to separate alcohol abusers from nonabusers. The implications for program development are discussed.
INTRODUCTION
Adolescent alcohol use and abuse remain significant public health problems in the United States, despite declines in consumption over the last 15 years. Alcohol use is virtually universal among adolescents. According to the 1993 national Monitoring the Future Survey, by eighth grade 67% students have tried alcohol, which increases to 87% by 12th grade. Heavy drinking is also widespread, with 27% of 12th graders reporting consumption of five or more drinks in a row in the 2-week period prior to the survey (National Institute on Drug Abuse, 1994). Alcohol use is even more widespread among out-of-school youth (Centers for Disease Control, 1994). Driving while intoxicated and riding with an intoxicated driver are common occurrences among adolescents and underlie the high rate of alcohol-related car accidents among this population (an estimated 50% of all fatal car crashes among drivers under 21 years old involve alcohol; Hingson, 1993). Results from an earlier Monitoring the Future Survey found that approximately one in four seniors had driven after drinking in the 2 weeks predating the survey, and two in five seniors had ridden with a drinking driver (Bachman, Johnston, & O’Malley, 1987). The purpose of this study was to examine the relative influence of three factors on patterns of adolescent alcohol use: the extent and nature of parental supervision of adolescents; adolescents’ perceived social norms regarding alcohol use; and the social contexts of adolescents’ alcohol use. Each of these factors is discussed below. Parental behaviors
Parents represent one of the most direct and immediate sources of influence over teens. Yet, among parents of adolescents, lack of awareness about underage drinking is evident, as is denial about their own teen’s drinking (Beck, 1990; Beck, Scaffa, Swift, & Ko, 1995; McDuffie & Bernt, 1993). Requests for reprints should be sent to Kenneth H. Beck. Department of Health Health and Human Performance, University of Maryland. College Park. MD 20742. 633
Education.
College
of
634
K. H. BECK and K. A. TREIMAN
Beck and Lockhart’s (I 992) model of parental involvement links parental monitoring and enforcement of family policies about drinking to adolescent alcohol abuse. Strong monitoring and enforcement are seen as factors limiting opportunities for alcohol abuse, whereas weak monitoring provides greater opportunities for abuse. Several recent studies support this model. Barnes and Farrell (1992) found that parents who engage in high levels of monitoring and support were least likely to have children who report that they drink regularly. Other research has identified proactive “family management, ” including monitoring and setting clear expectations, as an important protective factor against teen alcohol use (Peterson, Hawkins, Abbott, & Catalano. 1994). Williams, Lund, and Preusser (1986) found that high school students who reported more drinking were less likely to report parental rules and restrictions. Although most parents are concerned about underage drinking, many may fail to monitor and supervise their teens effectively, or enforce family rules about drinking inconsistently (Beck, 1990). For example, in a survey of parents in the Washington, DC, area, 43% of parents said they never, rarely, or only occasionally wait up for their teen to come home, and 37% said they did not have penalties for violating family rules about alcohol (Beck et al., 1995). Perceived social norms
Another line of research focuses on adolescents’ perceptions of drinking norms among their peers. Theoretical models of social behavior emphasize the importance of peer behavior as a modeling (Bandura, 1977) or normative (Fishbein & Ajzen, 1975) influence. Previous research has shown that adolescents’ alcohol use is most closely associated with perceptions of close friends’ drinking behaviors (Aas & Klepp, 1992; Baer, Stacy, & Larimer, 1991; Barnes, Farrell, & Banerjee, 1994; Downs, 1987). Laboratory experiments have also demonstrated the impact of drinking models on subjects’ rate and quantity of drinking (e.g., Collins, Parks, & Marlatt, 1985). However, there are conflicting research results about the influence of the broader peer group. For instance, Baer et al. (1991) found that, among college students, the “living group” (e.g., other students in the dormitory) influenced drinking behavior but to a lesser degree than did close friends, whereas Downs (1987) found that the larger social environment had no effect on adolescent drinking. Researchers have consistently found that adolescents and young adults exaggerate the drinking behaviors of their peers (Aas & Klepp, 1992; Baer et al., 1991; Thombs, Wolcott, & Farkash, 1994b). Such misperceptions are a risk factor for alcohol use. In a recent study of middle/high school and college students, Thombs and colleagues (1994b) found that drinking intensity, driving while intoxicated, and riding with an intoxicated driver were all associated with perceived norms variables. In a study of Norwegian adolescents, Aas and Klepp (1992) discovered that the estimated percentage of friends who drink was a better predictor of adolescent drinking than was grade in school, parents’ approval of drinking, or friends’ approval of drinking. Downs (1987) found that adolescents’ estimates of the prevalence of peer alcohol use was prospectively associated with the continuation of drinking. Social contexts of drinking
Another approach to understanding adolescent alcohol use and abuse is the social context of drinking. This construct was developed to explain the socio-psychological factors influencing adolescent alcohol use and abuse (Beck, Summons, & Thombs, 1991; Beck, Thombs, & Summons, 1993). Social context involves patterns of intraper-
Parental influence and adolescent drinking
635
sonal motivations occurring within specific social environments. Investigations have identified five distinct social context patterns of drinking among high school students (social facilitation, stress control, school defiance, peer acceptance, and parental control), which in combination have effectively discriminated between abusive and nonabusive adolescent drinkers (using a variety of indices for abusive drinking including binge drinking and alcohol-impaired driving), and among adolescents with different drinking patterns (Beck et al., 1991; Beck et al., 1993; Thombs & Beck, 1994; Thombs. Beck, Mahoney, Bromley, & Bezon, 1994a). The relative importance of these drinking contexts varies with the type of alcohol behavior in question. For instance, high-intensity and/or binge drinkers can be discriminated from low intensity, nonbinge drinkers by the frequency with which they drink in a context of social facilitation and for stress control. In contrast, students who engage in alcohol-impaired driving are more likely to drink in a context of school defiance than are nonimpaired drivers. These investigations have found these scales to be reliable, free from social desirability bias, and not highly redundant with frequency and quantity measures of alcohol consumption. The purpose of this study was to compare the relative relationship between these different social context measures of drinking, perceived social norms, and perceived parental influence with adolescent alcohol behaviors. Specifically, we wanted to assess the degree to which each of these measures alone, as well as in combination, could discriminate among teens with different levels of drinking intensity and involvement in high-risk alcohol behaviors.
METHODS
Subjects
An anonymous questionnaire was administered, as part of a school-initiated survey of alcohol use, to all students in a metropolitan Washington, DC, area high school in April 1994. The authors were allowed access to this data for research purposes upon the approval of the school officials. Students were informed of the voluntary nature of the study and assured that there was no way to connect them to their responses. The questionnaire was completed in homeroom periods. To ensure privacy, students recorded their responses on optical-scan answer sheets that were placed into envelopes. Instrumentation
School officials conducted the survey as part of a comprehensive needs assessment to determine the level of drug and alcohol use among students. The questionnaire included items to measure the independent and dependent constructs used in this research. In addition to the measures described below, the questionnaire also contained items measuring the frequency of illicit drug use. Copies of the instrument are available from the first author. Independent measures Social contexts of drinking. The instrument
contained the 32-item Social Context of Drinking Scales for high school students (Beck et al., 1993). Each item is worded to ask how often the respondent drinks in various contexts, such as at a party, with friends, to get drunk, or at school or on school grounds. Response categories are “never,” “seldom,” “ occasionally,” and “frequently” (scored O-3).
636
K. H. BECK and K. A. TREIMAN
Perceived social norms. The measurement of perceived social norms consisted of two sets of items, one focusing on what respondents think their “closest friends” do regarding alcohol use, and the other focusing on “other kids at school.” Each set contained six parallel items concerning drinking intensity (frequency of drinking, frequency of drunkenness, and usual number of drinks on one occasion), alcohol-impaired driving, riding with an alcohol-impaired driver, and binge drinking. These items were adopted from Thombs et al. (1994b). Parental behaviors. The nine items measuring parental influence addressed adolescent perceptions concerning family rules about alcohol (e.g., whether the teen is permitted to attend parties/bars where alcohol is served, how rules are enforced) and about parental supervision (e.g., whether parents wait up for teen to come home). These items were adopted from a survey of parent attitudes and behaviors regarding underage drinking and impaired driving (Beck et al., 1995) and were coded dichotomously where a positive score reflected greater perceptions of parental influence. Dependent measures Drinking intensity. Drinking intensity was measured
by three items-frequency of alcohol consumption, average number of drinks consumed per occasion, and frequency of drunkenness-whose summed score ranged from O-23. Respondents were categorized as low, moderate, or high-intensity drinkers, as follows: (1) Low-intensity drinkers (n = 150) reported their frequency of drinking was once a month or less, their quantity of consumption was two drinks or less per occasion, and they never get drunk; (2) moderate-intensity drinkers (n = 570) exceeded the low-intensity drinkers on one or more of these items, but they did not meet the criteria for a high-intensity drinker; and (3) high-intensity drinkers (n = 159) reported a frequency of drinking to be at least once a week or more, their quantity of consumption was at least five or more drinks per occasion, and they get drunk once a month or more. Binge drinking. Respondents were asked how many times during the past month they consumed five or more drinks on one occasion. Four response categories ranged from “none” (scored as 0) to “five or more times” (scored as 3). Respondents were categorized as “ever” (n = 449) or “never” (n = 424) binge drinkers. Alcohol-impaired driving. Respondents were asked how many times during the past year they had driven when they “have had perhaps too much to drink.” Seven response categories ranged from “never” (scored as 0) to “once a day or more” (scored as 7). Respondents were categorized as “ever” (n = 110) or “never” (n = 250) driving while under the influence (DWI). Only respondents holding drivers’ licenses (n = 364) were included in this measure. Riding with an alcohol-impaired driver. Respondents were asked how many times during the past year they had ridden with a driver “who has had perhaps too much to drink.” Response categories were as above. Respondents were categorized as “ever” (n = 420) or “never” (n = 466) riding with an impaired driver (RWID). Negative consequences. Respondents answered “yes” or “no” to eight items concerning whether they had ever experienced negative consequences of alcohol use (e.g.,
Parental
influence
and adolescent
drinking
637
been at a party that was busted, arrested for underage drinking, punished by school authorities for drinking on school grounds, injured in a drunk driving accident). Respondents were categorized as “consequence drinkers” (n = 419) if they had experienced one or more of these negative consequences, or as “no consequence drinkers” (n = 418) if they had not. RESULTS
Description of sample
A total of 1,378 students returned usable questionnaires, representing 93% of those in attendance that day. Of these, 897 students (65.1%) reported that they “ever drink” alcoholic beverages. Black students, students planning to attend college, and students living with two parents were less likely to be drinkers. Rates of ever drinking increased with age and by grade level, from 53% of 9th graders to 70% of 12th graders. All further analyses are based on the subsample of 897 drinkers, of which 54% were female and 46% were male. Sixty-five percent were White, 22% Black, and 13% were of some other race/ethnicity. The distribution by grade level was: 9th (21%) 10th (29%), 11th (27%), and 12th (24%) ( exceeds 100% due to rounding). Most (64%) reported living with two parents and planned to attend college (76%). Reliability assessments
Internal consistency of the various measures was assessed using Cronbach’s alpha (Table 1). All of these coefficients are acceptable (DeVellis, 1991) and are consistent with previous investigations (Beck et al., 1993; Thombs et al., 1994a). Perception of peers ’ drinking behaviors
The mean scores for the perceived behaviors of “closest friends” and “other kids at school,” related to drinking intensity, binge drinking, driving while intoxicated, and riding with an intoxicated driver, were calculated and compared to the mean scores for respondents’ actual behaviors. In each case, respondents reported these behaviors to be more frequent among their “closest friends” and even more so among “other kids at school.” The mean for the respondents’ drinking intensity was 8.6, compared to 8.8 Table 1. Coefficent alphas for social context. perceived social norms, parental behaviors. alcohol use intensity. and negative consequences scales Social context scales Social facilitation (11 items) Stress control (7 items) School defiance (6 items) Peer acceptance (5 items) Parental control (3 items) Perceived social normsa (12 items) Closest friends (6 items) Other kids at school (6 items) Parental behaviors (9 items) Drinking intensity (3 items) Negative consequences (8 items)
.93 .90 .91 .89 .6X .90 .91 .x5 .72 .X8 .86
“Regarding drinking intensity. binge drmking, driving while intoxicated. riding with an intoxicated driver.
638
K. H. BECK and K. A. TREIMAN
for “closest friends” and 11.7 for “other kids.” The mean for binge drinking was 0.9, compared to 1.1 for “closest friends” and 1.7 for “other kids.” The mean for riding with an intoxicated driver was 1.2, compared to 1.5 for “closest friends” and 2.2 for “other kids.” The greatest discrepancy occurred with driving while intoxicated. The mean was 0.6, compared to 1.4 for “closest friends” and 2.1 for “other kids.” Discriminant function analyses
To assess the degree of multicolinearity among the independent variable domains, each of the social context measures was correlated with social norms and parent behaviors. Because the items comprising each of these domains were internally consistent, each set was treated as a separate scale. The question of perceived social norms was examined in two ways. The first used the combined set of all 12 items, and the second used separate sets of six items each, for “closest friends” and “other kids” at school, respectively. Correlations of these measures with each of the social context scales ranged from .17 to .66, with an average of .34. All of the correlations of the social context scales and the parental scale and the “other kids” at school norm scale were at or below .36. All but one of the correlations of the social context scales with the total social norms scale and with the “closest friends” scale were below ~58. The highest correlation (.66) was between social facilitation and the “closest friends” norm scale. These results indicate that these measures were not highly intercorrelated and that multicolinearity was not a major problem with this data. Discriminant function analyses, along with classification analyses, were performed to determine the relative influence of the social context measures compared to the perceived social norms and parental behaviors to discriminate between the various types of drinkers. The ability of each of the three sets of predictors (social contexts, perceived social norms, and parental behaviors) alone, and in combination, was examined. All discriminant function analyses used the direct entry method, in which all predictor variables were entered simultaneously, and adjusted for the proportion of cases in each group. It should be noted that in each analysis, only the relevant perceived social norms items were used. For example, in the analysis of driving while intoxicated, only the perceived social norms regarding “closest friends” and “other kids” driving while intoxicated were used. Similarly, in the analysis of riding with an intoxicated driver, only the perceived social norms regarding riding with an intoxicated driver were used. However, in the analysis of experiencing negative consequences from drinking, all of the perceived social norms variables (intensity, driving while intoxicated, riding with an intoxicated driver, and binge drinking) were used. Table 2 presents the findings from the classification analyses. The first column lists the percentages of all drinkers correctly classified, using each of the three predictor measures alone and in combination. The second column lists the percentages of highrisk drinkers correctly identified. High-risk drinkers were those who ever (vs. never) drive while intoxicated, binge-drink, ride with an intoxicated driver, experience negative consequences of drinking, or report being a high-intensity drinker. The significance of the differences of these percentages was tested using chi-square analyses. The level of significance was adjusted to .005 for tests within each problem behavior. This was done to minimize the probability of a Type I error and was determined by dividing .05 by the number of tests to be performed (12) for each problem behavior. The social context measures were better at correctly classifying all drinkers than high-risk drinkers for drinking and driving, and drinking intensity. When used alone,
Parental influence and adolescent drinking
639
Table 2. Percent of all drinkers and high-risk drinkers correctly classified
Drinking and driving Social context Perceived social norms Parental behaviors All 3 combined Riding with impaired driver (RWID) Social context Perceived social norms Parental behaviors All 3 combined Binge drinking Social context Perceived social norms Parental behaviors All 3 combined Negative consequences Social context Perceived social norms Parental behaviors All 3 combined Drinking intensity Social context Perceived social norms Parental behaviors All 3 combined
Percent of all drinkers
Percent of high-risk drinkers
80.7 76.1” 76.7 84.7h
52.8 37.4” 43.0 S9.4”
69.3” 72.6” S9.G 74.6”
62.3” 60.0” 45.5” 64.7”
79.3” 69.5” 63.7h 79.7”
79.3” 51.4h 59.2” 78.7”
69.7 68.7 62.2 70.8
70.6 63.7 61.3 71.3
75.6” 71&J 68.7h 80/C2
S4.Y 54.8” 32.1h 62.1a
Nore. High-risk drinker = Ever DWI, ever RWID, ever binge-drink, experi1 or more negative consequences, high-intensity drinkers. a.h.l.2Foreach alcohol problem behavior, percentages with different superscripts differ at the p < .00.5level of significance. enced
the social context measures worked as well as when they were used in combination with perceived social norms and parental behaviors at classifying drinkers in either category, regardless of which behavior was considered. However, for riding with an impaired driver, binge drinking, and drinking intensity, the social context measures were significantly better predictors than were parental behaviors. The social context measures also were significantly better than social norms for classifying binge drinkers. They did not differ significantly from social norms at classifying drinking drivers or those who ride with an impaired driver. Thus. using all three measures in combination did not significantly increase the accuracy of prediction over that produced by the social context measures when used alone. Further. the social context measures offer a significant advantage over perceived parental behaviors for identifying adolescents who ride with intoxicated drivers, binge drinkers, and high-intensity drinkers. They also offer a significant advantage over social norms for identifying binge drinkers. No set of measures was found to be superior at identifying drinkers who had suffered a negative alcohol-related consequence. Table 3 presents the results of the discriminant function analysis of low, moderate, and high-intensity drinkers. Only those scales and variables with structure coefficients of .50 and above are included. Two statistically significant functions were extracted 0, < .OOOl).The first function accounted for most of the variance (86.78%). An examination of the structure matrix (the pooled within-group correlations between the variables and the derived functions) shows that Function 1 was clearly dominated by three
640
K. H. BECK and K. A. TREIMAN
Table 3. A multiple discriminant function analysis of three levels of drinking intensity, based on social contexts. social norms, and parental behaviors
Function 1 Function 2
Eigenvalue
% Variance
Canonical Correlation
Wilks’s lambda
1.26 0.19
86.78 13.22
0.75 0.40
0.37 0.84
Chi-square 582.27 103.14
df
Significance
40 19
0.001 0.001
Structure Matrix Scale/Variable Social facilitation School defiance Stress control Frequency of drunkenness close friends Usual number of drinks close friends Frequency of drinking close friends
Function I 0.834 0.632 0.617 0.577
Function 2 0.581
0.545 0.525
social context measures (social facilitation, school defiance, and stress control) and three perceived social norms relating to “closest friends” (frequency of drinking, frequency of drunkenness, usual number of drinks on one occasion). None of the parental behaviors contributed significantly to Function 1. Evaluation of the group centroids shows that Function 1 separated low- and high-intensity drinkers. Thus, high-intensity drinkers are best distinguished from low-intensity drinkers by their greater frequency of drinking in a context of social facilitation, school defiance, and stress control as well as by greater perceived social norms of closest friends’ drinking intensity. Function 2 accounted for a much smaller percentage of the variance (13.22%) and was largely associated with the social context of school defiance. None of the parental behaviors or perceived social norms contributed significantly to Function 2. An examination of group centroids shows that this function separated high and moderateintensity drinkers. Together, these results indicate a hierarchy of discriminations, such that high-intensity drinkers are separated from low-intensity drinkers by their frequency of drinking for social facilitation, stress reduction, defiance of school rules, and their perception of drinking intensity norms among their close friends. High-intensity drinkers can be further separated from moderate-intensity drinkers on the basis of their frequency of drinking in a context of school defiance. Table 4 presents the results of a series of pairwise discriminant analyses of drinkers who ever (vs. never) drive while impaired, ride with an impaired driver, binge-drink, and experience negative consequences. In each analysis, a statistically significant function was extracted 0, < .OOOl).As before, Table 4 includes only measures with structure coefficients of SO and above. Drinkers who drive while alcohol impaired were discriminated from nondrinking drivers by school-defiance (.732), social norms of closest friends’ DWI (.693), stress control (.693), social facilitation (.614), and peer approval (.550). Teens who ride with alcohol-impaired drivers were discriminated from those who don’t by their perceived social norms of closest friends’ RWID (863) social facilitation (.672), stress control (.588), and school-defiance (.545). Binge drinkers were discriminated from nonbinge drinkers by social facilitation (.930), stress control (.620) and perceived social norms of closet friends’ binge drinking (.602). Finally, drinkers who have experienced one or
Parental
influence
and adolescent
drinking
641
Table 4. A multiple discriminant function analysis of DWI. RWID, binge drinking, consequences based on contexts, social norms, and parental behaviors
Eigen value DWI” RWID Binge Consequences
0.71 0.38 0.72 0.39
Canonical Correlation
% Variance 100.00 100.00 100.00 100.00
0.64 0.53 0.65 0.53 Structure
Scale/Variable School Defiance Frequency of DWI close friends Stress Control Social Facilitation Peer Acceptance Frequency of RWID close friends Frequency of drunkenness close friends Frequency of binging close friends Frequency of drinking close friends Usual Number of drinks close friends aOnIy respondents with a driver’s license n/a = Variable not used for this analysis.
Wilks’s lambda 0.59 0.72 0.58 0.72
Chi-square 131.01 192.57 317.92 182.47
and negative
df
16 16 16 26
Significance 0.001 0.001 0.001 0.001
Matrix
DWI”
RWID
Binge
0.732 0.693
0.545 n/a
n/a nla
0.693 0.614 0.550 n/a
0.588 0.672
0.620 0.930
0.863
n/a
n/a
n/a
n/a
0.587
nla
nla
0.602
0.574
n/a
n/a
n/a
0.56’)
n/a
n/a
nla
0.566
(n = 364) are included
Consequences
0.539 0.793
in analysis.
more negative consequences were distinguished from nonconsequence drinkers by two social context measures-social facilitation (.793) and stress control (539). and four perceived social norms measures-closest friends’ frequency of drunkenness (.587), frequency of drinking (.569), frequency of binging (.574), and usual number of drinks (.566). DISCUSSION
Interpretation of these research findings must be qualified by two potential methodological limitations. First, the study used a convenience sample of students from one school. This procedure introduces uncertainty about the representativeness of this sample to other populations. However, the purpose of this investigation was to examine the relationship of these psychosocial variables to various drinking measures. not to establish the prevalence of these traits in the general population. Nevertheless, replication of these findings with diverse samples of adolescents is recommended. The second limitation was the reliance on self-report data. Though anonymous self-report measures of alcohol abuse are generally considered valid (Johnston & O’Malley, 1985), some degree of underreporting may have occurred. Earlier investigations have demonstrated that the Social Context of Drinking Scales are effective at discriminating among teens with different drinking patterns (Beck et al., 1993; Thombs & Beck, 1994; Thombs et al., 1994a). The present study confirmed this and found that social norms are also of some additional importance at distinguish-
642
K. H. BECKand K. A. TREIMAN
ing different types of adolescent drinkers. Classification analyses found that the overall predictive value, and the more sensitive prediction of the high-risk teens, did not improve when all three sets of measures were used. Social context works as well as if not better (in some circumstances) than social norms and perceived parental influence when used either separately or in combination. Therefore, social context of drinking, as a measurement domain, seems to offer a diagnostic advantage over normative or parental influence measures for identifying adolescent, high-risk drinkers. Social facilitation, stress control, and school-defiance were the most important of the social context measures that discriminated high-risk drinkers. Consistent with previous studies (Beck et al., 1993; Thombs et al., 1994a), social facilitation appeared to be more important for identifying teens who are high-intensity drinkers and who are prone to binge drinking, whereas school defiance was more strongly associated with impaired driving and experiencing negative consequences. In addition, the perceived social norms regarding close friends’ drinking intensity (frequency of drinking, quantity of drinking, and frequency of getting drunk) were also strongly associated with binge drinking and experiencing negative consequences. These findings are consistent with the recent findings of Thombs et al. (1994b), who found that, among middle/high school students, several social context variables and perceptions of close friends’ drinking intensity were strongly associated with teens’ drinking intensity. It is interesting to note that the social context of peer approval was generally not an important discriminator. Peer approval only played a moderate role in identifying teens who ever drive under the influence. These findings suggest that teens’ drinking behaviors are not driven so much by a need for peer approval or to be accepted by a group, but rather by what is perceived of as normal behavior among one’s close friends. The question then is, do teens seek out close friends whose drinking habits are similar to their own? Or do they adjust their own behavior according to their perception of what their close friends are doing? The cross-sectional nature of this study does not allow us to make causal inferences. Downs (1987) found the relationship between adolescent drinking and close friends’ drinking to be reciprocal. This indicates that the relationship between adolescent drinking and that of their close friends is mutually interdependent. None of the perceived parental influence measures (nor the social context of drinking under parental control) appeared to be important. The fact that perceived parental influence played a relatively weak role in discriminating among teens with different drinking behaviors was somewhat surprising. Previous theory and research suggest that parental monitoring and enforcement are associated with lower rates of alcohol use and abuse among teens (Beck & Lockhart, 1992; Barnes & Farrell, 1992). The discrepant results might be due to measurement issues, such as shortcomings in the parental behaviors scale, or that parental behaviors were as reported by teens, rather than directly by parents. Teens may not be aware of all parental behaviors or may misrepresent these behaviors. Future research should focus on the discrepancy between actual parental behaviors, as reported by the parents, and perceived parental influence, as reported by their adolescents, and whether the degree of discrepancy in parental style bears any relationship to adolescent alcohol misuse. The findings from this study suggest that perception of peer norms and social contexts of drinking are better, or at least more immediate, predictors of adolescents’ drinking intensity and alcohol abuse than are parental monitoring and enforcement behaviors, These findings have several implications for alcohol interventions aimed at teens. First, the challenge for program developers is to design interventions with an under-
Parental
influence
and adolescent
drinking
643
standing of the key social contexts in which teens prone to abuse alcohol drink (social facilitation, school-defiance, and stress control). These programs should not only seek ways of identifying such students who are at high risk for abusing alcohol, but should also address the needs of these targeted individuals for engaging in socialization, stress reduction, and rebellious behavior, without the concomitant use of heavy alcohol consumption. Another challenge is to find ways to influence teens’ perceptions of social norms regarding drinking. Teens’ exaggerated perceptions of their peers’ involvement with alcohol may be attributable in part to mass media portrayals of alcohol consumption. Television images of alcohol are pervasive and generally depict consumption as free of negative consequences (Atkin, 1989; Gerbner, 1990). Interventions can exploit the mass media to change perceived social norms, as was done to create social acceptance for the “designated driver” concept (Montgomery, 1993). However, the ultimate challenge for program developers will be to focus on changing the normative influence of the adolescents’ most proximal reference group---one’s closest friends. Clearly, mass media, when used in combination with other interpersonal delivery channels. show promise for accomplishing this. Finally, interventions targeting parents should also be emphasized. Despite the findings of this investigation, the potential of parental influence should not be dismissed. Further research is needed to understand the types of parental supervision and parenting style that are most effective in preventing teen alcohol misuse.
REFERENCES Aas, H., & Klepp, K. I. (1992). Adolescents’ alcohol use related to perceived norms. Scnndinovian Journal of Psychology, 33,315-325. Atkin. C. K. (1989). Mass communications effects on drinking and driving. In Surgeon General’s wwrkshop on drunk driving; Background papers. Rockville, MD: U.S. Department of Health and Human Services. Bachman, .I., Johnston, L., & O’Malley, I? (1987). Monitoring the future: Quesrionnaire responses from rhe nafion’s high school seniors. Ann Arbor, MI: Institute for Social Research. Baer. J. S., Stacy, A.. & Larimer, M. (1991) Biases in the perception of drinking norms among college students. Journal of Studies on Alcohol, 52(6), 580-586. Bandura. A. (1977). Social learning rheory. Englewood Cliffs, NJ: Prentice-Hall. Barnes, Cl. M., & Farrell, M. P (1992). Parental support and control as predictors of adolescent drinking. delinquency. and related problem behaviors. Journal of Marriage and the Family, 54.163-776. Barnes. Cl. M.. Farrell, M. P, & Banerjee, S. (1994). Family influence on alcohol abuse and other problem behaviors among black and white adolescents in a general population sample. Journal of Reseurch on Adolescence, 4(2). 183-201. Beck. K. H. (1990). Monitoring parent concerns about teenage drinking and driving: A random digit dial telephone survey. American Journal of Drug and Alcohol abuse. 16( l&2). 109-124. Beck. K. H., & Lockhart. S. J. (1992). A model of parental involvement in adolescent drinking and driving. Journal of Youth and Adolescence. 21(l), 35-51. Beck. K. H., Scaffa, M., Swift. R., & Ko, M. (1995). A survey of parent attitudes and practices regarding underage drinking. Juurnal of Youfh and Adolescence. 24(3). 315-334. Beck. K. H.. Summons,T. G., & Thombs, D. L. (1991). A factor analytic study of social context of drinking in a high school population. Psychology of Addictive Behaviors. 5.66-77. Beck, K. H., Thombs, D. L., & Summons, T. Cl. (1993). The social context of drinking scales: Construct validation and relationship to indicants of abuse in an adolescent population. Addictive Behaviors. 18,159169. Centers for Disease Control (1994). Health risk behaviors among adolescents who do and do not attend school-United States, 1992. Morbidity and Mortality Weekly Reporr, 43(8), 129-132. Collins, R. L., Parks, G. A., & Marlatt, G. A. (1985). Social determinants of alcohol consumption: The effects of social interaction and model status on the self-administration of alcohol. Journal of Consulring Clinical Psychology. 53,189-200. DeVellis, R. E (1991). Scale developmenr: Theory and applications. Newbury Park. CA: Sage. Downs, B. (1987). A panel study of normative structure, adolescent alcohol use and peer alcohol use. Journal of Studies on Alcohol 48(2), 167-175.
644
K. H. BECK
and K. A. TREIMAN
Gerbner. G. (1990). Stories that hurt: Tobacco. alcohol. and other drugs in the mass media. In Yourh and Drugs: Society’s Mixed Messages. (Office of Substance Abuse Prevention, Prevention Monograph No. 6). Rockville, MD: Office for Substance Abuse Prevention. Fishbein. M.. & Ajzen. I. (1975). Belief, altitude, intention, and behavior: An introduction [o fheory and research. Reading, MA: Addison-Wesley. Hingson. R. (1993). Prevention of alcohol-impaired driving. Alcohol, Health & Reseurch World. 17( 1). 2% 34. Johnston. L. D.. & O’Malley, P. M. (1985). Issues of validity and population coverage in student surveys of drug use. In B. S. Rouse, N. J. Dozel, & L. G. Richards (Eds.), Self-report methods ofesrimating drug use: Meeting currenl challenges of validity. (Publication X5-1402). Washington DC: U.S. Government Printing Office. McDuffie. T. E.. & Bernt, F. M. (1993) Contrasts between the perceptions of parents and their adolescent children regarding drug and alcohol use and prevention. Journal of Alcohol and Drug Education, 39(6). 92-106. Montgomery, K. C. (1993). The Harvard Alcohol Project: Promoting the designated driver on television. In T. E. Backer & E. M. Rogers (Eds.), Organizational aspects of health communication campaigns: What works:’ Newbury Park. CA: Sage. National Institute on Drug Abuse (1994). National survey resulfs on drug use from the Monitoring the Future Study, 1975-199.3: Vol. I. Secondary school students. Rockville, MD: U.S. Department of Health and Human Services. Peterson, P. L.. Hawkins, J. D., Abbott, R. D., & Catalano, R. F. (1994). Disentangling the effects of parental drinking, family management, and parental alcohol norms on current drinking by black and white adolescents. Journal of Research on Adolescence, 4(2), 203-227. Thombs, D. L.. & Beck, K. H. (1994). The social context of four adolescent drinking patterns. Healfh Educafion Research 9(l). 13-22. Thombs, D. L., Beck, K. H., Mahoney, C. A., Bromley, M. D., & Bezon, K. M. (1994a). Social context. sensation seeking. and teen-age alcohol abuse. Journal of School Heahh, 4(2), 73-79. Thombs. D. L.. Wolcott, B. J., & Farkash, L. G. E. (1994b). Canonical analysis of social context, perceived norms and drinking behavior in young people. Unpublished manuscript. Williams, A. E. Lund. A. K., & Preusser, D. F. (1986). Drinking and driving among high school students. International Journal of the Addicrions, 21(6), 643655.