- Email: [email protected]
Potentially modifiable psychosocial factors associated with alcohol use during early adolescence
Addictive Behaviors 29 (2004) 1503 – 1515
Potentially modifiable psychosocial factors associated with alcohol use during early adolescence Peter W. Callasa,*, Brian S. Flynnb, John K. Wordenb a
Medical Biostatistics, University of Vermont, Hills Building, 105 Carrigan Drive, Burlington, VT 05405, USA b Office of Health Promotion Research and Department of Family Practice, College of Medicine, University of Vermont, Burlington, VT 05405, USA
Abstract This study was conducted to identify factors associated with alcohol use among early adolescents. A survey was administered to all Grade 7 and 8 students in 16 Vermont school districts. The questionnaire covered demographics, alcohol, tobacco, and marijuana use, and measures of psychosocial mediators of alcohol use drawn from social cognitive theory. These included positive and negative expectancies about alcohol effects, perceived peer and parent alcohol norms, perceived prevalence of adolescent alcohol use, and confidence in ability to refuse alcohol. Of the 2919 respondents, 29% reported having at least one drink of beer in the preceding 30 days. In logistic regression, factors independently related to risk of drinking beer in the past 30 days were smoking (odds ratio [OR] = 2.3, 95% confidence interval [CI] 1.8 – 3.0), marijuana use (OR 3.9, 95% CI 3.0– 5.2), negative expectancies (OR 0.4, 95% CI 0.3– 0.6), parent norms (OR 1.4, 95% CI 1.1 – 1.7), and estimated percentage of high school students who drink (OR 1.3, 95% CI 1.1 –1.5). Gender, positive alcohol expectancies, and lack of confidence in ability to refuse alcohol all significantly interacted with peer norm, with these items more strongly associated with alcohol use when peer norm is toward ‘‘shouldn’t drink.’’ Modifiable perceptions of alcohol use were strongly associated with actual use in this adolescent sample, providing a basis for intervention program design. D 2004 Elsevier Ltd. All rights reserved. Keywords: Drug abuse prevention; Alcohol; Adolescent attitudes; Measurement; Predictability
* Corresponding author. Tel.: +1-802-656-3195; fax: +1-802-656-0632. E-mail address: [email protected] (P.W. Callas). 0306-4603/$ – see front matter D 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.addbeh.2004.02.028
1504
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1. Introduction Alcohol is the most commonly used drug among adolescents in the United States. In the year 2002, 20% of 8th graders reported having consumed an alcoholic beverage in the previous month (Johnston, O’Malley, & Bachman, 2003). This increased to 49% by 12th grade. Early alcohol use is associated with an increased likelihood of later heavier use patterns and alcohol-related social and behavioral problems, as well as progression to other substances (Chou & Pickering, 1992; Grant & Dawson, 1997; Gruber, DiClemente, Anderson, & Lodico, 1996). There is evidence that delaying alcohol initiation reduces risk of these later problems (Hawkins et al., 1997; Robins, 1984; Yamaguchi, 1990). Understanding the factors that lead to early use is one step in prevention. According to social cognitive theory, cognitive, behavioral, and environmental factors interact to produce particular behaviors in given situations (Bandura, 1986). For substance use among adolescents, behavior has been theorized to result from interaction between personality system, behavior system, and perceived social environment (Jessor & Jessor, 1977). Specific factors in these related systems include expectancies about substance use effects, perceived norms, perceived prevalence of use of the substance, and confidence in refusing offers of substances. If these are associated with substance use, prevention efforts can focus on targeting them, such as by using advertising to strengthen negative expectancies about use or to model refusal skills (Hawkins, Catalano, & Miller, 1992). We surveyed Grade 7 and 8 students, examining the relationship between alcohol use and cognitive, behavioral, and social factors. The objective was to identify factors associated with alcohol use among early adolescents that could potentially be modified by educational interventions.
2. Methods 2.1. Participants In spring 1997, a survey regarding alcohol behavior and attitudes was given to Grade 7 and 8 students in 16 Vermont school districts. The school districts were composed of the eight intervention and eight comparison districts for a mass media youth alcohol prevention study. This survey was conducted prior to implementing the interventions. The study was approved by the University of Vermont Committee on Human Research. Students from Vermont were surveyed because Vermont has a high level of alcohol misuse and has standard statewide grade K-12 alcohol and drug abuse education programs, ensuring comparable educational exposures across schools. Of the 69 school districts in Vermont, 16 were selected based on size, demographic characteristics, and distribution of television and radio channels (for the media intervention).
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1505
2.2. Measurement The questionnaire administered to the students consisted of 110 items covering current and past alcohol use, future intentions for alcohol use, perceived access to alcohol, exposure to alcohol education, tobacco and marijuana use, demographics, and several psychosocial measures based on social cognitive theory (Bandura, 1986). These measures included expectancies about alcohol effect, perceived peer and parent normative attitudes toward alcohol, perceived adolescent alcohol use, and confidence in ability to refuse alcohol. Expectancies were measured using 28 questionnaire items concerning potential physical, social, and psychological consequences of alcohol. The common stem for these items stated that, ‘‘If I had 2 or 3 whole drinks of beer, wine, or liquor, I would . . ..’’ Individual items included ‘‘get along better with friends,’’ ‘‘feel more confident,’’ ‘‘feel sick,’’ and ‘‘act stupid.’’ Possible responses for each item were (1) ‘‘unlikely,’’ (2) ‘‘possibly,’’ and (3) ‘‘likely.’’ These items were developed to represent a balance of positive and negative social, affective, and physical consequences expected from alcohol use. Alcohol social norm measures were based on work by Akers, Krohn, Lanza-Kaduce, and Radosevich (1979). The 14 items included questions such as ‘‘What does your best male friend think about girls your own age drinking?’’ and ‘‘What does your mother think about girls your own age drinking?’’ Response choices were (1) ‘‘shouldn’t drink at all,’’ (2) ‘‘it doesn’t matter very much,’’ (3) ‘‘it’s okay to drink sometimes,’’ and (4) ‘‘it’s okay to drink as much as you want.’’ These correspond to the categories labeled by Akers et al. as proscriptive, ascriptive, prescriptive, and permissive. Perceived adolescent alcohol use consisted of eight items covering each combination of boys or girls of the respondent’s age or high school age in the respondent’s school district or the United States. Items asked for the best guess as to the percentage in each group who had one or more drinks of beer, wine, or liquor in the past 30 days. Respondents marked estimates on a visual analog scale of 0–100%, which had labeled marks at 0%, 25%, 50%, 75%, and 100%. Responses were coded to the nearest 1% for data analysis. Confidence in ability to refuse alcohol was derived from the self-efficacy principles of Bandura (1986). Fifteen items included a number of social situations where the respondent might be offered alcohol. Items were preceded by the common stem, ‘‘If someone offered you a drink of beer, wine, or liquor, do you feel it would be easy to refuse or hard to refuse. . ..’’ Situations included ‘‘at a friend’s house after school and no adults are there’’ and ‘‘at a party with friends and most people are drinking.’’ Response choices were (1) ‘‘easy to refuse,’’ (2) ‘‘not sure,’’ and (3) ‘‘hard to refuse.’’ 2.3. Data collection The survey was conducted during normal class times, with the regular classroom teacher absent from the room. All Grade 7 and 8 students at each school completed the survey except those whose parents refused, special education students excluded by the school, and students who were absent on the day of the survey.
1506
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
2.4. Data analysis We conducted bivariate and multivariate analyses relating a number of independent variables to alcohol use. SAS (SAS Institute, 2000) was used for all data analyses. Alcohol use was measured by the question, ‘‘During the past 30 days, on how many days did you have at least one drink of beer?’’ Responses were dichotomized to ‘‘0 days’’ versus ‘‘1 or more days.’’ Independent variables included grade, gender, tobacco use, marijuana use, and several scales developed from the psychosocial measures on the questionnaire. Scales were constructed using principal components factor analysis with varimax rotation (Harman, 1976). Eigenvalues, scree plots, variance proportion, and judgment as to interpretability were used to determine the number of factors to retain. Internal consistency of each scale was measured by computing Cronbach’s coefficient alpha (Cronbach, 1951). Bivariate relationships were assessed via Fisher’s exact test for categorical variables and two-sample t test for continuous variables. All independent variables were evaluated simultaneously using logistic regression (Hosmer & Lemeshow, 2000), with relevant interaction terms considered as candidates for inclusion in the model after significant main effects were identified.
3. Results 3.1. Respondents The survey was completed by 2919 students from 24 schools in the 16 districts. This was nearly a complete census of the population attending school on the survey days, since only a few parents refused participation. Upon reviewing the questionnaires, eight were discarded due to clearly invalid responses, leaving a sample of 2911 for data analysis. Some characteristics of the respondents are given in Table 1. The age range was 11–16 years, with 81% either 13 or 14 years old. Only 14% of students had never had a sip or taste of beer, wine, or liquor, but 42% had never had a drink of these beverages (not just a taste or sip). For substance use in the previous 30 days, 29% reported drinking beer on at least one day, 26% reported smoking cigarettes, and 18% used marijuana. 3.2. Scale development Four separate factor analyses were conducted to identify factors from the multipleitem measures of perceptions concerning alcohol use. The resulting factors were used to construct factor-based scales from each subject’s mean score over the items in the factor. The first factor analysis, using the 28 questions pertaining to expected consequences of drinking alcohol, found two important factors, with 14 questions loading on Factor 1 and 14
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1507
Table 1 Characteristics of respondents No. (%) Grade Gender How many times have you had a drink of beer, wine, or liquor)not just a taste or sip?
During the past 30 days, on how many days did you have at least one drink of beer?
During the past 30 days, on how many days did you smoke cigarettes?
During the past 30 days, on how many days did you use marijuana?
7 8 Girl Boy Can’t remember Never 1 or 2 times 3 to 10 times More than 10 times 0 days 1 or 2 days 3 to 5 days 6 to 9 days 10 to 19 days 20 or more days 0 days 1 or 2 days 3 to 5 days 6 to 9 days 10 to 19 days 20 or more days 0 days 1 or 2 days 3 to 5 days 6 to 9 days 10 to 19 days 20 or more days
1438 (50) 1459 (50) 1471 (51) 1416 (49) 132 (5) 1221 (42) 586 (20) 456 (16) 503 (17) 2043 (71) 472 (16) 185 (6) 77 (3) 57 (2) 59 (2) 2136 (74) 238 (8) 108 (4) 69 (2) 78 (3) 258 (9) 2352 (82) 178 (6) 109 (4) 78 (3) 63 (2) 100 (4)
on Factor 2. The first factor, the negative expectancies scale (the mean of these 14 items), ranged from 1 (unlikely for negative things to happen) to 3 (likely for negative things to happen). The second factor, the positive expectancies scale, ranged from 1 (unlikely for positive things to happen) to 3 (likely for positive things to happen). In the second factor analysis, of 14 questions regarding perceived attitude norms of alcohol use, two factors were retained. Eight items loaded primarily on Factor 1 (peer norm), and six on Factor 2 (parent norm). The peer and parent norm scales both ranged from 1 (shouldn’t drink at all) to 4 (OK to drink as much as you want). The third factor analysis used eight questions on the student’s estimate of the percentage of Grade 7/8 and high school boys and girls in the student’s district and in the entire United States who had one or more drinks of beer, wine, or liquor in the past 30 days. The first of three factors consisted of four questions on the percentage of Grade 7 and 8 boys and girls estimated to have had a drink in the past 30 days. The second had two questions on high school boys and girls in the student’s district, and the third had two questions on high school boys and girls in the United States.
1508
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
Table 2 Respondent characteristics versus beer-drinking behavior
Grade Gender Days smoke cigarettes in past 30 days Days use marijuana in past 30 days Positive expectancies scale
Negative expectancies scale
Peer norm scale
Parent norm scale
Percent drink scale for high school students in United States Percent drink scale for high school students in student’s district Percent drink scale for grade 7/8 students
Confidence scale
a
n 7 8 Girl Boy 0 1+ 0 1+ 1.00 – 1.49 1.50 – 1.99 2.00 – 2.49 2.50 – 3.00 1.00 – 1.49 1.50 – 1.99 2.00 – 2.49 2.50 – 3.00 1.00 – 1.99 2.00 – 2.99 3.00 – 4.00 1.00 – 1.99 2.00 – 2.99 3.00 – 4.00 0 – 24% 25 – 49% 50 – 74% 75 – 100% 0 – 24% 25 – 49% 50 – 74% 75 – 100% 0 – 24% 25 – 49% 50 – 74% 75 – 100% 1.00 – 1.49 1.50 – 1.99 2.00 – 2.49 2.50 – 3.00
(unlikely) (possibly) (likely) (very likely) (unlikely) (possibly) (likely) (very likely) (shouldn’t drink) (doesn’t matter) (OK to drink) (shouldn’t drink) (doesn’t matter) (OK to drink)
(easy to refuse) (not sure) (hard to refuse) (very hard to refuse)
Total n
Drink beer 1 or more days, no. (%)a
2893 1432 1448 1467 1404 2121 749 2339 526 1062 855 684 261 376 618 928 945 1200 908 760 2575 249 47 109 583 1335 845 411 879 996 582 885 1095 699 189 1056 647 754 422
850 (29) 339 (24) 508 (35) 385 (26) 456 (33) 353 (17) 490 (65) 428 (18) 414 (79) 77 (7) 237 (28) 346 (51) 185 (71) 178 (47) 303 (49) 254 (27) 109 (12) 61 (5) 297 (33) 488 (64) 662 (26) 144 (58) 38 (81) 20 (18) 115 (20) 375 (28) 334 (40) 66 (16) 182 (21) 328 (33) 268 (46) 150 (17) 300 (27) 294 (42) 100 (53) 129 (12) 125 (19) 302 (40) 290 (69)
No. (%) reporting having had at least one drink of beer in the past 30 days.
OR
95% CI
1.7
1.5 – 2.1
1.4
1.2 – 1.6
9.5
7.8 – 11.5
16.5
13.1 – 20.9
4.9 13.1 31.1
3.7 – 6.5 9.9 – 17.3 21.9 – 44.4
1.1 0.4 0.2
0.8 – 1.4 0.3 – 0.5 0.1 – 0.2
9.1 33.5
6.8 – 12.2 24.9 – 45.1
4.0 12.2
3.0 – 5.2 5.9 – 25.4
1.1 1.7 2.9
0.7 – 1.7 1.1 – 2.9 1.8 – 4.8
1.4 2.6 4.5
1.0 – 1.9 1.9 – 3.5 3.3 – 6.1
1.9 3.6 5.5
1.5 – 2.3 2.8 – 4.5 3.9 – 7.7
1.7 4.8 15.8
1.3 – 2.3 3.8 – 6.1 12.0 – 20.8
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1509
For the fourth factor analysis, there were 15 questions on how hard it would be to refuse a drink of beer, wine, or liquor in a number of situations. Only one factor was found. The confidence scale ranged from 1 (easy to refuse) to 3 (hard to refuse). Cronbach’s coefficient alpha was greater than .80 for all of the above factors. 3.3. Bivariate analyses Table 2 presents the bivariate relationships between respondent characteristics and beer drinking behavior in the past 30 days. For the scales, findings were similar when they were treated as continuous and when they were categorized; only the categorized versions are presented. Risk of beer drinking was higher for Grade 8 students, boys, and respondents who smoked cigarettes or marijuana, had high positive expectancies, had low negative expectancies, perceived permissive peer and parent norms toward alcohol use, estimated high levels of drinking, and felt they would have difficulty refusing if offered alcohol. 3.4. Multivariable analyses In bivariate analysis, the negative expectancies scale did not show a linear change in the natural log of the odds across its range. The best representation of this variable for modeling was to combine Categories 1 and 2. The exponential assumption for continuous variables in logistic regression modeling was reasonable for all other scales, so they were not changed. For main effects, all independent variables were significantly associated with beer drinking in the past 30 days except grade (Wald P =.19), percentage high school district ( P =.46), and percentage Grade 7/8 ( P =.24). Three two-way interactions were significant: the interaction Table 3 Logistic regression model Variable Intercept Gender Smoking in past 30 days Marijuana in past 30 days Positive expectancies Negative expectancies Peer norm Parent norm Percent high school U.S. Confidence Peer Norm Gender Peer Norm Positive Expectancies Peer Norm Confidence a b
See text and Table 4. See text.
Comparison
Boy vs. girl 1+ vs. 0 days 1+ vs. 0 days 1 unit increase 1 unit increase 1 unit increase 1 unit increase 25% increase 1 unit increase
Parameter 10.83 1.78 0.84 1.37 2.03 0.93 2.88 0.32 0.23 1.21 0.49 0.46 0.28
Standard error
Odds ratio
0.43 0.13 0.15 0.45 0.21 0.45 0.11 0.08 0.38 0.16 0.16 0.13
a
2.3 3.9
95% Confidence interval
1.8 – 3.0 3.0 – 5.2
a
0.4
0.3 – 0.6
b
1.4 1.3 a
1.1 – 1.7 1.1 – 1.5
1510
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
Table 4 Odds ratios for variables interacting with peer norm Level of peer norm
Gender (boy vs. girl) OR
95% CI
OR
95% CI
OR
95% CI
1 2 3 4
3.7 2.2 1.4 0.9
2.1 – 6.4 1.7 – 3.0 1.1 – 1.8 0.5 – 1.4
4.8 3.0 1.9 1.2
2.7 – 8.6 2.2 – 4.2 1.4 – 2.6 0.7 – 2.0
2.5 1.9 1.5 1.1
1.5 – 4.2 1.5 – 2.6 1.2 – 1.8 0.7 – 1.7
(Shouldn’t drink at all) (It doesn’t matter very much) (OK to drink sometimes) (OK to drink as much as you want)
Positive expect (1 unit increase)
Confidence (1 unit increase)
between peer norm and gender ( P < .01), peer norm and positive expectancies ( P < .01), and peer norm and confidence ( P =.04). These interactions were judged to be meaningful, so all three were included in the model. The Hosmer–Lemeshow goodness-of-fit test for the final model (Table 3) indicated good fit ( P =.72). Odds ratios (ORs) for variables not involved in interactions are included in Table 3. Cigarette smoking and marijuana use were positively associated with risk of having a drink of beer in the past 30 days, but the association was not as strong as before adjusting for other variables (Table 2). Students with negative expectancies about alcohol use (‘‘likely for negative things to happen’’) were less likely to have had a drink of beer in the past 30 days; as parent norm increases toward ‘‘OK to drink as much as you want’’ risk of drinking increases; and those who estimate a higher percentage of high school boys and girls in the United States who drank in the past 30 days were themselves more likely to have had a drink of beer in the past 30 days. The interactions can be viewed from two perspectives: the ORs for the three variables that interact with peer norm at each level of peer norm and the OR for peer norm at each level of the three variables. Using the first approach gives the ORs in Table 4. These indicate that when peer norm is 1 (shouldn’t drink at all), boys, those with positive expectancies, and those who would find alcohol hard to refuse are more likely to have had a drink of beer in the past 30 days. As peer norm increases toward ‘‘OK to drink as much as you want,’’ gender, positive expectancies, and confidence are no longer associated with risk of drinking. Similarly, examining peer norm at each level of gender, positive expectancies, and confidence shows peer norm has a larger influence on girls than boys, and the influence of peer norm decreases as positive expectancies increases toward ‘‘likely for positive things to happen’’ and confidence increases toward ‘‘hard to refuse.’’
4. Discussion In this study, a number of intrapersonal and social environmental factors were found to be associated with early alcohol use. These findings provide a basis for designing targeted intervention programs. Increased negative expectancies, decreased perception of alcohol use among U.S. high school students, and more prohibitive parent norms are all associated with
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1511
reduced risk of drinking, and should be considered as primary educational objectives when intervening with young adolescents. The interaction terms indicate that reducing positive expectancies and increasing confidence are associated with reduced drinking among adolescents with peer norms of ‘‘shouldn’t drink at all,’’ and are potential objectives for intervening with this low-risk target group. However, these factors are of less importance when peer norm is toward ‘‘it’s okay to drink as much as you want,’’ indicating that for this higher risk target group, modifying peer norm is a more viable objective than attempting to change either positive expectancies or confidence. Peer norm is also a more important factor in determining drinking behavior in girls than in boys. Thus, interventions for altering peer norms are more likely to have an effect on girls. The associations we observed are generally consistent with the findings of other studies (Johnstone, 1994). Only a few previous studies have examined interactions among these factors. There appear to be different underlying social norms regarding alcohol use and misuse for boys and girls, and these may result in gender differences in the effects of peer norms on drinking (Fleming, Kellam, & Brown, 1982; Hawkins et al., 1997; Thomas, 1996). SimonsMorton et al. (1999) found that alcohol use was associated with having ‘‘problem-behaving friends’’ for girls but not boys. This is in agreement with our finding that peer norms have a stronger effect for girls. No gender effect was found by Epstein, Botvin, Diaz, and Schinke (1995) or Kosterman, Hawkins, Guo, Catalano, and Abbott (2000), but these studies may not have tested for interactions between gender and other risk factors. Expectancies about the effects of alcohol consumption are generally strong predictors of alcohol consumption, with positive and negative expectancies having independent effects (Goldman, Del Boca, & Darkes, 1999). Our results concur with this. Using factor analysis, Grube and Agostinelli (1999) constructed three expectancy scales, two positive and one negative. One positive scale was ‘‘social facilitation,’’ with items such as ‘‘feel more confident or sure of yourself’’ and ‘‘feel more outgoing or friendly.’’ The other was ‘‘affective enhancement,’’ consisting of ‘‘feel happy,’’ feel relaxed,’’ and ‘‘have a lot of fun.’’ They found significant interactions between the three expectancy scales. We did not find two positive scales, and did not find interactions between positive and negative expectancies. Some studies have found positive expectancies to have a stronger effect in males than females (Griffin, Botvin, Epstein, Doyle, & Diaz, 2000; Thomas, 1996), although the outcome for the latter of these was heavy drinking, which may have different risk factors than drinking onset (Scheier, Botvin, & Baker, 1997). We did not find a direct interaction between gender and either positive or negative expectancies, but both gender and positive expectancies interacted with peer norm. It has been hypothesized that peer norms indirectly affect alcohol use through effect on positive alcohol expectancies (Scheier & Botvin, 1997). Although there is some evidence of this, peer norms are generally found to also have a direct effect (Ennett & Bauman, 1991; Scheier & Botvin, 1997; Webb, Baer, Francis, & Caid, 1993). We did not directly test mediational models, but the fact that peer norms remained significant in the multivariable
1512
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
model is evidence that there was not complete mediation by expectancies (Baron & Kenny, 1986). In addition to gender and positive expectancies, the other factor interacting with peer norm was confidence in ability to refuse alcohol. This is related to self-control and to some extent self-esteem. In younger children (fifth grade), low self-regulation was found to be associated with increased likelihood of having used alcohol (Jackson, Henriksen, Dickinson, & Levine, 1997). Males have been found to have poorer refusal skills than females (Scheier, Botvin, Diaz, & Griffin, 1999), but interactions have not been reported between gender and confidence regarding alcohol behaviors, nor did we find one. The interaction we observed between confidence and peer norms has not been reported by others. A possible explanation for this interaction is that some students who regularly drink may have a different interpretation of confidence in ability to refuse alcohol. One such respondent who selected ‘‘hard to refuse’’ for every item in this scale noted on the questionnaire that she had no desire to refuse. In cases such as this, ‘‘hard to refuse’’ could be a result of drinking behavior rather than a cause. We found parent norms toward alcohol use to be an independent risk factor for alcohol use. Other studies that have examined interactions between risk factors have also reported independent parent norm effects (Barnes & Welte, 1986; Grube & Morgan, 1990). Although parental influences may decrease as children get older, parent constraining influences are still important during the early teen years (Johnson & Johnson, 2001). Adolescents vary widely in perception of alcohol use among others. This perception is positively associated with use, as we found. Alcohol users may overestimate use among friends as a way to rationalize behavior, or may seek friends with similar behaviors (Grube & Morgan, 1990). Most studies have focused on perceived use among friends or peers (Barnes & Welte, 1986; Epstein et al., 1995; Griffin et al., 2000; Hansen et al., 1987; Hundleby & Mercer, 1987; Kosterman et al., 2000; Simons-Morton et al., 1999; Weber, Graham, Hansen, Flay, & Johnson, 1989). We measured three separate factors regarding perceived alcohol use: percentage for high school students in the United States, percentage for high school students in the district, and percentage for Grade 7/8 students. Only the first of these was significant in the multivariable analysis, implying that reducing the perception of widespread alcohol use among older youths may be the more important intervention target. Studies have suggested that overall social deviance is a risk factor for alcohol use (Hawkins et al., 1992). While we did not include a general deviance measure due to questionnaire length limitations, our finding that tobacco and marijuana use are independently associated with alcohol use may be a reflection of this. Use of these three substances is often found to be correlated (Barnes & Welte, 1986; Hansen et al., 1987; Scheier et al., 1999). A strength of our study is the large sample size, providing power to detect interactions that might be missed in smaller studies. The surveys were anonymous and confidential, increasing the likelihood of valid and reliable responses (Gfroerer, 1985). In addition, the sample was population based, with every seventh- and eighth-grade student in 16 school districts in the sampling frame. A study limitation is that it was cross-sectional, so that the temporal relationship between risk factors and drinking initiation cannot be established. Another is that the respondents were
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1513
only representative of rural White youth, limiting generalizability. However, substance use is high among rural youth (Johnston et al., 2003), and research on risk behaviors in this population is lacking (Fahs et al., 1999). In summary, we found that factors related to perceptions of alcohol use were associated with actual use in this adolescent sample. These perceptions are potentially modifiable, providing important directions for prevention programs to consider.
Acknowledgements This study was funded by grant RO1 AA10725 from the National Institute on Alcohol Abuse and Alcoholism. We would like to thank Anne Dorwaldt and Barbara Branch for their assistance with conducting the study.
References Akers, R. L., Krohn, M. D., Lanza-Kaduce, L., & Radosevich, M. (1979). Social learning and deviant behavior: A specific test of a general theory. American Sociological Review, 44, 636 – 655. Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall. Barnes, G. M., & Welte, J. W. (1986). Patterns and predictors of alcohol use among 7 – 12th grade students in New York State. Journal of Studies on Alcohol, 47, 53 – 62. Baron, R. M., & Kenny, D. A. (1986). The moderator – mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173 – 1182. Chou, S. P., & Pickering, R. P. (1992). Early onset of drinking as a risk factor for lifetime alcohol-related problems. British Journal of Addiction, 87, 1199 – 1204. Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297 – 334. Ennett, S. T., & Bauman, K. E. (1991). Mediators in the relationship between parental and peer characteristics and beer drinking by early adolescents. Journal of Applied Social Psychology, 21, 1699 – 1711. Epstein, J. A., Botvin, G. J., Diaz, T., & Schinke, S. P. (1995). The role of social factors and individual characteristics in promoting alcohol use among inner-city minority youths. Journal of Studies on Alcohol, 56, 39 – 46. Fahs, P. S., Smith, B. E., Atav, A. S., Britten, M. X., Collins, M. S., Morgan, L. L., & Spencer, G. A. (1999). Integrative research review of risk behaviors among adolescents in rural, suburban, and urban areas. Journal of Adolescent Health, 24, 230 – 243. Fleming, J. P., Kellam, S. G., & Brown, C. H. (1982). Early predictors of age at first use of alcohol, marijuana, and cigarettes. Drug and Alcohol Dependence, 9, 285 – 303. Gfroerer, J. (1985). Influence of privacy on self-reported drug use by youths. In B. A. Rouse, N. J. Kozel, & L. G. Richards (Eds.), Self-report methods of estimating drug use: Meeting current challenges to validity. NIDA Research Monograph 57 ( pp. 22 – 30). Washington, DC: U.S. Government Printing Office (DHHS Publication No. ADM 85-1402). Goldman, M. S., Del Boca, F. K., & Darkes, J. (1999). Alcohol expectancy theory: The application of cognitive neuroscience. In K. E. Leonard, & H. T. Blane (Eds.), Psychological theories of drinking and alcoholism (2nd ed.) (pp. 203 – 246). New York: Guilford Press. Grant, B. F., & Dawson, D. A. (1997). Age at onset of alcohol use and its association with DSM-IV alcohol abuse and dependence: Results from the National Longitudinal Alcohol Epidemiologic Survey. Journal of Substance Abuse, 9, 103 – 110.
1514
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
Griffin, K. W., Botvin, G. J., Epstein, J. A., Doyle, M. M., & Diaz, T. (2000). Psychosocial and behavioral factors in early adolescence as predictors of heavy drinking among high school seniors. Journal of Studies on Alcohol, 61, 603 – 606. Grube, J. W., & Agostinelli, G. E. (1999). Perceived consequences and adolescent drinking: Nonlinear and interactive models of alcohol expectancies. Psychology of Addictive Behaviors, 13, 303 – 312. Grube, J. W., & Morgan, M. (1990). Attitude – social support interactions: Contingent consistency effects in the prediction of adolescent smoking, drinking, and drug use. Social Psychology Quarterly, 53, 329 – 339. Gruber, E., DiClemente, R. J., Anderson, M. M., & Lodico, M. (1996). Early drinking onset and its association with alcohol use and problem behavior in late adolescence. Preventive Medicine, 25, 293 – 300. Hansen, W. B., Graham, J. W., Sobel, J. L., Shelton, D. R., Flay, B. R., & Johnson, C. A. (1987). The consistency of peer and parent influences on tobacco, alcohol, and marijuana use among young adolescents. Journal of Behavioral Medicine, 10, 559 – 579. Harman, H. H. (1976). Modern factor analysis (3rd ed.). Chicago: University of Chicago Press. Hawkins, J. D., Catalano, R. F., & Miller, J. Y. (1992). Risk and protective factors for alcohol and other drug problems in adolescence and early adulthood: Implications for substance abuse prevention. Psychological Bulletin, 112, 64 – 105. Hawkins, J. D., Graham, J. W., Maguin, E., Abbott, R., Hill, K. G., & Catalano, R. F. (1997). Exploring the effects of age of alcohol use initiation and psychosocial risk factors on subsequent alcohol misuse. Journal of Studies on Alcohol, 58, 280 – 290. Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression (2nd ed.). New York: Wiley. Hundleby, J. D., & Mercer, G. W. (1987). Family and friends as social environments and their relationship to young adolescents’ use of alcohol, tobacco, and marijuana. Journal of Marriage and the Family, 49, 151 – 164. Jackson, C., Henriksen, L., Dickinson, D., & Levine, D. W. (1997). The early use of alcohol and tobacco: Its relation to children’s competence and parents’ behavior. American Journal of Public Health, 87, 359 – 364. Jessor, R., & Jessor, S. L. (1977). Problem behavior and psychosocial development: A longitudinal study of youth. New York: Academic Press. Johnson, P. B., & Johnson, H. L. (2001). Reaffirming the power of parental influence on adolescent smoking and drinking decisions. Adolescent and Family Health, 2, 37 – 43. Johnston, L. D., O’Malley, P. M., & Bachman, J. G. (2003). Monitoring the Future national results on adolescent drug use: Overview of key findings. Bethesda, MD: National Institute on Drug Abuse (NIH Publication No. 03-5374). Johnstone, B. M. (1994). Sociodemographic, environmental, and cultural influences on adolescent drinking behavior. In R. Zucker, G. Boyd, & J. Howard (Eds.), The development of alcohol problems: Exploring the biopsychosocial matrix of risk ( pp. 169 – 203). Rockville, MD: U.S Department of Health and Human Services (NIH Publication No. 94-3495). Kosterman, R., Hawkins, J. D., Guo, J., Catalano, R. F., & Abbott, R. D. (2000). The dynamics of alcohol and marijuana initiation: Patterns and predictors of first use in adolescence. American Journal of Public Health, 90, 360 – 366. Robins, L. N. (1984). The natural history of adolescent drug use. American Journal of Public Health, 74, 656 – 657. SAS Institute. (2000). SAS version 8. Cary, NC: SAS Institute. Scheier, L. M., & Botvin, G. J. (1997). Expectancies as mediators of the effects of social influences and alcohol knowledge on adolescent alcohol use: A prospective analysis. Psychology of Addictive Behaviors, 11, 48 – 64. Scheier, L. M., Botvin, G. J., & Baker, E. (1997). Risk and protective factors as predictors of adolescent alcohol involvement and transitions in alcohol use: A prospective analysis. Journal of Studies on Alcohol, 58, 652 – 667. Scheier, L. M., Botvin, G. J., Diaz, T., & Griffin, K. W. (1999). Social skills, competence, and drug refusal efficacy as predictors of adolescent alcohol use. Journal of Drug Education, 29, 251 – 278. Simons-Morton, B., Haynie, D. L., Crump, A. D., Saylor, K. E., Eitel, P., & Yu, K. (1999). Expectancies and other psychosocial factors associated with alcohol use among early adolescent boys and girls. Addictive Behaviors, 24, 229 – 238. Thomas, B. S. (1996). A path analysis of gender differences in adolescent onset of alcohol, tobacco and other drug
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1515
use (ATOD), reported ATOD use and adverse consequences of ATOD use. Journal of Addictive Diseases, 15, 33 – 52. Webb, J. A., Baer, P. E., Francis, D. J., & Caid, C. D. (1993). Relationship among social and intrapersonal risk, alcohol expectancies, and alcohol usage among early adolescents. Addictive Behaviors, 18, 127 – 134. Weber, M. D., Graham, J. W., Hansen, W. B., Flay, B. R., & Johnson, C. A. (1989). Evidence for two paths of alcohol use onset in adolescents. Addictive Behaviors, 14, 399 – 408. Yamaguchi, K. (1990). Drug use and its social covariates from the period of adolescence to young adulthood: Some implications from longitudinal studies. In M. Galanter (Ed.), Recent developments in alcoholism, vol. 8. Combined alcohol and other drug dependence ( pp. 125 – 143). New York: Plenum Press.
Potentially modifiable psychosocial factors associated with alcohol use during early adolescence Peter W. Callasa,*, Brian S. Flynnb, John K. Wordenb a
Medical Biostatistics, University of Vermont, Hills Building, 105 Carrigan Drive, Burlington, VT 05405, USA b Office of Health Promotion Research and Department of Family Practice, College of Medicine, University of Vermont, Burlington, VT 05405, USA
Abstract This study was conducted to identify factors associated with alcohol use among early adolescents. A survey was administered to all Grade 7 and 8 students in 16 Vermont school districts. The questionnaire covered demographics, alcohol, tobacco, and marijuana use, and measures of psychosocial mediators of alcohol use drawn from social cognitive theory. These included positive and negative expectancies about alcohol effects, perceived peer and parent alcohol norms, perceived prevalence of adolescent alcohol use, and confidence in ability to refuse alcohol. Of the 2919 respondents, 29% reported having at least one drink of beer in the preceding 30 days. In logistic regression, factors independently related to risk of drinking beer in the past 30 days were smoking (odds ratio [OR] = 2.3, 95% confidence interval [CI] 1.8 – 3.0), marijuana use (OR 3.9, 95% CI 3.0– 5.2), negative expectancies (OR 0.4, 95% CI 0.3– 0.6), parent norms (OR 1.4, 95% CI 1.1 – 1.7), and estimated percentage of high school students who drink (OR 1.3, 95% CI 1.1 –1.5). Gender, positive alcohol expectancies, and lack of confidence in ability to refuse alcohol all significantly interacted with peer norm, with these items more strongly associated with alcohol use when peer norm is toward ‘‘shouldn’t drink.’’ Modifiable perceptions of alcohol use were strongly associated with actual use in this adolescent sample, providing a basis for intervention program design. D 2004 Elsevier Ltd. All rights reserved. Keywords: Drug abuse prevention; Alcohol; Adolescent attitudes; Measurement; Predictability
* Corresponding author. Tel.: +1-802-656-3195; fax: +1-802-656-0632. E-mail address: [email protected] (P.W. Callas). 0306-4603/$ – see front matter D 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.addbeh.2004.02.028
1504
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1. Introduction Alcohol is the most commonly used drug among adolescents in the United States. In the year 2002, 20% of 8th graders reported having consumed an alcoholic beverage in the previous month (Johnston, O’Malley, & Bachman, 2003). This increased to 49% by 12th grade. Early alcohol use is associated with an increased likelihood of later heavier use patterns and alcohol-related social and behavioral problems, as well as progression to other substances (Chou & Pickering, 1992; Grant & Dawson, 1997; Gruber, DiClemente, Anderson, & Lodico, 1996). There is evidence that delaying alcohol initiation reduces risk of these later problems (Hawkins et al., 1997; Robins, 1984; Yamaguchi, 1990). Understanding the factors that lead to early use is one step in prevention. According to social cognitive theory, cognitive, behavioral, and environmental factors interact to produce particular behaviors in given situations (Bandura, 1986). For substance use among adolescents, behavior has been theorized to result from interaction between personality system, behavior system, and perceived social environment (Jessor & Jessor, 1977). Specific factors in these related systems include expectancies about substance use effects, perceived norms, perceived prevalence of use of the substance, and confidence in refusing offers of substances. If these are associated with substance use, prevention efforts can focus on targeting them, such as by using advertising to strengthen negative expectancies about use or to model refusal skills (Hawkins, Catalano, & Miller, 1992). We surveyed Grade 7 and 8 students, examining the relationship between alcohol use and cognitive, behavioral, and social factors. The objective was to identify factors associated with alcohol use among early adolescents that could potentially be modified by educational interventions.
2. Methods 2.1. Participants In spring 1997, a survey regarding alcohol behavior and attitudes was given to Grade 7 and 8 students in 16 Vermont school districts. The school districts were composed of the eight intervention and eight comparison districts for a mass media youth alcohol prevention study. This survey was conducted prior to implementing the interventions. The study was approved by the University of Vermont Committee on Human Research. Students from Vermont were surveyed because Vermont has a high level of alcohol misuse and has standard statewide grade K-12 alcohol and drug abuse education programs, ensuring comparable educational exposures across schools. Of the 69 school districts in Vermont, 16 were selected based on size, demographic characteristics, and distribution of television and radio channels (for the media intervention).
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1505
2.2. Measurement The questionnaire administered to the students consisted of 110 items covering current and past alcohol use, future intentions for alcohol use, perceived access to alcohol, exposure to alcohol education, tobacco and marijuana use, demographics, and several psychosocial measures based on social cognitive theory (Bandura, 1986). These measures included expectancies about alcohol effect, perceived peer and parent normative attitudes toward alcohol, perceived adolescent alcohol use, and confidence in ability to refuse alcohol. Expectancies were measured using 28 questionnaire items concerning potential physical, social, and psychological consequences of alcohol. The common stem for these items stated that, ‘‘If I had 2 or 3 whole drinks of beer, wine, or liquor, I would . . ..’’ Individual items included ‘‘get along better with friends,’’ ‘‘feel more confident,’’ ‘‘feel sick,’’ and ‘‘act stupid.’’ Possible responses for each item were (1) ‘‘unlikely,’’ (2) ‘‘possibly,’’ and (3) ‘‘likely.’’ These items were developed to represent a balance of positive and negative social, affective, and physical consequences expected from alcohol use. Alcohol social norm measures were based on work by Akers, Krohn, Lanza-Kaduce, and Radosevich (1979). The 14 items included questions such as ‘‘What does your best male friend think about girls your own age drinking?’’ and ‘‘What does your mother think about girls your own age drinking?’’ Response choices were (1) ‘‘shouldn’t drink at all,’’ (2) ‘‘it doesn’t matter very much,’’ (3) ‘‘it’s okay to drink sometimes,’’ and (4) ‘‘it’s okay to drink as much as you want.’’ These correspond to the categories labeled by Akers et al. as proscriptive, ascriptive, prescriptive, and permissive. Perceived adolescent alcohol use consisted of eight items covering each combination of boys or girls of the respondent’s age or high school age in the respondent’s school district or the United States. Items asked for the best guess as to the percentage in each group who had one or more drinks of beer, wine, or liquor in the past 30 days. Respondents marked estimates on a visual analog scale of 0–100%, which had labeled marks at 0%, 25%, 50%, 75%, and 100%. Responses were coded to the nearest 1% for data analysis. Confidence in ability to refuse alcohol was derived from the self-efficacy principles of Bandura (1986). Fifteen items included a number of social situations where the respondent might be offered alcohol. Items were preceded by the common stem, ‘‘If someone offered you a drink of beer, wine, or liquor, do you feel it would be easy to refuse or hard to refuse. . ..’’ Situations included ‘‘at a friend’s house after school and no adults are there’’ and ‘‘at a party with friends and most people are drinking.’’ Response choices were (1) ‘‘easy to refuse,’’ (2) ‘‘not sure,’’ and (3) ‘‘hard to refuse.’’ 2.3. Data collection The survey was conducted during normal class times, with the regular classroom teacher absent from the room. All Grade 7 and 8 students at each school completed the survey except those whose parents refused, special education students excluded by the school, and students who were absent on the day of the survey.
1506
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
2.4. Data analysis We conducted bivariate and multivariate analyses relating a number of independent variables to alcohol use. SAS (SAS Institute, 2000) was used for all data analyses. Alcohol use was measured by the question, ‘‘During the past 30 days, on how many days did you have at least one drink of beer?’’ Responses were dichotomized to ‘‘0 days’’ versus ‘‘1 or more days.’’ Independent variables included grade, gender, tobacco use, marijuana use, and several scales developed from the psychosocial measures on the questionnaire. Scales were constructed using principal components factor analysis with varimax rotation (Harman, 1976). Eigenvalues, scree plots, variance proportion, and judgment as to interpretability were used to determine the number of factors to retain. Internal consistency of each scale was measured by computing Cronbach’s coefficient alpha (Cronbach, 1951). Bivariate relationships were assessed via Fisher’s exact test for categorical variables and two-sample t test for continuous variables. All independent variables were evaluated simultaneously using logistic regression (Hosmer & Lemeshow, 2000), with relevant interaction terms considered as candidates for inclusion in the model after significant main effects were identified.
3. Results 3.1. Respondents The survey was completed by 2919 students from 24 schools in the 16 districts. This was nearly a complete census of the population attending school on the survey days, since only a few parents refused participation. Upon reviewing the questionnaires, eight were discarded due to clearly invalid responses, leaving a sample of 2911 for data analysis. Some characteristics of the respondents are given in Table 1. The age range was 11–16 years, with 81% either 13 or 14 years old. Only 14% of students had never had a sip or taste of beer, wine, or liquor, but 42% had never had a drink of these beverages (not just a taste or sip). For substance use in the previous 30 days, 29% reported drinking beer on at least one day, 26% reported smoking cigarettes, and 18% used marijuana. 3.2. Scale development Four separate factor analyses were conducted to identify factors from the multipleitem measures of perceptions concerning alcohol use. The resulting factors were used to construct factor-based scales from each subject’s mean score over the items in the factor. The first factor analysis, using the 28 questions pertaining to expected consequences of drinking alcohol, found two important factors, with 14 questions loading on Factor 1 and 14
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1507
Table 1 Characteristics of respondents No. (%) Grade Gender How many times have you had a drink of beer, wine, or liquor)not just a taste or sip?
During the past 30 days, on how many days did you have at least one drink of beer?
During the past 30 days, on how many days did you smoke cigarettes?
During the past 30 days, on how many days did you use marijuana?
7 8 Girl Boy Can’t remember Never 1 or 2 times 3 to 10 times More than 10 times 0 days 1 or 2 days 3 to 5 days 6 to 9 days 10 to 19 days 20 or more days 0 days 1 or 2 days 3 to 5 days 6 to 9 days 10 to 19 days 20 or more days 0 days 1 or 2 days 3 to 5 days 6 to 9 days 10 to 19 days 20 or more days
1438 (50) 1459 (50) 1471 (51) 1416 (49) 132 (5) 1221 (42) 586 (20) 456 (16) 503 (17) 2043 (71) 472 (16) 185 (6) 77 (3) 57 (2) 59 (2) 2136 (74) 238 (8) 108 (4) 69 (2) 78 (3) 258 (9) 2352 (82) 178 (6) 109 (4) 78 (3) 63 (2) 100 (4)
on Factor 2. The first factor, the negative expectancies scale (the mean of these 14 items), ranged from 1 (unlikely for negative things to happen) to 3 (likely for negative things to happen). The second factor, the positive expectancies scale, ranged from 1 (unlikely for positive things to happen) to 3 (likely for positive things to happen). In the second factor analysis, of 14 questions regarding perceived attitude norms of alcohol use, two factors were retained. Eight items loaded primarily on Factor 1 (peer norm), and six on Factor 2 (parent norm). The peer and parent norm scales both ranged from 1 (shouldn’t drink at all) to 4 (OK to drink as much as you want). The third factor analysis used eight questions on the student’s estimate of the percentage of Grade 7/8 and high school boys and girls in the student’s district and in the entire United States who had one or more drinks of beer, wine, or liquor in the past 30 days. The first of three factors consisted of four questions on the percentage of Grade 7 and 8 boys and girls estimated to have had a drink in the past 30 days. The second had two questions on high school boys and girls in the student’s district, and the third had two questions on high school boys and girls in the United States.
1508
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
Table 2 Respondent characteristics versus beer-drinking behavior
Grade Gender Days smoke cigarettes in past 30 days Days use marijuana in past 30 days Positive expectancies scale
Negative expectancies scale
Peer norm scale
Parent norm scale
Percent drink scale for high school students in United States Percent drink scale for high school students in student’s district Percent drink scale for grade 7/8 students
Confidence scale
a
n 7 8 Girl Boy 0 1+ 0 1+ 1.00 – 1.49 1.50 – 1.99 2.00 – 2.49 2.50 – 3.00 1.00 – 1.49 1.50 – 1.99 2.00 – 2.49 2.50 – 3.00 1.00 – 1.99 2.00 – 2.99 3.00 – 4.00 1.00 – 1.99 2.00 – 2.99 3.00 – 4.00 0 – 24% 25 – 49% 50 – 74% 75 – 100% 0 – 24% 25 – 49% 50 – 74% 75 – 100% 0 – 24% 25 – 49% 50 – 74% 75 – 100% 1.00 – 1.49 1.50 – 1.99 2.00 – 2.49 2.50 – 3.00
(unlikely) (possibly) (likely) (very likely) (unlikely) (possibly) (likely) (very likely) (shouldn’t drink) (doesn’t matter) (OK to drink) (shouldn’t drink) (doesn’t matter) (OK to drink)
(easy to refuse) (not sure) (hard to refuse) (very hard to refuse)
Total n
Drink beer 1 or more days, no. (%)a
2893 1432 1448 1467 1404 2121 749 2339 526 1062 855 684 261 376 618 928 945 1200 908 760 2575 249 47 109 583 1335 845 411 879 996 582 885 1095 699 189 1056 647 754 422
850 (29) 339 (24) 508 (35) 385 (26) 456 (33) 353 (17) 490 (65) 428 (18) 414 (79) 77 (7) 237 (28) 346 (51) 185 (71) 178 (47) 303 (49) 254 (27) 109 (12) 61 (5) 297 (33) 488 (64) 662 (26) 144 (58) 38 (81) 20 (18) 115 (20) 375 (28) 334 (40) 66 (16) 182 (21) 328 (33) 268 (46) 150 (17) 300 (27) 294 (42) 100 (53) 129 (12) 125 (19) 302 (40) 290 (69)
No. (%) reporting having had at least one drink of beer in the past 30 days.
OR
95% CI
1.7
1.5 – 2.1
1.4
1.2 – 1.6
9.5
7.8 – 11.5
16.5
13.1 – 20.9
4.9 13.1 31.1
3.7 – 6.5 9.9 – 17.3 21.9 – 44.4
1.1 0.4 0.2
0.8 – 1.4 0.3 – 0.5 0.1 – 0.2
9.1 33.5
6.8 – 12.2 24.9 – 45.1
4.0 12.2
3.0 – 5.2 5.9 – 25.4
1.1 1.7 2.9
0.7 – 1.7 1.1 – 2.9 1.8 – 4.8
1.4 2.6 4.5
1.0 – 1.9 1.9 – 3.5 3.3 – 6.1
1.9 3.6 5.5
1.5 – 2.3 2.8 – 4.5 3.9 – 7.7
1.7 4.8 15.8
1.3 – 2.3 3.8 – 6.1 12.0 – 20.8
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1509
For the fourth factor analysis, there were 15 questions on how hard it would be to refuse a drink of beer, wine, or liquor in a number of situations. Only one factor was found. The confidence scale ranged from 1 (easy to refuse) to 3 (hard to refuse). Cronbach’s coefficient alpha was greater than .80 for all of the above factors. 3.3. Bivariate analyses Table 2 presents the bivariate relationships between respondent characteristics and beer drinking behavior in the past 30 days. For the scales, findings were similar when they were treated as continuous and when they were categorized; only the categorized versions are presented. Risk of beer drinking was higher for Grade 8 students, boys, and respondents who smoked cigarettes or marijuana, had high positive expectancies, had low negative expectancies, perceived permissive peer and parent norms toward alcohol use, estimated high levels of drinking, and felt they would have difficulty refusing if offered alcohol. 3.4. Multivariable analyses In bivariate analysis, the negative expectancies scale did not show a linear change in the natural log of the odds across its range. The best representation of this variable for modeling was to combine Categories 1 and 2. The exponential assumption for continuous variables in logistic regression modeling was reasonable for all other scales, so they were not changed. For main effects, all independent variables were significantly associated with beer drinking in the past 30 days except grade (Wald P =.19), percentage high school district ( P =.46), and percentage Grade 7/8 ( P =.24). Three two-way interactions were significant: the interaction Table 3 Logistic regression model Variable Intercept Gender Smoking in past 30 days Marijuana in past 30 days Positive expectancies Negative expectancies Peer norm Parent norm Percent high school U.S. Confidence Peer Norm Gender Peer Norm Positive Expectancies Peer Norm Confidence a b
See text and Table 4. See text.
Comparison
Boy vs. girl 1+ vs. 0 days 1+ vs. 0 days 1 unit increase 1 unit increase 1 unit increase 1 unit increase 25% increase 1 unit increase
Parameter 10.83 1.78 0.84 1.37 2.03 0.93 2.88 0.32 0.23 1.21 0.49 0.46 0.28
Standard error
Odds ratio
0.43 0.13 0.15 0.45 0.21 0.45 0.11 0.08 0.38 0.16 0.16 0.13
a
2.3 3.9
95% Confidence interval
1.8 – 3.0 3.0 – 5.2
a
0.4
0.3 – 0.6
b
1.4 1.3 a
1.1 – 1.7 1.1 – 1.5
1510
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
Table 4 Odds ratios for variables interacting with peer norm Level of peer norm
Gender (boy vs. girl) OR
95% CI
OR
95% CI
OR
95% CI
1 2 3 4
3.7 2.2 1.4 0.9
2.1 – 6.4 1.7 – 3.0 1.1 – 1.8 0.5 – 1.4
4.8 3.0 1.9 1.2
2.7 – 8.6 2.2 – 4.2 1.4 – 2.6 0.7 – 2.0
2.5 1.9 1.5 1.1
1.5 – 4.2 1.5 – 2.6 1.2 – 1.8 0.7 – 1.7
(Shouldn’t drink at all) (It doesn’t matter very much) (OK to drink sometimes) (OK to drink as much as you want)
Positive expect (1 unit increase)
Confidence (1 unit increase)
between peer norm and gender ( P < .01), peer norm and positive expectancies ( P < .01), and peer norm and confidence ( P =.04). These interactions were judged to be meaningful, so all three were included in the model. The Hosmer–Lemeshow goodness-of-fit test for the final model (Table 3) indicated good fit ( P =.72). Odds ratios (ORs) for variables not involved in interactions are included in Table 3. Cigarette smoking and marijuana use were positively associated with risk of having a drink of beer in the past 30 days, but the association was not as strong as before adjusting for other variables (Table 2). Students with negative expectancies about alcohol use (‘‘likely for negative things to happen’’) were less likely to have had a drink of beer in the past 30 days; as parent norm increases toward ‘‘OK to drink as much as you want’’ risk of drinking increases; and those who estimate a higher percentage of high school boys and girls in the United States who drank in the past 30 days were themselves more likely to have had a drink of beer in the past 30 days. The interactions can be viewed from two perspectives: the ORs for the three variables that interact with peer norm at each level of peer norm and the OR for peer norm at each level of the three variables. Using the first approach gives the ORs in Table 4. These indicate that when peer norm is 1 (shouldn’t drink at all), boys, those with positive expectancies, and those who would find alcohol hard to refuse are more likely to have had a drink of beer in the past 30 days. As peer norm increases toward ‘‘OK to drink as much as you want,’’ gender, positive expectancies, and confidence are no longer associated with risk of drinking. Similarly, examining peer norm at each level of gender, positive expectancies, and confidence shows peer norm has a larger influence on girls than boys, and the influence of peer norm decreases as positive expectancies increases toward ‘‘likely for positive things to happen’’ and confidence increases toward ‘‘hard to refuse.’’
4. Discussion In this study, a number of intrapersonal and social environmental factors were found to be associated with early alcohol use. These findings provide a basis for designing targeted intervention programs. Increased negative expectancies, decreased perception of alcohol use among U.S. high school students, and more prohibitive parent norms are all associated with
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1511
reduced risk of drinking, and should be considered as primary educational objectives when intervening with young adolescents. The interaction terms indicate that reducing positive expectancies and increasing confidence are associated with reduced drinking among adolescents with peer norms of ‘‘shouldn’t drink at all,’’ and are potential objectives for intervening with this low-risk target group. However, these factors are of less importance when peer norm is toward ‘‘it’s okay to drink as much as you want,’’ indicating that for this higher risk target group, modifying peer norm is a more viable objective than attempting to change either positive expectancies or confidence. Peer norm is also a more important factor in determining drinking behavior in girls than in boys. Thus, interventions for altering peer norms are more likely to have an effect on girls. The associations we observed are generally consistent with the findings of other studies (Johnstone, 1994). Only a few previous studies have examined interactions among these factors. There appear to be different underlying social norms regarding alcohol use and misuse for boys and girls, and these may result in gender differences in the effects of peer norms on drinking (Fleming, Kellam, & Brown, 1982; Hawkins et al., 1997; Thomas, 1996). SimonsMorton et al. (1999) found that alcohol use was associated with having ‘‘problem-behaving friends’’ for girls but not boys. This is in agreement with our finding that peer norms have a stronger effect for girls. No gender effect was found by Epstein, Botvin, Diaz, and Schinke (1995) or Kosterman, Hawkins, Guo, Catalano, and Abbott (2000), but these studies may not have tested for interactions between gender and other risk factors. Expectancies about the effects of alcohol consumption are generally strong predictors of alcohol consumption, with positive and negative expectancies having independent effects (Goldman, Del Boca, & Darkes, 1999). Our results concur with this. Using factor analysis, Grube and Agostinelli (1999) constructed three expectancy scales, two positive and one negative. One positive scale was ‘‘social facilitation,’’ with items such as ‘‘feel more confident or sure of yourself’’ and ‘‘feel more outgoing or friendly.’’ The other was ‘‘affective enhancement,’’ consisting of ‘‘feel happy,’’ feel relaxed,’’ and ‘‘have a lot of fun.’’ They found significant interactions between the three expectancy scales. We did not find two positive scales, and did not find interactions between positive and negative expectancies. Some studies have found positive expectancies to have a stronger effect in males than females (Griffin, Botvin, Epstein, Doyle, & Diaz, 2000; Thomas, 1996), although the outcome for the latter of these was heavy drinking, which may have different risk factors than drinking onset (Scheier, Botvin, & Baker, 1997). We did not find a direct interaction between gender and either positive or negative expectancies, but both gender and positive expectancies interacted with peer norm. It has been hypothesized that peer norms indirectly affect alcohol use through effect on positive alcohol expectancies (Scheier & Botvin, 1997). Although there is some evidence of this, peer norms are generally found to also have a direct effect (Ennett & Bauman, 1991; Scheier & Botvin, 1997; Webb, Baer, Francis, & Caid, 1993). We did not directly test mediational models, but the fact that peer norms remained significant in the multivariable
1512
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
model is evidence that there was not complete mediation by expectancies (Baron & Kenny, 1986). In addition to gender and positive expectancies, the other factor interacting with peer norm was confidence in ability to refuse alcohol. This is related to self-control and to some extent self-esteem. In younger children (fifth grade), low self-regulation was found to be associated with increased likelihood of having used alcohol (Jackson, Henriksen, Dickinson, & Levine, 1997). Males have been found to have poorer refusal skills than females (Scheier, Botvin, Diaz, & Griffin, 1999), but interactions have not been reported between gender and confidence regarding alcohol behaviors, nor did we find one. The interaction we observed between confidence and peer norms has not been reported by others. A possible explanation for this interaction is that some students who regularly drink may have a different interpretation of confidence in ability to refuse alcohol. One such respondent who selected ‘‘hard to refuse’’ for every item in this scale noted on the questionnaire that she had no desire to refuse. In cases such as this, ‘‘hard to refuse’’ could be a result of drinking behavior rather than a cause. We found parent norms toward alcohol use to be an independent risk factor for alcohol use. Other studies that have examined interactions between risk factors have also reported independent parent norm effects (Barnes & Welte, 1986; Grube & Morgan, 1990). Although parental influences may decrease as children get older, parent constraining influences are still important during the early teen years (Johnson & Johnson, 2001). Adolescents vary widely in perception of alcohol use among others. This perception is positively associated with use, as we found. Alcohol users may overestimate use among friends as a way to rationalize behavior, or may seek friends with similar behaviors (Grube & Morgan, 1990). Most studies have focused on perceived use among friends or peers (Barnes & Welte, 1986; Epstein et al., 1995; Griffin et al., 2000; Hansen et al., 1987; Hundleby & Mercer, 1987; Kosterman et al., 2000; Simons-Morton et al., 1999; Weber, Graham, Hansen, Flay, & Johnson, 1989). We measured three separate factors regarding perceived alcohol use: percentage for high school students in the United States, percentage for high school students in the district, and percentage for Grade 7/8 students. Only the first of these was significant in the multivariable analysis, implying that reducing the perception of widespread alcohol use among older youths may be the more important intervention target. Studies have suggested that overall social deviance is a risk factor for alcohol use (Hawkins et al., 1992). While we did not include a general deviance measure due to questionnaire length limitations, our finding that tobacco and marijuana use are independently associated with alcohol use may be a reflection of this. Use of these three substances is often found to be correlated (Barnes & Welte, 1986; Hansen et al., 1987; Scheier et al., 1999). A strength of our study is the large sample size, providing power to detect interactions that might be missed in smaller studies. The surveys were anonymous and confidential, increasing the likelihood of valid and reliable responses (Gfroerer, 1985). In addition, the sample was population based, with every seventh- and eighth-grade student in 16 school districts in the sampling frame. A study limitation is that it was cross-sectional, so that the temporal relationship between risk factors and drinking initiation cannot be established. Another is that the respondents were
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1513
only representative of rural White youth, limiting generalizability. However, substance use is high among rural youth (Johnston et al., 2003), and research on risk behaviors in this population is lacking (Fahs et al., 1999). In summary, we found that factors related to perceptions of alcohol use were associated with actual use in this adolescent sample. These perceptions are potentially modifiable, providing important directions for prevention programs to consider.
Acknowledgements This study was funded by grant RO1 AA10725 from the National Institute on Alcohol Abuse and Alcoholism. We would like to thank Anne Dorwaldt and Barbara Branch for their assistance with conducting the study.
References Akers, R. L., Krohn, M. D., Lanza-Kaduce, L., & Radosevich, M. (1979). Social learning and deviant behavior: A specific test of a general theory. American Sociological Review, 44, 636 – 655. Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall. Barnes, G. M., & Welte, J. W. (1986). Patterns and predictors of alcohol use among 7 – 12th grade students in New York State. Journal of Studies on Alcohol, 47, 53 – 62. Baron, R. M., & Kenny, D. A. (1986). The moderator – mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173 – 1182. Chou, S. P., & Pickering, R. P. (1992). Early onset of drinking as a risk factor for lifetime alcohol-related problems. British Journal of Addiction, 87, 1199 – 1204. Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297 – 334. Ennett, S. T., & Bauman, K. E. (1991). Mediators in the relationship between parental and peer characteristics and beer drinking by early adolescents. Journal of Applied Social Psychology, 21, 1699 – 1711. Epstein, J. A., Botvin, G. J., Diaz, T., & Schinke, S. P. (1995). The role of social factors and individual characteristics in promoting alcohol use among inner-city minority youths. Journal of Studies on Alcohol, 56, 39 – 46. Fahs, P. S., Smith, B. E., Atav, A. S., Britten, M. X., Collins, M. S., Morgan, L. L., & Spencer, G. A. (1999). Integrative research review of risk behaviors among adolescents in rural, suburban, and urban areas. Journal of Adolescent Health, 24, 230 – 243. Fleming, J. P., Kellam, S. G., & Brown, C. H. (1982). Early predictors of age at first use of alcohol, marijuana, and cigarettes. Drug and Alcohol Dependence, 9, 285 – 303. Gfroerer, J. (1985). Influence of privacy on self-reported drug use by youths. In B. A. Rouse, N. J. Kozel, & L. G. Richards (Eds.), Self-report methods of estimating drug use: Meeting current challenges to validity. NIDA Research Monograph 57 ( pp. 22 – 30). Washington, DC: U.S. Government Printing Office (DHHS Publication No. ADM 85-1402). Goldman, M. S., Del Boca, F. K., & Darkes, J. (1999). Alcohol expectancy theory: The application of cognitive neuroscience. In K. E. Leonard, & H. T. Blane (Eds.), Psychological theories of drinking and alcoholism (2nd ed.) (pp. 203 – 246). New York: Guilford Press. Grant, B. F., & Dawson, D. A. (1997). Age at onset of alcohol use and its association with DSM-IV alcohol abuse and dependence: Results from the National Longitudinal Alcohol Epidemiologic Survey. Journal of Substance Abuse, 9, 103 – 110.
1514
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
Griffin, K. W., Botvin, G. J., Epstein, J. A., Doyle, M. M., & Diaz, T. (2000). Psychosocial and behavioral factors in early adolescence as predictors of heavy drinking among high school seniors. Journal of Studies on Alcohol, 61, 603 – 606. Grube, J. W., & Agostinelli, G. E. (1999). Perceived consequences and adolescent drinking: Nonlinear and interactive models of alcohol expectancies. Psychology of Addictive Behaviors, 13, 303 – 312. Grube, J. W., & Morgan, M. (1990). Attitude – social support interactions: Contingent consistency effects in the prediction of adolescent smoking, drinking, and drug use. Social Psychology Quarterly, 53, 329 – 339. Gruber, E., DiClemente, R. J., Anderson, M. M., & Lodico, M. (1996). Early drinking onset and its association with alcohol use and problem behavior in late adolescence. Preventive Medicine, 25, 293 – 300. Hansen, W. B., Graham, J. W., Sobel, J. L., Shelton, D. R., Flay, B. R., & Johnson, C. A. (1987). The consistency of peer and parent influences on tobacco, alcohol, and marijuana use among young adolescents. Journal of Behavioral Medicine, 10, 559 – 579. Harman, H. H. (1976). Modern factor analysis (3rd ed.). Chicago: University of Chicago Press. Hawkins, J. D., Catalano, R. F., & Miller, J. Y. (1992). Risk and protective factors for alcohol and other drug problems in adolescence and early adulthood: Implications for substance abuse prevention. Psychological Bulletin, 112, 64 – 105. Hawkins, J. D., Graham, J. W., Maguin, E., Abbott, R., Hill, K. G., & Catalano, R. F. (1997). Exploring the effects of age of alcohol use initiation and psychosocial risk factors on subsequent alcohol misuse. Journal of Studies on Alcohol, 58, 280 – 290. Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression (2nd ed.). New York: Wiley. Hundleby, J. D., & Mercer, G. W. (1987). Family and friends as social environments and their relationship to young adolescents’ use of alcohol, tobacco, and marijuana. Journal of Marriage and the Family, 49, 151 – 164. Jackson, C., Henriksen, L., Dickinson, D., & Levine, D. W. (1997). The early use of alcohol and tobacco: Its relation to children’s competence and parents’ behavior. American Journal of Public Health, 87, 359 – 364. Jessor, R., & Jessor, S. L. (1977). Problem behavior and psychosocial development: A longitudinal study of youth. New York: Academic Press. Johnson, P. B., & Johnson, H. L. (2001). Reaffirming the power of parental influence on adolescent smoking and drinking decisions. Adolescent and Family Health, 2, 37 – 43. Johnston, L. D., O’Malley, P. M., & Bachman, J. G. (2003). Monitoring the Future national results on adolescent drug use: Overview of key findings. Bethesda, MD: National Institute on Drug Abuse (NIH Publication No. 03-5374). Johnstone, B. M. (1994). Sociodemographic, environmental, and cultural influences on adolescent drinking behavior. In R. Zucker, G. Boyd, & J. Howard (Eds.), The development of alcohol problems: Exploring the biopsychosocial matrix of risk ( pp. 169 – 203). Rockville, MD: U.S Department of Health and Human Services (NIH Publication No. 94-3495). Kosterman, R., Hawkins, J. D., Guo, J., Catalano, R. F., & Abbott, R. D. (2000). The dynamics of alcohol and marijuana initiation: Patterns and predictors of first use in adolescence. American Journal of Public Health, 90, 360 – 366. Robins, L. N. (1984). The natural history of adolescent drug use. American Journal of Public Health, 74, 656 – 657. SAS Institute. (2000). SAS version 8. Cary, NC: SAS Institute. Scheier, L. M., & Botvin, G. J. (1997). Expectancies as mediators of the effects of social influences and alcohol knowledge on adolescent alcohol use: A prospective analysis. Psychology of Addictive Behaviors, 11, 48 – 64. Scheier, L. M., Botvin, G. J., & Baker, E. (1997). Risk and protective factors as predictors of adolescent alcohol involvement and transitions in alcohol use: A prospective analysis. Journal of Studies on Alcohol, 58, 652 – 667. Scheier, L. M., Botvin, G. J., Diaz, T., & Griffin, K. W. (1999). Social skills, competence, and drug refusal efficacy as predictors of adolescent alcohol use. Journal of Drug Education, 29, 251 – 278. Simons-Morton, B., Haynie, D. L., Crump, A. D., Saylor, K. E., Eitel, P., & Yu, K. (1999). Expectancies and other psychosocial factors associated with alcohol use among early adolescent boys and girls. Addictive Behaviors, 24, 229 – 238. Thomas, B. S. (1996). A path analysis of gender differences in adolescent onset of alcohol, tobacco and other drug
P.W. Callas et al. / Addictive Behaviors 29 (2004) 1503–1515
1515
use (ATOD), reported ATOD use and adverse consequences of ATOD use. Journal of Addictive Diseases, 15, 33 – 52. Webb, J. A., Baer, P. E., Francis, D. J., & Caid, C. D. (1993). Relationship among social and intrapersonal risk, alcohol expectancies, and alcohol usage among early adolescents. Addictive Behaviors, 18, 127 – 134. Weber, M. D., Graham, J. W., Hansen, W. B., Flay, B. R., & Johnson, C. A. (1989). Evidence for two paths of alcohol use onset in adolescents. Addictive Behaviors, 14, 399 – 408. Yamaguchi, K. (1990). Drug use and its social covariates from the period of adolescence to young adulthood: Some implications from longitudinal studies. In M. Galanter (Ed.), Recent developments in alcoholism, vol. 8. Combined alcohol and other drug dependence ( pp. 125 – 143). New York: Plenum Press.