Stepping Back from Social Norms Campaigns: Comparing Normative Influences to Other Predictors of Health Behaviors

Transcription

1 CAMERON STEPPING BACK AND CAMPO FROM SOCIAL NORMS CAMPAIGNS HEALTH COMMUNICATION, 20(3), Copyright 2006, Lawrence Erlbaum Associates, Inc. Stepping Back from Social Norms Campaigns: Comparing Normative Influences to Other s of Health Behaviors Kenzie A. Cameron Center for Communication and Medicine and Division of General Internal Medicine Northwestern University Shelly Campo Department of Community and Behavioral Health and Department of Communication Studies University of Iowa Recent health campaigns on college campuses have used a social norms approach, which suggests that one s perceptions of others attitudes and behaviors are the key components in attitude and behavior change. However, the efficacy of social norms campaigns has been mixed. This study was conducted to assess the relationships among sociodemographics, normative perceptions, and individual attitudes on 3 health behaviors. Students at 2 universities (N = 393) completed questionnaires assessing how these variables related to their consumption of alcohol, tobacco use, and exercise behaviors. Regressions indicated that each of these variables was associated with behavior, but varied independent variables emerged as the salient predictors among behaviors. In several conditions the effect of normative perceptions on behaviors was not significant, a finding in direct opposition to social norms marketing. In all 3 behavioral conditions, the variable accounting for the greatest variance was whether or not the individual liked participating in that particular behavior. Thus, although some social norms marketing may be meeting with success, it may be the case that predicted attitudinal and behavioral changes will not be found when applied across diverse health topics. Social norms marketing campaigns have become a favorite among many universities attempting to discourage unhealthy behaviors, particularly those related to binge drinking. Recent research suggests that social norms campaigns have been adopted by nearly half of all colleges and universities in the United States (Wechsler, Seibring, Liu, & Ahl, 2004) and have spread to state health departments (Frauenfelder, 2001). Social norms marketing was named an idea of the year in 2001 by the New York Times Magazine and has been described as the science of persuading people to go along with the crowd (Frauenfelder, 2001, p. 100). These campaigns are based on pluralistic ignorance (Allport, 1924; Fields & Schuman, 1976; Prentice & Miller, Correspondence should be addressed to Kenzie A. Cameron, Center for Communication & Medicine, Northwestern University, The Feinberg School of Medicine, Division of General Internal Medicine, 676 North St. Clair Street, Suite 200, Chicago, Illinois k-cameron@north western.edu 1993), which suggests that individuals are unable to accurately judge the social norm. The underlying assumption is that once you correct the perceived norm so that it matches the actual norm, individuals will alter their behavior accordingly (Perkins & Berkowitz, 1986). Social norms marketing originates from a Perkins and Berkowitz (1986) study indicating that students misperceived social norms regarding how much alcohol fellow students consumed. If the misperception were corrected, it was believed behavior change would follow. The idea first was used widely at Northern Illinois University in 1990, where newspaper ads, posters, and handouts conveyed that most students had five or fewer drinks when they partied. Northern Illinois University and the University of Arizona, which implemented social norms campaigns along with other harm reduction approaches, reported significant declines in student drinking behavior (Frauenfelder, 2001; Johannessen, Collins, Mills-Novoa, & Glider, 1999). Researchers differ in their opinions of the efficacy of social norms campaigns.

2 278 CAMERON AND CAMPO Some tout the observed reductions in college drinking (Perkins & Berkowitz, 1986; Perkins, Meilman, Leichlier, Cashin, & Presley, 1999; Stewart et al., 2002), whereas others question the discovered effects (Campo et al., 2003; Campo & Cameron, 2006; Campo, Cameron, Brossard, & Fraser, 2004; Keeling, 1999, 2000; Wechsler & Kuo, 2000; Wechsler et al., 2003). Research has begun to explore potential moderators between normative perceptions and behavior to better understand why social norms campaigns may succeed or fail (Rimal & Real, 2003). Yet, with so much current focus on social norms campaigns, as well as the influx of funds for such campaigns, administrators may have put the cart before the horse. Even though social norms marketing often is implicitly presented as being a theory unto itself, it is atheoretical and has not been rigorously tested, as Keeling (1999, 2000) states in a series of editorials. Rather, common assumptions are advanced in the social norms literature (e.g., people want to be like others, social norms are more salient than personal attitudes, normative perceptions are related to actual behavior); these assumptions are what public health researchers test (Campo et al., 2003, Perkins & Berkowitz, 1986; Wechsler et al., 2004). We propose the need to step back from the narrow focus of social norms campaigns to reexamine the influences, normative and others, on college student health behaviors. A social norms approach assumes attitude and behavior change are influenced by one s perception of others attitudes and behavior. Numerous persuasion theories suggest that one s own attitudes influence behaviors (cf. Ajzen, 1985; Kim & Hunter, 1993). The theory of reasoned action (TRA; Ajzen & Fishbein, 1980) and the theory of planned behavior (TPB; Ajzen, 1985) predict that normative beliefs affect behavior. These theories consider other influences, such as personal attitudes, and note that we are not always motivated to comply with what others think we should do. Demographic characteristics such as age, sex, and race also have been identified as related to college student health behaviors (Weitzman, 2004). Other variables shown to be related to drinking behaviors include sex (e.g., Gomberg, Schneider, & DeJong, 2000), fraternity or sorority membership (e.g., Larimer, Irvine, Kilmer, & Marlatt, 1997; Lo & Globetti, 1993), race (e.g., Chassin & DeLucia, 1996), and living arrangement such as residence halls or off-campus housing (e.g., Olds & Thombs, 2001). Perkins and Berkowitz s (1986) seminal study examined the influence of personal attitudes, various sociodemographics (e.g., sex, housing), and perceptions of the campus norm of student drinking. They found that students drinking behavior was significantly related to sex, housing, personal attitudes toward drinking, and the discrepancy between personal attitudes and perceptions of the campus norm. Their measure of normative perception was operationized by asking participants to choose the statement they believed best represented the general campus attitude out of five statements that ranged from very conservative ( Drinking is never a good thing to do ) to very liberal ( A frequent drunk is okay if that s what the individual wants to do ). These items measure perceptions of the attitudes of fellow college students, not student behavior. However, social norms campaigns such as those conducted at Northern Illinois University and Arizona (Frauenfelder, 2001; Gose, 1997; Haines & Spear, 1996; Johannessen et al., 1999) present messages related to student behavior (i.e., number of drinks students drink when they party, percentage of students who drink less than once a week). The nationwide College Alcohol Study reported by Perkins and Wechsler (1996) considered sociodemographics, perceived campus alcohol norm, and personal alcohol attitude in predicting alcohol abuse among college students. The perceived campus alcohol norm was measured using an index of responses to five Likert-type scale items (e.g., Students here admire non-drinkers and Drinking is an important part of the college experience, p. 965). These items are better characterized as general attitude items and do not have as direct a correspondence to the current social norms campaigns as would items querying the percentage of students who believe drinking is an important part of the college experience or items that more specifically assess one s belief about the drinking behaviors of the general campus population. Perkins and Wechsler measured personal alcohol attitude using an index, the items of which asked students to indicate their response to various situations as to How much do you think is appropriate for a college student to drink in each of the following situations (p. 964). Nine situations were provided (e.g., at a party, alone, on a week night), and response categories were no alcohol at all, only 1 to 2 drinks, enough to get high but not drunk, or enough to get drunk (p. 964). Again, these items do not correspond with the social norms messages provided in current campaigns. Fishbein and Ajzen (1975) stress the importance of scale items being both specific to the context and the situation. Such specificity increases accuracy and precision of measurement. We advocate the need for measures of multiple specific normative judgments and attitudes. For any given behavior, there are numerous normative judgments (e.g., how many days per week do undergraduates exercise, how many minutes do undergrads exercise per session, what percentage of undergrads believe exercising regularly is important). The variation in the influence of different normative judgments may account for some of the mixed outcomes in social norms studies, as they typically target only one normative judgment. As social norms campaigns are focused on the behavior of others, attitude measurements should include attitudes toward others attitude or behavior (e.g., undergraduates think drinking to get drunk is a bad idea). In addition, items measuring one s attitudes toward the behavior itself (e.g., I think drinking to get drunk is a bad idea), attitudes related to one s liking of participating in the specific behavior (e.g., I like to exercise), and attitudes related to the healthiness of the behavior (e.g., I believe that smoking is bad for health) increase context specificity of the attitudinal measurement.

3 STEPPING BACK FROM SOCIAL NORMS CAMPAIGNS 279 Theories such as the TPB provide guidance to variables believed to influence behavior. Recently, both the TRA and the TPB have been used to predict behavioral intentions and behavior in the health domain (e.g., Coleman, 2002; Greene, Hale, & Rubin, 1997). Van Ryn, Lytle, and Kirscht (1996) used the TPB and discovered that different models were predictive for the pro-health behaviors of breast self-exam and exercise. Campo et al. (2003) used the TPB to examine subjective norms versus general campus norms of alcohol consumption. Yet little has been done to assess the major influences on college students regarding various health behaviors. Social norms campaigns are being readily adopted and transferred to other health topics such as smoking and seatbelt use (Montana State University Bozeman, 2004). University administrations and the U.S. Department of Education continue to earmark extensive funds to develop such campaigns on college campuses, particularly to decrease binge drinking (Higher Education Center for Alcohol and Other Drug Prevention, 2004). However, there is a lacuna in the literature comparing influences across multiple health behaviors for college students, as well as a lack of comparison of influences related to unhealthy behaviors (e.g., smoking) and prohealth behaviors (e.g., exercise). Before more funds are spent on such controversial campaigns, we must scrutinize the effects of normative judgments as well as other factors on unhealthy and prohealth behaviors of college students. We can then ascertain relevant factors related to health behaviors, as well as learn the relative effects of sociodemographics, normative judgments, and attitudes on health behavior. Ergo, we propose these research questions. RQ1: What are the sociodemographic, normative judgment, and attitudinal influences on college students health behaviors? RQ2: What is the influence of normative judgments as compared to other influences on college students health behaviors? RQ3: How do sociodemographic, normative judgment, and attitudinal influences compare across positive and negative health behaviors? Selecting two universities, one of which had an extensive social norms campaign related to binge drinking and the other, which did not, we collected data to discover the effects of sociodemographics, normative judgments, and attitudinal factors on personal behavior related to alcohol consumption, smoking behavior, and exercise behavior. We measured those normative judgments related to student behavior, which better parallel messages advanced by social norms campaigns (e.g., Most students drink only four drinks when they party). We also included multiple measures of attitudes to assess personal behavior as well as other students behavior. METHOD Participants Participants (N = 393) were drawn from populations at two universities, one in the Southeast (n= 168) and one in the Northeast (n = 225), and were students in introductory and advanced communication courses. Communication majors represented 26% of the sample. Fifty-three percent of the participants were women and 47% were men. The age of the participants ranged from 18 to 31 (M = 20.16, SD = 1.61). African American/Blacks comprised 6.6% of the sample, 80.8% were White, 6.4% Asian/Asian American, 0.5% American Indian/Native American, 1.8% biracial, 2.6% Chicano/Hispanic, and 1.3% reported being of other racial backgrounds. Procedure Participants received either extra credit or course credit for their involvement in this study. They were administered the questionnaire in groups. Participants arrived at the classroom and were instructed that researchers wanted their responses to a series of health-related messages under consideration for use at their university. They were advised that the questionnaire would take approximately 30 min to complete and were given one of six versions of the questionnaire. Following completion of the questionnaire, participants were debriefed and were given the researchers contact information should any questions or concerns arise. Questionnaire Each student read and responded to two of the three health topics (i.e., alcohol use, smoking, exercise); we did not present all three topics to all participants to minimize fatigue. Questionnaires were randomized based on topic order (i.e., alcohol use, smoking, exercise). Six versions were developed so that each topic appeared in a questionnaire version first and second and was paired with every other topic at least once; each topic thus was given to two thirds of the participants (n = 260). The six versions were distributed in the same order; the order was repeated with every seventh student. The questionnaire included measures of the participants own behaviors related to alcohol use, smoking, and/or exercise; normative judgments of other undergraduates attitudes and behaviors; and scales assessing their attitudes toward their own health behaviors as well as their attitudes toward other undergraduates behaviors. The questionnaire concluded with sociodemographic questions. Sociodemographics. In addition to sex, age and race, participants were asked to indicate if they were a member of a fraternity or sorority (Greek affiliation) and their living arrangement. Participants were asked if they lived in a resi-

4 280 CAMERON AND CAMPO dence hall, off-campus house or apartment, program house (e.g., thematic living/learning environments), fraternity or sorority, university or cooperative housing, or other living arrangement. As some researchers have found that living in a dormitory is related to reduced drinking among college students (Baer, Stacy, & Larimer, 1991; Weitzman, 2004), for analyses this variable was recoded as dormitory or other. As the number of minorities was rather small, the race variable was recoded as White or other. Health behaviors. All participants were asked to estimate their behaviors regarding the three health topics. Alcohol behaviors were measured by a two-item scale regarding situational drinking behavior ( On average, how many alcoholic drinks do you consume when you socialize in a setting with alcohol? and Within the last 2 weeks, how many times have you had five or more alcoholic drinks in a sitting? ; α =.78), as well as a one-item measure assessing the total number of alcoholic drinks the student reported consuming per week. Smoking behaviors were measured with two items, one querying the number of cigarettes or cigars the participant smoked per month, the other asking the number of days in the last month the participant had smoked. Two items measured exercise behaviors: one asking the number of days per week the participant exercised, the other seeking an estimate of the number of minutes per session exercised. Normative judgments. For each topic, four normative judgment items were measured. Alcohol normative judgments included (a) how many alcoholic drinks do you think a typical Cornell University/University of Georgia (UGA) undergrad consumes when (s)he socializes in a setting with alcohol (number of drinks), (b) on average, how many alcoholic drinks do you think a typical Cornell/UGA undergrad consumes in a week, (c) what percentage of Cornell/UGA students think it is important to avoid drinking to get drunk, and (d) what percentage of Cornell/UGA students do you think drink less than 5 alcoholic drinks per week. Measures of tobacco normative judgments included (a) how many cigarettes/cigars do you think the typical Cornell/UGA undergrad student has used in the last 30 days, (b) how many days in the last month do you think the typical Cornell/UGA undergrad student has used tobacco products, (c) what percentage of Cornell/UGA undergraduates believe it is a good idea not to smoke, and (d) what percentage of Cornell/UGA undergrad students smoke less than three cigarettes per month. Exercise items included (a) how much do you think Cornell/UGA undergrads exercise (days per week), (b) on average, how much time do you think Cornell/UGA undergrads exercise in one session (in minutes), (c) what percentage of Cornell/UGA undergrads believe exercising regularly is important, and (4) what percentage of Cornell/UGA undergrads exercise at least twice per week. Attitude measures. All attitudes were measured on 5-point Likert-type scales from 1 (strongly disagree) to5 (strongly agree). Four types of attitudes were measured for each behavior. The degree to which someone liked to engage in excessive drinking behavior was measured using two items ( I like to get drunk and I enjoy getting drunk ; α =.97). This value was recoded for analysis, so that a higher score indicated healthier behavior, that is, not liking to get drunk. One item measured undergraduate drinking attitudes ( Cornell/UGA undergrads think that drinking to get drunk is a bad idea ). To measure attitude toward one s own drinking behavior, participants were asked to indicate on a 5-point scale the extent of their agreement to three items ( I don t have to get drunk to have a good time, I think drinking to get drunk is a bad idea, and I feel better when I do not drink ). These three items fit a singlefactor(α=.60).attitudestowardtheharmofdrinkingto get drunk were measured with two items: I believe that drinking to get drunk is harmful to good health and Drinking too much alcohol causes health problems (α =.66). Four items measured liking to smoke (e.g., Smoking is somethingienjoy ;α=.96).thisvaluewasrecodedforanalysis such that a higher score was related to a healthier behavior, that is, not liking to smoke. Four items were used to measure participants attitudes toward undergraduate smoking behaviors (e.g., Avoiding smoking is a good practice for Cornell/ UGA undergrads ; α =.88). To measure participants own attitudes toward smoking, participants were asked to indicate on a 5-point scale the extent of their agreement to three items (e.g., IbelievethatIshouldavoidsmoking ).Thesethreeitemsfita single factor (α =.89). Three items assessed the harm of smoking(e.g., Ibelievethatsmokingisbadforhealth ;α=.78). To measure participants attitudes toward liking to exercise, four items were used (e.g., I like to exercise ; α =.85). Four items were used to measure participants attitudes toward undergraduate exercising behaviors (e.g., Cornell/UGA undergrads need to exercise to maintain good health. ; α =.90). Four items assessed attitudes toward participants own exercising (e.g., I believe that I should exercise regularly ). These four items fit a single factor (α =.89). To assess attitudes toward the benefit (i.e., lack of harm) of exercising, four items fitting a single factor were used ( I believe that exercising is important to maintain good health ; α =.83). Table 1 provides the means and standard deviations for all of the measures. Data were analyzed using hierarchical regression, entering the sociodemographic variables in the first block, normative judgments in the second block, and attitudes in the final block. RESULTS We found that, although students at the Southeastern university smoked and exercised more, predictors of their drinking, smoking, and exercise behavior did not vary from those of

5 TABLE 1 Means and Standard Deviations of Measures STEPPING BACK FROM SOCIAL NORMS CAMPAIGNS 281 Measure M SD Normative judgment of how many drinks typical undergraduate consumes in a week Normative judgment of how many drinks typical undergraduate consumes in a setting Normative judgment of percentage of students who think it is important to avoid drinking to get drunk Normative judgment of percentage of students who drink less than 5 drinks a week Normative judgment of how many cigarettes/cigars typical undergraduate has used in last 30 days Normative judgment of how many days in the last month the typical undergraduate has smoked Normative judgment of percentage of undergraduates who think it is a good idea not to smoke Normative judgment of percentage of undergraduates who smoke less than 3 cigarettes a month Normative judgment of how many days per week undergraduates exercise Normative judgment of how many minutes per session undergraduates exercise Normative judgment of percentage of undergraduates who think exercising regularly is important Normative judgment of percentage of undergraduates who exercise at least twice per week Attitude toward liking to get drunk (recoded) Attitude toward undergraduates drinking behavior Attitude toward own drinking behavior Attitude toward harm of drinking to get drunk Attitude toward liking to smoke (recoded) Attitudes toward undergraduates smoking behavior Attitude toward own smoking behavior Attitude toward harm of smoking Attitude toward liking to exercise Attitude toward undergraduates exercising behavior Attitude toward own exercising behavior Attitude toward healthiness of exercising Situational drinking behavior Drinks consumed per week Number of cigarettes/cigars smoked in past month Number of days smoked in past 30 days Days per week exercised Minutes per exercise session the Northeastern university; the pattern of significant predictors was identical when separate regression analyses were run. University affiliation was not a significant predictor of behavior within the regression models. We found no topic order effects based on presentation order. Thus, the reported regression analyses were collapsed across universities and questionnaire versions. Hierarchical Regression Results Alcohol. Hierarchical regression indicated that sex, Greek affiliation, the participant s normative judgment of the amount it was believed other students drink per setting, his or her liking to drink to get drunk, and his or her attitude toward the harm of drinking to get drunk predicted both the student s binge drinking behavior and the number of drinks he or she consumed per week. Men and fraternity and sorority members were heavier binge drinkers and consumed more drinks per week. The more a participant believed that other students drank per setting and the more he or she liked to drink, the more the student binge drank or consumed per week, whereas the more the student believed that drinking to get drunk is harmful to health, the less the student binge drank or consumed per week. Tables 2 and 3 provide the beta weights, partial correlations, R 2, and R 2 change for each block. Smoking. Results indicated that sex, the participant s normative judgment of the number of days he or she believed others smoke per month, his or her attitude toward liking to smoke, his or her attitude toward undergraduate smoking behavior, and the participant s attitude toward his or her own smoking were predictive of the number of cigarettes students smoked per month. Men smoked more than women, and the more a participant believed that others smoked per month, the more he or she smoked. In addition, the more students smoked, the more they believed that other undergraduates should not smoke, but the less they felt they personally should avoid smoking. Table 4 provides the beta weights, partial correlations, R 2, and R 2 change for each block. Sex, liking to smoke, and attitude toward the participant s own smoking were significant predictors for number of days students smoked per month. Those who smoked on more days were more likely to be men, like to smoke more, and believe less strongly that they should avoid smoking. Table 5

6 282 CAMERON AND CAMPO TABLE 2 Hierarchical Regression Predicting Binge Drinking β Partial Correlation R 2 R 2 F Step 1 Sociodemographics.23.23** 19.05** Sex.24**.31 Greek.15*.19 White Dorm Step 2 Normative judgments.34.10** 15.53** How many drinks do you think a typical undergraduate consumes in a week? How many drinks do you think a typical undergraduate consumes in a setting?.22*.22 What percentage of students do you think believe it is important to avoid drinking to get drunk? What percentage of undergraduates drink less than 5 drinks per week? Step 3 Attitudes.49.15** 19.20** Liking to drink to get drunk (recoded).26**.29 Undergraduates think that drinking to get drunk is a bad idea Attitudes toward own drinking behavior Attitudes toward harm of drinking to get drunk.22**.26 Note. Betas and partial correlations are taken from the final regression equation. F is F from analysis of variance at step of entry. *p <.01. **p <.001. TABLE 3 Hierarchical Regression Predicting Drinks per Week β Partial Correlation R 2 R 2 F Step 1 Sociodemographics.24.24** 20.06** Sex.28**.32 Greek.19*.21 White Dorm Step 2 Normative judgments.31.07** 13.85** How many drinks do you think a typical undergraduate consumes in a week?.23*.21 How many drinks do you think a typical undergraduate consumes in a setting? What percentage of students do you think believe it is important to avoid drinking to get drunk? What percentage of undergraduates drink less than 5 drinks per week? Step 3 Attitudes.38.08** 12.65** Liking to drink to get drunk (recoded).17*.17 Undergraduates think that drinking to get drunk is a bad idea Attitudes toward own drinking behavior Attitudes toward harm of drinking to get drunk.17*.18 Note. Betas and partial correlations are taken from the final regression equation. F is F from analysis of variance at step of entry. *p <.01. **p <.001. provides the beta weights, partial correlations, R 2, and R 2 change for each block. Exercise. Hierarchical regression indicated that race, a participant s normative judgments related to the number of days per week and minutes per session he or she believed other students exercised, the participant s normative judgments of the percentage of students believing exercise is important, his or her attitude toward liking to exercise, and attitude toward his or her own exercise behavior were predictive of the number of days participants exercised per week. Whites were less likely to exercise than students of other races. The more days per week participants believed that other students exercised, the more days per week they exercised. However, the less time a participant believed other students exercised per session and the less he or she believed other students thought regular exercise was important, the more days per week he or she exercised. The more students liked to exercise and the more students believed that they should exercise, the more days per week they exercised. Ta-

7 STEPPING BACK FROM SOCIAL NORMS CAMPAIGNS 283 TABLE 4 Hierarchical Regression Predicting Number of Cigarettes Used in Last 30 Days β Partial Correlation R 2 R 2 F Step 1 Sociodemographics.06.06** 3.73** Sex.17**.20 Greek White Dorm Step 2 Normative judgments.10.04* 3.17** How many cigarettes/cigars do you think typical undergraduate has used in last 30 days? How many days in last month has typical undergraduate smoked?.20**.21 What percentage of undergraduates believe it is a good idea not to smoke? What percentage of undergraduates smoke less than 3 cigarettes per month? Step 3 Attitudes.33.26*** 10.90*** Liking to smoke (recoded).47***.43 Attitudes toward undergraduates smoking behavior.19*.15 Attitudes toward own smoking behavior.20*.16 Attitudes toward harm of smoking Note. Betas and partial correlations are taken from the final regression equation. F is F from analysis of variance at step of entry. *p <.05. **p <.01. ***p <.001. TABLE 5 Hierarchical Regression Predicting Number Days Smoked in the Past Month β Partial Correlation R 2 R 2 F Step 1 Sociodemographics.07.07* 4.50* Sex.14*.21 Greek White Dorm Step 2 Normative judgments * How many cigarettes/cigars do you think typical undergraduate has used in last 30 days? How many days in last month has typical undergraduate smoked? What percentage of undergraduates believe it is a good idea not to smoke? What percentage of undergraduates smoke less than 3 cigarettes per month? Step 3 Attitudes.56.45** 23.94** Liking to smoke (recoded).65**.62 Attitudes toward undergraduates smoking behavior Attitudes toward own smoking behavior.20*.18 Attitudes toward harm of smoking Note. Betas and partial correlations are taken from the final regression equation. F is F from analysis of variance at step of entry. *p <.01. **p <.001. ble 6 provides the beta weights, partial correlations, R 2, and R 2 change for each block. Sex, participants normative judgments related to minutes per session other students exercise, their normative judgments of the percentage of students believing exercise is important, their attitudes toward liking to exercise, and attitudes toward their own exercise behavior predicted the number of minutes students exercised per session. Men exercised longer than women. The more minutes per session participants believed that other students spent exercising, the longer they themselves exercised. The less participants believed other students thought exercise was important, the more they exercised. The more students liked to exercise as well as the more that students believed they should exercise, the longer they spent during each exercise session. Table 7 provides the beta weights, partial correlations, R 2, and R 2 change for each block. Relation of Normative Judgments to Behavior Normative judgments were related to participants behavior in five out of the six studied behaviors (see Tables 2 7). When observing the beta weights, it is evident that the type of normative judgment associated with each behavior dif-

8 284 CAMERON AND CAMPO TABLE 6 Hierarchical Regression Predicting Number of Days per Week of Exercise β Partial Correlation R 2 R 2 F Step 1 Sociodemographics Sex Greek White.14*.17 Dorm Step 2 Normative judgments.06.04* 1.85 How many days per week do you think undergraduates exercise?.17*.16 How many minutes/session do undergraduates exercise?.14*.16 What percentage of undergraduates do you think believe exercising regularly is important?.14*.15 What percentage of undergraduates exercise at least twice per week? Step 3 Attitudes.34.29*** 10.51*** Liking to exercise.47***.38 Attitudes toward undergraduate exercise behavior Attitudes toward own exercise behavior.26**.20 Attitudes toward healthiness of exercising Note. Betas and partial correlations are taken from the final regression equation. F is F from analysis of variance at step of entry. *p <.05. **p <.01. ***p <.001. TABLE 7 Hierarchical Regression Predicting Number of Minutes of Exercise per Session β Partial Correlation R 2 R 2 F Step 1 Sociodemographics Sex.11*.13 Greek White Dorm Step 2 Normative judgments.14.12** 4.88** How many days per week do you think undergraduates exercise? How many minutes/session do undergraduates exercise?.29**.31 What percentage of undergraduates do you think believe exercising regularly.14*.14 is important? What percentage of undergraduates exercise at least twice per week? Step 3 Attitudes.30.17** 8.74** Liking to exercise.40**.39 Attitudes toward undergraduate exercise behavior Attitudes toward own exercise behavior.17*.13 Attitudes toward healthiness of exercising Note. Betas and partial correlations are taken from the final regression equation. F is F from analysis of variance at step of entry. *p <.05. **p <.001. fers among behaviors; moreover, normative judgments are not remarkably strong predictors. In some cases, one s sex was a stronger predictor of behavior than was normative judgment. The normative judgment of behavior in a setting (e.g., drinks per setting, days smoked per month, minutes exercised per session) was the best predictor from the four normative judgment variables tested; it was predictive for four out of the six health behaviors. However, in all cases in which normative judgments were predictive of behavior, other sociodemographic or attitudinal variables were as predictive if not more so. Normative judgments as to the percentage of undergraduate students who engaged in the particular behavior (e.g., percentage of students who drink less than five alcoholic drinks per week) were never a significant predictor of behavior. Further, normative judgments as to the percentage of undergraduates who thought engaging in a particular behavior was a good idea were only predictive for the positive behavior of exercising; those who exercised more believed that other students found the behavior to be less important. Table 8 provides a

9 TABLE 8 Comparison of s STEPPING BACK FROM SOCIAL NORMS CAMPAIGNS 285 Binge Drinking Scale Drinks per Week No. of Cigarettes in Last 30 Days No. of Days Smoked in Past Month No. of Days per Week Exercise No. of Minutes per Exercise Session Sociodemographics Sex.24***.28***.17**.14**.11* Greek affiliation.15**.19** Whites compared to others.14* Dorm vs. other housing Normative judgments Drinks per week, exercise day per week, cigarettes per month.23**.17* Drinks per setting, days smoked per month, exercise time per session.22**.20**.14*.29*** Avoid drinking to get drunk, good idea not to smoke, exercise regularly.14*.16* important (health behavior) Attitudes Liking to engage in the behavior (coded such that higher score was related to.26***.17**.47***.65***.47***.40*** healthier behavior) Attitudes toward undergraduates behavior.19* Attitudes toward own behavior.20*.20**.26**.17* Attitude toward healthiness of behavior.22***.17** *p <.05. **p <.01. ***p <.001. comparative view of the significant predictors for each of the six behaviors. s Across Positive and Negative Health Behaviors The only predictor consistently significant across both positive and negative health behaviors was the participants liking to engage in that particular behavior. Sex was predictive for five of the six behaviors, both positive and negative: Men engaged in the behavior more often than women in all instances except the number of days they exercised per week (for which there was no sex difference). Greek affiliation was associated only with alcohol consumption. Living arrangement (e.g., dormitory or other housing) was not predictive for any of the behaviors under scrutiny. The effect of normative judgments varied across health behaviors. Participants attitudes toward their own behavior were predictive for all smoking and exercising behaviors but were not predictive for alcohol consumption. Attitudes toward the healthiness or harm of the behavior were predictive only for the negative behavior of excessive alcohol consumption. DISCUSSION This study took a step back from social norms campaigns to examine the relative effects of sociodemographic variables, normative judgments, and individual attitudes on personal behavior related to alcohol consumption, smoking, and exercising. We thus were able to assess predictors for both healthy and unhealthy behaviors. The data were collected at two universities, with no significant difference in predictors of behaviors found between locations. This lack of difference suggests that the results are generalizable to college students. Results indicated that different variables were predictive among behaviors, which is consistent with other research on prohealth behaviors (van Ryn et al., 1996). Normative judgments were not consistently predictive for the behaviors under scrutiny. Liking to engage in a particular behavior and participant sex were the only predictors that remained consistent across positive and negative health behaviors. Otherwise, no clear pattern of predictors, whether sociodemographic, normative judgment, or attitudinal, emerged in these data. Perkins and Wechsler (1996) note that personal alcohol attitude holds the lion s share of effect on personal alcohol abuse (p. 968) in their sample. They found the perceived campus alcohol norm to be a significant predictor independent of other sociodemographic variables, yet the majority of the effect was accounted for by personal alcohol attitude. Our results parallel theirs in all cases, across both positive and negative health behaviors, as attitudes were a more significant predictor than normative judgments. The participant s attitude toward liking to engage in the particular activity (drinking, smoking, exercising) was always a significant predictor of behavior and often was by far the most statistically significant predictor. Future research should explore the relationship between normative judgments and liking to engage in a behavior. The messages most advocated by pundits of social norms campaigns are those that provide receivers with an accu-

10 286 CAMERON AND CAMPO rate view of peers behaviors, for example, 2/3 of Rutgers students drink 3 or fewer drinks, and 1 in 5 don t drink at all (Higher Education Center for Alcohol and Other Drug Prevention, 2004). However, in no conditions, neither positive nor negative health behaviors, was the normative judgment of other students behaviors a significant predictor of own behavior. Another common formulation in social norms campaigns is the use of messages relating to the accurate perceptions of peers attitudes, for example a University of Oregon message that proclaims 87.6% of first year students report that drinking alcohol does not make them feel sexier (Higher Education Center for Alcohol and Other Drug Prevention, 2004). In our study, this variable was predictive only for exercising behaviors. Therefore, although normative judgments may have some effect on college students health behaviors, they are neither consistent nor the optimal predictors of such behaviors. Measures created for this study assessed multiple normative judgments and attitudes, providing a more extensive view of the effects of normative judgments and attitudes on behavior than past research. Our measurements of normative judgments (querying amount of engaging in behavior as well as percentage of students holding a particular attitude or engaging in a particular behavior) are in greater correspondence to current social norms messages. These current messages tend to present statements about specific student behavior, for example, Most of us at UW-Whitewater have 5 or fewer drinks per week (Higher Education Center for Alcohol and Other Drug Prevention, 2004). Our measures elicited participant judgments of such behaviors among peers, whereas past measurement failed to do so. Yet even with these more specific normative judgment measures, the data indicate that normative measures are not consistent predictors of student behavior as it relates to alcohol consumption, smoking, or exercising. Limitations The majority of attitudinal scales had reliabilities greater than.80. The lowest scale reliabilities were related to alcohol attitudes. The measure of participants attitudes toward undergraduate alcohol use was a one-item measure; future studies may seek to enhance this measure as well as increase the reliability of the measurements of alcohol attitudes. Based on this study, we do not know if there is an interaction between normative judgments and liking to engage in a behavior on likelihood of engaging in a behavior. Although results pointed to significant effects on individual health behavior, in all cases there remains a large portion of variance unaccounted for. Future studies need to ascertain other significant predictors of positive and negative health behaviors of college students and determine if interaction effects exist. Implications This research extended past studies of influences on college student behavior by considering positive and negative behaviors and furthered our understanding of behavioral influences by including sociodemographic variables and multiple measures of normative judgments and personal attitudes. As different variables were found to be predictive for the varying health behaviors, these results bring into question the effectiveness of social norms campaigns, not only for alcohol consumption but for other behaviors such as smoking, seatbelt use, and so forth. The results indicate that one s liking to engage in a particular behavior is consistently a significant predictor of both addictive and nonaddictive behaviors. Smokers in this study conceivably recognized the harmful effects of smoking, as the more they smoked the more they believed others should not smoke. This result may seem counterintuitive; however, it is well known that smoking cigarettes is hazardous to one s health. Although smokers may be unwilling to give up their own habit, they may recognize the dangers and addictive properties of smoking and believe that others should avoid use. The findings related to exercise (the more a participant exercised, the fewer minutes he or she thought other students exercised and the less he or she thought other students thought exercise is important) may be a result of participants regularly being in an exercise facility and seeing others exercising for a short time. For behavioral intervention campaigns to meet with success, it appears we must learn how best to influence individuals predilection for behaviors, both healthy behaviors we wish to encourage (e.g., exercise) and unhealthy behaviors we wish to diminish (e.g., excessive alcohol use). We caution against practitioners indiscriminately transferring campaigns from one topic (e.g., from alcohol to tobacco) or type of message (attitudinal or behavioral) to another without proper evaluation. Using untested campaigns, even with the best intentions, can have positive, no, or negative effects on audience behaviors (Fishbein, Hall-Jamieson, Zimmer, von Haeften, & Nabi, 2002). CONCLUSION The premise of social norms campaigns is that if you alter your normative judgments of others behavior, then you will alter your own attitudes and behavior. Although some normative judgments were statistically significant predictors of individual behavior, no one judgment (of the four types measured) was a consistent predictor. Rather, when normative judgments were significant, as in the case of binge drinking behavior, it was the judgment most closely related to the specific behavior (e.g., drinks per setting, drinks per week) that attained significance. Ergo, health campaigns, whether social norms or other persuasive campaigns, must be targeted at rel-

11 STEPPING BACK FROM SOCIAL NORMS CAMPAIGNS 287 evant attitudes and normative judgments. Simply using a message derived from social norms marketing is unlikely to consistently affect college student behavior, as the relevant attitudes and normative judgments change both within and across positive and negative health behaviors. ACKNOWLEDGMENTS The authors contributed equally to this article. This research was supported in part by the Cornell University Agricultural Experiment Station federal formula funds Project NYC , received from Cooperative State Research, Education and Extension Service, U.S. Department of Agriculture. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the view of the U.S. Department of Agriculture. We are grateful to Somjen Frazer, Dominique Brossard, Esther Baker, Baseema Banoo, Deanna Caputo, Sarah Doherty, Erin Payne, and Tommy Wood for their assistance. We also wish to thank Teri Thompson and two anonymous reviewers for their comments. A version of this manuscript was presented at the annual American Public Health Association Conference, San Francisco, November REFERENCES Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In J. Kuhl & J. Beckmann (Eds.), Action control: From cognition to behavior (pp ). New York: Springer-Verlag. Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice Hall. Allport, F.H. (1924). Social psychology. Boston: Houghton Mifflin. Baer, J. S., Stacy, A., & Larimer, M. (1991). Biases in the perception of drinking norms among college students. Journal of Studies on Alcohol, 52, Campo, S., Brossard, D., Frazer, M. S., Marchell, T., Lewis, D., & Talbot, J. (2003). Are social norms campaigns really magic bullets? Assessing the effects of students misperceptions on drinking behavior. Health Communication, 15, Campo, S., & Cameron, K. A. (2006). Differential effects of exposure to social norms campaigns: A cause for concern. Health Communication, 19, Campo, S., Cameron, K. A., Brossard, D., & Frazer, M. S. (2004). Social norms and expectancy violation theories: Assessing the effectiveness of health communication campaigns. Communication Monographs, 71, Chassin, L., & DeLucia, C. (1996). Drinking during adolescence. Alcohol Health and Research World, 20, Coleman, C. L. (2002). Examining influences of pharmacists communication with consumers about antibiotics. Health Communication, 15, Fields, J. M., & Schuman, H. (1976). Public beliefs about the beliefs of the public. Public Opinion Quarterly, 40, Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley. Fishbein, M., Hall-Jamieson, K., Zimmer, E., von Haeften, I., & Nabi, R., (2002). Avoiding the boomerang: Testing the relative effectiveness of antidrug public service announcements before a national campaign. American Journal of Public Health, 92, Frauenfelder, M. (2001, December 9). Social-norms marketing. New York Times Magazine, 100. Gomberg, L., Schneider, S. K., & DeJong, W. (2000). Evaluation of a social norms marketing campaign to reduce high-risk drinking at the University of Mississippi. American Journal of Drug and Alcohol Abuse, 27, Gose, B. (1997, October 24). Colleges try to curb excessive drinking by saying moderation is okay. Chronicle of Higher Education, 44 (9), A61 A62. Greene, K., Hale, J. L., & Rubin, D. L. (1997). A test of the theory of reasoned action in the context of condom use and AIDS. Communication Reports, 10, Haines, M., & Spear, S. F. (1996). Changing the perception of the norm: A strategy to decrease binge drinking among college students. Journal of American College Health, 45, Higher Education Center for Alcohol and Other Drug Prevention. (2004). Social norms and social marketing. Retrieved May 4, 2004, from Johannessen, K., Collins, C., Mills-Novoa, B., & Glider, P. (1999). A practical guide to alcohol abuse prevention: A campus case study in implementing social norms and environmental management approaches. Tucson: Campus Health Services, The University of Arizona. Keeling, R. P. (1999). Proceed with caution: Understanding and changing norms. Journal of American College Health, 47, Keeling, R. P. (2000). Social norms research in college health. American Journal of College Health, 49, Kim, M. S., & Hunter, J. E. (1993). Relationships among attitudes, behavioral intentions, and behavior: A meta-analytic view of past research, Part 2. Communication Research, 20, Larimer, M. E., Irvine, D. L., Kilmer, J. R., & Marlatt, G. A. (1997). College drinking and the Greek system: Examining the role of perceived norms for high-riskbehavior. JournalofCollegeStudentDevelopment, 38, Lo, C. C., & Globetti, G. (1993). A partial analysis of the campus influence on drinking behavior: Students who enter college as non-drinkers. The Journal of Drug Issues, 23, Montana State University-Bozeman, (2004). Most of us campaign. Retrieved May 4, 2004, from Olds, R. S., & Thombs, D. L. (2001). The relationship of adolescent perceptions of peer norms and parent involvement to cigarette and alcohol use. Journal of School Health, 71, Perkins, H. W., & Berkowitz, A. D. (1986). Perceiving the community norms of alcohol use among students: Some research implications for campus alcohol education programming. International Journal of the Addictions, 21, Perkins, H. W., Meilman, P. W., Leichlier, J. S., Cashin, J. R., & Presley, C. A. (1999). Misperceptions of the norms for the frequency of alcohol and other drug use on college campuses. Journal of American College Health, 47, Perkins, H. W., & Wechsler, H. (1996). Variation in perceived college drinking norms and its impact on alcohol abuse: A nationwide study. Journal of Drug Issues, 26, Prentice, D. A., & Miller, D. T. (1993). Pluralistic ignorance and alcohol use on campus: Some consequences of misperceiving the social norm. Journal of Personality and Social Psychology, 64, Rimal, R. N., & Real, K. (2003). Understanding the influence of perceived norms on behavior. Communication Theory, 13, Stewart, L. P., Lederman, L. C., Golubow, M., Cattafesta, J. L., Goodhart, F. W., Powell, R.L., et al. (2002). Applying communication theories to prevent dangerous drinking among college students: The RU SURE campaign. Communication Studies, 53, van Ryn, M., Lytle, L. A., & Kirscht, J. P. (1996). A test of the theory of planned behavior for two health-related practices. Journal of Applied Social Psychology, 26,