AM. J.

DRUG ALCOHOL ABUSE, 27(2), 375389 (2001)

EVALUATION OF A SOCIAL NORMS MARKETING CAMPAIGN TO REDUCE HIGH-RISK DRINKING AT THE UNIVERSITY OF MISSISSIPPI*
Laura Gomberg,1 Shari Kessel Schneider,1 and William DeJong 2,
1

Education Development Center, Inc., Newton, Massachusetts 2 Harvard School of Public Health, Cambridge, Massachusetts

ABSTRACT A social marketing campaign to change perceptions of peer drinking norms was conducted by the National Golden Key Honor Society at the University of Mississippi during the 19951996 school year. To assess the campaigns impact on perceptions of student drinking norms and alcohol consumption, Golden Keys national ofce administered a survey three times during the school year to all students enrolled in a random sample of required freshmen English courses. Regression analyses suggest that exposure to the

* The views expressed in this article are those of the authors and do not necessarily reect the ofcial position of the Department of Education. Reprints: Dr. William DeJong, Department of Health and Social Behavior, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115. E-mail: dejong@ hsph.harvard.edu 375
Copyright 2001 by Marcel Dekker, Inc. www.dekker.com

376

GOMBERG, SCHNEIDER, AND DEJONG

marketing campaign may be associated with lower (and more accurate) estimates of student drinking norms. While offering promising results, this study was limited due to shortcomings in the research design. Future evaluations of social norms marketing campaigns should adhere to basic evaluation principles, such as using comparison groups, collecting contextual data, using a valid and reliable survey instrument, and ensuring proper survey administration techniques. Key Words: Alcohol; Evaluation; Prevention; Social marketing; Social norms.

INTRODUCTION Heavy episodic drinking on college and university campuses is a widespread phenomenon (1). High-risk drinking creates a host of problems for both the students who engage in such behaviors and their peers, including academic difculties, injury, property damage, unplanned and unprotected sexual activity, or trouble with campus police (2). Finding effective ways of addressing the problem of high-risk drinking has been and remains a prominent issue for many institutions of higher education. Traditional responses to the problem of high-risk drinking have included student education and counseling (3). While helping to increase individual knowledge and awareness, these approaches have had little apparent effect on students alcohol consumption. Recognizing the limitations of past efforts, new approaches to reducing high-risk drinking have begun to focus on changing the physical, social, economic, and legal environment that drives alcohol use (4). A major factor inuencing students decisions about alcohol consumption is their perceptions of campus drinking norms (5). Typically, college students greatly overestimate the number of their peers who engage in high-risk drinking, both nationally and on their own campus. The idea that many other students drink excessively may cause students to feel both justied and pressured to consume more alcohol than they would if they believed instead that their peers drank more moderately. Thus, shifting these perceptions of social norms to reect drinking patterns more realistically could be an effective means of reducing high-risk drinking on college campuses (6). Working from this idea, various prevention programs have sought to use social marketing strategies to change students perceptions of campus drinking norms. Social marketing can be dened as the application of basic marketing principles to the design and implementation of programs and information campaigns that advance social causes such as alcohol and other drug prevention (7).

SOCIAL NORMS CAMPAIGN TO REDUCE DRINKING

377

The rst such campaign was conducted at Northern Illinois University (NIU) (Dekalb, IL). Students were informed through a series of campus newspaper advertisements and on-campus skits about the true rate of high-risk alcohol consumption among their peers, which was far less than what students typically estimated. Survey data showed that, after the campaigns rst year, students offered lower (and more accurate) estimates of the high-risk drinking rate among NIU students. Students also reported a sizable decline in high-risk drinking, even while national rates continued relatively unchanged (8). These promising results have inspired social norms marketing campaigns at several other colleges and universities in the United States. It should be noted, however, that the evaluation of the NIU campaign has several shortcomings. First, the surveys were administered to convenience samples of students. Second, there are no data from carefully selected comparison schools. The comparison against national data is informative, but not denitive. Third, process data were not collected to monitor campus-, community-, or state-level contextual factors that might have affected the measured outcomes. Based on the NIU campaign, the Golden Key National Honor Society developed a social norms marketing campaign, Just the Facts (JTF), which was pilot tested on eight university campuses across the United States, including the University of Mississippi (Ole Miss), during the 19951996 school year (9). (Other pilot test sites included Colorado State University, Fort Collins; Eastern Michigan University, Ypsilanti; Kennesaw State University, Marietta, GA; State University of New York at Geneseo; Tufts University, Boston, MA; University of California at Los Angeles; and University of Wisconsin at La Crosse.) This article reports an in-depth analysis of the results of the Ole Miss JTF campaign. This particular campaign was chosen due to the exhaustive sampling methodology used for the student surveys (see Methods section below) that was not feasible at the other seven campuses.

CAMPAIGN OVERVIEW The Ole Miss chapter of the Golden Key National Honor Society implemented the Just the Facts campaign during the 19951996 school year following guidelines developed by Golden Keys national ofce (9). To begin, the University of Mississippi conducted the Core Alcohol and Drug Survey in April 1994 to assess baseline levels of student drinking. These data were then used to create specic messages about actual student drinking norms. The campaign was implemented twice during the 19951996 school year, once in the fall semester and again in the spring semester, with each phase lasting 8 consecutive weeks. The campaign messages did not address freshmen explicitly, but freshmen were intended to be the primary target audience.

378

GOMBERG, SCHNEIDER, AND DEJONG

The campaign relied primarily on advertisements in the campus newspaper, The Daily Mississippian. Focus groups had veried that students saw the paper as a credible and reliable source of information. Surveys conducted for the evaluation (see below) conrmed that the paper is widely read. In the pretest, 61.1% of the surveyed freshmen reported reading the paper every time or most of the time, with these gures rising to 70.0% and then 74.1% in the two posttests conducted later in the school year. Very few students (4.6% at pretest, 1.2% at posttest 1, and 0.9% at posttest 2) reported never reading the school newspaper. To reach students before the weekend, a time of especially heavy alcohol consumption, the JTF advertisements appeared in each Thursdays issue of the newspaper during both phases of the campaign. The advertisements presented several statistical facts about the University of Mississippi, including some related to current levels of student alcohol use. For example, one advertisement used the following fact: About 8000 Ole Miss students have not used alcohol 3 or more times this week. Another advertisement stated, 60% of the students at Ole Miss use alcohol once a week or less. Student members of the Golden Key chapter posted yers that announced the Just the Facts campaign and encouraged students to look for the newspaper advertisements. In addition, the campus radio station was provided with live copy public service announcements (PSAs) to read on the air. During the campaign, Golden Key also ran a contest in which students who accurately reported statements of fact from the campaign were entered in a drawing for cash prizes. The contest feature was added to increase awareness of the campaign and to encourage attention to the messages conveyed in the newspaper advertisements.

METHODS During the 19951996 school year, Golden Keys national ofce administered a written survey three times to all students enrolled in randomly selected required freshman English courses. The pretest was administered in September 1995. The rst posttest, using the same set of selected class sections, was in November 1995, just after the end of the rst 8-week campaign phase. The second posttest, using a new set of randomly selected class sections (due to the change in semester), was in April 1996, just after the end of the second 8-week campaign phase. While the English classes were required for freshmen, several nonfreshmen also attended those classes and completed the survey. All nonfreshmen and students with missing data about class year were eliminated from the data analysis. The numbers of freshman students completing the surveys were 785 for the pre-

SOCIAL NORMS CAMPAIGN TO REDUCE DRINKING

379

test, 698 for the rst posttest, and 583 for the second posttest. Response rates are unknown. The self-administered survey consisted of 42 questions to assess demographics, alcohol use, various attitudes toward alcohol and its associated effects, and exposure to the campaign and other alcohol advertising. Exposure to the JTF campaign was measured by whether students recognized the JTF logo (yes versus no/not sure). Three survey questions measured student alcohol use. Students were rst asked, Do you drink alcohol? If they answered yes, they were asked two additional questions: If yes, how many drinks do you normally have per week? and How many days do you normally drink per week? For each of the last two questions, a blank space was provided for students to write in the appropriate number. The survey did not measure high-risk drinking specically. Instead, for each student, the number of drinks consumed per week was divided by the number of days per week on which alcohol was consumed, yielding the mean number of drinks per day that each student consumed when drinking alcohol. Females who drank a mean of 4 or more drinks per day and males who drank a mean of 5 or more drinks per day were classied as high-risk drinkers. Of the questions, 11 concerned specic facts about alcohol use at the University of Mississippi that were featured in the Just the Facts campaign. Of these, 5 questions were selected for this analysis as appropriate indicators of perceived drinking norms. Other questions were not analyzed because of ambiguous wording or lack of direct relevance to student alcohol use. Students were asked to answer true, false, or not sure to the following statements, all of which are true: 1. Students with an A grade average drink less than 3 drinks per week. 2. Over half of college students drink less than once per week. 3. More than 75% of students have not used alcohol 3 or more times this week. 4. Almost 2/3 (67%) of college students report that they have never missed class due to alcohol use. 5. Over half of college students do not binge drink. The last item was the primary indicator of perceived drinking norms at Ole Miss, with responses categorized as true versus false/not sure. However, the survey instructions are unclear whether students were being asked specically about norms at Ole Miss students or about norms of college students in general. For continuous variables, two-sample independent t tests were used to compare means between the pretest and each of the two posttest surveys. Chi-square analyses were used to compare categorical responses across surveys. Linear regression was used to determine predictors of drinks per week, while logistic re-

380

GOMBERG, SCHNEIDER, AND DEJONG

gression was used to determine predictors of high-risk drinking behavior and perceptions of student drinking norms.

RESULTS Demographics Table 1 presents demographic data for the survey respondents for each of the three surveys. Men made up less than half of the survey participants, with a range from 47.6% for the pretest to 43.1% for the second posttest; a comparison 2.84, df 1, p of these two extremes was not statistically signicant ( 2 .05). For each survey, about nine-tenths of the surveyed students were white. The mean age of students increased slightly from survey to survey, as would be expected with the passage of time.

Alcohol Use Table 2 reports four measures of alcohol use. The percentage of students who reported drinking any alcohol increased slightly across the three surveys, from 79.9% at the pretest to 82.5% at the second posttest, although this difference was not statistically signicant (2 1.39, df 1, p .05). More males than females reported alcohol use at the pretest (83.9% versus 76.4%; 2 6.80, df 1, p .01) and rst posttest (85.4% versus 78.8%; 2 4.96, df 1, p .05), but with the second posttest, slightly more females reported

Table 1.

Demographics of Participants by Survey Administration

Pretest (n 785) Posttest 1 (n 697)
46.3 53.7 90.4 9.6 18.35

Posttest 2 (n 583)
43.1 56.9 90.5 9.5 18.72

Male (%) Female (%) White (%) Nonwhite a (%) Mean age (years)

47.6 52.4 88.6 11.4 18.21

Note: The JTF Survey was administered in September 1994 (pretest), November 1994 (Posttest 1), and April 1995 (Posttest 2). a Non-white refers to any participants who did not indicate that they are white. Other response categories included African-American, Latino-Chicano, Asian, and Other, Specify.

Table 2. Pretest Male 83.9 21.81 3.25 65.6 2.66 40.5 2.96 52.9 3.12 61.0 2.53 40.6 9.90 15.80 20.11 9.48 76.4 79.9 85.4 78.8 81.8 14.69 2.82 50.6 Female Total Male Female Total Male 81.6 17.61 2.89 58.4 Posttest 1

Alcohol Use by Gender by Survey Administration Posttest 2 Female 83.1 9.00 2.47 34.7 Total 82.5 12.61 2.65 44.7

Students who drink alcohol (%) Mean number of drinks students normally have per week Mean number of days on which students normally drink per week Students who are high-risk drinkers a (%)

SOCIAL NORMS CAMPAIGN TO REDUCE DRINKING

Note: The JTF Survey was administered in September 1994 (pretest), November 1994 (Posttest 1), and April 1995 (Posttest 2). a For men, high-risk drinking is dened as having 5 or more drinks in a row; for women, it is having 4 or more drinks in a row.

381

382

GOMBERG, SCHNEIDER, AND DEJONG

alcohol use than males (83.1% versus 81.6%; 2 0.23, df 1, p .05). For males, there was a slight, but nonsignicant, drop in the percentage of students who reported drinking alcohol from the pretest to the second posttest (83.9% versus 81.6%; 2 0.55, df 1, p .05). For females, there was over a 6 percentage point increase in reported alcohol use from the pretest to the second posttest (76.4% versus 83.1%; 2 5.09, df 1, p .05). The mean number of drinks students consumed per week decreased signicantly from the September pretest to the second posttest in April (15.80 drinks versus 12.61 drinks; t 3.74, df 942, p .001), as did the mean number of days on which students drink per week (2.96 days versus 2.65 days; t 3.70, df 1039, p .001). On each survey, men reported drinking approximately twice as many drinks per week as women. At the pretest, men reported drinking 21.81 drinks per week and women reported drinking 9.90 drinks per week (t 10.62, df 515, p .001). At the rst posttest, men reported drinking 20.11 drinks per week, and women reported drinking 9.48 drinks per week (t 8.64, df 492, p .001). Finally, at the second posttest, men reported drinking 17.61 drinks per week, and women reported drinking 9.00 drinks per week (t 8.08, df 425, p .001). The mean number of drinks consumed per week by male students decreased by 4.20 drinks from the pretest to the second posttest (21.81 drinks versus 17.61 drinks; t 2.80, df 433, p .01). The mean number of drinks consumed by women decreased over this period, but only by 0.90 drinks (9.90 drinks versus 9.00 drinks; t 1.32, df 507, p .05). Likewise, the mean number of days per week on which students drank decreased from September to April, by 0.36 days for men (3.25 days versus 2.89 days; t 2.71, df 493, p .01) and 0.19 days for women (2.66 days versus 2.47 days; t 1.87, df 544, p .05). At the pretest, just over half of students (52.9%) were categorized as highrisk drinkers, but by the second posttest, fewer than half (44.7%) were so categorized ( 2 6.31, df 1, p .05). As shown in Table 2, rate decreases were seen for both men (7.2%; 2 2.31, df 1, p .05) and women (5.8%; 2 1.78, df 1, p .05). At all three survey administrations, the proportion of men who could be categorized as high-risk drinkers was signicantly higher than the proportion of women. At the pretest, 65.6% of men and 40.5% of women reported high-risk drinking ( 2 32.36, df 1, p .001), at the rst posttest, 61.0% of men and 40.6% of women reported high-risk drinking ( 2 20.46, df 1, p .001), and at the second posttest, 58.4% of men and 34.7% of women reported high-risk drinking ( 2 23.49, df 1, p .001). White students consistently drank more than non-white students across all survey administrations and with respect to all measures of drinking. For example, at the pretest, 85.0% of white students and 39.8% of non-white students reporting drinking alcohol ( 2 99.41, df 1, p .001), and at the second posttest, 86.4% of whites and 47.3% of non-whites reported drinking alcohol ( 2 107.09, df

SOCIAL NORMS CAMPAIGN TO REDUCE DRINKING

383

1, p .001). White students also reported more high-risk drinking across all tests, with 54.7% of whites and 13.6% of non-whites reporting high-risk drinking at the pretest ( 2 14.27, df 1, p .001), and 45.9% of whites and 19.0% of non-whites reporting high-risk drinking at the second posttest ( 2 5.81, df 1, p .05).

Perceived Drinking Norms Table 3 shows the percentage of students answering true to each of ve factual statements about current drinking norms at the University of Mississippi. Comparing the pretest with the second posttest, there were signicant increases in the percentage of students who correctly answered questions about these norms, all of which were the subject of JTF newspaper advertisements. Of particular interest is perception of the high-risk drinking norm, indicated by students answering true to the statement, Over half of college students do not binge drink. For all students, accurate reporting of this norm increased over time, with 23.5% of students at the pretest and 31.6% of students at the second posttest

Table 3.

Correct Perceptions of Drinking by Survey Administration Pretest (%) a Posttest 1 (%) a 28.2 Posttest 2 (%) a 31.6 b (df 2)

Over half of students do not binge drink. Almost 2/3 (67%) of college students report that they have never missed a class due to alcohol use. More than 75% of students have not used alcohol three or more times this week. Over half of college students drink less than once per week. Students with an A grade average drink less than 3 drinks per week.

23.5

11.29 c

22.1

24.4

27.3

6.86 c

20.3 6.2 23.9

30.2 14.4 35.7

35.2 15.1 43.4

38.97 d 30.31 c 59.47 c

Note: The JTF Survey as administered in September 1994 (pretest), November 1994 (posttest 1), and April 1995 (posttest 2). a Indicates proportion of students correctly responding that the statement is true. b Chi-square test comparing pretest and posttest 2. c p .01. d p .05.

384

GOMBERG, SCHNEIDER, AND DEJONG

who believed correctly that high-risk drinking is not the norm ( 2 11.29, df 2, p .01). For all three surveys, a greater proportion of males than females correctly perceived the high-risk drinking norm, but the difference by gender was signicant only for the rst posttest (32.2% of males versus 25.2% of females; 2 6.11, df 2, p .05). For men, the percentage of students who correctly believed that high-risk drinking is not the norm increased from 27.0% at the pretest to 32.7% at the second posttest, although this increase was not signicant ( 2 2.43, df 2, p .05). For women, it increased signicantly, from 21.0% at the pretest to 30.8% at the second posttest ( 2 9.43, df 2, p .01). This shift in perceptions occurred concurrently with the decrease in alcohol consumption among both men and women (see Fig. 1).

Exposure to the Just the Facts Campaign Logo recognition was used to assess exposure to the JTF campaign. Reported recognition of the JTF logo increased signicantly, from 6.2% at the pretest to 55.4% at the rst posttest and 78.5% at the second posttest (comparing pretest to posttest 2, 2 782.73, df 2, p .001). Neither gender nor highrisk drinking status was associated with recognition of the JTF logo at either the

Figure 1. Drinking behaviors and correct perception of drinking norms. Note: The JTF Survey was administered in September 1994 (pretest), November 1994 (posttest), and April 1995 (posttest 2). *For men, high-risk drinking is dened as having 5 or more drinks in a row; for women, it is dened as having 4 or more drinks in a row. **Measured by the percentage of students who correctly responded that the statement, Over half of college students do not binge drink, is true.

SOCIAL NORMS CAMPAIGN TO REDUCE DRINKING

385

pretest or rst posttest. At the second posttest, however, signicantly fewer highrisk drinkers (72.6%) recognized the logo compared to low-risk drinkers (82.3%) ( 2 6.37, df 2, p .05).

Regression Analyses For each of the three survey administrations, regression analyses were conducted to determine whether race, gender, and self-reported recognition of the JTF logo were signicant predictors of either alcohol consumption or accurate perceptions of the high-risk drinking rate. A linear regression model (see Table 4) was estimated to determine what factors predict the number of drinks per week consumed by students (Model 1). For all three surveys, being male was a signicant predictor of consuming a higher number of drinks per week. Being white was a signicant predictor of higher consumption for both the pretest and rst posttest, although not for the

Table 4. Three Regression Models for Drinking Behaviors and Norms Model 1 (Number of Drinks per Week) Slope Coefcient Male Pretest Posttest 1 Posttest 2 White Pretest Posttest 1 Posttest 2 Logo recognition Pretest Posttest 1 Posttest 2
a

Model 2 (High-Risk Drinking a Behavior) Odds Ratio 2.85 c 2.26 c 2.82 c 7.65 e 8.34 e 3.53 d 1.43 1.12 0.67

Model 3 (Perceived Drinking Norm b) Odds Ratio 1.41 d 1.40 d 1.11 1.34 0.45 d 0.72 1.15 1.42 d 1.11

11.79 c 10.25 c 8.75 c 8.02 e 9.56 e 4.27 2.71 1.96 1.48

For men, high-risk drinking is dened as having 5 or more drinks in a row; for women, it is having 4 or more drinks in a row. b Measured by the percentage of students who correctly responded that the statement, Over half of college students did not binge drink, is true. c p .001. d p .05. e p .01.

386

GOMBERG, SCHNEIDER, AND DEJONG

second posttest. Logo recognition was not a signicant predictor for any of the three surveys. Logistic regression techniques were used to determine which factors were signicant predictors of being classied as a high-risk drinker, dened for females as drinking a mean of 4 or more drinks per day and for males as drinking a mean of 5 or more drinks per day (Model 2). For all three surveys, being male and white were signicant predictors of being classied as a high-risk drinker. Again, logo recognition was not a signicant predictor for any of the three surveys. Finally, a logistic regression model was also used to determine which factors were signicant predictors of correctly perceiving the high-risk drinking norm, as indicated by a true response to the statement, Over half of college students do not binge drink (Model 3). Being female was a signicant predictor at the pretest and rst posttest, but not at the second posttest. At the rst posttest, being non-white was a signicant predictor of correctly perceiving the norm against high-risk drinking. Controlling for gender and race, students who reported seeing the JTF logo at the rst posttest were signicantly more likely to report correctly the norm against high-risk drinking (odds ratio [OR] 1.42, df 3, p .05). This nding did not hold for the second posttest, however, perhaps because there was less variation in logo recognition by that time. As noted above, at the rst posttest, 55.4% of students recognized the logo, compared to 78.5% at the second posttest.

DISCUSSION The ndings reported here suggest that the University of Mississippis Just the Facts campaign helped correct students perceptions of high-risk drinking norms and may also have been associated with decreases in alcohol consumption among freshman students. Recognition of the JTF logo (an indicator of exposure to the campaign) was a signicant predictor of accurately perceiving norms that do not support high-risk drinking, even when controlling statistically for gender and race. The association between apparent exposure to JTF and changes in perceived social norms supports the theory behind the programnamely, that a social norms marketing campaign can lead to changes in perceived norms about drinking. In time, changes in perceived norms are expected to lead to decreases in alcohol consumption. It should be noted, however, that while decreases in alcohol consumption were observed over the course of the school year, exposure to JTF was not a signicant predictor of alcohol use for either of the two posttests. Therefore, we may not conclude that the campaign accounts for this observed decrease in alcohol use, which may reect general trends among Ole Miss freshmen during the school year.

SOCIAL NORMS CAMPAIGN TO REDUCE DRINKING

387

Limitations of the Study An advantage of this study, compared to others reported in the literature, is the inclusion of regression analyses to control for key variables, such as race and gender, that could inuence observed outcomes. In addition, the regression analyses included a measure of exposure to the campaign as a potential predictor of both perceived norms and drinking behavior. Even so, the results of this evaluation should be interpreted with caution due to several methodological shortcomings. First, the data were gathered from two different groups of students on the basis of random classroom selection. This means that the two groups might differ in some systematic way related to these outcomes, which could lead to a mistaken estimate of the JTF programs true impact. Moreover, the sample sizes varied across the two groups of students. With the rst group, there were 785 students in the pretest sample and 698 in the rst posttest sample. With the second group, there were 583 students in the second posttest sample. The greatly reduced sample size for the second posttest means that its results are somewhat less reliable. As noted previously, response rates for all three surveys are unknown, but if the reduced sample size for the second posttest was due to a lower response rate, its results would also be less valid. Second, although the pretest and rst posttest were administered to the same students, the data collection procedure did not allow for matching surveys from the same individuals. As a result, it is impossible to assess changes in perceived norms and drinking behavior at the individual level. For future evaluations, it should be noted that survey respondents can be asked to generate unique data codes that allow repeated surveys to be linked, while still preserving their anonymity. Third, exposure to the campaign was measured by newspaper readership and JTF logo recognition, but students were not asked directly if they recalled seeing any JTF advertisements in the newspaper, the campaigns primary means of media communication, or if they could recall any of the campaigns key messages. Fourth, alcohol use was measured by the number of drinks consumed per week and the number of days on which students normally drank. As a result, the dichotomous high-risk drinking variable is an imprecise measure of such drinking, which is normally assessed as the number of times students had a certain number of drinks on a single occasion. Fifth, there are no comparison data from matched universities. Without data from comparison schools that did not implement the campaign, it is impossible to prove that the observed effects were due to the JTF campaign and not to some other external factors that may have similarly affected comparable schools.

388

GOMBERG, SCHNEIDER, AND DEJONG

Finally, no contextual data were available to monitor other programs, policies, or critical events that could affect the high-risk drinking rate at Ole Miss. For any evaluation of this sort, both newspaper content analysis and interviews with key informants are needed to track changes at the campus, community, state, and national levels.

CONCLUSION Results of the Just the Facts (JTF ) social norms marketing campaign conducted at the University of Mississippi are promising, but inconclusive. Among the surveyed freshmen, apparent exposure to the campaign predicted more accurate perceptions of student drinking norms. The JTF campaign also coincided with a decrease in freshmens self-reported alcohol consumption, although this effect could not be attributed singly to the campaign. More rigorous evaluations are necessary to conclude that this approach is effective and warrants replication. Designing a rigorous, valid, and generalizable evaluation study may be costly and time consuming, but in the long run, condently assessing the impact of a program with a well-designed evaluation may be more cost-effective than continuing to implement a program with questionable impact.

ACKNOWLEDGMENT The authors thank Stephen C. Pracht, Director of Alumni and Development for the Golden Key National Honor Society, and Dr. R. Daniel Duquette, Chair, Health Education Department, University of Wisconsin at La Crosse, for providing the data used in this analysis and additional information about the University of Mississippis social norms marketing campaign. Preparation of this article was supported by U.S. Department of Education contract SS95013001 to EDC.

REFERENCES 1. Wechsler, H.; Dowdall, G. W.; Maenner, G.; Gledhill-Hoyt, J.; Lee, H. Changes in binge drinking and related problems among American college students between 1993 and 1997: results of the Harvard School of Public Health College Alcohol Study. J. Am. College Health 1998, 47, 5768. 2. Wechsler, H.; Austin, B.; DeJong, W. Secondary Effects of Binge Drinking

SOCIAL NORMS CAMPAIGN TO REDUCE DRINKING

389

3.

4.

5.

6.

7.

8.

9.

on College Campuses; U.S. Department of Education, Higher Education Center for Alcohol and Other Drug Prevention: Washington, DC, 1996. Perkins, H. W. College and student misperceptions of alcohol and other drug norms among peers: exploring causes, consequences, and implications for prevention programs. In Designing Alcohol and Other Drug Prevention Programs in Higher Education; U.S. Department of Education, Higher Education Center for Alcohol and Other Drug Prevention: Washington, DC, 1997; 177206. DeJong, W.; Vince-Whitman, C.; Colthurst, T.; Cretella, M.; Gilbreath, M.; Rosati, M.; Zweig, K. Environmental Management: A Comprehensive Strategy for Reducing Alcohol and Other Drug Use on College Campuses; U.S. Department of Education, Higher Education Center for Alcohol and Other Drug Prevention: Washington, DC, 1998. Perkins, H. W.; Wechsler, H. Variation in perceived college drinking norms and its impact on alcohol abuse: a nationwide study. J. Drug Issues 1996, 26, 961974. Perkins, H. W.; Berkowitz, A. D. Perceiving the community norms of alcohol use among students: some research implications for campus alcohol education programming. Int. J. Addictions 1986, 21, 961976. Zimmerman, R. Social Marketing Strategies for Campus Prevention of Alcohol and Other Drug Problems; U.S. Department of Education, Higher Education Center for Alcohol and Other Drug Prevention: Washington, DC, 1997. Haines, M.; Spear, A. F. Changing the perception of the norm: a strategy to decrease binge drinking among college students. J. Am. College Health 1996, 45, 134140. Golden Key National Honor Society. Just the Facts Implementation Guide; Golden Key: Atlanta, GA, 1996.

Copyright of American Journal of Drug & Alcohol Abuse is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.