A randomized test of a small-group interactive social norms intervention
|Please reference to source, this paper for note only|
Social norms interventions are a common approach to addressing the problem of college student drinking. An increasingly popular but not yet well-validated social-norms-based intervention consists of providing normative feedback to students in small groups. Objective, Participants, and Methods: In this study, the authors used a randomized design to test an interactive form of small-group social norms correction with 502 first-year students during September and October 2001. Because the unit of random assignment was at the level of the classroom, the authors used hierarchical linear modeling to estimate variability. They investigated whether small-group interactive social norms correction could influence alcohol perceptions and behaviors above and beyond a non-interactive social norms education approach. Results indicate that the approach has a fairly substantial influence on student perceptions; however, the findings do not support an influence of interactive small-group social norms correction on measures of alcohol use behaviors. Conclusions. Given these findings, the use of interactive small-group social norms approach to influence student misperceptions may be considered as a primer for population-level preventive interventions.
Keywords: alcohol use, college health, social norms intervention
Recent research has identified college student drinking as an important health issue. Estimates indicate that approximately 1,700 students die, 500,000 are injured, and 600,000 are assaulted while using alcohol or other drugs each year.1 Since 1993, the Harvard School of Public Health has measured levels of heavy episodic-or binge-drinking among college students with the College Alcohol Survey (CAS).2 In this study, heavy episodic drinking is defined for men as the consumption of 5 or more drinks in a single drinking session within the past 2 weeks and for women as 4 of more drinks within the same time frame.2 CAS researchers have consistently found that approximately 2 out of 5 students drink at binge levels.3 In addition, they monitor frequent heavy episodic drinking, defined as binge drinking 3 or more times in the past 2 weeks. Analyses with this subgroup reveal that frequent heavy episodic drinkers were 21 times more likely than were nonfrequent heavy episodic drinkers to experience 5 or more of the 12 alcohol-related problems included in the survey. The Monitoring the Future Survey produced similar results.4 In this study, 40% of students reported consuming 5 or more drinks on a single occasion at least once in the previous 2 weeks, a greater proportion than same-age, noncollege-attending peers (35%) or high school seniors (31%).
Although most college student drinkers begin drinking regularly prior to college, drinking rates increase significantly during the transition from high school to college.5-7 Moreover, students entering college show marked increases in alcohol use compared with those who enter the workforce after graduation from high school.4
These findings have both theoretical and practical implications. Theoretically, although some first-year students arrive on campus with high-risk drinking patterns, many others develop these patterns during the first-year transition. Thus, as suggested by the National Institute on Alcoholism and Alcohol Abuse (NIAAA) Task Force, preventive interventions that target incoming freshmen are of critical importance.8
Overview of Social Norms Theory
One popular approach to addressing this high-risk drinking culture is social norms prevention. Initial correlational and theoretical articles from the late 1980s and early '90s established a conceptual and theoretical basis for social norms approaches on college campuses. Researchers9-11 in these studies found that college students have strong misperceptions of peer drinking behaviors, typically over-estimating the frequency and amount that other students drink. In a review of peer influences on college student drinking, Borsari and Carey12 noted that in 29 of 30 studies, investigators observed misperceptions between actual and perceived alcohol use rates. These misperceptions can be influenced by several factors, including the type of norm (injunctive or descriptive), sex, reference group, or campus demographics.13 Injunctive norms are perceptions of social approval or disapproval of behavior and reference the cultural acceptance of behavior.14 Conversely, descriptive norms are perceptions of what others do. In this case, descriptive norms refer to students' perceptions of other students' drinking patterns.
Perkins and Berkowitz10 theorized that providing students with feedback about the actual norms could reduce drinking levels. On the basis of this premise, many university prevention programs implement social norms interventions focusing on correcting misperceptions. Schools have typically implemented 3 types of social norms interventions: social marketing media campaigns,15 normative correction in the context of individualized feedback or brief motivational interventions,16-18 and small-group social norms interventions.19-21 Findings from each of these 3 types of interventions are described in the next sections.
Social Marketing Media Campaigns
Media campaigns are, by far, the most widely used of these approaches. Using local survey data, these universal preventive interventions advertise accurate information about the typical quantity of alcohol consumed per drinking occasion via campus newspaper ads, posters, flyers, and other mass media venues. It is theorized that corrected misperceptions will lead to reductions of normative misperception of heavy drinking, which will, in turn, lead to reduced levels of drinking and alcohol-related negative consequences.
Following the Haines and Spear15 publication of promising case study results at one campus, social norms marketing has arguably become the most highly publicized preventive intervention on college campuses. It has been cited in The New York Times Magazine,22 and Wechsler et al23 found that as many as 1 in 9 schools implement this prevention approach. A handful of universities have reported similar success with this approach. For example, the University of Arizona observed a 29% reduction in binge drinking over a 4-year period.24 Hobart and William Smith colleges reported a 12% decrease over 2 years, and Western Washington University recorded an 8% decline in binge drinking during the first year the program was initiated.25
Despite social marketing interventions' strong popular appeal, peer-reviewed, rigorous tests of such an approach are lacking, prompting the NIAAA Task Force to rate this approach as a Tier 3 Intervention (ie, displaying "evidence of logical and theoretical promise but requiring more comprehensive evaluation").8(p21) Studies applying controlled experimental designs have suggested less promise. In a study implementing a social norms intervention to first-year students, Werch et al26 observed nonsignificant differences in alcohol use and alcohol use risk factor measures at posttest between intervention and comparison groups. In a test of a social norms marketing campaign within university residence halls at treatment and comparison sites, Clapp et al27 observed no intervention effects on drinking behaviors. Thombs and Hamilton28 obtained similar results when comparing student athletes who had been exposed to a social norms media campaign. In all 3 studies, students exposed to the social norms media campaigns experienced a greater decrease in alcohol-related misperceptions than did comparisons, but none of the researchers observed significant reductions in alcohol use behaviors.
Individual-Level Preventive Interventions
Brief motivational interventions have been the most promising individual-level approach for reducing alcohol abuse and alcohol-related problems among college students. 29 These interventions, as well as the more pared-down individualized feedback approaches, focus on correcting normative misperceptions of alcohol use and problems.30,31 These approaches give students individualized feedback, comparing their drinking rates with typical student perceptions and drinking patterns. Consistent with social norms theory, it is hypothesized that personalized normative feedback will correct misperceptions, which will then motivate the individual to decrease drinking levels. Researchers in 2 studies16,30 have observed evidence consistent with such a mediational role for perceived norms. This type of personalized feedback has shown promise using both mail and computer delivery.31-34 Despite the effectiveness of brief motivational interventions and individualized feedback, however, they are comparatively resource- and time-intensive, prompting some investigators to explore the utility of correcting normative misperceptions in small-group settings.
Small-Group Social Norms Interventions
The least used social-norms-based intervention consists of providing normative feedback to students in groups. Common settings typically consist of classrooms or presentation venues with fewer than 50 students.20,35 The approach merges direct feedback and media campaigns to reduce participants' resistance to social norming messages. This approach often seeks to capitalize on a readily available reference group. Facilitators present correct normative information and encourage students to discuss the information. This approach recognizes that misperceptions may be longstanding and that resistance to the messages will most likely be the initial response. Small-group social norms intervention researchers theorize that discussion of correct information with a reference group can reduce participants' resistance to and facilitate their acceptance of the messages.
As with social marketing media campaigns, few peerreviewed publications have published studies on smallgroup social norms interventions. In one test with an experimental design, Barnett et al35 found associations between perceptions and drinking patterns but did not find reductions in drinking behaviors associated with the social norms treatment approach at 4-month follow-up. Similarly, Smith20 described implementation of a small-group social norms approach with classroom-level random assignment to treatment and comparison groups. The author implied that he performed analysis as if assignment was at the individual level. However, after clarifying conversations, Smith stated that he performed interclass correlations but that the results did not suggest the need for hierarchical linear modeling (HLM; B.H. Smith, personal communication, October 18, 2004). In posttests performed 1 month after intervention, Smith found that the treatment was effective at reducing students' misperception of other students' drinking but observed no evidence for reductions in drinking behaviors. Stamper et al21 replicated these findings 1 year later.
The Current Study
Theoretically and anecdotally, campuswide media campaigns appear to be a relatively cost-effective means to reduce binge-drinking risks with the college-aged population. However, peer-reviewed evaluations have yet to support their effectiveness. Brief motivational interventions and individualized feedback approaches provide experimentally measured effectiveness but are expensive, are labor-intensive, and reach a limited number of students. Thus, we sought to test a hybrid approach in the current study. Specifically, we investigated whether a small-group interactive social norms correction could achieve the behavioral outcomes of brief motivational interventions at greatly decreased cost and labor levels. As described in the Methods section, the interactive component is students' completion of the survey at the beginning of the class period and provision of feedback on survey results in the same class period. With these efficacy and cost-effectiveness goals in mind, we sought to answer 2 research questions. First, can a small-group interactive social norms intervention reduce normative alcohol use misperceptions above and beyond a noninteractive social norms education approach? Second and more important, can a small-group social norms intervention decrease alcohol use as compared with a noninteractive social norms education approach?
We recruited participants from a 1-credit orientation course required of all first-year students at a medium-sized New England public university. The course-which is broken up into sections of fewer than 35 students-acquaints new students with the university. Some sections discuss specific majors; all aim to promote a sense of community among students.
One component of the course is a health and safety presentation. Trained graduate students or peer leaders facilitate this presentation, which the university's director of Substance Abuse Prevention Services supervises. The facilitators provide prevention and education information on alcohol and drugs, sexual assault, and violence, as well as information on campus police, health services, and the counseling center. In the current study, implementation of intervention and comparison conditions occurred during the health and safety presentation.
We sent letters to the course's instructors requesting participation. These letters informed instructors that we would randomly assign sections to either an intervention or a comparison condition. Students in the standard social norms correction (SSNC) health and safety presentation condition received the standard health and safety presentation. During the presentation, the section leaders provided students with prevention information focusing on alcohol and drug issues, sexual assault and violence, and health and personal safety. They also provided students in this condition with campuswide statistics on average levels of alcohol use (70% of students have 5 or fewer drinks when they consume), marijuana use (80% of students abstain or use less then once a month), and sexual activity (70% of students reported 0 or 1 sexual partner in the past year). We obtained this information from data on previous-year, first-semester freshmen.36 Our intent in providing this information was to illustrate the normalcy of sexual abstinence and low-risk use of alcohol and marijuana.
Prior to the presentation, the section leaders administered the survey, which queried students on alcohol use, marijuana use, and the number of past-year sexual partners, in addition to perceptions of other students' patterns on these same items. Distribution and collection of the classroom survey lasted 5 to 10 minutes. For SSNC students, section leaders did not encourage them to discuss the normative information presented, nor did they provide feedback on the survey results.
Students in the interactive small-group social norms correction (ISNC) health and safety presentation condition received the standard health and safety presentation plus feedback and discussion of section-specific survey results. During the presentation, upper-class mentors tallied classroom survey results. After the presentation, the section leaders provided students with their section-specific norms (generated by the same students just 30 minutes prior). The section leaders encouraged discussion as to why observed discrepancies between actual and perceived alcohol use norms may exist.
Both conditions received the classroom survey at the beginning of the class period. However, only ISNC students received their section-specific results. The ISNC method consisted of approximately 5 more minutes of data feedback and 5 more minutes of class discussion than the comparison condition. The university's institutional review board approved all procedures and measures.
Forty-five of 90 (50%) section instructors consented to participate in the study. We randomly assigned 22 sections to the SSNC and 23 to the ISNC. Not all section instructors who agreed to participate returned pretests. Forty-two sections (93%; 22 SSNC and 20 ISNC) returned pretests. We received 905 completed pretests: 471 SSNC participants (52%) and 434 ISNC participants (48%).
We analyzed data for 502 students. This number represents the usable number of matched responses. The high attrition rate (44.5%) was due to either nonparticipation at follow-up (4 sections, n = 100, 24.8%) or to student absence at baseline (n = 101, 25.1%), intervention (n = 80, 19.8%) or follow- up (n = 89, 22.1%). In addition, we used self-generated codes (described in the Measures section) to anonymously link pre- and posttest data, and some students provided unmatchable codes (n = 33, 8.2%). The final analyzed dataset consisted of 259 (51%) SSNC students and 243 (49%) ISNC students. A majority of the sample was female (55%) and white (87%), followed by Hispanic (5.3%), Asian (3.8%), African American (3.6%), Native American (1.2%), and 1.6% reporting another ethnicity. These numbers closely match first-year student demographics at this university (56% female, 78% white, 3.9% Hispanic, 3.9% Asian, 3.7% African American, 0.5% Native American, and 10% not reporting ethnicity). Comparisons between the final sample and the incoming freshman population revealed nonsignificant differences by sex or African American, Hispanic, Asian, and Native American ethnicities (all ps > .05). White students, however, were significantly overrepresented in the sample (χ^sup 2^[1, N = 437] = 7.55, p < .01), and the proportion of the campus population not reporting ethnicity was significantly greater than in the study sample (χ^sup 2^[1, N = 50] = 19.58, p < .0001). We suspect but cannot verify that the significant overrepresentation of white students in our sample (compared with overall university data) may be due to the disproportionate number of white students in the campus population that opted not to report race or ethnicity data. However, the final sample is a close approximation, demographically, of the population of first-year students from which it was drawn.
Section leaders administered pre- and posttest batteries in a fixed order, as detailed later, using Teleforms-created (Teleforms Inc, Manitoba, Canada) surveys that enabled scanning of participant responses. Participants read, signed, and provided informed consent through survey completion. Section leaders instructed students not to place their names or student numbers on the surveys. A self-generated unique code number procedure allowed for linking pre- and posttest surveys while maintaining participants' anonymity: Students wrote down the first letter of their mother's maiden name, followed by the month, day, and year of their birth. Using written procedures that we provided, class instructors administered pre- and posttest surveys. The pre- and posttest self-report surveys took approximately 10 minutes to complete. Section leaders administered pretest surveys in the first class period and collected posttests in the last class period. Class durations were 6, 10, or 12 weeks depending on how long each class met and how many times a week the class was held. All health and safety presentations occurred at least 2 weeks after pretest and 3 weeks before posttest. The focus of the surveys was on alcohol use and perceptions of alcohol use by peers. The survey contained additional content related to overall student health and safety that are not included in the analyses presented here.
We assessed alcohol use with 2 items-which have demonstrated reliability (eg, internal consistency, 18-month test-retest) and validity (strong associations with other measures of consumption) in a number of our previous studies37-39-regarding typical quantity and frequency of alcohol consumption per week. We then multiplied these items to produce a weighted quantity-frequency index of weekly consumption. We used number of drinks consumed in a typical week as our primary measure. All surveys provided clear definitions of a standard drink: 12 oz of beer, 4 oz of wine, and 1.25 oz of liquor in a shot or a mixed drink.
Perceptions of Others' Alcohol Use
We measured perceived norms with a widely used measure, the Drinking Norms Rating Form,9 which has demonstrated reliability and validity in our own work,37,39 as well as in several previous studies.9,40,41 The items assessed perceptions of others' drinking practices for the typical same-sex student at the study university (eg, "How often do you think a typical student of your gender consumes alcohol?"). We assessed perceived quantity (number of drinks when drinking, ranging from 1 to 12 or more) and perceived frequency (number of drinking days, ranging from 0 to 7) of alcohol consumption. We multiplied these items to produce a similar quantity-frequency perception index of weekly consumption.
Overview of Data Analysis
Because the unit of random assignment was at the level of the classroom and not the individual, we used HLM to estimate classroom-level variability.42 We used 2 two-level models to test whether the ISNC condition influenced normative misperceptions and behavior. We constructed models for posttest outcomes of alcohol perception and alcohol use. All models contained Level 1 covariates of sex and corresponding individual pretest scores. Level 2 models contained the type of intervention assigned.
Descriptive Statistics for Participant Alcohol Use
We observed no baseline differences between the 2 intervention groups on either self-reported alcohol use (t = 0.40, ns) or perceived norms (t = -0.49, ns). Therefore, we aggregated these data across the sample to describe normative perceptions and alcohol use. Pretest data revealed significant sex differences in both reported drinks per week (t = -4.83, p < .0001) and perceived drinks per week (t = -6.05, p < .0001). Men reported an average of 11.74 drinks (SD = 15.59) per week versus women's reported 6.03 drinks (SD = 8.51). Both men and women substantially overestimated how much alcohol other students consumed; men estimated that other male students drank an average of 18.10 (SD = 12.20) and women estimated that other female students drank an average of 13.16 (SD = 7.32) drinks per week.
We conducted attrition analyses to examine whether students who were not available at follow-up (n = 403) differed on variables of interest at baseline from those who remained in the study. We observed no significant differences between attriters and nonattriters on baseline alcohol use measures, but some differences did emerge on normative perceptions. We observed group differences for drinking frequency perceptions (t = 2.059, p < .040, d = .14) and past-week alcohol use perceptions (t = 2.451, p < .014, d = .17), with the attriters reporting more perceived drinks per week (M = 18.03, SD = 0.54) than did nonattriters (M = 16.33, SD = 0.44). Cross-tabulation of attriters by treatment type revealed no evidence of experimental mortality (F[1, 903] = 0.134, p < .714), with 207 (44%) attriters in the SSNC condition and 196 (45%) in the ISNC condition.
Tests of Substantive Research Questions
RQ1: Can an Interactive Social Norms Correction Influence Misperceptions?
We obtained a substantial interclass correlation for alcohol perceptions, which indicated variability for alcohol perceptions by section (ρ = .15) and justified HLM. The interclass correlation for alcohol use was less extreme (ρ = .07); nonetheless, we used HLM for analyses with each of our outcome variables.
As anticipated, the Level 1 covariate of baseline perceptions was significantly related to drinking perceptions at posttest (t = 6.05, p < .0001), such that higher baseline perceptions were associated with higher perceptions at follow-up. There was a trend toward sex differences in perceptions at posttest (t = 1.76, p < .08), with men reporting a mean of 12.1 (SD = 7.24) and women reporting a mean of 10.3 (SD = 7.9) perceived drinks per week. After controlling for pretest alcohol use perceptions and sex, results of the full 2-level HLM model show that interactive social norms intervention for small groups effectively reduces perceptions of drinking amounts (t = -3.976, p < .001). As can be seen in Figure 1, we observed pre- to posttest reductions in the ISNC condition, falling from 16.8 (SD = 10.91) to 11.1 (SD = 7.56) drinks, whereas the SSNC condition decreased from 16.3 (SD = 9.61) to 15.4 (SD = 9.79) drinks. Computation of the intervention effect size using the posttest means and a pooled standard deviation43 yielded a d index of .41. Cohen defined effect sizes44 as small (d = .2), medium (d = .5), and large (d = .8); thus, the intervention effect for correcting normative misperceptions would be considered moderate.
RQ2: Can an Interactive Social Norms Correction Influence Behaviors?
Consistent with expectations and the previous analyses, the Level 1 covariate of baseline alcohol use was significantly related to reported alcohol use at posttest (t = 7.67, p < .0001), such that higher reported baseline use was associated with higher reported postintervention use. We also observed significant sex differences in alcohol use at posttest (t = 2.95, p < .005), with men reporting a posttest mean of 10.6 (SD = 12.30) and women reporting a mean of 5.9 (SD = 7.9) drinks per week. After we controlled for pretest alcohol use perceptions and sex, RQ2 was not supported (t = 0.433, p < .665). We observed a modest decrease in reported drinks per week among SSNC students, from a pretest mean of 8.74 drinks (SD = 13.19) to a posttest mean of 7.71 drinks (SD = 9.80; see Figure 2). Reported drinks per week did not decrease for ISNC students, with a mean of 8.31 (SD = 11.50) at pretest and 8.31 (SD = 10.21) at posttest. Computation of the effect size using posttest means and a pooled standard deviation yielded an index of d = .06, well below an effect size determined to be small (d = .2). These results do not support the major intervention aim of decreasing alcohol use.
Our results support the general belief that misperceptions of alcohol use exist among the college student population and that small-group interactive social norms correction can have a fairly substantial influence on these perceptions. However, consistent with an emerging body of research, we found no evidence that small-group interactive social norms correction effectively reduced alcohol use among incoming college students. It is important to note that we specifically compared interactive and noninteractive social norms approaches. Thus, we gave students in both conditions some normative feedback, but we observed changes in misperceptions only in the INSC condition, in which section leaders provided and interactively discussed survey results.
The SSNC condition closely approximated the interventions done by Barnett et al35 and by Smith,20 which involved disseminating campuswide statistics highlighting the norms of low-risk alcohol use. In our interactive social norms comparison condition, however, we introduced the novel innovation of using the classroom to provide real-time feedback on class norms. The large reduction in misperceptions among this group supports the heightened salience of this approach in correcting normative misperceptions.
Participation in the ISNC (vs the SSNC) condition did not result in lower levels of alcohol use (RQ2). These results add to a small but increasing body of research suggesting that although small-group social norms influences appear to affect normative perceptions, they do not affect actual drinking behavior.19-21,35 Although college students' alcohol use trajectories vary, the transition into college is typically associated with increased alcohol use and alcoholrelated problems.45,46 Our intervention conditions may have altered this trajectory somewhat, as evidenced by the relatively stable levels of alcohol consumption we observed; however, given that we did not include a nonintervention condition-which would have provided a comparison from which to better support this conclusion-we make this suggestion tentatively.
A particular strength of this study was the use of grouplevel random assignment in conjunction with a statistical approach (ie, HLM) to compare variability within sections. Previous researchers20,21,35 have reported using grouplevel random assignment and individual-level data analyses without consideration of the importance of controlling for within-group variability. As noted, with respect to the intraclass correlations observed in our study, random grouplevel assignment necessitates statistical modeling of withingroup variability.42
Several study limitations merit discussion. First, there was a considerable degree of attrition in this study. Student absence for this 1-credit course was consistently about 20% per section per time point (pretest, presentation, posttest). Absence at any time point resulted in incomplete data. This degree of attrition could potentially have major effects on the validity of our findings. These concerns are somewhat mitigated by attrition analyses demonstrating no differences between attriters and nonattriters on the main outcome measure, alcohol use. However, we observed significant differences when comparing attriters and nonattriters on normative perceptions, with attriters having a higher perception of drinks per week at baseline than did nonattriters. Given the lack of differential attrition by group, however, the threat to internal validity appears modest.
An additional limitation is the use of self-reported data, although self-report procedures are by far the most common method to obtain alcohol use data from students. Researchers47,48 have recommended that investigators make assurances of anonymity or confidentiality, stress the importance of truthful responding, and adopt a nonjudgmental data-collection perspective as ways to increase the reliability and validity of self-reports. Moreover, recent research has suggested that students' self-reports of alcohol use are underestimated.49 However, White et al49 found that providing students with specific definitions of a standard drink reduced this influence. In the present study, we used definitions of a standard drink for all survey instruments. Most importantly, we had no reason to suspect that tendencies to under- or overreport alcohol use would interact with intervention condition, which would be necessary to constitute a threat to internal validity.
An additional limitation includes the relative lack of ethnic and racial heterogeneity in our sample. Although our sample closely approximated the population of first-year students at the predominantly white campus, future researchers must examine the potential efficacy of small-group social norms approaches with more ethnically diverse populations. Furthermore, although the use of an experimental design in this study is a strength, the lack of a true control group is a limitation. Most notably, it prohibits comparison between the 2 experimental conditions and a natural history18 comparison group during a developmental transition (college matriculation) typified by increased drinking.
As stated earlier, neither the population-level social norms media campaigns nor small-group approaches have yet demonstrated efficacy in studies using experimental design or quasiexperimental approaches that adequately control for threats to internal validity.50 Future researchers examining small-group social norms or population-level social marketing studies would profit from using experimental or well-designed quasiexperimental designs, to allow for strong inferences about the causes of any observed effects. Incorporation of theory-based mediators (eg, changes in normative misperceptions) into study designs and analyses would further strengthen researchers' ability to make causal inferences about mechanisms of effect.51 Reductions in alcohol use and problems observed in studies of brief motivational interviewing29,31 and individualized feedback31-34 suggest potential efficacy for social norms approaches in group and environmental preventive interventions. Even stronger inferences can be drawn about the potential importance of social norms from the few studies in which investigators have identified changes in normative perceptions as mediators of intervention effects.16,30
In conclusion, our results do not support small-group interactive social norms intervention as being effective as a stand-alone intervention to achieve behavioral change. Given these findings, along with the methodological issues raised previously, one strategy for future work may be the use of interactive small-group social norms approaches to influence student misperceptions as a primer for populationlevel (universal) preventive interventions.
Finding ways to influence college student drinking and associated negative consequences remains an important health concern in our society. Current research and theory suggest that multiple interventions at the individual, group, and environmental levels are needed to yield meaningful progress in reducing alcohol-related harm among the college population. Currently, beyond the individual level, little is known about effective interventions. Therefore, identifying the most effective approaches at the group and population level through the use of theory-based empirical research is of critical importance. Social norms approaches appear to be an important component of successful individual-level preventive interventions, but much research remains to be done in validating and transferring these successes to larger groups and the overall college population.
Hingson R, Heeren T, Winter M, Wechsler H. Magnitude of alcohol-related mortality and morbidity among US college students ages 18-24: changes from 1998 to 2001. Annu Rev Public Health. 2005;26:259-279.
Wechsler H, Dowdall GW, Davenport A, Rimm EB. A genderspecific measure of binge drinking among college students. J Am Coll Health. 1995;85:982-985.
Wechsler H, Lee JE, Kuo M, Seibring M, Nelson TF, Lee H. Trends in college binge drinking during a period of increased prevention efforts. Findings from 4 Harvard School of Public Health College Alcohol Study surveys: 1993-2001. J Am Coll Health. 2002;50:203-217.
Perkins HW, Berkowitz AD. Perceiving the community norms of alcohol use among students: some research implications for campus alcohol education programming. Int J Addict. 1986;21:961-976.
Prentice DA, Miller DT. Pluralistic ignorance and alcohol use on campus: some consequences of misperceiving the social norm. J Pers Soc Psychol. 1993;64:243-256.
Borsari B, Carey KB. Peer influences on college drinking: a review of the research. J Subst Abuse. 2001;13:391-424.