Enhancing a Norms Program to Reduce High-Risk Drinking - CiteSeerX

4 downloads 0 Views 1MB Size Report
“Whether it's Thursday, Friday, or Saturday night, 2 out of 3 UNC ...... fewer individuals live in a typical Greek house than in any of the residence halls, members ...

Enhancing a Norms Program to Reduce High-Risk Drinking Among First Year Students

Robert Foss, PhD Shane Diekman, M.P.H. Arthur Goodwin, M.A. Christopher Bartley, M.A.

University of North Carolina – Chapel Hill

U.S. Department of Education Award No. S184H000056

August, 2003

Table of Contents Executive Summary ......................................................................................................... 1 Background ...................................................................................................................... The problem of college student alcohol use ............................................................... Addressing student drinking ...................................................................................... Measuring student drinking .......................................................................................

5 5 6 7

The UNC-CH Social Norms Program ............................................................................. 9 Program implementation ............................................................................................ 10 First year student orientation sessions (“CTOPS”) .................................................... 11 Posters ........................................................................................................................ 13 Campus-wide zipper-pull promotion ......................................................................... 14 Informational web site ................................................................................................ 15 Fall Fest activity booth ............................................................................................... 16 The “BAC Guy” ......................................................................................................... 17 Curriculum infusion ................................................................................................... 17 Coordination with the UNC-CH alcohol program ..................................................... 18 The BAC Survey .............................................................................................................. 19 Sampling procedure ................................................................................................... 19 Interview procedure ................................................................................................... 20 Precautions for students potentially at risk ................................................................ 21 Survey Results .................................................................................................................. 22 Study participation and sample characteristics .......................................................... 22 Study participation ............................................................................................... 22 Sample characteristics .......................................................................................... 24 Sample representativeness ................................................................................... 25 Campaign awareness, understanding & believability ................................................ 27 Campaign awareness ............................................................................................ 27 Campaign understanding ...................................................................................... 28 Belief in accuracy of the alcohol fact ................................................................... 28 Measures of student drinking ..................................................................................... 29 General alcohol use .............................................................................................. 29 Heavier drinking .................................................................................................. 30 Variations in drinking by time ............................................................................. 33 Summary of changes in all drinking measures .................................................... 35 Comparison to national trends ............................................................................. 36 Discussion and Conclusions ............................................................................................. 38 Benefits of using BAC data – triangulation of measurement .................................... 39 Intensive program implementation ............................................................................ 39 Non-believers ............................................................................................................. 41 Attributing changes to the program ............................................................................ 41 i

Representativeness of BAC survey results ................................................................ 42 Conclusion ................................................................................................................. 43 References ........................................................................................................................ 44

ii

Tables and Figures Tables 1. Participation rates by survey year ........................................................................ 23 2. Selected characteristics of respondents, by survey year ...................................... 24 3. Selected characteristics of BAC survey samples and corresponding UNC-CH undergraduate student populations, by survey year ............................................. 26 4. BAC distributions among all respondents, by survey year .................................. 30 5. BAC distributions among respondents who reported drinking on the night of the interview, by survey year ........................................................................... 30 6. Self-reported heavy drinking on the night of the interview among all respondents, by survey year ................................................................................. 32 7. Self-reported heavy drinking among respondents drinking on the night of the interview, by survey year ..................................................................................... 32 8. BAC distribution by time of week and survey year ............................................. 34 9. BAC distribution by time of night and survey year ............................................. 35 10. Summary of changes in all measures of drinking ................................................ 36

Figures 1. Percent of residence hall rooms with the “2 out of 3” poster displayed, by year................................................................................................................... 14 2. Proportion of backpacks observed with zipper pull attached, by semester ......... 15 3. Campaign awareness (heard of campaign), by survey year ................................. 27 4. Campaign understanding, by survey year ............................................................ 28 5. Campaign believability, by survey year ............................................................... 28 6. Reported and measured drinking on night of interview, by survey year ............. 29 7. Self-reported heavy drinking and frequent heavy drinking during the past two weeks, by survey year; all respondents ................................................. 31 8. Proportion of respondents with BAC above each of four thresholds by hour of night (2002 only) ..................................................................................... 34 9. Trends in student self-reports of frequent heavy drinking, UNC-CH vs. national survey data ............................................................................................. 37

iii

Executive Summary Excessive alcohol use by college students is a persistent concern among public health and university officials. To address this problem, many college campuses have implemented “social norms” marketing programs. These efforts are designed to reduce misperceptions about student drinking by presenting accurate information about student alcohol use. When students understand the reality of drinking on their campus, they should feel less pressure to drink which, in turn, should reduce actual drinking. The present report details the long-term effects of a social norms program to reduce student alcohol use among students at the University of North Carolina - Chapel Hill (UNC-CH). To gain a clearer picture of drinking among UNC-CH students, a field survey was conducted in the fall of 1997. The sampling procedure was designed to obtain a representative sample of students as they returned to their residences in the late evening. A unique feature of this survey was the collection of breath samples, allowing for a direct measurement of blood alcohol concentration (BAC). An important study finding was that two-thirds of students returned home with a zero BAC on the traditional “party” nights. This finding ran counter to reports in the national media that portrayed college student drinking as a rampant epidemic. Through a series of discussions with students, a primary message was developed that was clear and easily understood by students:

“Whether it’s Thursday, Friday, or Saturday night, 2 out of 3 UNC students return home with a .00 blood alcohol concentration” A multi-faceted approach was developed to integrate this message into the routine functioning of the entire campus. The resulting “2 out of 3” social norms program focused initially on first year students, who are being immersed in the campus culture for the first time. Subsequent efforts were expanded to reach the entire campus community. Several activities were conducted to promote the 2 out of 3 program messages on campus, including (but not limited to) presentations at first-year student orientation 1

sessions, dissemination of print materials, poster incentive campaigns, curriculum infusion, coordination with the university’s general alcohol program and maintenance of a 2 out of 3 web site.

Program Evaluation To determine the effects of the 2 out of 3 program, additional field survey data were obtained from students during the fall semesters of 1999 and 2002. Since the 2 out of 3 program was initiated in the summer of 1999, the fall 1999 data reflect possible early effects of the program, while the 2002 data represent enduring changes that may have occurred as the result of a sustained program. The analyses examined data for 6,352 undergraduate students interviewed during 1997, 1999 or 2002. A breath alcohol measurement was obtained for 6,108 (96%) of these individuals.

Campaign Awareness, Understanding, and Believability Awareness •

In 1999, 72% of all respondents indicated that they had heard of the campaign, including 92% of first year students.



In 2002, 91% of all respondents indicated that they had heard of the campaign, including 92% of first year students.

Understanding •

In 1999, 72% of all respondents understood the general meaning of the campaign, including 80% of first year students.



In 2002, 82% of all respondents understood the general meaning of the campaign, including 83% of first year students.

Believability •

In 1999, 35% of respondents who had heard of, and understood the message, found it to be believable, including 34% of first year students.



In 2002, 45% of respondents who had heard of, and understood the message, found it to be believable, including 51% of first year students.

2

These findings indicate high awareness for the 2 out of 3 campaign, even from the beginning. Although many students found the message difficult to believe, the percent of non-believers decreased noticeably from 1999 to 2002.

Measurement of Student Drinking A number of changes in student drinking were observed over the course of the study. Some of the most noteworthy changes from 1997 to 2002 include the following:

General Alcohol Use •

Any alcohol use on the night of the interview – determined by both self-report and direct BAC measurement – decreased from 1997 to 2002.



Among those drinking on the night of the interview, the mean number of selfreported drinks consumed decreased from 5.1 to 4.3.

Heavier Drinking •

The percentage of respondents who could be classified as heavy drinkers on the night of the interview decreased from 14% to 10%.



Among respondents who were drinking on the night of the interview, the proportion with a BAC above .05% decreased from 60% to 52%.



Using the traditional self-report criterion for heavy drinking (five or more drinks on at least one occasion in the past two weeks), the proportion of heavy drinkers decreased from 50% to 45%.



Self-reported frequent heavy drinking (5 or more drinks on 3 or more occasions in the past two weeks) decreased from 24% to 20%.

Conclusion These findings provide support for the effectiveness of the 2 out of 3 campaign in reducing student alcohol use on the UNC-CH campus. The high degree of awareness and understanding of the 2 out of 3 message among UNC-CH students indicates that the social norms program was successful in reaching its target audience. The increased believability of the campaign suggests that the primary message, which specifies that on “party” nights most students drink little, if at all, is becoming more accepted as the reality

3

of student drinking on campus. Finally, the decreases observed in both self-report and direct measures of student drinking suggest that social norm marketing programs can affect actual drinking behavior and not merely self-reports of drinking. One limitation of this study was the lack of a control group, which would have helped rule out alternate explanations for our findings. However, evidence from national surveys that indicate stable or increased drinking among students provides reason to believe that the decrease in drinking among UNC-CH students is not simply the reflection of a general downward trend in college student drinking.

4

1 Background The Problem of College Student Alcohol Use The past decade has seen a substantial increase in attention to college student drinking from alcohol researchers, policy-makers and the media. Many campuses have increased efforts to inform students of the risks of excessive drinking. Despite these efforts, heavy drinking among college students has remained stable. Two multi-year national surveys report that throughout the 1990s, about two of five U.S. college students had consumed five or more drinks (four or more for females) on a single occasion at least once in the 2 weeks before the survey – a behavior usually labeled ‘heavy episodic’ or ‘binge’ drinking (O’Malley & Johnston, 2002; Wechsler et al., 2002). Furthermore, about one of five U.S. college students could be classified as ‘frequent heavy episodic’ drinkers (three or more heavy drinking episodes in a two-week period) during the 1990s. Although heavy drinking has remained relatively stable there has been some polarization since 1997, with both more abstainers and more of the heaviest drinkers (Wechsler et al., 2002). Excessive consumption of alcohol by college students exposes them to a wide range of risks resulting from impaired cognitive functioning. In addition to the well-known risks of driving after excessive drinking (Hingson et al., 2002; Sleet, Wagenaar & Waller, 1989), students also experience increased risk of injury as pedestrians, bicyclists, and in falls and fires (Hingson & Howland, 1993). Excessive drinking is associated with greater probabilities of physical and sexual assault, health problems, unsafe and unplanned sexual activity, sexual harassment, impaired sleep, and interpersonal problems (Presley, Meilman & Lyerla, 1993, Roizen, 1997; Harrington & Leitenberg, 1994). In addition to increasing a variety of risks for themselves, students who drink excessively also constitute a substantial nuisance to other students (Wechsler et al., 1995).

5

Addressing Student Drinking Along with the dramatic increase in quality research attention to student drinking during the past decade, there has been a substantial increase in the development of sophisticated programs to reduce alcohol-related problems among college students. Throughout North America, programs to address excessive student drinking have evolved from individuallyoriented “educational” efforts into multi-faceted projects that address several aspects of the larger socio-cultural and physical environment within which drinking occurs. In recent years, it has become common to hear statements from academic leaders about the need to “change the culture” on college campuses – that is, to reduce the sense that alcohol use is the central feature of college student life (NIAAA, 2002). Although a small number of U.S. campuses are engaged in serious, coalition-based comprehensive programs to begin that task, most seem to still lack the resources (or the will) to attempt a large-scale transformation to serious, research-based programs (NIAAA, 2002). Many institutions have, however, embraced programs that address one ubiquitous feature of campus life: the mistaken belief that drinking is common, widespread and excessive (Perkins, 2003). During the 1980s researchers began to recognize that there are substantial misperceptions about the amount of drinking that occurs among college students (Perkins & Berkowitz, 1986). This led to a realization that it might be possible to reduce actual drinking simply by disabusing students of these misperceptions, thereby lowering a strong perceived pressure to conform to a norm that doesn’t match actual student drinking. A study among middle school students suggested that such an approach could be effective (Hansen, 1993). Following these early efforts, a number of sophisticated “social marketing” programs have been developed to help college students understand that, despite all the reports they have heard decrying the dire state of affairs concerning student alcohol use, the norm for alcohol use among college students (at least on most campuses) is actually moderation, rather than excess (Haines, 1996; Johannessen et al., 1999). Unlike earlier efforts to address student drinking, so-called “social norms” programs are (1) based on a sound theoretical underpinning (Bandura, 1977; Haines & Spear, 1996) and (2) have been empirically demonstrated to produce changes in selfreported drinking behavior over a multi-year period across different campus settings (Haines, 1996; Johannessen, 1999; Perkins 2002). As a result of the apparent dramatic success of some 6

early programs (Haines, 1996; Johannessen et al., 1999), a large number of college campuses have now developed social norms programs (National Social Norms Resource Center, 2003; http://www.socialnorm.org).

Measuring Student Drinking Despite their popularity and apparent success, social norms programs have not yet been clearly shown to reduce student drinking. A number of studies indicate that following introduction of a program to correct misperceptions, students generally report less heavy drinking. To date, however, there are little or no objective data to document the effects of these programs on actual drinking behavior. A common weakness in research on student drinking is the near exclusive reliance on selfreports of drinking, which often requires detailed recall over long periods. Although selfreports of drinking are generally considered valid (Midanik, 1988), they cannot directly assess an individual’s level of impairment on a given occasion since self-reports are inherently subjective. In fact, recent research suggests that a sizable proportion of individuals meeting the criteria for ‘binge’ drinking on a specific occasion do not reach BAC levels that are typically associated with observable impairment (Foss, 2001; Lange & Voas, 2002; Perkins, DeJong & Linkenbach, 2001). A more precise, occasion-specific measure of alcohol impairment can be obtained using direct measurement of a respondent’s breath alcohol concentration (BAC). BAC data provide a highly desirable adjunct to self-reports of alcohol use and contribute to ‘triangulation’ of measurement of student drinking. To address concerns about relying exclusively on self-reports of drinking, we designed a study to obtain more objective information about student alcohol use among undergraduate students at the University of North Carolina - Chapel Hill (UNC-CH) during the fall 1997 term. To do this we adapted procedures that have been used to study drinking among the general nighttime driving population (Foss, Beirness & Sprattler, 1994), and among recreational boaters (Foss, Bartley & Smith, 2000). Research teams interviewed a random sample of individuals as they approached residences late at night (10 p.m. to 3 a.m.) on all

7

nights of the week. As part of the interview, a BAC measurement was obtained using a portable breath test device. A “social norms” program, informing students about the findings from the 1997 survey, was developed and initiated during the summer of 1999. A second BAC measurement survey was then conducted during the fall of 1999, providing an opportunity to evaluate the initial effects of the program (Foss, Marchetti & Holladay, 2000). In order to assess the long-term impact of the program on drinking norms and actual behavior, a third BAC measurement survey was conducted during the fall of 2002. This report describes the multi-year social norm program and presents the findings from all three waves of the survey to assess changes in directly measured and self-reported drinking from 1997 to 2002.

8

2 The UNC-CH Social Norms Program A social norms approach cannot be expected to produce substantial changes in student drinking behavior on a short term, or one-shot basis. Truly changing misperceptions about a social norm, which is part of the larger social environment on campus, requires a sustained long-term effort that is integrated with other programs and elements of campus life. To this end, a comprehensive social norms program was developed at UNC-CH. This program was built around the primary message: "Whether it’s Thursday, Friday, or Saturday night, 2 out of 3 UNC students return home with a .00 BAC." A multi-faceted, social marketing approach was used to deliver this message to students through several channels and to encourage them to remember it. The program became known, and is referred to throughout this document, as the “2 out of 3” program. The present project was undertaken to examine the long-term effects of a social norms marketing program aimed to reduce student alcohol use. The 2 out of 3 program was sustained from its inception in June 1999 through December 2002. During the 3½ years of the program, various materials were developed centered around the 2 out of 3 message. A variety of avenues and methods were used to deliver information about the reality of student alcohol use and to integrate normative information with other university functions. By the fall of 2002, most current students had been exposed to the social norms program throughout their entire student careers at UNC-CH. This provided ample time for misconceptions about student drinking (based on myth, rumor, conventional wisdom, media portrayals of college students, and other origins) to be counterbalanced by factual information from a reasonably credible source. That is, if the social norms marketing program is able to correct misperceptions, and those perceptions contribute significantly to students’ behavior, there should be measurable evidence of a change in students’ measured BACs by the fall of 2002.

9

Program Implementation To achieve the overall goal of reducing heavy drinking throughout the UNC-CH student population, the existing 2 out of 3 social norms program was continued for an additional three-year period. A variety of program activities were used to help incoming and returning students understand that excessive use of alcohol is atypical on the UNC-CH campus. Although posters were a key component of the 2 out of 3 program, efforts were made to develop a multi-faceted approach to integrate this single message into the routine functioning of the entire campus. The central tenets of this program were as follows: •

Continued Integration: Continue the program to communicate facts about the reality of student alcohol use, using creative approaches to maximize the likelihood that students attend to, remember, understand and internalize this message.



Early Exposure: Communicate accurate information about drinking norms at UNC-CH to incoming freshmen students and their parents before students arrive on campus.



Coordination: Coordinate the social norms program with other elements of the UNC-CH alcohol program to take advantage of the many program components that can convey the normative message.



Feedback: Pay special attention to student reactions and receptivity to the various approaches developed to communicate messages about what is normative concerning alcohol use on campus.

The following sections describe the various program activities designed to accomplish these objectives.

10

First Year Student Orientation Sessions (“CTOPS”) To reach students and parents before the beginning of the school year, project representatives participated in summer orientation sessions in 1999, 2000, 2001, and 2002. UNC-CH freshmen and transfer students were introduced to the 2 out of 3 program at these mandatory 3-day orientation sessions, commonly referred to as “CTOPS” (Carolina Testing and Orientation Program Sessions). Young project team members gave a featured 10-minute presentation at each CTOPS opening session.1 At these sessions the Vice Chancellor for Student Affairs welcomed incoming students and their parents, then introduced the representatives of the 2 out of 3 program to talk about a research project we had undertaken on the UNC-CH campus. Couching the presentation as research rather than “preaching about drinking” helped to set the tone for the entire 2 out of 3 program, which was simply to provide students with research-based facts. Central to the 2 out of 3 presentation was a short video showing interviews with actual UNCCH students. To create this video, students were approached on campus and asked if they would be willing to participate in an interview that would be used during CTOPS presentations. They were then asked what they thought were the biggest “party nights” at UNC-CH. Next, they were presented with the 2 out of 3 alcohol fact and asked for their honest opinion about it. Student reactions to these questions were then compiled to create the aforementioned video. In the spring of 2002, because the sound quality of the video was deteriorating, a new video was created following the same format.

1

During 2002 CTOPS was restructured to reduce its duration to 2 days. This resulted in the “2 out of 3” presentations being moved to a later session during the first day, which students attended alone rather than with parents. Although missing parents, these smaller sessions had the advantage of allowing direct interaction with students. This substantially increased the staff time needed to reach all students (32 vs. 10 sessions), but we generally felt that this format was preferable because of the opportunity to directly address students’ questions.

11

We began the 2 out of 3 presentation by showing the first section of the video where students talked about the big party nights at UNC-CH. The video was then stopped and the audience participated in a brief demonstration to illustrate the meaning of the program’s main message: “Whether it’s Thursday, Friday, or Saturday night, 2 out of 3 UNC students return home with a .00 BAC.” As individuals entered the hall where the opening session was held, each had been given a colored card. Two-thirds of these cards were blue; the remaining cards (one-third) were equally divided between pink and yellow. Although all cards were blank, the blue cards represented students who returned home on party nights with a zero BAC. The pink and yellow cards represented students who had non-zero BACs: the pink cards corresponding to those who had consumed 1 to 4 drinks, and the yellow cards corresponding to those who had consumed 5 or more drinks. During the interactive demonstration, individuals with a blue card were asked to stand up and look around to see how many others were standing; it was explained that those standing represented the proportion of students (2 out of 3) who returned home on party nights with a zero BAC. After those with the blue cards were asked to sit down, individuals with yellow and pink cards were asked to stand. It was explained that this represented the proportion of students who were drinking. Finally, those with yellow cards were asked to sit, leaving a rather small-looking group with pink cards standing. At this point, the students were told – and the strong visual representation emphasized – how few students actually have 5 or more drinks on traditional “party nights.” At the conclusion of this demonstration, students and their parents were told about cash incentives for displaying posters and for simply knowing the 2 out of 3 fact. A card was also distributed as a reminder of the 2 out of 3 fact.

12

Posters As with most social norms programs, posters featuring factual evidence about student drinking were the primary mechanism used to establish an enduring presence for accurate information about local student drinking norms. New 2 out of 3 posters were created each year. While developing the posters we solicited student feedback to determine how best to frame potential messages. Prototype materials were pre-tested with students to ensure the message was clear and was understood to mean what we intended. Using a new poster each year helped to keep the message delivery fresh, thereby reducing the possibility that students would tune out the information as outdated in subsequent years. Before students arrived on campus each fall, project staff members met with UNC-CH senior residence life staff to discuss the 2 out of 3 campaign and to solicit their assistance in distributing the posters to student rooms. Resident assistants (RAs) also received information cards with the posters to remind them about the BAC study, the purpose of the 2 out of 3 program and the distribution procedures. At the beginning of the fall 1999 and 2000 semesters, posters were distributed to the rooms of all 3,600 first-year students living in 34 campus residence halls, and one large private residence hall. For the fall 2001 and 2002 semesters, an attempt was made to distribute posters to all 8,200 undergraduate students living in residence halls.2

2

Although posters were provided to all RAs and Residence Life Directors, they did not reach every residence hall room as planned. Discussion with students during “prize patrol” trips (see below) revealed that certain residence hall staff did not distribute the posters. In these instances, we sent email reminders to RAs encouraging them to distribute the materials. Additionally, we suggested that these students obtain a poster from their residence life office or we sent posters via intercampus mail to students (if they desired).

13

Students were initially told about the “poster prize patrol” during their CTOPS orientation session. In addition, a card accompanied each poster explaining how putting the poster up could result in a cash reward. Each week throughout the academic year, project representatives visited randomly selected residence hall rooms to see whether the poster was displayed. If a poster was found displayed in a clearly visible location in the room, the residents of the room were given $50 to share; if they chose, their picture was taken and displayed on the 2 out of 3 web site among the list of winners. During spring semesters, the poster reward was increased to $100; students received an email message before the end of the fall term announcing this and reminding them to keep their posters displayed. A detailed record was maintained of the number of

Figure 1. Percent of residence hall rooms with the "2 out of 3" poster displayed, by year.

rooms visited and whether the

70%

poster was displayed to help

60%

track the extent to which the

50%

message was reaching students

40% 30%

by virtue of displayed posters.

20%

Data collected from prize patrol

10%

trips indicated that a substantial

0%

proportion of students had

1999-2000*

2000-01

2001-02

Fall 2002

* Freshman rooms only.

displayed the posters in their

rooms, as illustrated in Figure 1. The slight increase in displayed posters in 2001-02 probably reflects improvements in the distribution of posters that year.

Campus-Wide Zipper-Pull Promotion To better reach students not living in residence halls we also distributed “zipper pulls” with the 2 out of 3 logo on them. Attached to the zipper pull was an information card explaining the “alcohol fact” and indicating how students could obtain a small reward ($5) by displaying the zipper pull, as well as $1 for simply knowing the fact. The campus-wide zipper pull component of the program began in January 2001, and continued through the fall semester of 14

2002. Zipper pulls were distributed at various campus locations with heavy student traffic. The primary location for distribution was the area surrounding a popular student hangout, an outdoor area called the “Pit,” which is situated next to the main campus dining hall, undergraduate library, student bookstore, and student union. Beginning in the fall of 2001, we also included zipper pulls with the posters that were sent to residence halls. To gauge the extent to which

Figure 2. Proportion of backpacks observed with zipper pull attached, by semester.

zipper pulls were being worn, we observed their prevalence

8%

on backpacks (the commonly

7%

chosen method of display) on a

6%

few occasions (see Figure 2).

5% 4%

The proportion of students on

3%

campus displaying a zipper pull

2%

was small, but their prevalence

1% 0%

increased markedly from spring to fall semesters in 2001 as

Spring 2001 (n=1,270)

Fall 2001 (n=326)

Spring 2003 (n=190)

improvements were made in distribution.

Informational Web Site The 2 out of 3 web site (http://www.2outof3unc.org), was established in November, 2000 as a resource for students to read about the poster and zipper pull incentives, to learn about the alcohol study itself, and to see photos of weekly winners. This site also provided an email address to which questions about the BAC study or the 2 out of 3 program could be directed. The site address was included on all program materials.

15

For the spring 2001 semester, we created an additional incentive program via the web site: the “Find the Sign” contest. Every few weeks, students were given the opportunity to identify the location of a 2 out of 3 neon sign which was photographed and displayed somewhere on campus. The purpose of this contest was to draw traffic to the web site. Those who gave the correct answer were entered into a drawing for $50. Unfortunately, after a high number of initial hits for the Find the Sign contest, visits to the site dropped substantially. This incentive program was abandoned after one semester on the belief that we had already gotten the attention of most of those who could be reached this way.

Fall Fest Activity Booth At the beginning of each fall semester, UNC-CH hosts a non-alcoholic block party, known as “Fall Fest”, which takes place the first weekend students arrive on campus (preceding the start of classes). Fall Fest was designed to compete with the “traditional” alcohol-oriented parties on the same weekend. For this event, roads are closed to traffic from 9 p.m. until 2:00 a.m. and the central campus is transformed into a carnival-like street party for students. During each Fall Fest, a booth was set up displaying posters, the 2 out of 3 neon sign, zipper pulls and information about the BAC study. Fall Fest provided an excellent opportunity for program representatives to interact with students, distribute materials, and answer questions. During Fall Fest 2000, the 2 out of 3 display and sign was featured in the local NBC affiliate’s coverage of Fall Fest and the 2 out of 3 project manager was interviewed for the story.

16

The “BAC Guy” At Fall Fest 2002, we introduced a promotional character – the “BAC Guy”3 – drawing increased attention to the information booth. We developed this promotional character in conjunction with the university’s Department of Dramatic Art to draw attention to our efforts and to provide comic relief. Students in a costume design course designed and constructed the outfit as a class project. In addition to providing the character costume, this also infused the 2 out of 3 message into student class work. The final product was completed during the spring 2002 semester.

Curriculum Infusion Although a formal curriculum infusion program was well beyond the capabilities of this project, we took advantage of all opportunities to encourage and support curriculum infusion on a “guerrilla” basis. Guest lectures about the BAC research project were given in several graduate classes, with the hope that some of this would “trickle down” into classes taught by these students. Results from the BAC survey were provided to a few instructors who wanted to use this information to illustrate analytic and research design points using “meaningful” data to which students could easily relate. A member of the research team who was also a Sociology instructor devoted a full class session on research methods to the BAC Survey. He also discussed the study in several Sociological research methods classes. From May of 2000 through May of 2002, approximately 300 students in seven different classes participated in these discussions. Finally, we assisted a number of undergraduate students with class projects that focused on student drinking, the social norms program or the BAC survey.

3

Students had begun referring to the BAC interview teams as “the BAC guys.”

17

Coordination with the UNC-CH Alcohol Program The 2 out of 3 program was integrated with on-going activities of the university’s alcohol program in a number of ways. First, the program was featured prominently during student orientation sessions, replacing the brief alcohol education that had been provided in previous years. Second, project representatives conducted an hour-long workshop with residence hall/student life managers to familiarize them with social norms theory and its practical application. Third, project representatives attended eight meetings with residence hall advisors, during which program representatives presented an overview of the 2 out of 3 campaign, discussed the distribution of posters, and answered questions (e.g., dispelling myths). This helped reduce the likelihood of students receiving conflicting messages, as they may reduce the effectiveness of a social norms program (Perkins, 2002; NIAAA, 2002).

18

3 The BAC Survey The use of blood alcohol concentration data obtained from a representative sample of students distinguishes the UNC-CH project from the many other social norms programs that have been implemented on campuses around the United States. These data informed the development of the 2 out of 3 social norms program and aided in its evaluation. The details of the survey procedure are briefly described here. A more extensive description of the data collection procedure is available in Foss, Marchetti and Holladay (2000). Sampling Procedure Data were collected in the fall semesters of 1997, 1999, and 2002. Between early October and mid-November, teams worked approximately 20 nights during each wave of data collection. Interviewing began at 10 p.m. and ended at 3 a.m. Over the course of each survey period, all residences were visited repeatedly at different times of the week: both before and after midnight on weeknights (Monday-Wednesday), Thursdays, and weekends (Fridays & Saturdays). The sampling procedure was designed to obtain a representative sample of students who were returning home in the evening. To do this, the campus was divided into four geographically distinct routes along which interview teams walked. These routes passed by every residence hall and Greek house on or near campus. Three separate interview teams, consisting of a supervisor and three interviewers, worked on scheduled data collection nights. Starting locations on each interview route were randomly selected to ensure that no particular location was more likely than others to be used for interviewing. Teams moved continuously between locations, typically collecting data at a single location for no more than 10 - 15 minutes on a single visit.

19

Interviews were also conducted at five apartment complexes that, according to UNC-CH records, housed large concentrations of university students. At these locations, sampling and interviewing resembled techniques typically used in roadside surveys (Foss, Beirness & Sprattler, 1994).

Interview Procedure At each residence, interview teams randomly sampled individuals or groups who were either approaching or standing in front of the residence; those who were merely passing by were not eligible for selection. One member of the interview team approached selected groups/individuals and asked if they would do a brief interview. For groups of three or fewer, all who consented were interviewed. For larger groups, three individuals were randomly sampled. When possible, the team supervisor conducted brief interviews with remaining group members. Interviewers and their respondents stepped away from any other individuals in the area to maintain confidentiality. Interviews took approximately 5-6 minutes to complete and requested information about students’ activities during the night, drinking (where, when, what, how much), their perceptions about alcohol use among students, and their knowledge of the 2 out of 3 campaign. Copies of the survey instruments are available upon request. At the conclusion of the interview, respondents were asked to provide a breath sample, which was taken and analyzed using a portable breath test device, the Lion S-D2 Intoxilyzer™. Interviewers then showed respondents their BAC readings and provided a brief explanation of the risks associated with any non-zero BAC level. Persons who declined to be interviewed were asked if they would simply provide a breath sample and many did so. At apartment complexes, vehicles entering the parking lot were motioned to stop. Interviewers introduced themselves 20

and asked if anyone in the vehicle was a UNC-CH student. If a student was present, that person(s) as well as the driver (if a non-student) was interviewed while they remained in their vehicles.

Precautions for Students Potentially at Risk Although BAC surveys rarely encounter individuals who have consumed potentially dangerous amounts of alcohol, a protocol was established to deal with any person who might be a risk to him/herself or others by virtue of excessive alcohol consumption. Interviewers were trained to attend to several sources of information that would indicate if there was reason for concern about a student’s welfare, and especially whether there was any possibility of alcohol poisoning.4 Any time an interviewer had a concern, the team supervisor spoke with the student and assessed the situation to decide what, if any, protective action should be taken. Supervisors carried mobile telephones so they were able to contact other teams, the project supervisor, the UNC-CH Student Health Service, the local emergency department, and the campus or local police at any time. In addition, an administrator from the Office of the Dean of Students was always on call if supervisors needed assistance in determining how to proceed.

4

Since students were interviewed when returning home, rather than at a location away from home, there was little concern about the risk to them of driving after drinking. Nonetheless, those persons who registered a BAC above .05% were explicitly told that they should not even consider driving should they decide to go back out because it was both dangerous and illegal to do so.

21

4 Survey Results This section is organized into three main parts. The first part describes study participation, characteristics of the sample(s), and sample representativeness. The second part details awareness and believability of the 2 out of 3 campaign among UNC-CH students. The third part focuses on changes in student alcohol use over the course of the campaign. Results presented here are based on data obtained in face-to-face interviews conducted during the fall semesters in 1997, 1999, and 2002. The 2 out of 3 program was initiated in the summer of 1999. Accordingly, the 1997 data represent a baseline measurement, the fall 1999 data reflect possible early effects of the program, and the 2002 data represent enduring changes that may have occurred as the result of a sustained program.5 Except where noted, the Likelihood Ratio Chi-square (χ2) statistic was used to test for the statistical significance of distributional changes.

Study Participation and Sample Characteristics Study Participation Across the three data collection periods, 8,454 persons were sampled and asked to participate in the BAC survey. One hundred and three of these persons were younger than 18 and were ineligible to participate because of age restrictions. Participation rates are based on this sample of 8,351 individuals, and are presented in Table 1. Overall, 7,109 individuals (85%) agreed to participate to some degree, and 6,910 (83%) provided a breath sample (meaning that 97% of those who participated provided a breath sample). In total, 5,872 individuals (70%) completed the full interview and provided a breath sample. Individuals who declined the initial interview request were offered an alternative, briefer interview (which took 1 to 2

5

For the data presented in this report, values 10% and greater are rounded to the nearest percentage point. Those values under 10% are listed to the nearest decimal place to avoid loss of precision.

22

minutes); 757 individuals (9%) completed this abbreviated version and provided a breath sample. Table 1. Participation rates† by survey year.

Overall (N= 8,351)

Survey Year 1997 1999 2002 (N= 2,536) (N= 3,358) (N= 2,457)

Overall*

85.2%

86.0%

85.0%

84.1%

Breath Long form with breath sample Short form with breath sample Breath sample only Total Breath Participation

70.3 9.1 3.4 82.8

74.9 4.2 4.3 83.4

69.0 9.7 3.8 82.5

67.4 13.1 1.8 82.3

No Breath Long form, no breath sample Short form, no breath sample

2.0 0.4

2.4 0.2

2.2 0.3

1.2 0.5



The sample used to determine participation rates includes all individuals approached, except those under 18 years of age. * Overall rate may not equal the combined percentages due to rounding.

Table 1 also presents details of participation rates by survey year. Overall participation rates remained relatively stable across the study periods. There was a decrease in the proportion of individuals who completed the full interview from 1997 to 2002, with a corresponding increase in those completing the brief interview. This was due, at least in part, to changes in survey procedures across the three years.6 In general, students who declined participation did so simply because they were in a hurry. Students also cited the late time of night and being cold as reasons for not wanting to participate. Among the 16% who refused participation in 2002, 21% showed evidence of drinking – typically either the smell of alcohol or obvious lack of coordination.7 Although this assessment is based on interviewer observation and judgment, it is comparable to the prevalence of drinking found among those who did provide a breath measurement.

6

The ‘short-form’ option was introduced part way through the 1997 survey. In addition, mainly supervisors used the short form during the first two waves of the survey, whereas all interviewers were allowed to use it in 2002. Hence, it was more often available as an option during the later waves of data collection. 7 Reliable observations regarding evidence of alcohol consumption were not available prior to 2002.

23

Sample Characteristics Although sampled persons were interviewed regardless of their student status, results presented here represent only those 6,352 individuals interviewed during 1997, 1999 or 2002 who were explicitly identified as undergraduate students at UNC-CH, age 18 or older. A breath alcohol measurement was obtained for 6,108 (96%) of these individuals. Table 2 presents the distributions of selected characteristics of respondents for each wave of the survey. Table 2. Selected characteristics of respondents, by survey year. 1997 %

1999 (n)

%

2002 (n)

%

(n)

Age (in years) 18 19 20 21+

33.2 27.0 20.4 19.4

(587) (478) (361) (343)

30.1 32.7 20.8 16.4

(739) (805) (511) (404)

31.7 31.4 17.9 19.0

(578) (571) (326) (346)

Class Year Freshman Sophomore Junior Senior

34.9 27.2 21.4 16.5

(618) (481) (378) (292)

33.9 30.6 21.9 13.6

(835) (754) (540) (335)

37.3 30.5 16.0 16.2

(680) (557) (292) (296)

Sex Male Female

49.0 51.0

(865) (901)

52.4 47.6

(1,290) (1,173)

49.8 50.2

(909) (915)

Race White Black Other

84.4 10.2 5.4

(1,454) (175) (93)

82.1 12.2 5.7

(1,986) (295) (139)

80.0 12.7 7.3

(1,424) (226) (129)

Greek Status Member Non-member

26.2 73.8

(458) (1,291)

24.8 75.2

(606) (1,833)

23.8 76.2

(428) (1,370)

Residence Residence hall Greek house Off-campus

78.4 8.8 12.8

(1,298) (146) (218)

79.4 6.4 14.2

(1,922) (156) (352)

80.7 8.9 10.4

(1,423) (157) (195)

24

A majority of respondents were either 18 or 19 years old and nearly two-thirds were either freshmen or sophomores; the 2002 study sample had a somewhat greater proportion of freshmen and sophomores than the two previous samples. Approximately equal proportions of males and females participated each year, although slightly more males than females participated in the 1999 survey. More than 80% of all respondents were white and approximately 12% were black, with the remaining respondents representing other races. Each year of the study there was a slight decrease in the proportion of white respondents. About one-quarter of all respondents were members of Greek-letter organizations (fraternities & sororities) in each survey year. More than three-quarters of respondents lived in residence halls and about 8% lived in a Greek house. In sum, although there were slight variations by year, the demographic composition of the respondent samples remained relatively constant across the three waves of the survey.

Sample Representativeness The sampling procedure concentrated primarily on individuals returning home to residences on and near campus. Because nearly half of undergraduate students live off campus, the composition of the BAC study sample is somewhat different from that of the UNC-CH undergraduate student populations during the same years (Table 3). For example, younger students are over-represented in the BAC study samples. This was expected, given that younger students are more likely to live on or near campus. The overrepresentation of males may indicate that they are more likely than females to be outside late at night. Lastly, the overrepresentation of members of Greek organizations probably results from the sampling strategy. Each residence was visited with similar frequency and for approximately the same length of time; however, the number of individuals living at each residence varied. Since fewer individuals live in a typical Greek house than in any of the residence halls, members of Greek organizations were proportionately more likely to be sampled. In sum, the study samples differed somewhat from the undergraduate population in that respondents were slightly younger, and were more likely to be males and members of Greek organizations.

25

Table 3. Selected characteristics of BAC survey samples and corresponding UNC-CH undergraduate student populations, by survey year. 1997

1999

2002

BAC Study

Student Population

BAC Study

Student Population

BAC Study

Student Population

Age (years)† < 18 § 18 19 20 21+ Total N

-33.2% 27.0 20.4 19.4 (1,769)

(165) 20.1% 21.2 21.6 37.1 (15,156)

-30.1% 32.7 20.8 16.4 (2,459)

(422) 21.3% 23.2 22.8 32.7 (15,012)

-31.7% 31.4 17.9 19.0 (1,821)

(421) 20.8% 23.5 22.7 33.1 (15,540)

Class Year†† Freshman Sophomore Junior Senior Total N

34.9 27.2 21.4 16.5 (1,769)

23.5 21.6 26.4 28.5 (14,806)

33.9 30.6 21.9 13.6 (2,464)

23.1 23.5 25.7 27.8 (14,968)

37.3 30.5 16.0 16.2 (1,825)

22.6 24.8 24.9 27.6 (15,468)

Sex† Male Female Total N

49.0 51.0 (1,766)

39.9 60.1 (15,321)

52.4 47.6 (2,463)

39.4 60.6 (15,434)

49.8 50.2 (1,824)

40.4 59.6 (15,961)

Race† White Black Other Total N

84.4 10.2 5.4 (1,722)

81.6 11.2 7.2 (14,798)

82.1 12.2 5.7 (2,420)

81.2 11.2 7.6 (15,434)

80.0 12.7 7.3 (1,779)

77.8 11.1 11.1 (15,961)

Greek Status† Member Non-member Total N

26.2 73.8 (1,749)

18.1 81.9 (15,321)

24.8 75.2 (2,439)

14.0 86.0 (15,434)

23.8 76.2 (1,798)

11.9 88.1 (15,961)

Note. Percentages for each category may not total 100% due to rounding. § The number of students under the age of 18 is reported for the UNC-CH population only; these individuals were ineligible to participate in the BAC Survey. † Student demographic data were obtained from UNC-CH Office of Institutional Research. †† Class year information was obtained from UNC-CH registrar.

A more pertinent consideration is whether drinkers were properly represented in the study sample. The high response rate, which far exceeds that of most telephone or mailed questionnaire surveys, provides some assurance that the sample is a good representation of the population of interest. However, it is conceivable that sampling only persons who were 26

outdoors may have resulted in our missing drinkers, who might have remained inside for the entire evening. Similarly, the underrepresentation of students who lived off campus is a concern if their drinking patterns differ from on-campus residents. Information available from a self-report mailed questionnaire (the Core alcohol and drug survey) administered by the university at the same time as the 1997 BAC survey helps to address this issue. Both of these surveys included the following question about heavy drinking: "About how many times in the past two weeks have you had five or more drinks in a row?" In the Core survey, 46% of respondents indicated they had done this at least once. By comparison, 50% of respondents in the BAC survey indicated that they had consumed five or more drinks in a row during the past two weeks. The similar percentage of heavy drinkers increases our confidence that drinkers were adequately represented in the BAC survey.

Campaign Awareness, Understanding & Believability Campaign Awareness Campaign awareness was quite high among respondents from the beginning (see Figure 3). In 1999, after only a few months of program activities, 72% of respondents indicated that they had heard about the campaign, including

Figure 3. Campaign aw areness (heard of campaign), by survey year.

All Respondents Freshmen

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1999

92% of first year students

2002

(freshmen).8 By 2002, all class years had been exposed repeatedly to the 2 out of 3 program, and awareness had increased to 91% for all respondents (p < .001). Campaign awareness among first year students remained unchanged at 92% in 2002.

8

There are no data about campaign awareness from the 1997 baseline survey, which was conducted before the campaign was initiated.

27

Campaign Understanding Respondents who had heard

Figure 4. Campaign understanding, by survey year

of the campaign were asked what they thought the 2 out of 3 message meant (see Figure 4). In 1999, 72% of respondents understood the general meaning, including 80% of freshmen. By 2002,

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

All Respondents Freshmen

1999

correct understanding

2002

increased to 82% for all students (p < .001), and remained unchanged at 83% for freshmen.

Belief in Accuracy of the Alcohol Fact Respondents who had heard of the campaign and also understood its message were asked whether they believed the 2 out 3 fact accurately represented student drinking at UNC-CH (see Figure 5). The percent of respondents who believed the fact was accurate increased

Figure 5. Campaign believability, by survey year 60% All Respondents 50%

significantly from 35% in

40%

1999 to 45% in 2002 (p
.05%, compared to 14% in 2002 (p < .01; see Table 4). A small number (2.3%) of respondents who reported not drinking registered a positive BAC. Most (71%) of these individuals had a BAC below .05%.

29

Table 4. BAC distributions among all respondents, by survey year.

BAC Zero .002 - .049% .05 - .099% > .10%

1997 (N=1,716) 71.3% 10.0 8.5 18.6 10.1

1999 (N=2,389) 74.3% 10.5 9.1 15.1 6.0

2002 (N=1,791) 75.6% 10.0 7.2 7.2

14.4

Overall: LR P2 (2002 vs. 1997) = 12.85, df = 3, p < .01 Using .05% BAC threshold: LR P2 (2002 vs. 1997) = 11.47, df = 1, p < .01

Focusing on respondents who reported drinking on the night of the interview, there was a decrease in the level of likely impairment as measured by BAC (Table 5). Specifically, those registering a BAC of .05% or greater decreased from 60% in 1997 to 52% in 2002 (p < .05).

Table 5. BAC distributions among respondents who reported drinking on the night of the interview, by survey year.

BAC Zero .002 - .049% .05 - .099% > .10%

1997 (N=470) 10.4% 29.4 27.9 60.2 32.3

1999 (N=557) 14.0% 36.3 31.1 50.8 18.7

2002 (N=371) 13.5% 34.2 26.7 25.6

52.3

Overall: LR P2 (2002 vs. 1997) = 6.53, df = 3, p < .10 Using .05% BAC threshold: LR P2 (2002 vs. 1997) = 5.30, df = 1, p < .05

Heavier Drinking Self-report. Excessive alcohol use among college students has typically been measured as self-reported consumption of five or more drinks on at least one occasion during the past two weeks,9 a behavior often referred to as “binge” drinking. Substantial divergence of this behavior from the more commonly understood meaning of a “binge” has generated debate about the appropriateness of that term (Journal of Studies on Alcohol, 2003). Accordingly, we

9

We did not use the sex-specific measure of heavy episodic drinking that is now commonly used (> 5 for males, > 4 for females); this measure was not yet widely embraced when we began this study.

30

will simply use ‘heavy drinking’ to refer to self-reports of having 5 or more drinks on an occasion in this report. Using a heavy drinking criterion of two-week prevalence of five or more drinks on one occasion, approximately half of all respondents across the three study periods were heavy drinkers (Figure 7). The proportion of heavy drinkers in 2002 (45%) was significantly lower than in both 1997 (50%; p < .05) and 1999 (50%; p < .01). Frequent heavy drinking (i.e., consuming five or more drinks on three or more occasions in the past two weeks) is a particular cause for concern. These individuals are at higher risk for experiencing a variety of problems such as injuries, unsafe and unplanned sexual

Figure 7. Self-reported heavy drinking and frequent heavy drinking during the past tw o w eeks, by survey year; all respondents Heavy drinking 60% Frequent heavy drinking 50% 40% 30% 20% 10% 0% 1997

1999

2002

activity, and interpersonal problems (c.f., Wechsler et al., 1994). Looking across the three survey waves, approximately one of five UNC-CH respondents met the criterion for frequent heavy drinking. Once again, however, there were significantly fewer frequent heavy drinkers in 2002 (20%) than in 1997 (24%; p < .01) and 1999 (23%; p < .05). In the present study, we were able to determine whether respondents could be classified as heavy drinkers on the night of the interview based on the reported number of drinks consumed that evening. As shown in Table 6, there was a 30% decrease from 1997 to 2002 in the number of respondents who reported consuming five or more drinks on the night of the interview (p < .001). A similar pattern was revealed when using the gender-specific measure of heavy drinking (> 5 for males, > 4 for females)10, although the overall percentages are slightly higher because of the lower cut-point for females.

10

For these analyses it was possible to use the sex-specific “5+/4+” criterion for heavy episodic drinking since respondents were asked how many drinks they had consumed on the night they were interviewed.

31

Table 6. Self-reported heavy drinking on the night of the interview among all respondents, by survey year. 1997 (N=1,678) Heavy drinking measure Five or more 13.8% (males & females) 5 or more (males) 4 or more (females)

15.5

1999 (N=2,149)

2002 (N=1,520)

11.9%

9.6%

12.8

11.0

Proportionate Change (%)† p ’02 vs ’97 LR P2 -30.4

13.41

< .001

-29.0

13.89

< .001



Proportionate change was determined by dividing the difference in percentages between years by the baseline year.

A separate analysis was conducted focusing only on respondents who reported drinking on the night of the interview (Table 7). There was a 20% decrease from 1997 to 2002 in the number of respondents who reported consuming five or more drinks on the night of the interview (p < .01), and a 19% decrease when using the gender-specific measure of heavy drinking (p < .01). Table 7. Self-reported heavy drinking among respondents drinking on the night of the interview, by survey year. 1997 (N=475)

1999 (N=565)

2002 (N=375)

Proportionate Change (%)† p ’02 vs ’97 LR P2

Heavy drinking measure Five or more (males & females)

48.6%

45.1%

38.9%

-20.0

8.01

< .01

5 or more (males) 4 or more (females)

54.8

48.8

44.5

-18.8

8.76

< .01



Proportionate change was determined by dividing the difference in percentages between years by the baseline year.

Direct measurement of drinking. Of particular interest is the relationship of BAC measurements to self-reports of heavy drinking. Examining the BACs of those respondents who reported at least one heavy drinking occasion (5 or more drinks) in the past two weeks, 54% had a zero BAC on the night they were interviewed in 1997, and approximately 60% of these respondents had a zero BAC in both 1999 and 2002. That is, over half of the individuals 32

who would be labeled “binge drinkers,” by some standards, had no trace of alcohol in their system at the time of the interview. Not surprisingly, self-reported heavy drinking on the night of the interview was closely associated with respondents’ measured BACs. For all survey years combined, among those who met the gender-specific criterion for heavy drinking on the night of the interview, 83% had a BAC above .05% and a substantial proportion (42%) had a BAC of .10% or greater. On the other hand, a noteworthy percentage (17%) had a BAC of less than .05%, suggesting little or no impairment at the time of the interview despite having consumed at least 4 (females) or 5 (males) drinks.

Variations in Drinking by Time Drinking by time of week. As illustrated in Table 8, drinking was less common on weeknights (as evidenced by non-zero BACs) compared to the traditional “party” nights (all years combined: LR P2 = 269.33, p < .001). Although drinking was found to decrease between 1997 and 2002 on weekday nights, there was a small, nonsignificant decrease in drinking on party nights. Moreover, examining only those drinking on the night of the interview, there is no clear association of BAC with time of week. That is, although drinking is less common on weekday nights, those who do drink reach similar BAC levels regardless of time of week (data not shown).

33

Table 8. BAC distribution by time of week and survey year.

1997

Survey Year 1999

2002

LR P2

p

(‘02 vs. ‘97)

Weekday Nights (Sun-Wed) Zero .002 - .049% .05 - .099% > .10%

Thursday Nights Zero .002 - .049% .05 - .099% > .10%

Weekend Nights (Fri. & Sat.) Zero .002 - .049% .05 - .099% > .10%

85.5% 4.2 4.2 6.1 (n=592)

85.2% 4.4 5.9 4.4 (n=698)

90.4% 4.3 2.7 2.6 (n=582)

11.20

< .05

66.2 9.6 12.1 12.1 (n=355)

64.6 14.3 12.6 8.4 (n=593)

68.7 12.7 9.3 9.3 (n=387)

4.51

.21

62.8 14.7 10.1 12.4 (n=769)

72.7 12.2 9.4 5.7 (n=1,098)

68.4 12.8 9.4 9.5 (n=822)

6.19

.10

Note. Thursday, Friday and Saturday nights constitute “party nights.”

Drinking by time of night. Obtaining data in situ allowed us to examine how

Figure 8. Proportion of respondents with BAC above each of four thresholds by hour of night (2002 only)

drinking and, more

60%

importantly, BAC levels

50%

change as a function of time

40%

of night. Figure 8 indicates

30%

the proportion of respondents

20%

in 2002 with a BAC above

10%

each of four thresholds (zero,

> .00% > .05% > .10% > .15%

0% 10-11pm

11pm-12am

.05%, .10% and .15%), indicating the clear association of BAC with time of night.

34

12-1am

1-2am

2-3am

Table 9. BAC distribution by time of night and survey year.

1997

Survey Year 1999

2002

LR P2

p

(‘02 vs. ‘97)

Early (10 p.m. – 12:30 a.m.) Zero .002 - .049% .05 - .099% > .10%

Late (12:30 a.m. – 3 a.m.) Zero .002 - .049% .05 - .099% > .10%

84.9% 8.0 3.5 3.6 (n=1,050)

86.3% 8.0 4.4 1.4 (n=1,407)

85.1% 8.0 4.1 2.8 (n=1,115)

1.71

.64

50.0 13.2 16.4 20.4 (n=666)

57.2 14.1 16.0 12.7 (n=982)

59.9 13.3 12.3 14.5 (n=656)

16.71

< .001

Analyses of the changes from 1997 to 2002 in the BAC distributions of interviewed students reveal that there was little if any change in the hours before midnight (Table 9). On the other hand, the overall BAC distribution clearly shifted downward among students interviewed during the second interview shift (from 12:30 a.m. to 3 a.m.), with 10% more having a zero BAC and 10% fewer having BACs above .05%.

Summary of Changes in All Drinking Measures Table 10 presents a summary of changes in the several measures of student drinking reported above, indicating the proportionate change between each of the survey waves. Considering the overall pattern of changes, two things are readily apparent. First, the direct BAC measurements declined quickly. More generally, it appears that the more precise nightspecific self-report measures declined more markedly by 1999 than the less precise retrospective measures. Second, virtually every measure declined between successive survey waves. In addition, it appears that changes in drinking reflect both fewer drinking occasions and less drinking on those occasions when students do drink. The overall decline in nightspecific measures of heavier drinking could indicate either fewer drinking occasions or less consumption on drinking occasions (or both). The decline in measures of heavy drinking 35

among drinkers clearly indicates that there is less consumption on drinking occasions, which themselves are less frequent.

Table 10. Summary of changes in all measures of drinking. 1997 vs. 1999 % p change† Night-specific measurement BAC – all respondents > .00% > .05% BAC – drinkers only > .05% Self-report – all respondents Any drinking Heavy drinking (5+) Heavy drinking (5/4) Self-report – drinkers only Heavy drinking (5+) Heavy drinking (5/4) Retrospective Accounts Self-report – all respondents Heavy drinking past 2 weeks Frequent heavy drinking past 2 weeks †

1999 vs. 2002 % change† p

1997 vs. 2002 % change† p

- 10 - 19

< .05 < .01

-5 -5

.35 .48

- 15 - 23

< .01 < .01

- 16

< .01

+3

.45

- 13

< .05

-7 - 14 - 17

.18 .08 < .05

-6 - 19 - 14

.27 < .05 .09

- 13 - 30 - 29

< .05 < .001 < .001

-7 - 11

.26 .06

-14 -9

.06 .19

- 20 - 19

< .01 < .01

+1 -2

.81 .69

-9 -15

< .01 < .05

-9 - 17

< .05 < .01

% change was determined by dividing the difference in percent drinking between years by the baseline year.

Comparison to National Trends To ensure that changes in alcohol use were not simply the result of decreases in drinking among college students more generally, it is important to examine trends in student alcohol use at other campuses. Although there are no longitudinal BAC data from other campuses, we can compare some of the standard self-report items from the UNC-CH survey with those on national surveys. The Harvard College Alcohol Survey (CAS) indicates that from 1997 to 2001 there were small increases in student reports of heavy and frequent heavy drinking (Wechsler et al, 2002). By comparison, we found substantial decreases in both these measures over essentially the same time period (1997 - 2002).

36

Figure 9 illustrates the different trajectories of the clearest indicator of

Figure 9. Trends in student self-reports of frequent heavy drinking, UNC-CH vs. national survey data 30%

problematic student

Harvard CAS

consumption – frequent heavy

UNC - CH

drinking – nationally and at

25%

UNC-CH. Although these national data do not constitute

20%

an ideal control group, they do provide assurance that the declines in drinking at UNC-

0% 15%

1997

1998

1999

2000

2001

2002

CH do not simply reflect a general downward trend among college students. Although data were not reported by region for the 2001 CAS, reported heavy drinking by students in the South was nearly identical to the national rates in 1997 and 1999, suggesting that the national rates provide a reasonable comparison group for a southern university.

37

5 Discussion & Conclusions It is clear that student drinking declined in conjunction with the 2 out of 3 social norms program. By every measure, whether self-report or direct BAC measurement, there was significantly less drinking during the fall of 2002 than during the baseline year of 1997. Moreover, the program appears to have affected drinking occasions, as well as the amount consumed among those who had been drinking, and heavy drinking appears to have been affected more than light to moderate drinking. Although the overall focus of the analyses was on the total change in alcohol use from the 1997 baseline to the 2002 follow-up, inspecting the interim changes in 1999 reveals an interesting pattern. The more direct measures of drinking showed relatively quick changes in response to the program. By fall 1999 fewer students had been drinking on the night they were interviewed, fewer students had a high BAC (> .05%), and among those who were drinking the proportion with a high BAC was lower. Self-reports of heavy drinking (> 5 for males, > 4 for females) the night of the interview also had declined by 1999. On the other hand, the more general measures of heavy drinking typically used in research on college student drinking (i.e., retrospective accounts of heavy drinking or frequent heavy drinking over a two-week period) showed no change. However, by 2002 all measures showed a statistically significant decrease, with many reflecting substantial changes of 20% or more from their baseline values. The rapid decline in some measures of drinking (by 1999) might have occurred in response to the substantial visibility of the program, and perhaps its novelty as well, during the initial months. The high degree of program awareness among students – even among upperclassmen, who initially were a secondary target group – indicates that such an early effect was possible. The fact that the initial changes were sustained and some

38

other measures continued to decline throughout the program period is encouraging and indicates that the early effects were not simply the fairly common short-term effect that many programs can have, only to lose their gains over time.

Benefits of Using BAC Data – Triangulation of Measurement Blood alcohol concentration measurements are precise, time-specific measures of alcohol use that provide more direct information about an individual’s level of alcohol impairment than do retrospective self-reports about number of drinks consumed. Accordingly, BAC data can add unique detail to our understanding of student drinking, providing a richer picture than can be obtained only from self-report information and data concerning relatively rare incidents (e.g., alcohol-related deaths, injuries or contacts with authorities). As noted above, several self-reports of alcohol use did not change from 1997 to 1999, while there were pronounced declines in measured BACs. When comparing 1997 and 2002 data, both self-report and direct measures of alcohol use significantly declined. This suggests that self-report data may be somewhat insensitive to initial changes in drinking behavior, although it is not clear why this would be the case. It may be that the size or alcoholic content of drinks declined such that the number of “drinks” was the same, but the resulting consumption of alcohol nonetheless decreased; or perhaps drinks were consumed over a longer period of time than before, resulting in lower BACs. More likely is that retrospective reports of drinking – especially with regard to a measure such as five or more drinks on an occasion in the past two weeks – are relatively insensitive because they are so imprecise. Whatever the explanation, it is clear that heavy drinking, as measured by BAC values, declined and this was not reflected in most of the 1999 self-report data.

Intensive Program Implementation Along with evidence documenting the effectiveness of social norm programs, there are also reports of social norm programs that have failed to produce the expected effects (Mosher et al., 2002; Wechsler et al., 2003; Werch et al., 2000). Consequently the question is why some programs work and others do not. That is not easily answered, but presently there does appear to be one difference between more and less successful

39

programs: fidelity of implementation. In a corollary to this, the recent Wechsler et al. (2003) study failed to assess the quality of the norms programs upon which they based their assessment of the effects of such programs. Those programs that have been documented to produce changes in reported student drinking have been multi-year efforts that use a variety of tactics to disseminate accurate normative information to entire student populations. That was also a significant feature of the UNC-CH program. The 2 out of 3 program relied on a multi-faceted approach to reach all campus members with the message that excessive alcohol use is not the norm on the UNC-CH campus, even on traditional “party” nights. This comprehensive approach to reach students through multiple channels with mutually supportive information – and to induce student discussions of the information – was sustained for several years and was integrated with other programs and elements of campus life. The responses of students interviewed during the fall 1999 semester, a few months after the program was initiated, indicated an extremely high recognition rate. Fully 92% of first year students and 72% of all students had heard of the program. Also, a majority of the students (80% of first year students and 72% among all respondents) understood the message the campaign was attempting to convey. Both the degree of recognition and the extent of understanding are quite high, especially since the program had been in place for a relatively short period of time. We believe the main reasons for the success in getting this information into the student population resulted from the way the program was structured and implemented, with integrated mutually supportive elements. This degree of message recognition and comprehension is exceptional for any public information campaign. We believe that the effort devoted to crafting a clear and compelling single statement to convey the notion that drinking is overestimated was important. The steps taken to present that message in a credible manner may also have paid dividends. Including of a video of candid comments from real students about the validity of the message in presentations to incoming students was intended to enhance its credibility. The incentive campaigns in which students were rewarded for knowing or

40

displaying the message appears to have been a low cost, highly effective way to ensure that the program message was ubiquitous on campus. The quick spread of program awareness beyond the groups initially targeted suggests that a good deal of informal discussion of the program took place among students, which was part of the overall program strategy. The very high degree of program recognition and understanding would not likely have been achieved with a simple program that merely distributed posters around campus to compete with the thousands of other visual messages. Hence, although social norm programs may seem relatively easy to develop and implement, with modest commitments of staff time and financial resources, such an approach can probably not be expected to produce the effects that the more successful norms programs have achieved.

Non-Believers Student reluctance to believe accurate normative information is common to social norm programs. Although the high visibility of the BAC data collection may have contributed to believability of the 2 out of 3 fact, there remained a substantial degree of skepticism even three years after the program was initiated. This skepticism is to be expected when a program seeks to overturn deeply held “cultural truisms.” Nonetheless, efforts to counter reasons for disbelief are possible and are important to pursue. We made a number of efforts to do that during the present program, but these clearly did not have a major effect. The increase in belief from 1999 to 2002 may have resulted partly from these efforts, but more likely this reflects a changing recognition of actual drinking norms as the 2 out of 3 message became a permanent, pervasive – although low-key – fixture of campus life.

Attributing Changes to the Program Without a control group, it is impossible to conclude with certainty that changes in drinking among UNC-CH students were directly the result of the 2 out of 3 campaign. However, some of the more obvious alternative explanations can be ruled out. First, although the sample did differ somewhat from the university population as a whole, the

41

demographic composition of the survey samples remained relatively constant across the three waves of the survey. Thus, the measured changes in drinking cannot be attributed to changes in the composition of the survey samples. Second, both national survey data and data from the southern region of the U.S. indicate little change in student drinking from 1997 to 2001. Consequently, it is unlikely that the declines at UNC-CH simply represent a general decrease in student drinking independent of the program. Finally, there were no other major university initiatives focused on student drinking that might have produced or contributed to changes. Indeed, in a report to the University Board of Governors the Vice Chancellor for Student Affairs attributed a dramatic decrease in residence hall damage in 1999-2000 to the 2 out of 3 program. There was one other alcohol-oriented effort on campus that partially overlapped with the 2 out of 3 program. This was sponsored by the Center for Science and Public Interest and was designed to encourage students to advocate for changes in alcohol policies at UNC and another campus in the northeastern U.S. During 1998 a few ads, developed by an outof-state agency, appeared in the student newspaper. After that the program had little visibility on campus, probably due to internal problems with the other campus and the need to change advertising agencies. The only information the program provided about UNC-CH specific alcohol use was information from the 2 out of 3 program. Among the thousands of interviews we conducted with UNC-CH students, few if any ever mentioned that project. It seems unlikely that any changes in student drinking resulted from that modest effort.

Representativeness of BAC Survey Results Although extensive efforts were made to ensure that the interview sample was representative of the general student campus population, it is important to keep in mind that we were not entirely successful in that regard. The sampling procedure concentrated primarily on individuals returning home in the evening to residences on and near campus. As a result, younger students, males and members of Greek organizations were somewhat overrepresented (see Chapter 4). Studies have consistently shown that males and Greek members consume more alcohol than their respective counterparts. In fact, when the 1997

42

BAC survey data are weighted to reflect the proportion of males and Greek members found in the overall student population, the rate of reported heavy drinking in the past two weeks decreases slightly, from 50% to 48%. Thus, if anything, the data reported here probably over-represent UNC-CH student drinking somewhat, although it appears the effect is not substantial. The similarity of the 1997 BAC survey results to a mailed questionnaire (Core) survey conducted that same year, for which respondents more closely resemble the demographic characteristics of the overall student population, provides additional confirmation that the BAC survey is fairly representative of drinking by the undergraduate student population.

Conclusion The findings reported here provide support for the effectiveness of the 2 out of 3 campaign in reducing student alcohol use on the UNC-CH campus. The high degree of awareness and understanding of the 2 out of 3 message among UNC-CH students indicates that the social norms program was successful in reaching its target audience. The increased believability of the campaign suggests that the primary message, which specifies that on “party” nights most students drink little, if at all, is becoming more accepted as the reality of student drinking on campus. Finally, the decreases observed in both self-report and direct measures of student drinking suggest that social norm marketing programs can affect actual drinking behavior and not merely self-reports of drinking. One limitation of this study was the lack of a control group, which would have helped rule out alternate explanations for our findings. However, evidence from national surveys that indicate stable or increased drinking among students provides reason to believe that the decrease in drinking among UNC-CH students is not simply the reflection of a general downward trend in college student drinking.

43

REFERENCES Bandura A (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall. Foss RD (2001). Blood alcohol concentration levels of college student "binge" drinkers. Presented at annual meeting of the American Public Health Association, Atlanta GA, (October). Foss RD, Bartley CL, Smith GS (2000). Alcohol and boating-related fatalities in North Carolina. Proceedings of the 15th International Conference on Alcohol, Drugs, and Traffic Safety (CD ROM). Borlänge, Sweden: Swedish National Road Administration. http://www.vv.se/traf_sak/t2000/312.pdf Foss RD, Beirness DJ, & Sprattler K (1994). Seat belt use among drinking drivers in Minnesota. American Journal of Public Health;88(11): 1732-1737. Foss RD, Marchetti LJ & Holladay KA (2000). Development and evaluation of a comprehensive program to reduce drinking and impaired driving among college students (Final report). Washington: National Highway Traffic Safety Administration; DOT HS 809 396. http://www.nhtsa.dot.gov/people/injury/research/ReduceCollege/ Haines M (1996). Social norms approach to preventing binge drinking at colleges and universities. (Pub. No. ED/OPE/96-18). Newton, MA: U.S. Department of Education. Haines M & Spear AF (1996). Changing the perception of the norm: A strategy to decrease binge dinking among college students. Journal of American College Health;45: 134-140. Hansen WB (1993). School-based alcohol prevention programs. Alcohol Health and Research World;17(1):54-60. Harrington NT & Leitenberg H (1994). Relationship between alcohol consumption and victim behaviors immediately preceding sexual aggression by an acquaintance. Violence and Victims;9(4):315-324. Hingson R & Howland J (1993). Alcohol and non-traffic unintended injuries. Addiction;88:877-883.

44

Hingson R, Heeren T, Levenson S, Jamanka A & Voas R (2002). Age of drinking onset, driving after drinking, and involvement in alcohol related motor-vehicle crashes. Accident Analysis and Prevention;34(1):85-92. 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, AZ: University of Arizona Campus Health Service. Journal of Studies on Alcohol (JSA) (2003). Guidance for authors on the policy of the Journal of Studies on Alcohol regarding the appropriate use of the term ‘binge’. Retrieved on June 18, 2003, from: http://www.rci.rutgers.edu/~cas2/journal/Binge.html Lange JE & Voas RB (2001). Defining binge drinking quantities through resulting blood alcohol concentrations. Psychology of Addictive Behaviors;15(4):310-316. Midanik LT (1988). Validity of self-reported alcohol use: A literature review and assessment. British Journal of Addiction;83(9):1019–1030. Mosher KL, Berkley-Patton JY, Towns C, McCluskey-Fawcett K & Hillman K (2002). Seeing is not believing for college students: A report on a four-year social norms media project to reduce alcohol use. Poster presentation at the 130th Annual APHA Conference Meeting, Philadelphia, PA; Social Norms Media Project: Lawrence, KS; Abstract 42757. National Highway Traffic Safety Administration (NHSTA) (1999). Traffic safety facts, 1998. Washington, DC: National Center for Statistics and Analysis. National Institute on Alcohol Abuse and Alcoholism (NIAAA) (2002). A call to action: Changing the culture of drinking at U.S. colleges. Final report prepared by the Task Force on College Drinking for the National Advisory Council (NAC) to the National Institute on Alcohol Abuse and Alcoholism (NIAAA). (NIH Pub. No: 02-5010). National Social Norms Resource Center (2003). Retrieved on August 15th, 2003, from: http://www.socialnorm.org O’Malley PM & Johnston LD (2002). Epidemiology of alcohol and other drug use among American college students. Journal of Studies on Alcohol Supplement;14:23-39.

45

Perkins HW (2002). Social norms and the prevention of alcohol misuse in collegiate contexts. Journal of Studies on Alcohol Supplement;14:164-172. Perkins HW (2003). The social norms approach to preventing school and college age substance abuse: A handbook for educators, counselors, and clinicians. San Francisco: Jossey-Bass. Perkins HW & Berkowitz AD (1986). Perceiving the community norms of alcohol use among students: Some research implications for campus alcohol education programming. International Journal of Addiction;21:961-976. Perkins H, DeJong W & Linkenbach J (2001). Estimated blood alcohol levels reached by ‘binge' and ‘non-binge' drinkers: A survey of young adults in Montana. Psychology of Addictive Behaviors;15(4):317-320. Presley CA, Meilman PW & Lyerla R (1993). Alcohol and drugs on American college campuses: Use, consequences, and perceptions of the campus environment. Carbondale, IL: Core Institute. Roizen, J (1997). Epidemiological issues in alcohol-violence. In Recent developments in alcoholism, vol. 13: Alcoholism and violence. Marc Galanter (Ed.). New York: Plenum Press, pp.7-40. Rosenthal R (1966). Experimenter effects in behavioral research. New York: AppletonCentury-Crofts. Sleet DA, Wagenaar AC & Waller PF (1989). Introduction: Drinking, driving and health promotion. Health Education Quarterly;16:329-333. Wechsler H, Dowdall GW, Davenport AE & Castillo S (1995). Correlates of college student binge drinking. American Journal of Public Health;85:921-926. Wechsler H, Lee JE, Kuo M, Seibring M, Nelson TF & Lee H (2002). 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. Journal of American College Health;50(5):203-217. Wechsler H, Nelson TF, Lee JE, Seibring M, Lewis C & Keeling RP (2003). Perception and reality: A national evaluation of social norms marketing interventions to reduce college students’ heavy alcohol use. Journal of Studies on Alcohol;64(4):484-494.

46

Werch CE, Pappas DM, Carlson JM, DiClemente CC, Chally PS & Sinder JA (2000). Results of a social norm intervention to prevent binge drinking among first-year residential college students. Journal of American College Health;49(2):85-92.

47

Suggest Documents