Research on social norms approaches to drinking on college campuses indicate that they result in a reduction in student drinking (Haines, 1996; Johannessen et al., 1999). However, because of the mechanism by which the social norm approach produces effects, these changes tend to be gradual. Over a six year period, Haines (1996) found roughly a 3% yearly decrease in self-reported "binge" drinking among students at Northern Illinois University. Since we expected only a small change during the first year of the program, we designed the follow-up survey to examine student awareness and understanding of the program, as well as the ultimate criterion of drinking.
The"2 out of 3" program is designed to become a permanent part of the ongoing alcohol awareness program at UNC-CH, thus allowing sufficient time to produce the desired behavioral changes. While outside funding for the project was completed in March, 2000, the program will continue as part of UNC-CH’s ongoing alcohol programs for the 2000-20001 academic year. Figure 4.1 presents the timeline of the project to date.
During the fall of 1999, the BAC survey was repeated using the same procedures employed for the original 1997 survey. This was conducted at the same time of the semester as the 1997 survey to minimize effects due to possible changes in drinking throughout the course of the semester. As a result of somewhat better weather, and because of experience with the original survey, we were able to obtain a greater number of interviews during the same time period. Data were weighted to adjust for oversampling of Thursday, Friday and Saturday nights.
As can be seen in Table 4.1, the basic demographic make-up of the 1999 sample was quite similar to that in 1997, which is to be expected given that the same procedures were used. The one noteworthy difference is that males were a slight majority in the 1999 sample, with 4% more than in the 1997 sample. The 1999 sample was also somewhat younger, with 3% more respondents under the legal drinking age.
Because a program to promote accurate understanding of campus drinking norms would not be expected to have a dramatic, immediate effect on drinking behavior, the issues of primary interest in this survey concern how extensively the message reached the target population and how well it was understood. Unlike most messages, a norms message is not prescriptive. That is, rather than directly telling individuals what they ought to do, a norms message simply helps to correct a particularly important perception, that itself sends an indirect prescriptive message. Hence, even if a large proportion of the target audience has heard, and remembers a normative message, no measurable effect is likely unless the implication of that information is recognized by individuals. Accordingly, we asked several questions during the interview about student awareness of the "2 out of 3" information, how they had learned about it, what they thought it meant and whether they believed it (a copy of the questionnaire is included in Appendix 4.a.).
Among 2,535 students interviewed, 2,279 were asked whether they had heard of the "2 out of 3" program.8 Well over two-thirds of those interviewed (71%) had heard of the "2 out of 3" campaign. The primary initial target of the program was first-year students. In particular, the two program elements designed to directly reach individual (the orientation presentation and posters distributed to individual rooms) focused on first year students. Other elements targeted the campus community generally rather than individuals. Among those who identified themselves as freshmen, 92% had heard of the program. This compares with 60% of Sophomores, Juniors and Seniors (p < .01).
Those persons who indicated they knew about the "2 out of 3" program were asked where they had learned about it. Their responses were coded into several categories shown in Table 4.3. Because first year students were a primary target, responses are broken out separately for freshmen. Respondents were prompted to mention all sources through which they had learned about the program rather than only one, so these categories are not mutually exclusive.
The program was designed to reach the student population through multiple, mutually supportive channels with the hope that this information would take on a more normative "feel" than a message that comes only from a single source. Accordingly, it is interesting to look at these reported information sources in some detail. Table 4.3 shows the percent of persons who mentioned each of the various channels by which students learned about the "2 out of 3" fact.
*Columns do not sum to 100% because individuals cited multiple sources.
**The "prize patrol" refers to the project staff members who visited campus looking for posters and stickers in exchange for the promised rewards. When talking with a group about the stickers and the "2 out of 3" fact, they became a source of information about the program for those who had not yet heard about it.
*** Fall Fest is the campus-wide celebration held on the Sunday night before fall classes begin. "2 out of 3" stickers were handed out by project staff at this event as one way of reaching students, other than freshmen, early in the semester.
It appears that there were five productive channels for message distribution. Seventy-two percent of first year students recalled the "2 out of 3" message from their orientation session. Given that all first year students are required to attend an orientation session, it is somewhat surprising that only about three-quarters reported hearing the "2 out of 3" message there. Part of the slippage is probably due to inattentiveness during the session; another part probably results from the fact that a number of students were late to the opening orientation session where the "2 out of 3" message was presented early in the program. The 3% of older students who reported hearing the "2 out of 3" information at orientation probably results from their attendance at an orientation session that was held for transfer students.
Nearly two-thirds (63%) of first year students reported seeing a "2 out of 3" poster. This was precisely the purpose of first making it available to as many freshman students as possible by putting it on the bed in their residence hall room before they checked in, then using the prospect of a financial reward to induce them to actually put it up in their room (so they and others would see it). The fact that 32% of students who would not have had a poster themselves saw one attests to the success of this mechanism for getting the word out. Reports of seeing the poster clearly indicate the age-grouping of students on campus. Whereas 39% of Sophomores reported learning about the "2 out of 3" program from a poster, that declined to 25% of Juniors & Seniors (p < .01).
Another indication of the extent to which the poster was displayed in student rooms comes from records kept by the "Prize Patrol" when they visited randomly selected rooms each week to hand out $50 to persons who had the poster displayed in their room. Since rooms were visited randomly, it is possible to estimate the proportion that had a poster displayed from the number of rooms visited before a poster was found. For example, if all rooms had a poster, then the first room visited would always receive an award. If 50% of rooms had a poster displayed then, on average, two rooms would have to be visited before finding a poster. During 7 of the first 8 weeks of the program, the very first room visited had displayed the poster, suggesting that nearly 90% of freshman residence hall rooms had the poster displayed. During the next 8 weeks, the display rate declined to around 42%. During the spring semester more than a third of rooms (37%) continued to display the poster.9
It is surprising how few first year students reported seeing anything about the program in the student newspaper. Whereas nearly four in ten (39%) older students reported learning about the program from the newspaper, only 6% of freshmen mentioned that source of information (p < .01). It may be that with so many other more salient sources of information, first year students simply forgot that they had also seen something in the newspaper. Alternatively, it may be that this group of students had not yet gotten into the habit of reading the newspaper regularly -- only three ads were run throughout the semester.
An interesting finding that again points to the effectiveness of the efforts to generate discussion about the program through various mechanisms is that nearly one in five (18%) Sophomore, Junior and Senior students mentioned hearing about the program from a friend. This clearly suggests that the information was delivered in a non-passive fashion. Program elements appear to have stirred up discussion among the target group. This is critical for a norms campaign, because social norms are embedded in the everyday lives of individuals; they don’t reside in formal messages, emerging instead from social interaction. Evidence in the present case seems to indicate that we have made steps in the direction of the amount of student drinking being a topic of discussion. Moreover, that discussion now revolves, in part, around the "2 out of 3" fact that we discovered, rather than only the myths that existed previously.
As mentioned earlier, it is important for persons to understand the implication of the simple factual information that is at the core of a social norms program. To assess this, during the BAC survey we asked respondents who had heard about the "2 out of 3" program "What do you think the message of the campaign means?" Responses were coded into four categories: Understood the point of the fact (e.g., that drinking is less common than people think, there is less pressure to drink etc.), misunderstood it (e.g., drinking is dangerous or illegal, the university is cracking down on drinkers), didn’t know, or simply responded that they didn’t believe it without explaining what they thought it meant. Seventy percent of respondents were coded as having understood the general point that drinking norms are different (or associated pressures to drink are less) than are typically thought to be the case on the UNC-CH campus. Reflecting their greater exposure to the point of the message through the orientation sessions, and perhaps greater discussion of the fact with parents and/or friends as a result, freshmen were more likely than older students to understand this point (78% vs. 64%, p < .01).
Heavy media coverage of alcohol-related tragedies typically lead readers to believe that drinking and driving is more common than is actually the case. As a result, it is not unusual for individuals to view the results of BAC surveys with some skepticism. We expected the student BAC survey results to be met with skepticism as well. When initially told of our findings about student drinking, nearly everyone – media representatives, other researchers, members of the general public, individual students – was surprised, if not disbelieving. Many questioned the methodology until learning the care with which the study was conducted. Hence, we were not surprised to learn that a substantial proportion (54%) of students did not believe "2 out of 3" accurately represented student alcohol use at UNC-CH. Another 11% were unsure, leaving only 35% who thought the "2 out of 3" fact was accurate.
Interestingly, Seniors were more likely to believe "2 out of 3" (47% vs. only 32% of other classes, p < .01). This is despite the fact that Seniors in this sample reported as much drinking and had similar measured BACs similar to those from other class years. Hence, this difference in belief about campus drinking norms may reflect a maturing of students who have spent three years in college, with the somewhat greater perspective that brings.
We were not surprised by the substantial proportion of respondents who expressed skepticism that two-thirds of UNC-CH students have little or nothing to drink on traditional "party nights." Nonetheless, it is a concern that needs to be addressed. Before attempting to deal with this disbelief, we examined characteristics of those persons who were most likely to indicate they didn’t believe the "2 out of 3" fact.
Figure 4.1
Non-Belief in Accuracy of "2 out of 3" by Number of Drinking
and Heavy Drinking Nights in Past Two Weeks
Those who did not believe that "2 out of 3" is an accurate representation of UNC-CH are the more extreme drinkers (not merely drinkers or occasional heavy drinkers). Belief in the accuracy of "2 out of 3" is related to number of drinking nights and heavy drinking (> 5 drinks) nights during the past two weeks, as shown in figure 4.1. In addition, persons who had consumed the greatest amount of alcohol (measured by multiplying the number of drinking occasions in the past two weeks by the typical amount consumed) were less likely to believe the accuracy of "2 out of 3." Among those in the upper quartile of alcohol consumption, 73% were non-believers vs. 49% of those who drank less or not at all in the past two weeks (p < .01). Looking at measured drinking and acceptance of the "2 out of 3" fact, those who had a BAC above .08 when we interviewed them were more skeptical than those with a lower or .00 BAC (68% vs. 53%, p < .02). Also of interest is that in every analysis, those persons who were in the lighter drinking category – whether measured as amount, number of times, or simply being in a category that drinks less – were more likely to indicate that they were unsure whether "2 out of 3" was accurate rather than stating definitively that the believed it was true or not.
Another indication of how those who believed that the fact is accurate are different from those who don’t is that non-believers were more likely to have begun drinking at an early age (15 or younger). Whereas 67% of those who began drinking by age 15 did not believe "2 out of 3" was accurate, only 52% of those who started drinking at a later age found this fact hard to believe (p < .01). Among those who did not begin drinking until they had reached college age (18) only 46% did not believe the fact was accurate.
A particularly interesting finding is shown in Figure 4.2. When we compared the BAC distributions on weekend nights (Thursday - Saturday) for persons who did not believe "2 out of 3" was accurate, 68% of them had a .00 BAC when we interviewed them. That is, as a group, the non-believers evinced the very thing they found difficult to believe.
Figure 4.2
BAC distributions comparing those who believed
the "2 out of 3" fact was accurate and those who did not.
In an effort to begin addressing student disbelief about the veracity of "2 out of 3," we created a full page newspaper ad to provide feedback to students about findings from the 1999 survey (see Figure 4.3). In addition to detailing the survey method (e.g., time and place of interviews, number of interviews) and characteristics of persons interviewed (e.g., age, percent male vs. female, percent Greek vs. non-Greek) this ad highlighted the basic BAC distribution for party nights. It also included a mention of the "strange but true" fact that among non-believers, the BAC distribution was almost exactly "2 out of 3" with no alcohol the night they were interviewed. Unfortunately, since this ad was based on the second (1999) survey data, we have no information on what effect it may have had on students’ beliefs.
Although it is not possible to clearly attribute changes in student behaviors between 1997 and 1999 to any intervention, since we do not have a control group, it is instructive to see whether changes have occurred. Internal analyses of the data may help to isolate reasons for any changes that are detected.
Table 4.4 shows the proportion of students found to have a positive BAC as well as BACs above .08 and .15. It is clear that student drinking declined from the fall of 1997 to the fall of 1999. It appears that heavier drinking declined more than the act of drinking. The decrease in very high BACs (> .15) is not statistically significant, although the proportion of respondents with BACs in that range is so small that it would be difficult to obtain a statistically significant effect with anything less than a major change in behavior.10
| BAC | 1997 | 1999 | Percent Decline |
p |
|---|---|---|---|---|
| > .00 | 23.7 | 21.5 | 9.3 | < .10 |
| > .08 | 10.7 | 8.3 | 22.4 | < .01 |
| > .15 | 1.9 | 1.3 | 31.6 | > .15 |
| n = 1,786 | n = 2,451 |
Subgroup analyses found declines in positive and high BACs (> .08) in virtually every subgroup, although many of these were not large enough to be considered statistically reliable. There were no statistically significant changes among males, but females were less likely to have a positive BAC (p < .05) and a BAC > .08 (p < .01). Within racial subgroups, whites showed a decline in BACs over .08 (p < .01) but there were no other significant changes. There were declines at both levels among Greeks and non-Greeks alike, but only the decrease in BACs above .08 among Greeks was statistically significant (p < .02). Examining class years, only the declines among Sophomores were large enough to reach traditional levels of statistical significance (" = .05). This finding was paralleled by a significant decrease among persons age 19.
When looking at self-reported heavy drinkers (i.e., those who reported having 5 or more in a row at least once during the past two weeks), there was a significant decline in the proportion who had a measured BAC above .00 the night they were interviewed (from 19% to 13%, p < .01). Similarly, those who reported frequent heavy drinking (five or more on at least 3 occasions) also were less likely to have a BAC over .08 (declining from 25% in 1997 to 19% in 1999, p < .05).
Changes in the BACs of persons who reported they had driven (or who were interviewed as drivers) showed the same decrease found among students in general. The proportion of drivers with a BAC of .08 or higher declined from 2.6% to 1.3% (p > .21) and the proportion with any measurable alcohol declined from 13% to 9.7% (p > .18). These changes are not statistically significant, but a relatively small number of drivers were interviewed. Given that the magnitude of these changes is consistent with those among students in general, it is probable that this lack of significance is due to the relatively small sample size.
Student perceptions of their drinking in comparison to that of the "typical" UNC-CH student moved only slightly in the direction to be expected. These were not statistically significant in any subgroup, including the one group where such a change should have been most apparent – first year students. There was no change in self-reported heavy drinking or frequent heavy drinking. Finally, reports of any drinking and number of drinks also remained the same.
Official reports of alcohol-related incidents are systematically collected at UNC-CH. The three most pertinent of these are: (1) Alcohol-related incidents in and around residence halls, which are reported to the Department of Housing and Residential Education, (2) on-campus incidents not involving residence halls, which are reported to the Office of the Dean of Students, and (3) off-campus incidents that come to public attention (e.g., by report to local police), which are reported to the Center for Healthy Student Behaviors. Table 4.5 shows the change in incidents from the Fall 1998 to the Fall 1999 semesters.11 Although there are a variety of possible explanations for these dramatic decreases, as a result of changes in various university programs, policies, and enforcement activities, they have been attributed to the "2 out of 3" program by the Vice Chancellor for Student Affairs in her recent report to the Board of Trustees.
