Health Education Research, Vol. 19, No. 1, 71-84,
February 1, 2004
© 2004 Oxford University Press
Preventing alcohol-related harm in college students: Alcohol-related Harm Prevention program effects on hypothesized mediating variables
1 Department of Biobehavioral Health, Pennsylvania State University, University Park, PA 16802, USA 2 Correspondence to: J. W. Graham; e-mail: jgraham{at}psu.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodResultsDiscussionSupplementary materialAcknowledgmentsReferences |
|---|
The Alcohol-related Harm Prevention (AHP) program is a normative education and skill-acquisition program designed to reduce serious, long-term alcohol-related harm in college students. Without admonishing students not to drink, which is likely to fail in many student populations, the AHP program attempts to give students the necessary perceptions, motivation and skills to intervene within their peer group, and to make proactive harm-avoidance plans with friends prior to social occasions that involve using alcohol. The AHP program is a two-session, in-class intervention that corrects misperceived norms regarding levels of alcohol use, caring about friends, acceptability of risky behaviors and willingness to intervene. The program also makes use of interactive discussions with students and a graded, peer interview assignment to identify and promote harm-prevention strategies. The AHP program was implemented during fall 1999 at a large northeastern university. The program was received very well by students and showed significant effects on the proximal outcomes hypothesized to mediate more distal health-relevant outcomes.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionSupplementary materialAcknowledgmentsReferences |
|---|
Student alcohol use is implicated in a multitude of undesirable and harmful events on university campuses throughout the US. During the past 30 years, numerous studies have been conducted to determine the levels of student use, the underlying motivations for alcohol use, and the relationship between use and negative outcomes, and to develop prevention programs [see, e.g. (Journal of Studies on Alcohol, 2002
), Special Issue on college alcohol research]. The literature indicates that 80–95% of college students drink at least some alcohol and that 45% engage in what Wechsler et al. (Wechsler et al., 1994, 1998) have referred to as ‘binge drinking’ (five or more drinks in a row for men; four or more in a row for women). Studies have also documented the relationship between the use of alcohol by college students and a number of harmful events (Perkins and Berkowitz, 1986; Kraft, 1988; Wechsler et al., 1994, 1998).
In response to these problems, colleges and universities offer an array of alcohol education programs and services to students. Typically, campus alcohol programs focus on ways to decrease student alcohol use. Many universities use ‘top-down’ approaches in which university policy makers mandate guidelines for alcohol (non-) use on campus. Other environmental approaches involve reducing the supply of alcohol to underage students, increasing enforcement of drinking laws and promoting on-campus, no-alcohol social activities (Zimmerman, 1997).
Mandates from university officials that attempt to curtail student freedom tend to be very unpopular with the undergraduate student population. Students appear to know the ‘facts’ about the dangers of alcohol, yet they consume alcohol anyway (Marlatt et al., 1993). It is illegal for underage students to buy, possess or consume alcohol, yet the very high levels of student drinking indicate that the typical underage student is ready, able and willing to engage in all three of these activities (Marlatt et al., 1993).
‘Bottom-up’ approaches to alcohol education reflect a different viewpoint and offer an alternative to ‘top-down’ approaches. A ‘bottom-up’ approach involves listening to and involving students or focusing on individual students in finding solutions to problems of the student population [e.g. (Larimer and Cronce, 2002)]. A normative education program at Northern Illinois University was shown to be effective in reducing binge-drinking levels at that campus (Haines and Spear, 1996). Other alcohol programs have concentrated directly on the prevention of the harmful effects of alcohol use rather than drinking behavior itself (Marlatt et al., 1993, 1998). Harm-prevention programs that transcend judgments about drinking behavior, and focus on promotion of realistic intervention and avoidance strategies may ultimately provide better results.
This article describes a ‘bottom-up,’ normative education, harm-prevention intervention implemented at Penn State in fall 1999. We describe the program and present data relating to program effects on the variables hypothesized to be affected directly by the program.
The AHP program
The ultimate goals of the AHP program were to reduce alcohol-related harm and one of the major precursors to harm: alcohol-related risk taking. But how should these goals be reached? In his classic discussion of the naive analysis of action, Heider (Heider, 1958) suggested that the two main predictors of behavior are motivation and ability (‘try’ and ‘can’ in Heider’s terms). Thus, it is not surprising that most disease prevention and health promotion programs have at their very heart, an attempt to increase motivation and/or ability [e.g. (Larimer and Cronce, 2002)]. The AHP program is no different.
Our main approach focused on harm-prevention behavior. We theorized that risk taking would be reduced for each college student to the extent that individual students were willing and able to make harm-prevention plans relating to their or their friends’ alcohol use. We also theorized that risk taking (and harm) would be reduced in the student population to the extent that students were willing and able to take harm-prevention action when potentially harmful situations arose. Thus, the main focus of the AHP program was to increase student motivation and skill regarding these two kinds of action: harm-prevention planning and intervention.
We provide details in Methods regarding the intervention. In brief, our major focuses with respect to ability were to increase the ability of students (1) to make plans with their friends and (2) to intervene in certain situations. We also provided students with information about student norms that we believed would increase their motivation to make plans to and to intervene as necessary to prevent harm from coming to their friends. In this regard, we provided knowledge of student norms for caring about one another and student norms for alcohol-related risk taking.
Our second approach to preventing harm focused on the reduction of alcohol use, per se. This was not the main focus of our intervention, thus we attempted only to increase motivation to drink less, but did not specifically address the skill for drinking less. To this end, we provided participants with information about student norms regarding levels of drinking [(Hansen and Graham, 1991; Haines and Spear, 1996); also see (Perkins, 2002)].
Hypotheses
We hypothesized program effects on several variables theorized to mediate more distal, health-relevant outcomes. We hypothesized that the program would (1) increase the perception of students caring about the well-being of friends, (2) increase the perception that risky behaviors are generally unacceptable, and (3) increase skills relating to harm-prevention planning and intervention. Further, we hypothesized that the program would (4) increase intention to intervene in potentially harmful situations and (5) increase intention to make harm-prevention plans with friends. Finally, we hypothesized that the program would (6) decrease the perception of general peer use of alcohol and (7) increase the perception of peer non-use of alcohol.
|
Method |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionSupplementary materialAcknowledgmentsReferences |
|---|
Description of the AHP Intervention
The AHP program was implemented in two successive 50-min class periods in 30 small sections of a large, required, general education course. Session 1 required approximately 30 min; Session 2 filled an entire 50-min class period. The facilitators (four graduate students and one faculty member; median age approximately double that of the student participants) were recruited on the basis of having excellent presentation skills and belief in the harm-prevention message. These trained facilitators followed a carefully constructed presentation script and used an identical set of MS PowerPoint slides to increase uniformity across classes.
Session 1
The facilitator’s first objective in Session 1 was to establish credibility with the students. The topic of alcohol use by college students has been a controversial one at our university, possibly because of the relatively high levels of use (Wechsler et al., 1994). Any statement from an authority figure that sounded like an exhortation not to consume alcohol was likely to be received poorly by students. Therefore, the facilitator started the presentation by making mild fun of him/herself as an exhorting authority figure and stating the main project goals.
(1) ‘The goal is NOT to tell students not to drink’. This statement was important in transcending a point of controversy between students and health educators. (2) ‘The goal is NOT to keep students from having fun’. The facilitator promoted the idea that the social aspects of the college experience are important. (3) ‘The goal IS to prevent long-term, serious harm from happening to undergraduate students’. This was a goal upon which students, faculty and university administrators could all agree.
The second objective of Session 1 was to collect data (1) that students would regard as believable and (2) that could be used to correct misperceptions on several alcohol-relevant dimensions. This objective followed directly from the Adolescent Alcohol Prevention Trial (Hansen and Graham, 1991; Hansen et al., 1991). The facilitator discussed ways in which scientists gather valid information. Primary objectives were to get students to believe that the (anonymous) survey would provide believable data and to ask them to provide truthful answers to survey questions (Babor et al., 1987; Graham et al., 2002). The 38-question, anonymous survey took about 10 min, and asked questions regarding personal alcohol use, perceptions of peer alcohol use, caring about friends, willingness to intervene to prevent harm and acceptability of risky behaviors related to alcohol use. Students completed the survey using generic scan forms. The university’s testing service scanned the forms just a few hours after completion of the last session, providing ample time to obtain means and frequencies, and to make changes to the presentation materials for Session 2.
Session 2
The second session began with some attention-getting facts regarding the negative consequences of alcohol use. The facilitator reiterated the AHP project goals to re-establish credibility and to give a brief reading of the project goals for students who may have missed Session 1.
In-class survey results for all 30 classes were combined to increase the representativeness of the results and to protect the anonymity of students in these small classes. The facilitator revealed the results of the survey, showing that students (1) slightly overestimated levels of alcohol use by male and female students (the degree of overestimation might be larger in other contexts); (2) slightly underestimated the levels of non-use; (3) clearly underestimated the level of caring about friends and the willingness to intervene to prevent harm; (4) clearly overestimated the acceptability of drunk driving; and (5) dramatically underestimated the acceptability of using a condom while having sex. The overwhelming majority of students cared ‘a lot’ about their friends, would do something to prevent harm, found drunk driving to be UNacceptable and found condom use to be acceptable.
After presentation of the survey results, Session 2 became interactive. The facilitator presented three scenarios involving college students in risky situations stemming from alcohol use. The facilitator solicited student input regarding ways to intervene in such situations, ways to plan to avoid such situations and ways to start avoidance-planning conversations. During this discussion, students described pragmatic methods of intervening and planning to prevent harm. Solicitation and praise of student comments was an important component of our ‘bottom-up’ approach.
Facilitators were trained to listen at this point in Session 2. Where possible, students should be learning from other students. On the other hand, although all comments were welcomed, the facilitator had a set of suggestions to be articulated (see the eight goals appearing in Table I). These suggestions were a cumulative list of reasonable strategies derived from student comments during pilot testing. The facilitator guided the discussion to include these suggestions. Any strategy not suggested by a student was mentioned by the facilitator as a student suggestion from another class.
|
We also covered a fourth scenario in which a student was already drunk and announced that he or she was going to take part in a drinking game. The facilitator apologized, saying that he/she had promised not to preach, but said that he/she must say something about this. ‘Any time a person consumes large quantities of alcohol in a short period of time, the person is at serious risk for coma and death. I won’t say any more about that’. Because of the sensitive nature of this comment, we scripted this part of the presentation and delivered it as close to verbatim as possible.
Another important point presented, but not discussed, involved sexual behavior and condom use after drinking. We presented the last slide of results from the in-class survey, showing how people responded to the question, ‘Is it OK for a woman to require the man to wear a condom when they have sex?’. The data were rather dramatic: over 99% responded ‘definitely yes’ or ‘yes’, whereas the normative perception was that only 75% would find it acceptable. The facilitator concluded by saying, ‘I‘m not going to say very much about this. But, if you should find yourself in this kind of situation, please remember...99% of you, women and men, thought it was OK’.
The facilitator then distributed and briefly explained the homework assignment. This two-part homework assignment was a graded part of the course curriculum. The student was first asked to give an eight-question quiz to three other students (close friends, if possible). The eight questions were the same as in-class survey questions regarding perceptions of alcohol use, caring about friends and willingness to intervene. The interviewing student was then to describe the in-class survey results and compare the interviewee’s responses with the average intervention group responses.
In the second part of the homework assignment, the student was to ask each of the three interviewees to describe a harm-related incident that was a direct result of someone’s alcohol use. If the interviewee could not come up with a real incident, several scenarios were provided on the homework assignment, any one of which could be used. Then the two students were to devise a strategy or plan that could have prevented the incident and a way to start the conversation to develop such a plan. The student was then to write a 2-page reaction paper describing (1) the interviewees’ reaction to the interview, (2) the harm-related incidents, (3) the proposed prevention strategies and (4) the conversation starters (the assignment was lengthened to 3 pages to accommodate material relating directly to the course content).
Session 2 concluded with an explicit request that students make a decision. The facilitator reiterated that virtually all students do care about their friends and are willing to take action to prevent harm from coming to their friends. The facilitator then asked the students to make a conscious decision to take action, or not to take action, to prevent harm within their peer group. We banked on the idea that if they did make a conscious decision, it most likely would be the decision to take preventive action.
Evaluation method
The AHP program was implemented during the sixth week of the fall 1999 semester. A pre-test survey was administered during the fourth and fifth weeks of that semester, and the first follow-up survey was administered during the 11th and 12th weeks of the semester. Two additional follow-up surveys were administered in February and April of the semester following the intervention, and a final follow-up survey was administered in November 2000, approximately 13 months after the intervention. All aspects of this research were conducted in compliance with the local institutional review board.
Participant characteristics
The AHP program was implemented in a large Speech Communication course required of nearly all undergraduate students. In total, 1702 students were registered for this course. Based on random assignment, 30 sections of the course received the program. The remaining 45 sections served as controls. An estimated 681 students were enrolled in the sections in which the program was delivered. Of these, 555 were present at the first program session (Monday or Tuesday of the implementation week) and 492 were present for Session 2 (Thursday or Friday). We estimate that 444 students (65%) took part in both program sessions, that 170 (25%) took part in just one session and that 67 (10%) took part in neither session. Those not taking part were absent that day.
During the second and third weeks of the semester, students from all 75 sections were recruited to participate in what was described as a measurement study. The questionnaire was administered at a scheduled time outside of class and each participant received a $10 cash payment for participation ($20 per session for the last three measurement waves). Of the 1702 students enrolled in the course, 634 completed the pre-test survey and 489 completed the survey at the immediate post-test. Participation rates were 707 for February 2000, 714 for April 2000 and 628 for November 2000.
In total, N = 1023 students took part in at least one wave of measurement. Most students participated in some, but not all waves. There were 31 patterns of participation and non-participation. The largest single pattern (N = 255) involved students who took part in all five waves of measurement. The next three largest patterns (N = 68, 64 and 63) involved, respectively, those who missed only the last wave, those who missed the first two waves and those who missed only the second wave. A table of patterns of participation for the five measurement waves is available as ‘Supple mentary material’ at http://her.oupjournals.org/content/vol19/issue1/index
A somewhat smaller percentage of program than control students attended the measurement sessions for the first two waves (33 versus 40% for wave 1; 25 versus 31% for wave 2). However, for the final three waves, participation rates averaged about 40% from both program and control groups, and the rates were not significantly different. A lower percentage of program students attended all five (12 versus 17%), or four or more measurement sessions (24 versus 28%). However, the percentages attending three or more (35 versus 38%), two or more (47 versus 49%) and one or more sessions (60 versus 61%) were not significantly different between program and control groups.
Students taking part in the measurement were quite representative of those enrolled in the general education course. Except for the fact that sophomores were over-represented in the course, and freshmen and seniors were under represented, those enrolled in the course were reasonably representative of the entire student population. Of those taking part, 55% were female and 21% were members of a fraternity or sorority. The class breakdown was 65% sophomore, 24% junior, 7% senior and 3% freshman. The student population was 85% white and came from a wide variety of majors.
Measures
At the end of each AHP program session, students completed a 10-question, anonymous survey evaluating the quality of session. Included were questions about the degree to which the two main goals were achieved (goal not to tell students not to drink, goal to prevent harm), the degree to which the facilitator appeared to believe in the harm-prevention message, and the facilitator’s expertise, trustworthiness and enthusiasm. Students also indicated the degree to which they thought the facilitator was preaching, how receptive they were to the subject matter in general and how this presentation compared to others they had seen on the subject.
The questions in the main evaluation survey included measures of skill for making plans and intervening, perceptions of norms for caring, perceptions of norms for risk taking, intentions to make plans, intention to intervene in potential alcohol-related harm situations, and perceptions of peer alcohol use and non-use. Program effects were hypothesized to occur on all of these outcomes. Details regarding the items were reported in Graham et al. (Graham et al., 2002).
The following eight composite scales were used in the analyses: Skill at Harm-Prevention Planning (five items; 
= 0.89); Perceived Norm for Caring (four items; = 0.82); Perceived Norm for (Un)Acceptability of Risky Behavior (three items; = 0.67); Perceived Peer Alcohol Use (three items; = 0.85); Perceived Peer Non-Use (three items; = 0.72) and Intention to Intervene (four items; = 0.77). Intention to Make Plans was divided into two parts: general plans (three items; = 0.84) and vehicle-related plans (two items; = 0.86). All of these scales were judged to have acceptable reliability (Graham et al., 2002).
For program effects analyses described below, several variables were included as covariates, including gender, age, fraternity membership, GPA and measures of each of the Big-Five personality constructs: ‘neuroticism,’ ‘introversion,’ ‘conscientiousness’, ‘agreeableness’ and ‘intellect’ (McCrae and Costa, 1987; Costa and McCrae, 1988; Digman, 1990; Goldberg, 1990, 1992; Costa, 1991). Also included in all analyses was the pre-test of a three-item alcohol use scale ( = 0.96).
Strategy for analysis with missing data
The main analysis strategy was implemented in two steps. First, we used multiple imputation with NORM software (Schafer, 1997) to deal with the missing data (Rubin, 1987; Schafer, 1997; Schafer and Olsen, 1998; Graham and Schafer, 1999; Graham and Hofer, 2000; Collins et al., 2001; Schafer and Graham, 2002; Graham et al., 2003). Second, we analyzed the resulting imputed datasets using a standard, complete-cases regression procedure. Additional technical details relating to the imputation used in this study are available as ‘Supple mentary material’ at http://her.oupjournals.org/content/vol19/issue1/index
All results reported here were based on statistically sound imputation procedures (Little and Rubin, 1987; Rubin, 1987; Graham et al., 1994, 1996, 1997, 2003; Schafer, 1997; Graham and Schafer, 1999; Graham and Hofer, 2000; Schafer and Graham, 2002). These procedures provided reasonable estimates of regression parameters and standard errors, allowing us to draw realistic statistical conclusions despite the missing data. In addition, these procedures corrected most, if not all, of the biases associated with the missing data (Graham et al., 1997, 2003). It has been shown in numerous publications that multiple imputation, and similar procedures, are clearly superior to using complete cases, mean substitution or pairwise deletion (Graham et al., 1994, 1996, 1997, 2003; Wothke, 2000).
Strategy for main program effects analyses
Multiple regression analysis was used to test outcomes hypothesized to be affected by the program. In order take into account the possible effects of the multilevel structure of the data (students within classes), all multiple regression analyses were done using the multilevel program VARCL (Aitkin and Longford, 1986; Longford, 1987, 1988). Eight separate dependent variables were tested. In addition to the 10 covariates described above, the pre-test measure of each dependent variable was included as a covariate in each regression model.
The program was hypothesized to have direct effects on: planning skills, perceived norm for caring, perceived norm for unacceptability of risk-taking, perceived peer alcohol use, perceived peer alcohol non-use, intentions to make general plans, intentions to make plans relating to vehicle use and intentions to intervene. To summarize the overall program effect on each scale, we averaged each scale over the four post-test waves. In order to see the results in more detail, we also examined regression models involving each of the eight dependent variables at each of the follow-up waves of measurement.
|
Results |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionSupplementary materialAcknowledgmentsReferences |
|---|
Implementation quality
High-quality implementation does not guarantee program effects, but implementation of sufficient quality is a necessary condition for program effects (Hansen et al., 1991
). Seven of the items from the anonymous implementation-quality survey were averaged to form a single scale, which could be thought of ‘facilitator credibility’ or ‘session quality’ ( = 0.86). The items making up the scale were recoded where necessary such that for all items, the ‘3’, ‘4’ and ‘5’ responses all had the same meaning (3 = ‘moderately’ positive, 4 = ‘very’ positive and 5 = ‘extremely’ positive). Results for the seven-item ‘session quality’ scale, along with results from a single item asking how this session compared to others the students had seen on this topic, show that students were generally receptive to the content and approach of the AHP intervention. The students’ average response (3.96 after Session 2; 3.82 after Session 1) corresponded to the response ‘very’ positive on the seven questions relating to facilitator credibility. In addition, 94% of the students, overall, found our presentations to be better than others they had seen on this topic.
Program effects on hypothesized proximal outcomes
Mean differences between program and control groups for the pre-test measures of all eight dependent variables were non-significant. The results for the program effects analysis on each of the eight outcomes, averaged over the four follow-up waves of measurement, appear in Table II. The means for the program and control groups were in the hypothesized direction for all eight of the outcome variables. Significant beneficial program effects were found on five of these composite variables: perceived norm for risk-taking, perceived norm for caring, intention to intervene, perceptions of peer alcohol use and intention to make vehicle-related plans. Marginally significant program effects were found for the remaining three composite variables: perceived peer alcohol non-use, intention to make general plans and planning skills. Effect sizes (R2-improvement) for each of the eight dependent variables also appear in Table II.
|
The results for the more detailed analyses on each of the eight outcomes separately at each of the follow-up waves of measurement appear in Table III. Only t-values and significance levels appear in Table III. In referring to strength of effect in this section, we refer to the R2-improvement at the individual level of analysis. For four of the dependent variables (Perceived Norm for Risk-Taking, Perceived Norm for Caring, Skill in Making Plans and Intent to Make General Plans), the strongest program effect was observed at the first follow-up wave of measurement. For Perceived Norm for Risk-Taking and Perceived Norm for Caring, there was a slight decrease in the program effect over the remaining waves of measurement. For Skill in Making Plans, following the first follow-up wave, there appeared to be somewhat inconsistent, but reduced effects over time. For Intent to Make Plans (General), the program effects for each of the follow-up waves were small, and none of them reached statistical significance.
|
For Perceived Peer Alcohol Use, the strongest effect was observed at the second follow-up wave and there was a decrease in the program effect after that. For Intent to Intervene, the strongest effect was observed at the third follow-up wave. For this variable, the program effect was relatively constant over the first three follow-up waves, with a slight fall off in effect at the fourth follow-up wave.
For Intent to Make Vehicle-Related Plans and Perceived Peer Alcohol Non-Use, the strongest program effect was observed at the last follow-up wave of measurement. For Intent to Make Vehicle-Related Plans, the program effect was relatively constant over time. For Perceived Peer Alcohol Non-Use, the program effect appeared to be increasing in strength over the four follow-up waves.
For six of the eight dependent variables (excluding Intent to Make General Plans and Intent to Make Vehicle-related Plans), the strongest effect in terms of R2-improvement was also statistically significant. For Intent to Make Vehicle-related Plans, the strongest effect (at the fourth follow-up wave) was marginally significant.
The R2-improvements at the class level of analysis, although considerably higher, showed roughly the same patterns across the four follow-up waves of measurement. The t-values [which, when divided by 
n/2, are equivalent to Cohen’s (Cohen, 1977) d statistic for effect size] also showed roughly the same pattern across the four follow-up waves of measurement.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionSupplementary materialAcknowledgmentsReferences |
|---|
Student response to the program
The AHP program was well received by the students. The results relating to the quality of program delivery showed that 94% of those receiving the program thought it was better than other treatments they had seen on this topic; the average rating of ‘session quality’ corresponded to ‘very’ good. The results relating to student response to the program sessions are important for two reasons. First, any program in which students are being persuaded about something will succeed only if the material is received well by the students and if they find the facilitator reasonably credible. Thus, our results are important because they demonstrate that we have met the minimum requirement for program success. Second, college-student drinking has been a highly charged topic on many college campuses. Our results relating to the student response are also important because they demonstrate that it is possible to address this general topic with college students in a way that allows them to be receptive. This supports our contention that a non-judgmental, ‘bottom-up’ approach that establishes common ground (e.g. prevention of serious harm) is a viable way to deal with this highly charged topic.
Program effects on hypothesized outcomes
The results for the eight outcomes hypothesized to be affected by the AHP program were very encouraging. The program had a statistically significant effect on five of the eight proximal outcomes averaged over the four follow-up waves of measurement. Looking at the program’s effect separately at each of the four follow-up waves revealed that for five of the outcomes, the strongest effect was observed at one of the first two follow-up waves. For two other outcomes, the effects seemed to be roughly comparable over the four follow-up waves of measurement. For one of the outcomes (Perceived Peer Alcohol Non-use), the strongest program effect was observed at the fourth follow-up wave. For six of the eight outcome measures, the strongest effect was statistically significant. Counting the effect on Intent to Make Vehicle-related Plans, which was significant when averaged over all four follow-up waves, statistically significant program effects were observed for seven of the eight dependent variables.
Overall, these findings bode well for longer-term health benefits of interventions like the one described here. This program increased awareness (1) that it is OK for students to care about one another, (2) that risk-taking is generally not acceptable, (3) that students in general drink less than previously thought and (4) that for students in general, non-use is more prevalent than previously thought. In addition, the program had a direct effect on students’ intentions (5) to intervene to prevent harm from coming to their friends and (6) to make vehicle-related plans. Finally, there was some evidence that the program (7) increased students’ skills in making plans related to preventing alcohol-related harm.
Some of these outcomes (e.g. reduced perceptions of peer alcohol use) have long been known to mediate longer-term health benefits (e.g. reductions of immoderate alcohol use) (Donaldson et al., 1994; Graham et al., 1994; Haines and Spear, 1996; Perkins 2002). We are hopeful that changes in the other measures will also translate into beneficial effects on the more distal health-relevant outcomes. For example, changes in perceptions about the unacceptability of risk-taking behavior (e.g. about drunk driving, risky sex) will likely have corresponding effects on this sort of risk-taking behavior itself. Similarly, changes in perception about the acceptability of caring about one’s friends should make caring behavior (e.g. prevention behavior in a general sense) more likely as well. Finally, the increases in intention to intervene may well translate into actual intervention, and harm reduction, as potential harm situations arise. We are also hopeful that this intention to intervene, which may be a proxy for a more general ‘intent to do the right thing’, will have a broader impact on the longer-term health behaviors of these students.
A key hope of the AHP program was to increase student skill and motivation to make plans. We argued (in program Session 2) that it is often difficult to deal with potentially harmful situations once they have arisen and that by making harm-prevention plans, one is better able to avoid the sometimes harmful consequences of immoderate alcohol consumption. There was evidence that the program did increase planning skill and intentions, but it is fair to say that these effects were not strong. Because of the importance of these outcomes for the AHP program, it is important to explore the possible reasons for the relative weakness of the program effect on them.
There are several possible explanations for observed program effects relating to skill and intention to make plans. One possible explanation is that the measures were either unreliable or not valid. We have shown that all of these measures appear to have good reliability (e.g. = 0.89, 0.86 and 0.84 for Skill, Intent to Make Vehicle Plans and Intent to Make General Plans, respectively) and that all have reasonable construct validity (Graham et al., 2002). This evidence would seem to argue against lack of data quality as an explanation for the observed program effects.
A second possible explanation for the observed effect is that the AHP program, as implemented, was relatively weak with respect to changing planning skills and motivation. This aspect of the program was, after all, the newest major component. Consistent with this possibility is the fact that the two program components that appeared (anecdotally) to facilitators to be the most successful were also the aspects of these three outcomes for which program effects were strongest. In addition, although program facilitators were supposed to follow a script during Session 2 and were supposed to cover the eight goals of Session 2 (see Table I), which were highly relevant to the harm-prevention planning skills, variability across sessions (within and between facilitators) was likely with respect to the degree to which, and quality with which, these eight goals were met. It is possible that such variability watered down the program’s effect on these three outcomes. These possibilities suggest that we could improve the program’s impact on planning skill and motivation by improving the examples used in the presentation itself, and by improving the quality of the program delivery.
A feature of the program that was intended to enhance planning skill and motivation was the homework assignment. The homework assignment was intended to give students an initial opportunity to devise plans with their friends. It is possible that the homework assignment failed, in part, for reasons outside the program itself. For example, although the level of overall cooperation of the course leader was excellent, the level of cooperation for the individual section leaders was variable. It is likely that the section leaders’ statements to their students about the importance of the homework assignment varied substantially. Improvements on this front would include establishing better rapport, not just with the department head and course leader (where we had excellent support), but also with the individual section leaders. Also, the program would be better accepted by section leaders if the AHP program itself were better integrated into the course activities and requirements.
Strengths and weaknesses
One weakness of the current study is that the results are based on self-report measures. However, with this general population, we have previously found substantial validity of self-reported cigarette smoking behavior (Taylor et al., 1998). It seems reasonable that self-reports relating to these and other, less-sensitive domains are also valid. In addition, the generally high level of reliability and good convergent validity observed on these same data (Graham et al., 2002) are encouraging on this front.
Effect sizes for the program effects (individual as unit of analysis) on these proximal outcomes were generally not large. For example, for effects on perceived peer alcohol use, perceived peer non-use and intention to intervene, the R2-improvements were less than 0.01. On the other hand, for the composite measure for perceived norm for caring, the R2-improvement due to program was 0.025 and for perceived norm for risk-taking, the R2-improvement due to program was 0.121.
When the class mean was the unit of analysis, the program effect was significant for five of the eight dependent variables and the corresponding effect sizes were much larger (Perceived Norm for Risk-Taking: t = 10.51, R2-improvement = 0.56; Perceived Norm for Caring: t = 3.85, R2-improvement = 0.13; Intent to Intervene: t = 3.21, R2-improvement = 0.08; Perceptions of Peer Alcohol Use: t = 2.57, R2-improvement = 0.07; Intent to Make Vehicle-related Plans: t = 2.39, R2-improvement = 0.06).
A major strength of this study is that we were able to assign classes randomly to receive the program or to serve as controls. In addition, because 30 classes were assigned to receive the program and 45 classes served as controls, the chances are good that random assignment did produce reasonable equivalence between the two groups on any number of extraneous variables. Random assignment was particularly important in this study, because it argues for strong internal validity of our study results, despite the fact that 60% of the enrolled students actually took part in at least one of our measures.
Summing up
Without focusing on alcohol consumption, per se, the AHP project sought to change students’ skills, perceptions and motivations thought to mediate longer-term health- and harm-relevant behaviors. Although the effects relating to skill and motivation to plan must be strengthened in future versions of the program, the present version of the program was successful in changing the variables thought to mediate more distal health-relevant outcomes. Although it remains to be seen the extent to which changes on these more distal student health outcomes will manifest themselves, we remain optimistic that this general approach is a good one. We argue that working with students is a viable approach in preventing harm to our students and that this approach has many advantages over approaches that change the environment without student participation or consent.
|
Supplementary material |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionSupplementary materialAcknowledgmentsReferences |
|---|
Supplementary material can be found at http://her.oupjournals.org/content/vol19/issue1/index
|
Acknowledgments |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionSupplementary materialAcknowledgmentsReferences |
|---|
The authors wish to thank Michael Hecht, Nancy Mahon, Matt Bumpus, Gerard Hoefling, Jennifer Shaffer, Bonnie Taylor, and (alphabetically) Veronica Barb, Marie Calvano, Yadi Colon, Alycia Darcangelo, Jamie Dart, Jessica Dart, Angie Gomez, April Kunsman, Vanessa Lori, Megan Martin, Raina Olexa, Katie O‘Toole, Dana Ratsprecher, Jessica Ray, Zetta Schuver, and Christina Widmann for their invaluable assistance on various aspects of this project. This research was funded in part by grants from the Hanley Family Foundation and The J. M. Foundation. Portions of this research were presented at the Annual Meeting of the Society for Prevention Research, Montreal, June 2000.
|
References |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionSupplementary materialAcknowledgmentsReferences |
|---|
Aitkin, M.A. and Longford, N. (1986) Statistical modeling in school effectiveness studies. Journal of the Royal Statistical Society, 149A, 1–43.
Babor, T.F., Stephens, R.S. and Marlatt, G.A. (1987) Verbal report methods in clinical research on alcoholism: response bias and its minimization. Journal of Studies on Alcohol, 48, 410–424.[Web of Science][Medline]
Cohen, J. (1977) Statistical Power Analysis for the Behavioral Sciences. Academic Press, New York.
Collins, L.M., Schafer, J.L. and Kam, C.M. (2001) A comparison of inclusive and restrictive strategies in modern missing data procedures. Psychological Methods, 6, 330–351.[Web of Science][Medline]
Costa, P.T. (1991) Clinical use of the five-factor model: an introduction. Journal of Personality Assessment, 57, 393–398.[CrossRef][Web of Science][Medline]
Costa, P.T. and McCrae, R.R. (1988) From catalog to classification: Murray’s needs and the five-factor model. Journal of Personality and Social Psychology, 55, 258–265.[CrossRef][Web of Science]
Digman, J.M. (1990) Personality structure: emergence of the five-factor model. Annual Review of Psychology, 41, 417–440.[Web of Science]
Donaldson, S.I., Graham, J.W. and Hansen, W.B. (1994) Testing the generalizability of intervening mechanism theories: understanding the effects of adolescent drug use prevention interventions. Journal of Behavioral Medicine, 17, 195–216.[CrossRef][Web of Science][Medline]
Goldberg, L.R. (1990) An alternative ‘description of personality’: the Big-Five factor structure. Journal of Personality and Social Psychology, 59, 1216–1229.[CrossRef][Web of Science][Medline]
Goldberg, L.R. (1992) The development of markers for the Big-Five factor structure. Psychological Assessment, 4, 26–42.
Graham, J.W. and Hofer, S.M. (2000) Multiple imputation in multivariate research. In Little, T.D., Schnabel, K.U. and Baumert, J. (eds), Modeling Longitudinal and Multiple-group Data: Practical Issues, Applied Approaches and Specific Examples. Erlbaum, Hillsdale, NJ, pp. 201–218.
Graham, J.W. and Schafer, J.L. (1999) On the performance of multiple imputation for multivariate data with small sample size. In Hoyle, R. (ed.), Statistical Strategies for Small Sample Research. Sage, Thousand Oaks, CA, pp. 1–29.
Graham, J.W., Hofer, S.M. and Piccinin, A.M. (1994) Analysis with missing data in drug prevention research. In Collins, L.M. and Seitz, L. (eds), Advances in Data Analysis for Prevention Intervention Research. National Institute on Drug Abuse Research Monograph Series 142. National Institute on Drug Abuse, Washington, DC.
Graham, J.W., Hofer, S.M. and MacKinnon, D.P. (1996) Maximizing the usefulness of data obtained with planned missing value patterns: an application of maximum likelihood procedures. Multivariate Behavioral Research, 31, 197–218.[CrossRef]
Graham, J.W., Hofer, S.M., Donaldson, S.I., MacKinnon, D.P. and Schafer, J.L. (1997) Analysis with missing data in prevention research. In Bryant, K., Winkle, M. and West, S. (eds), The Science of Prevention: Methodological Advances from Alcohol and Substance Abuse Research. American Psychological Association. Washington, DC, pp. 325–366.
Graham, J.W., Roberts, M.M., Tatterson, J.W. and Johnston, S.E. (2002) Data quality in evaluation of an alcohol-related harm prevention program. Evaluation Review, 26, 147–189.
Graham, J.W., Cumsille, P.E. and Elek-Fisk, E. (2003) Methods for handling missing data. In Schinka, J.A. and Velicer, W.F. (eds), Research Methods in Psychology. Wiley, New York, pp. 87–112.
Hansen, W.B. and Graham, J.W. (1991) Preventing alcohol, marijuana and cigarette use among adolescents: peer pressure resistance training versus establishing conservative norms. Preventive Medicine, 20, 414–430.[CrossRef][Web of Science][Medline]
Haines, M. and Spear, S. (1996) Changing the perception of the norm: a strategy to decrease binge drinking among college students. Journal of American College Health, 45, 134–140.[Web of Science][Medline]
Hansen, W.B., Graham, J.W., Wolkenstein, B.H. and Rohrbach, L.A. (1991) Program integrity as a moderator of prevention program effectiveness: results for fifth grade students in the Adolescent Alcohol Prevention Trial. Journal of Studies on Alcohol, 52, 568–579. [Web of Science][Medline]
Heider, F. (1958) The Psychology of Interpersonal Relations. Wiley, New York.
Journal of Studies on Alcohol (2002) College drinking, what it is and what do to about it: review of the state of the science. Special Issue. Journal of Studies on Alcohol, Suppl. 14, 1–250.
Kraft, D.P. (1988) The prevention and treatment of alcohol problems on a college campus. Journal of Alcohol and Drug Education, 34, 37–51.
Larimer, M.E. and Cronce, J.M. (2002) Identification, prevention and treatment: a review of individual-focused strategies to reduce problematic alcohol consumption by college students. Journal of Studies on Alcohol, Suppl. 14, 148–163.
Little, R.J.A. and Rubin, D.B. (1987) Statistical Analysis with Missing Data. Wiley, New York.
Longford, N.T. (1987) A fast scoring algorithm for maximum likelihood estimation in unbalanced mixed models with nested effects. Biometrika, 74, 817–827.
Longford, N.T. (1988) VARCL: Software for Variance Component Analysis of Data with Hierarchically Nested Random Effects (Maximum Likelihood) [Computer Program Manual]. Educational Testing Service, Princeton, NJ.
Marlatt, G.A., Baer, J., Kivlahan, D., Dimeff, L., Larimer, M., Quigley, L., Summers, J. and Williams, E. (1998) Screening and brief intervention for high-risk college student drinkers: results from a two-year follow-up assessment. Journal of Consulting and Clinical Psychology, 66, 604–615.[CrossRef][Web of Science][Medline]
Marlatt, G.A., Larimer, M., Baer, J. and Quigley, L. (1993) Harm reduction for alcohol problems: moving beyond the controlled drinking controversy. Behavior Therapy, 24, 461–504.
McCrae, R.R. and Costa, P.T. (1987) Validation of the five-factor model of personality across instruments and observers. Journal of Personality and Social Psychology, 52, 17–40.
Perkins, H.W. (2002) Social norms and the prevention of alcohol misuse in collegiate contexts. Journal of Studies on Alcohol, Suppl. 14, 164–172.
Perkins, H.W. and Berkowitz, A.D. (1986) Perceiving the community norms of alcohol use among students: some research implications for campus alcohol education programming. International Journal of the Addictions, 21, 961–976.[Web of Science][Medline]
Rubin, D.B. (1987) Multiple Imputation for Nonresponse in Surveys. Wiley, New York.
Schafer, J.L. (1997) Analysis of Incomplete Multivariate Data. Chapman & Hall, New York.
Schafer, J.L. and Graham, J.W. (2002) Missing data: our view of the state of the art. Psychological Methods, 7, 147–177.[CrossRef][Web of Science][Medline]
Schafer, J.L. and Olsen, M.K. (1998) Multiple imputation for multivariate missing data problems: a data analyst’s perspective. Multivariate Behavioral Research, 33, 545–571.[CrossRef]
Taylor, B.J., Graham, J.W., Palmer, R. and Tatterson, J.W. (1998) Validating self-reported cigarette smoking with saliva cotinine and expired air carbon monoxide. Paper presented at the Annual Meeting of the Society for Prevention Research, Park City, UT.
Wechsler, H., Davenport, A., Dowdall, G., Moeykens, B. and Castillo, S. (1994) Health and behavioral consequences of binge drinking in college: a national survey of students at 140 campuses. Journal of the American Medical Association, 272, 1672–1677.
Wechsler, H., Dowdall, G., Hoyt, J. and Lee, H. (1998) Changes in binge drinking and related problems among American college students between 1993 and 1997: results of the Harvard School of Public Health College Alcohol Study. Journal of American College Health, 47, 57–68.[Web of Science][Medline]
Wothke, W. (2000) Analysis of multivariate and longitudinal data with missing values. In Little, T.D., Schnabel, K.U. and Baumert, J. (eds), Modeling Longitudinal and Multiple-Group Data: Practical Issues, Applied Approaches and Specific Examples. Erlbaum, Hillsdale, NJ, pp. 219–240.
Zimmerman, R. (1997) Social Marketing Strategies. The Higher Education Center for Alcohol and Other Drug Prevention, Washington, DC.
Received on May 28, 2002; accepted on December 3, 2002
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||