xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001498252
007 cr mnu|||uuuuu
008 041209s2004 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000482
Hunt, William Michael.
Effects of participant engagement on alcohol expectancies and drinking outcomes for a computerized expectancy challenge intervention
h [electronic resource] /
by William Michael Hunt.
[Tampa, Fla.] :
University of South Florida,
Thesis (Ph.D.)--University of South Florida, 2004.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 108 pages.
ABSTRACT: The purpose of the present study was to examine the effect of varying the amount of participant engagement on alcohol expectancy and drinking outcomes during a social/sexual expectancy challenge based on Darkes and Goldman's (1993, 1998) protocol. This study was also intended to provide a test of the efficacy of administering an alcohol/placebo expectancy challenge outside of a live drinking scenario through video presented as part of a computerized intervention.One hundred fifty-eight male participants across three sites were randomized into a no-intervention control group that received non alcohol-related information in a minimally interactive computerized format, a low-level engagement experimental group that received minimally interactive computerized expectancy-related information, and a high-level engagement experimental group that received the same expectancy-related information presented in a more interactive computerized format that included games and audiovisual elements such as video clips, graphics, live narrations, and music. It was hypothesized that high-level engagement participants would report being more engaged in their computerized program and demonstrate greater decreases in social/sexual alcohol expectancies and drinking levels relative to control and low-level engagement participants.Results indicated that while high-level engagement participants reported being more engaged in their interventions, none of the groups exhibited changes in the alcohol expectancies measured. In addition, all three groups experienced significant but comparable decreases in drinking levels. Exploratory follow-up analyses were also conducted to provide suggestions for future directions.
Adviser: Goldman, Mark.
analysis of variance.
t USF Electronic Theses and Dissertations.
Effects of Participant Engagement on Alcohol Expectancies and Drinking Outcomes for a Computerized Expectancy Challenge Intervention By William Michael Hunt A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Psychology College of Arts and Sciences University of South Florida Major Professor: Mark Goldman, Ph.D. Thomas Brandon, Ph.D. Paul Jacobsen, Ph.D. Douglas Rohrer, Ph.D. Kristen Salomon, Ph.D. Date of Approval: November 4, 2004 Keywords: alcohol intervention, analysis of variance, colleg e students, male drinkers, Timeline Follow-Back Copyright 2004, William Michael Hunt
i Table of Contents List of Tables iii List of Figures iv Abstract vi Introduction 1 The Expectancy Concept 2 The Expectancy Concept Applied to Alcohol 3 Scientific Support for the Util ity of Alcohol Expectancies 4 Altering Alcohol Expectancies to Change Drinking Behavior 5 Variations on the Expectancy Challenge Theme 10 Challenging Alcohol Exp ectancies Vicariously 14 Computers and Multimedia Technology in the Enhancement of Learning and Change 18 Rationale for the Present Study 22 Method 25 Participants 25 Design 26 Equipment 27 Instruments 27 Procedure 31 Computerized Interventions 32 Control group program 33 Low-level engagement program 33 High-level engagement program 33 Module 1 34 Module 2 35 Module 3 36 Module 4 37 Results 38 Participant Characteris tics by Group and Site 38 Age 39 Ethnicity 39 Sexual orientation 41 Distribution among grou ps at each site 42
ii Participant Attenti on and Engagement 44 Alcohol Expectancies 45 Positive/Arousing octant scores 46 Arousing octant scores 47 Positive octant scores 48 Alcohol Consumption 49 Mean drinks consumed per day 51 Quantity/Frequency 52 Proportion of binge days 53 Exploratory Analyses 55 UCSD 58 UCSD with spring break in pretest 59 UCSD with spring break in follow-up 61 USF summer 2003 and summer 2004 64 Heavier drinkers 66 Lighter drinkers 68 Discussion 70 References 77 Appendices 83 Appendix A. Demographics and Drinking Styles Questionnaire 84 Appendix B. Alcohol Expectancy Circumplex 88 Appendix C. Timeline Follow-Back 89 Appendix D. Driving Practices 90 Appendix E. Driving Beliefs 91 Appendix F. Level of Engagement Questionnaire 92 Appendix G. Defensive Driving Content Questionnaire 96 Appendix H. Alcohol Program Content Questionnaire 97 Appendix I. Debriefing Form for University of South Florida 98 Appendix J. Debriefing Form fo r San Diego State University and University of California, San Diego 99 About the Author End Page
iii List of Tables Table 1 Participant Ages by Group and Site 39 Table 2 Mean Participant Performan ce on 5-point Content Summary Score 44 Table 3 Data Groupings for Exploratory Analyses 57
iv List of Figures Figure 1. Participant ethnicity by group. 40 Figure 2. Participant ethnicity by site. 41 Figure 3. Participant sexual orientation by group. 41 Figure 4. Participant sexua l orientation by site. 42 Figure 5. Participant gr oup assignment by site. 43 Figure 6. Participant engagement by group (min = 0, max = 18). 45 Figure 7. Positive/Arousing octant sc ores (max=18) as a function of group. 47 Figure 8. Arousing octant sc ores (max=18) as a function of group. 48 Figure 9. Time by site interaction for positive octant score (max=18; shown with non transformed data). 49 Figure 10. Mean drinks consumed per da y at pretest and follow-up for each group (shown with non transformed data). 52 Figure 11. Quantity/Frequency of drinking at pretest and follow-up for each group (shown with non transformed data). 53 Figure 12. Proportion of binge days at pretes t and follow-up for each group. 54 Figure 13. Positive octant scores (max=18) and mean drinks per day for UCSD subsample. 59 Figure 14. Positive octant scores (max=18) and positive/arousing octant scores (max=18) for UCSD subsampl e with spring break in pretest. 60 Figure 15. Mean drinks per day and ar ousing octant scores (max=18) for UCSD subsample with spring break in follow-up. 62 Figure 16. Positive octant scores (max=18) for UCSD subsample with spring break in follow-up. 63
v Figure 17. Quantity/Frequency of drinking and proportion of binge days for USF subsample for summer 2003 and summer 2004. 65 Figure 18. Positive octant expectancy score (max=18) for heavier drinker subsample (mean daily drinks = 2.16 or higher). 66 Figure 19. Positive octant expectancy score (max=18) for heavier drinker subsample (mean daily drinks = 2.16 or higher) comprised of Caucasians aged 18-25 years. 67 Figure 20. Positive octant expectancy score (max=18) for lighter drinker subsample (mean daily drinks < 2.16). 68
vi Effects of Participant Engagement on Alcohol Expectancies and Drinking Outcomes for a Computerized Expectancy Challenge Intervention William Michael Hunt ABSTRACT The purpose of the present study was to ex amine the effect of varying the amount of participant engagement on alcohol expectancy and drinking outcomes during a social/sexual expectancy challenge ba sed on Darkes and GoldmanÂ’s (1993, 1998) protocol. This study was also intended to provide a test of the efficacy of administering an alcohol/placebo expectancy challenge out side of a live drinking scenario through video presented as part of a comput erized intervention. One hundred fifty-eight male participants across three sites were randomized into a no-intervention control group that received non alcohol -related information in a minimally interactive computerized format, a low-level engagement experimental group that received minimally inte ractive computerized expectan cy-related information, and a high-level engagement experimental group th at received the same expectancy-related information presented in a more interactive computerized format that included games and audiovisual elements such as video clips, gr aphics, live narrations, and music. It was hypothesized that high-level engagement par ticipants would report being more engaged in their computerized program and demonstrat e greater decreases in social/sexual alcohol
vii expectancies and drinking levels relati ve to control and low-level engagement participants. Results indicated that while high-level engagement participants reported being more engaged in their interventions, none of the groups exhibited changes in the alcohol expectancies measured. In additi on, all three groups experienced significant but comparable decreases in drinking levels. Exploratory follow-up analyses were also conducted to provide suggestions for future directions.
1 Introduction Alcohol use has been linked to traffic cras h fatalities, shown to cause cirrhosis of the liver, and indicated as the primary cause of a variety of shortterm hospitalizations (e.g., alcohol dependence syndrome, alcohol ic psychoses, and nondependent abuse of alcohol). Additionally, alcohol abuse contribut es to a wide range of legal, social, and occupational problems (National Institute on Alcohol Abuse and Alcoholism, 2000). A short list of such problems can include absenc es from work, lost wages, losses in work productivity, work-related injuries, marita l distress, and disorderly conduct. Alcohol problems flourish in our institutions of higher learning as well. In fact, in the most recent update of a periodic nationa l college drinking survey (Wechsler, Lee, Kuo, & Lee, 2000), it was found that the propor tion of students engaging in frequent binge drinking has been increasing on campus es over the past several years. These drinkers are responsible for drinking two-th irds of the alcohol consumed by college students and account for over thr ee-fifths of the most serious alcohol-related problems on college campuses (e.g., drinking and driving, alcohol-related in juries, and vandalism). The survey also found that the proportion of student drinkers who had been drunk three or more times in the previous month (prior to the survey) had increased, as had the proportion of those who drank on ten or more occasions and who drank for the sole purpose of getting drunk.
2 Given the large and growing number of problems associated with excessive alcohol use, particularly on co llege campuses, it is not surprising that significant effort has been dedicated to the prevention and reduc tion of drinking. To this end, one concept that has received increasing scientific interest over the last several decades is that of alcohol expectancies, which are de fined and discussed below. The Expectancy Concept The general expectancy con cept dates back a considerab le way and has undergone significant changes in meaning over the years from an early affiliation with behaviorism (Tolman, 1932) to a more cognitive bent in more modern times. Goldman (1999) has recently expanded upon the expectancy concept in an effort to take it more fully into the cognitive realm. He views expectancy, in a broad sense, as patterns or templates of information that are stored in memory and as the use of this information to produce behaviors. These stored memories serve to help an organism deal (usually) more efficiently with new situations that are si milar to ones previously experienced. New information that is perceived is compared to existing information templates (expectancies). This comparison helps the orga nism to organize, interpret, and structure and enact behaviors accordingly. Goldman ( 1999) also maintains that expectancies are an active system, operating automatically as well as under conscious control of the organism. Alcohol expectancies are best understood as a specific type of general expectancies people may hold.
3 The Expectancy Concept Applied to Alcohol Inasmuch as alcohol expectancies describe a small subset of an individualÂ’s total expectancy network, a discussion of the alc ohol expectancy concept would seem simply to require the specific applicati on of general expectancy theory to the behavior of alcohol consumption. Applying GoldmanÂ’s (1999) gene ral definition of expectancies, alcohol expectancies are information stored in oneÂ’s memory about the way alcohol consumption affects behavior. Taken previously from di rect experience, or observation of others, these stored memories cause individuals to anticipate certain consequences from the consumption of alcohol. Depending on whet her an individual finds the anticipated consequences of drinking reinforcing, that individual may or may not engage in the behavior. Problems arise for people when their alc ohol expectancies lead them to drink excessively and to engage in other behavior s while drinking that have a negative impact upon their lives (e.g., driving while intoxica ted, having unsafe sex, and getting into fights). If theory holds, the n, it is not difficult to conclude that finding ways to alter peopleÂ’s alcohol expectancies in such a way that they no longer anticipate positive consequences from alcohol consumption (or anti cipate positive consequences to a lesser degree) should result in a decrease in their le vel of drinking and/or the problems that arise from drinking. Because expectancies ope rate largely without conscious control, Goldman (1999) suggests that, after determ ining which erroneous expectancies are influencing a behavior (such as problematic drinking), helping i ndividuals bring those
4 expectancies into conscious awareness and disconfirm them should have a significant impact on that behavior. Scientific Support for the Utili ty of Alcohol Expectancies In the past several decades, a growing body of research has begun to clarify the relationship between expectanci es and alcohol use. For ex ample, alcohol expectancies have been shown to form in children befo re drinking is initiated (Dunn & Goldman, 1996), to predict concurrent drinking over and above prediction using background variables alone (Brown, 1985a; Christiansen & Goldman, 1983), to discriminate problem and nonproblem drinkers (Brown, Goldma n, & Christiansen, 1985; Christiansen, Goldman, & Brown, 1985), to be related to al coholism treatment outcome and predictive of abstinence after treatment (Brown, 1985b), and to predict future drinking behavior (Christiansen, Smith, Roehling, & Goldman, 1989; Goldman, Greenbaum, & Darkes, 1997). Alcohol expectancies and drinking have also been dem onstrated to influence each other in a reciprocal fashion over time (Sher, Wood, Wood, & Raskin, 1996; Smith, Goldman, Greenbaum, & Christiansen, 1995). This existence of a reci procal relationship supports the theory that modify ing alcohol expectancies might lead to a change in alcohol consumption. Naturally, the greatest utility in modification of alcohol expectancies would lie in the control and reduction of problematic drinking. The following three sections will introduce a procedure develope d for changing alcohol expectancies and drinking behavior, called the expectancy chal lenge, and will attempt to identify possible key components of this methodology.
5 Altering Alcohol Expectancies to Change Drinking Behavior In 1987, Henderson and Goldman attempte d to decrease drinking through the manipulation of alcohol expect ancies. In a pre-post desi gn spanning two weeks, three groups of female college students were submitte d to either an alcohol education program, a no treatment control, or an expectancy modification program. The expectancy modification procedure involved administration of placebo alcohol in a social situation that pulled for behaviors like those found in real-life drinking s ituations. After the manipulation, participants were told of the true (no alcohol) nature of their beverage, and it was explained that their intoxicated be havior was caused not by pharmacology but by their expectancies for the e ffects of alcohol on them. Measures of participants Â’ alcohol expectancies taken at pre-treatment and posttreatment showed decreases in alcohol expect ancies only for the expectancy modification group. However, measures of alcohol consum ption taken at baseline and two-week post treatment showed decreases in drinking fo r both the expectancy modification group and the alcohol education group. The decrease in drinking, unfortunately, was not maintained for the expectancy modification group at four-week post treatm ent assessment. Massey and Goldman (1988) expanded Henderson and GoldmanÂ’s (1987) paradigm the following year. In this study fema le participants were again divided into an alcohol education group, a no treatment cont rol group, and an expectancy modification program, each of which met for a total of f our sessions. The first session of this expectancy program, however, involved giving some of the pa rticipants real alcohol in addition to giving some participants placebo alcohol. The participantsÂ’ task was to
6 observe each other during a social interacti on (a game of charades ) and then to guess afterward, based on behavior, which of them had received the real alcohol. The remainder of the first session was spent di scussing why the participants could not accurately identify the women who had imbibed alcohol and discussing other aspects of expectancy theory. Sessions two through four involved similar expectancy theory lessons along with a review and discussion of expect ancy logs which the expectancy program participants completed each day between session s. These logs were intended to help the participants become aware of the myriad of sources (e.g., advertisements, television, and movies) in their daily environments from which they received alcohol expectancy information. Results at four-week follow-up for th is study showed decreases in alcohol expectancies for all three groups of participants. However, while decreases in drinking were found for both the high-level and low-le vel drinkers in the expectancy program, none of the control group and only low-level drinkers in the alc ohol education program showed decreases in drinking. Although th ese results provided stronger support for the mediational role of alcohol expectancies in drinking behavior, the decrease in alcohol expectancies for all participants, particularly for those in the control group who displayed no changes in drinking pattern, tended to weak en the position. In explanation, however, the researchers did advance the theory that the measurement and alteration of expectancies that were more specific to a particular situation may have led to more definitive results (e.g., assessing and altering social alcohol expectancies for a social
7 situation rather than assessing and altering more general alcohol expectancies in a social situation). This issue of targeting expectancies that are specific to a particular situation was addressed in a similar study conducted by Darkes and Goldman (1993). In this case, they focused on social and sexual expectancies in situations that pulled for them. Undergraduate, male drinkers were randomly assigned to one of three groups in a pre-post design spanning one month. One group, the control, rece ived pre-treatment assessment and post-treatment assessment on a variety of measures, including drinking behavior and alcohol expectan cies, and was asked to monitor their daily drinking for a month. The two experimental groups were treated identically to the control group. Additionally, one of the groups received three sessions of a traditi onal college prevention program. The other group, the expectancy challenge group, receiv ed three sessions designed to challenge their ex pectancies about alcohol by de monstrating that the sexual and social effects they attribute to the pharm acological effects of al cohol are actually due to the placebo effect (i.e., their expectancies about alcohol). They were also assigned homework for the month involving keeping a log of situations in their environments (e.g., media advertisements) that served to reinfo rce their alcohol expect ancies. The idea for the homework assignment was to bring their expectancies more into conscious deliberation and to challenge their expectancies further by helping them find instances when the environment reinforced their previous, erroneous expectancies. The first two sessions for the expect ancy challenge group were conducted by having participants of legal age consume e ither two alcohol or two placebo alcohol
8 beverages without being told which type of drink they we re receiving. The groupÂ’s task was to identify which group members had act ually imbibed alcohol, basing their guesses on observations of the individualsÂ’ behavior during situations invol ving socially-related or sexually-related content (playing Win-Lose-or-Draw or rating slides of women for attractiveness, respectively). The groupÂ’s in ability to discriminate the individuals who had actually received alcohol from the individuals receiving placebo alcohol was used to demonstrate to them that many of the behavior al effects of alcohol are actually due to peopleÂ’s expectations rather than the pharmaco logy of the drug. In the third session the group leaders reiterated the lessons of th e first two sessions, provided information on expectancy theory and related research fi ndings, and facilitated a group discussion of what had been learned by participating in the sessions and by completing the homework assignment. All three groups had equivale nt levels of drinking and social/sexual expectancies during pre-assessment. At follow-up, only th e expectancy challenge group had lower levels of social/sexual expectancies. The expectancy challenge group also showed the greatest level of decrement in drinking, followed by the trad itional prevention condition. The assessment control group showed no decreases in drinking at follow-up. Although considered by Darkes and Goldma n (1993) mainly as strong evidence for the mediational role of al cohol expectancies in drinking behavior, the results of this study also suggested that drinking reductions (over a nd above those achieved by traditional interventions) could be achieved in males by reducing their expectancies for
9 social facilitation and sexual enhancement fr om alcohol. The aut hors noted two other particularly interesting findings as well. The first was that the expectancy challe nge intervention appeared to be equally effective for underage individuals who were not allowed legally to consume a beverage on the chance that it could be alcoholic. Th is result implies that such an intervention could be effective as a treatment or prev ention for individuals (e.g., adolescents or younger children who have or may soon develop expectancies associated with future problematic drinking) without ac tually having them partake in the experiential aspect of the challenge (i.e., drinking the alcohol or placebo beverage). Such a finding has ramifications for dissemination of this intervention to a far wider audience. Second, although the traditional prevention condition had levels of success equivalent to the expectancy challenge in tervention for lighter drinkers, only the challenge intervention was eff ective for the heavier drinkers (15.88 or more drinks per week). The authors hypothesi zed that the reason only the challenge intervention was effective for the heavier drinkers was that, even though they were the most committed to drinking, the challenge decrease d alcohol expectancies of in creased arousal and energy Â– expectancies which seem to dr ive heavy drinking the most. This finding suggests that the expectancy challenge may be a useful interv ention for drinkers who are traditionally the most difficult to influence. In 1998, Darkes and Goldman conducted an other expectancy challenge study of male college drinkers. This time, howev er, they expanded the scope of the study by increasing the treatment duration, adding a treatment booster session, and extending the
10 evaluation of alcohol expectancies and dri nking to six weeks beyond the initial treatment phase. They also added a second exp ectancy challenge group, addressing affective/cognitive arousal, with which they could compare the social/sexual expectancy challenge group designed in Darkes and Gold man (1993), and an attention/monitoring control group. Finally, as an added test of the efficacy of the expectancy challenge procedure, the study was timed so that th e post-treatment drinking assessment phase would coincide with spring break, a period of known increase in alcohol consumption for college populations. By study end, both challenges had resulted in a decrease in expectancies and alcohol consumption as compared with the control group, with expe ctancies generally being lowest for the social/sexual group. Inte restingly, alcohol consumption had actually increased during the post-treatment assessment phase (overlapping with spring break) for the control group but not for either challenge group. Variations on the Expectancy Challenge Theme A number of attempts have been made to reproduce the alcohol/placebo expectancy challenge procedure establis hed by Darkes and Goldman (1993, 1998), both to identify its boundary conditions and to determine its key components. Corbin, McNair, French, and Black (1998), for inst ance, conducted a study in which they did away with the alcohol/placebo drinking co mponent of the Darkes and Goldman (1993, 1998) expectancy challenge. Instead, they gave groups of college students several lectures on alcohol expectan cies and asked them to Â“ch allengeÂ” items from alcohol expectancy measures after each lecture. While this exercise led to a decrease in alcohol
11 expectancies later measured with the same questionnair es, there was no corresponding decrease in drinking levels. Corbin and colleagues conducted a simila r study more recently that included mixed gender groups, most notably heavy dr inking females (Corbin, McNair, & Carter, 2001). As in the previous study, they found d ecreases in alcohol expectancies for those individuals participating in the expectancy challenge. Interestingl y, however, they found trends toward better drinking outcomes for th e males in their expectancy challenge but increases in alcohol consumption for the females. Despite some methodological disadvantages (i.e., the dubious value of using the same questionnaires for follow-up that were used during the interv ention and the high potential for experimental demand effects), th e findings from these studies suggest that decreases in drinking may not come about simply by changing peopleÂ’s explicit views about the effects of alcohol. Perhaps exposure to an expectancy-disconfirming experience such as the alcohol/placebo ad ministration of Darkes and Goldman (1993, 1998) and efforts to get participants to adopt new beliefs about the effects of alcohol are instrumental in effecting a change in drinking. In 1999, an attempt was made to administer a social/sexual e xpectancy challenge in a single session to mixed-sex groups (Maddock, Wood, Davidoff, Colby, & Monti). While the challenge was adapted from the Darkes and Goldman (1993, 1998) methodology, it was only somewhat successful in altering alcohol expectancies and proved ineffective in decreasing drinking. Li ttle detail was given, however, regarding what material was sacrificed to reduce th e Darkes and Goldman challenge contact time
12 by half and regarding what adaptations were made to accommodate female participants in the mixed-sex groups. (The Darkes and Gold man challenge material addressing sexual expectancies only discusses male sexual functio ning under the influence of alcohol.) Another attempt to test an expectan cy challenge with women was made the following year. In 2000, Dunn, Lau, and Cruz borrowed the full Darkes and Goldman (1993, 1998) protocol for challenging social/sex ual expectancies, applying it to men as well as adapting it for use with women (tr eated separately from the men). They demonstrated the challengeÂ’s effectiven ess in significantly changing hypothesized patterns of organization of expectancies in memory and in decreasing drinking for men. They were unable, however, to demonstrate these changes in wome n. They hypothesized that the challengeÂ’s lack of effect on womenÂ’s drinking might have been due to the relatively small amount of change the women exhibited in their expectancy patterns. This minimal change in expectancy patterns, it was suggested, may have been the result of a failure to translate the challenge protoc olÂ’s material enough to address issues of concern for women (e.g., a discussion of the negative sexual side effects of drinking had less of an impact on women than it did on men). Musher-Eizenman and Kulick (2003) attemp ted further study of the findings of Dunn et al. (2000) by conducting th eir own expectancy challenge with female drinkers. Unlike Dunn et al. (2000), these authors de tected short-lived d ecreases in alcohol expectancies for the challenge group. They also found decreases in drinking for the challenge and control groups alik e. Interestingly, the authors reported that the follow-up
13 assessment period fell during the week before final exams and that drinking levels may have been suppressed in both gr oups as a result. In 2004, Wiers and Kummeling also ad apted the Darkes and Goldman (1993) protocol for administration to a mixed gender group. In contrast to Dunn et al. (2000), however, they detected reductions in pos itive alcohol expectancies and alcohol consumption in heavy drinking women in the challenge group but not their male counterparts. Both challenge and control gr oups showed significant decreases in drinking by follow-up. Findings from these studies suggest a few tentative conclusions that bear further examination. Perhaps modification of alcohol expectancies and a corresponding change in drinking requires exposure to one or more salient events such as the expectancydisconfirming experience of the alcohol/pl acebo administration of Darkes and Goldman (1993, 1998). Without this powerful experien ce as a demonstrati on of the disconnect between alcohol expectancies and pharmacol ogy, participants may fa il to buy in to the message sufficiently to effect a behavioral change. Another point to consider is that even with a salient demonstration of the disconnect between expectancies and pharm acology, participants may require help identifying the personal relevance of the info rmation. If the explan ation of how higher doses of alcohol can result in effects contrary to those expected and desired is too brief, or if effects are discussed that do not apply di rectly to the particip ants (such as pointing out to women how alcohol consumption can lead to erectile failure), the impact of the expectancy challenge may be diminished. If this is the case, an overly abbreviated
14 presentation or a lengthier presentation directed to an extremely diverse group (e.g., mixed sex) may dilute the power of th e expectancy challenge procedure. Finally, a growing number of studies s upport the efficacy of an experientially based expectancy challenge in altering alcohol expectancies and dr inking for groups of men. However, the effect of adapting such an intervention for mixed gender groups and all female groups remains less clear. Challenging Alcohol Ex pectancies Vicariously Another interesting variation of the Darkes and Goldman (1993, 1998) expectancy challenge procedure involves the presentation of the alcohol/placebo expectancy-disconfirming experience vicari ously through a multimedia presentation. In addition to the obvious benefit of increased ease of dissemination (w ithout the need of a mock bar and bartenders to run an inte rvention), a vicarious experience of the alcohol/placebo administration would allow use of the intervention wi th individuals for whom alcohol consumption would be problematic (e.g., underage drinkers or alcoholics). In 1995, Wooten attempted this variation of the Darkes and Goldman (1993, 1998) expectancy challenge with eighth grade students -an entirely underage sample. Because none of the participants were old enough to consume alcohol, they were exposed to the alcohol/placebo challenge component vicariously by helping to plan and then viewing an expectancy challenge session for college student drinke rs. By allowing the eighth graders to participate in the planning phase of the chal lenge, it was hoped that they would find the challenge more belie vable when viewing it later.
15 The adolescents were educated about the differences between expectancy effects and pharmacological effects of alcohol. They were also asked to help generate related Win-Lose-or-Draw phrases for the group activity of the expectancy challenge intervention with college students. This cha llenge intervention was videotaped for later viewing by the eighth graders. The video s howed the college students consuming either alcohol or placebo alcohol beverages before participating in the group activity. In spite of the fact that her sample only experienced the alcohol/placebo expectancy challenge vicariously, Wooten (1995) found decreases in expectations for changes in social behavior and arousal in th e adolescents in her (vicariously) modified expectancy challenge group, as compared with those in an alcohol education group and those in a no-treatment control group. Sh e was unable to study the effect on drinking patterns due to the small number of dr inkers in her sample, however. In 1999, Keillor, Perkins, and Horan atte mpted a vicarious presentation of the alcohol/placebo challenge in college students. They compared videotaped expectancy challenge followed by live didactic informati on with an alcohol information condition. The content for the expectancy plus didact ic information group was based on Darkes and GoldmanÂ’s (1993, 1998) protocol. Male college students atte nding an alcohol education program as a result of a single alcohol offense participated in two 90-minute alcohol information sessions or two 90-minute expectancy challenge sessions. For each expectancy challenge session, participants viewed a 25-mi nute videotape of male college students drinking either alcohol or placebo-alcohol drinks and inte racting by playing Win-Lose-or-Draw or by
16 rating slides of women on level of attractiven ess (one interaction pe r session). After the video, the participants spent five minutes wr iting down their guesses as to which drinkers on the video had consumed alcohol. The part icipants then viewed a seven minute video of the drinkers being told the correct answers and discus sing their own identification errors. Finally, the videos were follow ed by a presentation on the development, maintenance, and operation of alcohol expectancies. Results indicated that while the alcohol information gr oup demonstrated increased knowledge of the effects of alcohol, neith er group exhibited changes in alcohol expectancies at posttest. Additionally, neither group exhi bited decreases in drinking behavior. Thus, the videotaped expectancy challenge plus live dida ctic presentation on alcohol expectancies failed to result in ch anges in alcohol exp ectancies or drinking behavior. The authors suggested a number of possi bilities for a lack of expectancy and drinking modification in the vi deo challenge group. Two possibilities had to do with the nature of their sample. The participants we re adjudicated rather than volunteers and, as such, may have been less open to the interven tion than previous samples of volunteers. Additionally, because they were still dr inking in spite of experiencing negative consequences (i.e., being charged with an alc ohol offense), the author s suggested that the participants might have had more ingrained dr inking patterns than typical college student volunteers. In addition to characteristics of the sample, methodological issues may have affected the results. It is possible that me rely viewing a video of the drinking component
17 of the expectancy challenge was insufficient to alter expectancies. However, Wooten (1995) demonstrated some success in modifyi ng expectancies of children with a similar video and didactic format. It may be that th e children were more amenable to expectancy modification than adjudicated college students due to their relative lack of experience with alcohol and its potentia l effects on drinkers, but othe r factors could have been involved as well. For instance, a 25-minute video of other college students drinking and playing a game may have been too long to hold the interest of, and thus make an impact on, the college-aged viewers, particularly if they had other risk factors for alcohol problems potentially affecting attention span or were experiencing al cohol or drug-related cognitive difficulties. The theory that the vicarious expectancy challenge of the Keillor, Perkins, and Horan (1999) study was ineffective because th e participants were not engaged by the lengthy video may hold some merit, especially given that the production values of the study video were not of part icularly high quality (J. J. Horan, personal communication, June 3, 2002). As such, future investigations of expectancy challe nges should consider the role of engagement in the efficacy of th e intervention. Additionally, in spite of the disappointing findings of Keillor et al., the potential benefits of a vicariously administered expectancy challenge, along with the partial success of Wooten (1995) in administering one to a younger population, argue for further investigation of vicarious administration of expectancy challenges. In summary, research on the expectan cy challenge procedure has identified several possible key components that individually, or in co mbination, may contribute to
18 its efficacy. One is exposure to a salient e xpectancy-disconfirming experience such as an alcohol/placebo administration. Another is th e inclusion of conten t that is personally relevant to the participants. Additionally, characteristics of the population receiving the intervention may play a role in how effective it is. The intervention seems to work best on a homogenous population of relatively hea vy-drinking college males. Finally, the level of participant engagement may play a ro le in the effectiveness of the challenge by varying the Â“doseÂ” that participants receive as a result of their ability to attend to and be influenced by the intervention. Computer technology offers an ideal tool for examining these key components, in particular the role of participant engageme nt as a variable affecting expectancy and drinking outcomes. One benefit of using com puter software is that intervention content (i.e., a salient experience and relevance to the intended audience) can be kept similar across groups while participant engagement can be manipulated by varying the level of program interactivity. As will be demonstrated in the following section, computers provide the opportunity to vary the style of presentation of material, making it more or less engaging through the use of audiovisu al technology such as videos, graphics, narrations, music, and sound effects. Computers and Multimedia Technology in the Enhancement of Learning and Change The use of computers and multimedia technology for education, prevention, and treatment has exploded in recent years. For ex ample, they have been used to administer cognitive-behavioral therapy for depression (Selmi, Klein, Greist, Sorrell, & Erdman, 1990), for AIDS and sexually transmitted disease education (Seidner, Burling, &
19 Marshall, 1996), to treat agoraphobia (G hosh & Marks, 1987) and panic disorder (Newman, Kenardy, Herman, & Taylor, 1997), to treat (Winzelberg, Taylor, Sharpe, Eldredge, Dev, & Constantinou, 1998) or reduce the risk of eating disorders (Zabinski, Pung, Wilfley, Eppstein, Winzelberg, Celio, & Taylor, 2001), and to improve weight loss (Taylor, Agras, Losch, Plante, & Burnett, 1991). They have also been used and evaluated extensively as alcohol and drug education aids and to teach drug refusal skills (e.g., Alterman & Baughman, 1991; Rickert, Graham, Fisher, Gottlieb, Trosclair, & Jay, 1993; Duncan, Duncan, Beauchamp, Wells, & Ary, 2000; Bryson, 1999). There have been few published studies, however, examining the efficacy of computers and multimedia technology in the reduction of drinking or dr ug use. Nevertheless, recent research in interactive multimedia learning has contributed a wealth of suggestions for creating and maximizing the effects of such interventions. Recently, experts in interactive multimed ia learning have begun to set forth conditions as well as principles of software construction that enha nce user engagement, motivation, and learning. These sugges tions encompass both the modalities of information presentation (e.g., audio, graphic, and text-based) and th e types of learning activities believed to be the most effectiv e in engaging and motivating software users. Mayer and Moreno (2002) recently propos ed a cognitive theory of multimedia learning, which they adapted from dual coding theory, cognitive load theory, and constructivist learning theory. In essence, th eir theory states that multimedia learning is most enhanced when the learner is able to select, organize, and integrate new knowledge (constructivist learning) best processed through two different systems, the visual and
20 verbal processing systems (dual coding theory ), without exceeding the learnerÂ’s cognitive capacity (cognitive load theory). Defining multimedia as the combination of words (written or narrated) and pictures (animation, video, or static graphics ) they conducted a series of studies testing the tenets of their theory. Based on these stud ies, they set forth five principles they found significantly enhanced computer-based multi media learning, specifica lly the learning of the step-by-step operation of cause-and-effect systems (e.g., how biological systems work or the processes involved in the creation of lightning). These princi ples are as follows: 1. Multiple representations Â– learning is improved when material is presented in words and pictures rather than just in words 2. Contiguity Â– learning is improved wh en corresponding words and pictures are presented simultaneously rather than separately 3. Coherence Â– material is better unders tood and learned when few extraneous words and sounds are include d in the presentation 4. Modality Â– it is better to present words as narration than on-screen text 5. Redundancy Â– the addition of on-scr een text to concise animation and narration diminishes understanding When attempting to enhance engage ment and learning, however, some researchers believe that addres sing the userÂ’s affective state is as important as cognitive considerations. Lamenting that adult inter active multimedia often lacks the appeal and engaging interfaces and content of childrenÂ’s software, Stoney and Oliver (1998) asserted
21 that multimedia materials must address both th e cognitive and affectiv e needs of adults. After a review of the relevant literature, they described several factors believed to enhance motivation and engagement of adults during interactive multimedia learning. The first of these factors is Â“immersionÂ” where the learner becomes absorbed and engaged in the content of the program. Im mersion is best accomplished through the use of games and by avoiding gender bias and aski ng the learner to adopt a foreign persona (e.g., having a female learner assume the iden tity of a male character). Related to immersion are the concepts of Â“play and flow Â”, where boredom and anxiety in the learner are diminished by making the learning process more like a fun game than a learning assignment; Â“fantasyÂ”, best accomplished by simulating a scenario where knowledge must be applied in order to succeed in a ta sk; and Â“curiosityÂ”, which is increased through the incorporation of nov elty and surprises throughout the program. Stoney and Oliver (1998) also state that engagement ca n be increased by letting the learner set the pace and order of the learning activities (Â“learner controlÂ”), encouraging Â“collaborationÂ” with either ot her humans, the computer, or a computersimulated expert, and by challenging the lear ner to demonstrate competence with the material, preferably through its application as opposed to through testing (Â“challengeÂ”). Finally, engagement can be increased by enc ouraging Â“reflectionÂ” in learners during which they apply new knowledge to perform an activity and then receive feedback that furthers their understanding of the concept.
22 Rationale for the Present Study The purpose of the present study was to ex amine the effect of varying the amount of participant engagement on alcohol expectancy and drinking outcomes during a social/sexual expectancy challenge ba sed on Darkes and GoldmanÂ’s (1993, 1998) protocol. This study was also intended to provide a test of the efficacy of administering an expectancy challenge outside of a live dr inking scenario because the alcohol/placebo expectancy-disconfirming experience was pres ented vicariously (via video presented as part of a computerized chal lenge program). Efforts were taken to replicate, and maintain across groups, the other three previously discussed possible key components of the expectancy challenge: presenting the original protocol content to a homogenous sample of heavy-drinking college male volunteers along with a video of an al cohol/placebo expectancy-disconfirming experience. Only level of participant e ngagement was intentionally manipulated by presenting the intervention via computer soft ware that, using tec hniques previously discussed, varied the level of interactivity re quired to complete it. For the purposes of this study, the definition of Â“to engageÂ”, as adapted from WebsterÂ’s Ninth New Collegiate Dictionary (1990), is to attract and hold attentio n; to involve; to encourage active participation. As explained in more detail in the met hod section, three groups were included in this study. The first was a no-intervention co ntrol group that receiv ed non alcohol-related information in a computerized format that was minimally engaging. The second was a similarly low-level engagement group. Part icipants in this group, however, received
23 minimally engaging computerized expectancy -related information (PowerPoint slides containing program information in text format with minimal graphics) and were provided with a description of an in vivo alcohol/p lacebo expectancy-disc onfirming experience. The third group, a high-level engagement gr oup, viewed a video of a group undergoing an in vivo alcohol/placebo expectancy-disconfirming experience and received the same computerized content as the low-level e ngagement group; however, the high-level engagement group received the program information in a more engaging format. Presentation of text-based material was broken up by games and questions requiring active application of the information by the participants. Additionally, the program was made more interesting through the use of audiovisual elements such as video clips, graphics, live narrations, and music. It was hypothesized that how engagi ng participants found the computer interventions would affect how much attenti on they paid to the information provided and how much they processed that informati on. Information that is processed more thoroughly (i.e., that presente d in the high-level engageme nt group) was expected to result in a greater impact on the participan ts with regard to both changes in alcohol expectancies and changes in drinking levels More specifically, it was hypothesized that: 1. Participants in the high-level enga gement group would display greater engagement in their computerized interv ention than low-level engagement and control groups.
24 2. At one-month follow-up, the high-level engagement group would display greater changes in alcohol expectancies (decreases in social/sexual ex pectancies) than low-level engagement and control groups. 3. Level of drinking would decrease for the high-level engagement group relative to low-level engagement and cont rol groups at one-month follow-up.
25 Method Participants WoodÂ’s (1997) meta-analysis of the effect sizes obtaine d in Darkes and Goldman (1993, 1998) suggested .75 as an appropriate effect size for in vivo expectancy challenges. However, given the possibility th at a computerized intervention may be less efficacious than an in vivo intervention, a more conservative effect size of .40 was chosen. Thus, a total of 156 participants spl it into three groups were needed to provide power of .80 for the two-tailed analyses a nd alpha of .05 used in this study (Cohen, 1992). College student drinkers were recruite d from the psychology department on-line participant pools at the University of South Florida, San Diego State University, and the University of California San Diego. In each case, the university Institutional Review Boards approved the study, and students receive d course extra credit in exchange for their participation in the study. To create the be st chances of success for the computerized intervention, the sample used in this study wa s composed solely of male college student volunteers. In this way, th e protocol used successfully by Darkes and Goldman (1993, 1998) was kept in as nearly an unaltered form at as possible. This was an attempt to eliminate the problems experienced by other re searchers (Corbin et al., 2001; Dunn et al., 2000; Maddock et al., 1999; Musher-Eizen man & Kulick, 2003; Wiers & Kummeling, 2004) when they tried to alter the protocol to accommodate more diverse groups (i.e.,
26 females and mixed sex groups) and when they tried to work with adjudicated students (Keillor, Perkins, & Horan, 1999). Although the study was designed for drinke rs, no alcohol was served (the alcohol/placebo expectancy-disconfirming e xperience was presented vicariously via video). Thus, students aged 18 years and older were eligible for part icipation as long as they consumed between 6 drinks per week and less than 6 drinks per day, criteria used in Darkes and GoldmanÂ’s (1993) orig inal expectancy challenge pr ocedure. Finally, students who had previously participated in simila r expectancy challenge research were not eligible for this study. Design The study utilized a between groups desi gn with three groups. These groups varied along two dimensions: level of engage ment (low vs. high) and type of information provided (alcohol expectancy vs. informa tion unrelated to alcohol). The group designated the control group received informati on unrelated to alcohol that was presented in a low-level engagement format. Both intervention conditions received the same information intended to challenge their alcohol expectancies. However, for the low-level engagement expectancy challenge group, the information was presented in a minimally engaging format that did not require active pr ocessing of the information. For the other condition, the high-level engagement expect ancy challenge group, the information was presented in a more engaging format that re quired active use of the information provided. To this end, software design suggestions for improving participant learning and
27 engagement were incorporated into the highlevel engagement software program (Mayer & Moreno, 2002; Stoney & Oliver, 1998). Del Boca, Darkes, Greenbaum, and Goldma n (2004) observed that college student drinking varies considerably throughout th e school year depending on environmental influences such as exam periods and holidays. In an attempt to reduce the effect of this variation on drinking outcomes, participants were assigned ra ndomly in blocks to one of the three conditions. In addi tion, because data were collect ed over the period of a full calendar year and at three separate univer sities with different academic holidays, participant drinking data was reviewed before being analyzed, and drinking falling on local or national holidays was removed from th e data. This procedure was intended to reduce site and time difference s in the sample (e.g., because the data collection proceeded for a full year, some participantsÂ’ pretest dr inking fell during spring break whereas other participantsÂ’ posttest data fell during spring break). Equipment Participants in each group completed their computerized programs using personal computers with headphones when sound was in cluded in the program. The control and low-level engagement group programs were created using Micros oft CorporationÂ’s PowerPoint 2000 software. The high-level engagement group program was created using MacromediaÂ’s Authorware 5 Attain software. Instruments 1. The Demographics and Drinking Styles Questionnaire is a series of questions created to faci litate collection of demogr aphic information about
28 respondents as well as a rough measur e of their alcohol consumption patterns. Completion time is less than fi ve minutes. See Appendix A. 2. The Alcohol Expectancy Circumplex (AEC) was formerly known as the Alcohol Expectancy Inventory (R ather & Goldman, 1994; Rather, Goldman, Roehrich, & Brannick, 1992). It is a 24-item list of single word adjective descriptions of possible ef fects of alcohol. Respondents are asked to indicate on a seven point Likert-type scale ranging from 0 (Never) to 6 (Always) how often Â“Drinking alcohol makes one _____Â” where the list word is inserted in th e blank. These adjectives comprise eight factors or octants of alcohol expectancies including those from positive, negative, arousing, sedating, positive-arousing, negative -arousing, positive-sedating, and negative-sedating expectancies. Approximately 10 minutes are needed to complete this questionnaire. The AEC has been shown to predict drinking concurrently, accounting for 30% of drinking variance, and at one -year post assessment, accounting for 24% of drinking variance (Goldman & Darkes, submitted). Additionally, reliability and validity have been demonstrated for the AEC (Darkes, Sheffield, & Goldman, 2001; Goldman & Darkes, submitted; Sheffield, Darkes, Del Boca, & Goldman, 2001). See Appendix B for this measure. 3. The Timeline Follow-Back (TLFB; S obell & Sobell, 1992) is a method of obtaining self-reports of drinking. It takes the form of a calendar upon which drinkers write the number of standard alcoholic drinks consumed
29 each day of the time period in questi on. It takes approximately 15 minutes to complete a 30-day time period and has been used extensively in the college student population. The Timeline Follow-Back has been shown to be both reliable and valid as well as id eally suited to evaluating specific changes in drinking before and after tr eatment. Test-retest reliability is high for the college student populati on ranging from .87 to .97 for 30-day time periods (Sobell, Sobell, Leo, & Ca ncilla, 1988). Moderate to high correlations with collateral report s of drinking and m oderate to high concurrent validity have also been demonstrated for the Timeline FollowBack questionnaire (Sobell & Sobell 1992). See Appendix C for this measure. 4. The Driving Practices Questionnaire is a series of questions created for the control group that asks whet her and how often they engage in particular activities while driving that might decrease their ability to drive safely. The purpose of this questionnai re was to increase the range of questions the control group receives beyond th e scope of alcohol consumption and alcohol expectancies so as to make it less obvious that they were a control group. Completion time is less than five minutes. See Appendix D for this measure. 5. The Driving Beliefs Questionnaire follows the same format as the Alcohol Expectancy Circumplex (see above) and asks participants how important they think it is to avoid engaging in behavi ors that may interfere
30 with safe driving. This questionnaire was being used for the control group to increase the range of questions they received beyond the scope of alcohol consumption and alcohol expect ancies so as to make it less obvious that they were a control group. Completion time is less than five minutes. See Appendix E for this measure. 6. The Level of Engagement Questionnaire was created specifically for this study to assess participantsÂ’ opinions about how engaging they found their particular tasks on the computer to be. Respondents are also asked to indicate how much time they typica lly spend on a computer and the internet throughout the week. The por tion of the questi onnaire assessing participant engagement showed adequa te reliability (CronbachÂ’s Alpha = .91). Completion time is approximately five minutes. See Appendix F for this measure. 7. The Defensive Driving Content Questi onnaire was created sp ecifically as a manipulation check for this study to assess particip antsÂ’ level of understanding and recognition of the information provided by the defensive driving program. Completi on time is less than five minutes. See Appendix G for this measure. 8. The Alcohol Program Content Questionna ire was created spec ifically as a manipulation check for this study to assess particip antsÂ’ level of understanding and recognition of the information provided by both of the alcohol programs (low-level engage ment and high-level engagement).
31 Completion time is less than five minutes. See Appendix H for this measure. 9. The Debriefing Form was provided to participants after completion of the computerized interventions. See Appendix I and Appendix J for this form. Procedure Potential participants we re recruited directly from psychology classes or via psychology department participan t pool websites. Those mee ting the study criteria were allowed to sign up for a time to complete the fi rst part of the study in person and agreed to participate in a brief follow-up phone interview one month after participation in the study. When participants arrived for the study, th ey were assigned randomly in blocks to the control group or to one of the two experi mental groups. Participants in all three groups completed an informed consent and the Demographics and Drinking Styles Questionnaire followed by the Alcohol Expectan cy Circumplex as a measure of their alcohol expectancies prior to participation in the study. Members of the control group additionally completed the Driving Practi ces Questionnaire and the Driving Beliefs Questionnaire. All participants next comple ted a Timeline Follow-Back of their drinking for the 30 days prior to th e day of participation. After the Timeline Follow-Back, members of the control group completed a lowlevel engagement computerized training on sa fe driving practices for approximately 12 minutes (see below for details). At this time, members of the low-level engagement experimental group completed a computerized training of equal duration and number of
32 slides that covered the alcohol expectancy ch allenge interven tion (see below for details). Also during this time, members of the high-level engagement experimental group completed an alcohol expectancy challenge intervention with similar content but with the addition of interaction components and two videos (see below for details). This intervention lasted approximately 20 minutes. After participation in their respective com puterized interventions, participants in each of the groups completed the Level of E ngagement Questionnaire as an indicator of how engaging each program was for them a nd either the Defensive Driving Content Questionnaire (control group) or the Alcohol Program C ontent Questionnaire (both experimental groups). All participants were then debriefed, awarded extra credit points for their participation in the study, and reminde d of their appointment to receive a followup phone interview in one month for additional extra credit. At the one-month follow-up, all partic ipants again completed the Alcohol Expectancy Circumplex along with a sec ond Timeline Follow-Back (covering 30 days beginning with the day of their first partic ipation in the study) to measure potential changes in their patterns of alcohol e xpectancies and drinking behavior. Computerized Interventions As previously stated, members of th e control group completed a low-level engagement tutorial on safe driving practi ces. Members of the experimental groups completed either a low-level engagement or a high-level engagement computer-based expectancy challenge intervention.
33 Control group program. Participants in the control group watched a 12-minute PowerPoint slide presentation on safe driving practices. The slides were self-timed, included some pictures, and covered material to improve de fensive driving techniques such as scanning, managing oneÂ’s space and speed, communicatin g oneÂ’s intent while driving, and the utility of headlights in preventing accidents. Low-level engagement program. Participants in the low-level enga gement group also watched a 12-minute PowerPoint slide presentation. The slides we re self-timed, included some pictures, and covered alcohol expectancy challenge material based on that used by Darkes and Goldman (1993, 1998). The presentation bega n by describing an in vivo alcohol expectancy challenge interven tion that had been conducted in a mock bar with collegeage males (during the filming of the video used in the High-level Engagement Program described below) and the in vivo participan tsÂ’ verbal reactions to the intervention. Comments made by the participants after thei r in vivo alcohol expect ancy challenge were used to segue into the next phase of the co mputerized intervention consisting of a series of text-based PowerPoint slides describing th e alcohol expectancy concept and the effect of alcohol expectancies on drinking behavior. High-level engagement program. Participants in the high-level engageme nt group received the same information that was presented to the low-level engageme nt group except that it was presented in a more interactive format. They also spent an additional eight and a half minutes viewing
34 two videos, one of an in vivo alcohol exp ectancy challenge inte rvention conducted in a mock bar with college-age males, and a s econd of the in vivo participantsÂ’ verbal reactions to the intervention. (During participant debriefi ng after the study, all but one who saw the video reported believing the pe ople in the video were real as opposed to actors.) Comments made by the in vivo participants after their in vivo alcohol expectancy challenge were used to segue into the ne xt phase of the computerized intervention consisting of a series of modul es describing the alcohol expect ancy concept and the effect of alcohol expectancies on drinking behavior. With th e intention of making the intervention more engaging for the participan ts, however, this information was broken up and interspersed with gamelike exercises and questions designed to encourage deeper processing of the expectancy information. It was intended that completion of these exercises and answering these questions w ould entail drawing conclusions from the information provided and applying what they had learned about alc ohol expectancies. The video, information, interactive elements, and goals of the program are described as sequential modules below. Module 1. The first module began by showing the participant a video of a group of males drinking either real alcohol or placebo alcohol and interacting in a bar setting. After a few minutes the video stopped and the partic ipant was asked to guess which individuals in the video received real alcohol. The participant guessed by clicking his mouse on the photo of each the individuals he thought wa s intoxicated. After the participant had
35 finished guessing, the video resumed and the in dividuals in the video discussed who they thought were drinking and why. The video conc luded with the video individuals and the participant being told who actually consumed alcohol. Following that, the participantÂ’s guesses we re compared with the correct answers, and he was told how he performed. The part icipant was told that, on average, guessers do no better than chance in sel ecting the correct drinkers. It was explained that the poor performance is due to the fact that people ofte n act in contradictory ways when they have been drinking and that these contradict ory behaviors are th e result of alcohol expectancies, not the real e ffects of alcohol. Module 1 ended by posing the question, Â“Why do people start to act in certain ways (e.g., happy or social) when theyÂ’ve been drinking or when they think theyÂ’ve been drinking but didnÂ’t consume alcohol?Â” Module 2. The second module began by introducing a ga me intended to help the participant discover the answer to the question posed at the end of module 1. The participant was given a list of words describi ng different ways people can act (i.e., mellow, Â“pumped upÂ”, quiet, extroverted, friendly, a ggressive, sleepy, and sexually aroused). The participant was asked to select the behaviors he thought people often attribut e to the effects of drinking alcohol. After he made his selections, the par ticipant was told that most people think alcohol has all of those effects on people. It was then pointed out that the behaviors listed can often contradict each other (Â“Does alc ohol make us mellow or Â‘pumped upÂ’? Quiet
36 or extroverted? Friendly or aggressive? Sleepy or sexually aroused?Â”), and it was asked how the same drug can have opposite effects. The program continued by maintaining th at alcohol cannot have all of these opposite effects on people and by explaining that alcohol expect ancies are responsible for these behaviors, not pharmacology. The modul e ended by telling the participant that, because they affect how people act when they drink, alcohol expectancies also influence peopleÂ’s decisions to drink. Module 3. The third module introduced another game for the participant in which he had to distinguish between expectancy effects of alcohol and real, phys iological effects of alcohol. The program provided a list of words describing effects from alcohol consumption and asked the participant to drag the words either to a body (if they were real, physical effects) or a t hought bubble (if they were inaccu rate, expectancy effects). The program provided correctiv e feedback if the participant chose incorrectly. Following that exercise, the participant was asked a multiple choice question requiring him to characterize the inaccurate, ex pectancy effects of alcohol. (Â“Which of the following characterizes the inaccurate be liefs/expected effects of alcohol?Â”) The program provided feedback ei ther reinforcing his correct response or correcting his incorrect response. Module 3 ended with the program providing an example of how peopleÂ’s inaccurate beliefs about the effects of alcohol di ffer from the real effects of alcohol. The
37 example contrasted the facts that while men often expect alcohol to make them more sexually aroused, it actually diminishes th e ability for men to perform sexually. Module 4. The fourth module introduced Â“labeli ngÂ” as one mechanism by which people acquire and maintain alcohol expectancies (Alcohol consumption causes a general mental slowing that facilitate s the attribution of certain e ffects and outcomes to alcohol. These attributions are dependent on the situati on and oneÂ’s expectations for the effects of alcohol.) After working through two exampl es where people often attribute social facilitation and increased sexual arousal to alcohol consumption (Â“labelingÂ”), the program asked where these inaccurate alcohol expectancies originate. In answer to the question, the program explained that alcohol expectancies are often learned as children grow up hearing adu lts talk about their expectations or watching them drink and act certain ways. Additionally, the program also implicated advertising as a source for acquiring and main taining alcohol expectancies. After presenting a number of alcohol advertisements and pointing out their implied messages (e.g., that consuming a pa rticular beverage will lead to sexual encounters with attractive pe ople or fun parties), the module summarized the information presented during the program. It concluded by asserting that the goal was not to keep participants from consuming alcohol but to help them realize that because many of the desirable effects are due to e xpectancies and not alcohol itsel f, it is possible to drink less alcohol and enjoy oneself while reducing th e risks for negative consequences.
38 Results Participant Characteristics by Group and Site Research findings suggest that levels of alcohol consumption differ by gender and among ethnic groups, and they vary within those ethnic groups by drinker age (See Gerstein, Grat, Epstein, & Ghadialy, 1994; Jackson, Williams, & Gomberg, 1998; or the National Institute On Alcohol Abuse and Alcoho lismÂ’s Tenth Special Report to the U. S. Congress on Alcohol and Health: Highlights fr om Current Research, 2000). As a result, comparisons were made to determine whet her participant age and ethnicity were similarly represented in each experimental group and at each study si te. As previously stated, only males were included in this study to keep the protocol used successfully by Darkes and Goldman (1993, 1998) in as nearly an unaltered format as possible. Although the content of the experimental interventions targeted expectancies related more to social facilitation and se xual arousal and performance (as opposed to sexual orientation), the distribution of partic ipants of differing sexual orientations across sites and groups was examined. The distributi on of participants into groups at each site was also checked for comparability. In this case, comparisons were made to insure that similar proportions of participants from each st udy site were randomized into each of the study groups.
39 Age. Table 1 presents mean ages of part icipants broken down by group condition and site. There were no significant age differen ces across groups within the full sample or across groups within individual s ites. There were also no si gnificant differences across sites for any individual group. However, wh en comparing across sites, the USF site (22.12 years) as a whole had si gnificantly older participants than both the SDSU sites (19.53 years) and the UCSD sites (20.64 years) as a whole F (2, 155) = 5.84, p = .004. Thus, USF participants were on average 2.59 ye ars older than their counterparts at SDSU and 1.48 years older than their UCSD counter parts. While this difference among sites was not large, it was considered a potential source of variance in the main analyses. Table 1 Participant Ages by Group and Site Participant Ages in Years Full Sample USF SDSU UCSD M SD M SD M SD M SD Control group 20.89 (n=54) 2.98 21.65 (n=26) 3.86 19.43 (n=7) 1.72 20.43 (n=21) 1.54 Low level group 21.13 (n=52) 2.56 21.96 (n=26) 2.66 20.33 (n=6) 3.83 20.30 (n=20) 1.59 High level group21.69 (n=52) 4.73 22.73 (n=26) 6.36 18.83 (n=6) 1.17 21.20 (n=20) 1.47 Combined groups 21.23 (n=158) 3.54 *22.12 (n=78) 4.52 *19.53 (n=19) 2.41 *20.64 (n=61) 1.56 significant at the p < .05 level Ethnicity. Figure 1 displays the distri bution of participants acro ss ethnicities that were randomized into each of the studyÂ’s groups. A two-way contingency table analysis was
40 conducted to evaluate whether the same pr oportion of participants across groups was from each ethnicity. The two variables we re ethnicity (African American, Asian, Hispanic/Latino, Native American, White/C aucasian, and Other) and group (Control, Low level, and High level). Proportions of participants from each ethnicity were not significantly different across groups, Pearson 2 (8, N = 158) = 8.24, p = .410. Figure 1. Participant ethnicity by group. 13 10 4 15 6 12 11 8 6 54 65 73 7 12 6 0 10 20 30 40 50 60 70 80 90 100 ControlLow levelHigh level Study GroupPercentage African American Asian Hispanic/Latino White/Caucasian Other Figure 2 displays the distribu tion of participants across ethnicities that were in each of the studyÂ’s sites. A two-way continge ncy table analysis indicated that ethnicity and site were significantly related, Pearson 2 (8, N = 158) = 32.66, p = .000. Follow-up pairwise comparisons conducted using the HolmÂ’s sequential Bonferroni method to control for Type I error at the .05 level across all three co mparisons indicated that the only significant pairwise difference was between USF and UCSD, Pearson 2 (4, N = 139) = 26.43, p = .000. The USF sample was comprised of roughly 17% African American, 1% Asian, 10% Hispanic/Lati no, 67% White/Caucasian, and 5% Â“OtherÂ” participants. In contrast, UCSDÂ’s samp le was roughly comprised of 2% African American, 25% Asian, 5% Hispanic/Lati no, 59% White/Caucasian, and 10% Â“OtherÂ”
41 participants. As shown in Figure 2, the most notable differences were in the greater proportion of African American s at USF (17% vs. 2%) and Asians at UCSD (25% vs. 1%). Thus, while there were no group differenc es based on ethnicity, two of the sites reliably differed in their ethnic distributions. As a result, ethnicity was also considered a potential source of variance in the main analyses. Figure 2. Participant ethnicity by site. 17 0 2 1 5 25 10 11 5 67 68 59 5 16 10 0 10 20 30 40 50 60 70 80 90 100 USFSDSUUCSD Study SitePercentage African American Asian Hispanic/Latino White/Caucasian Other Sexual orientation. Figure 3 displays the distri bution of participants acro ss sexual orientations that were randomized into each of the studyÂ’s groups. Results of a two-way contingency Figure 3. Participant sexual or ientation by group. 94 9898 6 22 0 10 20 30 40 50 60 70 80 90 100 ControlLow levelHigh level Study GroupPercentage Heterosexual Homosexual
42 table analysis indicated that proportions of participants fr om each orientation were not significantly different across gr oups, Pearson Chi-Square (2, N = 157) = 1.50, p = .472. Figure 4 displays the distri bution of participants acro ss sexual orientations that were in each of the studyÂ’s sites. Again, a two-way contin gency table anal ysis indicated that the proportions of participants from each sexual orientation were not significantly different across sites, Pearson Chi-Square (2, N = 157) = 0.74, p = .691. Given the similarity across both groups a nd sites in participant sexual orientation and the relatively small numbers of homosexual participants, the impact of this variable on study outcomes was considered minimal. Therefore, sexual orientation was not treated as a source of additional variance in the main analyses, nor were participants excluded from analysis based on sexual orientation. Figure 4. Participant sexual orientation by site. 100 97 7 0 3 96 0 10 20 30 40 50 60 70 80 90 100 USFSDSUUCSD Study SitePercentage Heterosexual Homosexual Distribution among groups at each site. As depicted in Figure 5, a two-way conti ngency table analysis indicated that the proportion of participants from each of the groups was not significantly different across
43 sites, Pearson Chi-Square (4, N = 158) = .086, p = .999. As a result, the random distribution of participants to groups within each site was considered successful. Figure 5. Participant group assignment by site. 33 37 34 33 32 33 33 32 33 0 10 20 30 40 50 60 70 80 90 100 USFSDSUUCSD SitePercentage Control Low level High level In summary, no reliable di fferences were found among gr oups for age, ethnicity, or sexual orientation. In addi tion, there were no site differen ces for sexual orientation or in terms of the distribu tion of participants into each of the study groups. However, there was variation among sites with regard to bot h participant age and ethnicity with USF having significantly older students and an ethni c distribution that di ffered from that of UCSD. As a result of these findings, both age a nd ethnicity were c onsidered potential sources of additional variance in the main analyses. Given that both variables contributed to site differences, and, in an attempt to di minish the impact of differences among sites with regard to time of year when the interventions were administered (discussed in more detail below), study site was incorporated into the main analyses as a between subjects variable to clarify the role it played in th is study. (The analyses were also conducted assigning age and ethnicity as c ovariates, but this did not re sult in different findings.)
44 Participant Attention and Engagement The first of the study hypotheses was that participants in the high-level engagement group would display greater levels of engagement in their computerized intervention than partic ipants in the low-level engageme nt and control groups. Before assessing engagement, however, it was necessary to ascertain whether participants were attending to their com puterized programs. To this end, every participant completed a content quiz after viewing his program as a manipulation check to determine roughly if he was attending to his program. As shown in Table 2, participants in each of the study groups generally did well on the quiz, correctly answering between 3 and 4 questions out of 5. Table 2 Mean Participant Performance on 5-point Content Summary Score Group Mean (SD) Site Mean (SD) Control (n=54) 3.63 (.81) USF (n=78) 3.64 (.64) Low level (n=52) 3.52 (.70) SDSU (n=19) 3.37 (.90) High level (n=52) 3.65 (.62) UCSD (n=61) 3.62 (.73) Group total (n=158) 3.60 (.71) S ite total (n=158) 3.60 (.71) To test the engagement hypothesis, a 3 (g roup) X 3 (site) analysis of variance was also conducted on the summary engagement sc ores (minimum score = 0, maximum score = 18). Results indicated a significant main effect for group F(2, 149) = 74.63, p < .001, partial 2 = .500, but not for site F (2,149) = 1.10, p = .335, partial 2 = .015, or for the interaction of group and site F (4,149) = 1.94, p = .107, partial 2 = .050. As can be seen in Figure 6, post hoc analyses conducted with DunnettÂ’s C indicated that the high-level engagement group (mean engagement score = 13.38) reliably reported being more engaged in their intervention than the low-level engagement group whose mean
45 engagement score was 10.42 ( p < .001) and control group ( p < .001) whose mean engagement score was 5.46. In addition, the low-level engagement group reliably reported being more engaged in their intervention than the control group ( p < .001). Figure 6. Participant engagement by group (min = 0, max = 18). 5.46 10.42 13.38 0 2 4 6 8 10 12 14 16 GroupMean Engagement Score Control Low level High level Thus, it appeared that the participants overall attended fairly well to their respective interventions In addition, there was support for the first study hypothesis in that participants in the high-level engagement group re ported being more engaged by their intervention than other group members. Although no predictions were made with regard to the relative levels of engagement for the low-level engagement and control groups, findings suggest that low-level engagement group members were generally more engaged in their interventi on than control group members. Once support was found for the first hypothesis, analyses continued with those addressing the outcome variables. Alcohol Expectancies The second study hypothesis was that th e high-level engagement group would display greater decreases in so cial/sexual expectancies by foll ow-up than either of the
46 other groups. To address this, participant scores for the AEC positive/arousing octant, arousing octant, and positive octant were ex amined because they are comprised of the social/sexual expectancies addressed by the e xpectancy challenge material. (Examples of positive/arousing octant expectancies are that drinking alcohol makes one erotic, horny, and lustful. Examples of arousing octant expectancies are that drinking alcohol makes one appealing, attractive, and beautiful. Examples of positive octant expectancies are that drinking alcohol makes one outgoing, soci able, and social.) Because there was no missing data, sample sizes remained consta nt across groups and sites (USF n=78, SDSU n=19, UCSD n=61) for every analysis. Positive/Arousing octant scores. A 2 (time) X 3 (group) X 3 (site) mixed model analysis of variance was conducted to detect changes in participant po sitive/arousing expectanci es from pretest to follow-up. The results for the ANOVA indicated no significant main effect for time, WilksÂ’ lambda = .981, F (1,149) = 2.82, p = .095, multivariate 2 = .019, time by group interaction, WilksÂ’ lambda = .994, F (2,149) = .45, p = .641, multivariate 2 = .006, time by site interaction, WilksÂ’ lambda = .967, F (2,149) = 2.57, p = .080, multivariate 2 = .033, or three-way interaction of time by group by site, WilksÂ’ lambda = .993, F (4,149) = 0.26, p = .902, multivariate 2 = .007. Thus, positive/arousing octant expectancy scores appeared not to change as a result of a ny of the group interventions. See Figure 7.
47 Figure 7. Positive/Arousing octant scores (max=18) as a function of group. 11.27 10.23 9.6 11.38 10.92 10.36 0 1 2 3 4 5 6 7 8 9 10 11 12 PretestFollow-up Assessment TimePositive/Arousing Octant Score Control Low level High level Arousing octant scores. A 2 (time) X 3 (group) X 3 (site) mixed model analysis of variance was also conducted to detect changes in participant arous al expectancies from pretest to follow-up. As with the positive/arousing expectancy sc ores, the results for the ANOVA indicated no significant main effect for time, WilksÂ’ lambda = .997, F (1,149) = 0.52, p = .473, multivariate 2 = .003, time by group interaction, WilksÂ’ lambda = .980, F (2,149) = 1.55, p = .216, multivariate 2 = .020, time by site interaction, WilksÂ’ lambda = .975, F (2,149) = 1.94, p = .147, multivariate 2 = .025, or three-way time by group by site interaction, WilksÂ’ lambda = .988, F (4,149) = 0.44, p = .779, multivariate 2 = .012. Again, arousing octant expectancy sc ores appeared unaffected by any of the interventions. See Figure 8.
48 Figure 8. Arousing octant scores (max =18) as a function of group. 7.89 6.75 7.8 6.69 7.65 6.42 0 1 2 3 4 5 6 7 8 9 PretestFollow-up Assessment TimeArousing Octant Score Control Low level High level Positive octant scores. When conducted on positive octant scor es, which were normalized by squaring them, the 2 (time) X 3 (group) X 3 (site) mi xed model analysis of variance indicated slightly different results. There was still no significant main effect for time, WilksÂ’ lambda = .983, F (1,149) = 2.60, p = .109, multivariate 2 = .017, interaction of time and group assignment, WilksÂ’ lambda = .970, F (2,149) = 2.33, p = .100, multivariate 2 = .030, or three-way interaction of time by group by site, WilksÂ’ lambda = .991, F (4,149) = 0.35, p = .845, multivariate 2 = .009. There was, however, a significant time by site interaction (WilksÂ’ lambda = .930, F (2,149) = 5.65, p = .004, multivariate 2 = .070), suggesting that the change in positive octant scores over time was different for different sites. As shown in Figure 9, the change in positiv e octant scores over time for USF was significantly different from the change in scores for both SDSU, WilksÂ’ lambda = .957, F (1,91) = 4.09, p = .046, multivariate 2 = .043, and UCSD, WilksÂ’ lambda = .933,
49 F (1,133) = 9.48, p = .003, multivariate 2 = .067. In this case the positive octant scores increased over time for USF while they d ecreased for both SDSU and UCSD suggesting that, by follow-up, USF particip ants reported increases in th e frequency with which they believed that alcohol makes one outgoing, soci al, or sociable. The frequency of these beliefs decreased for SDSU and UCSD participants by follow-up. Figure 9. Time by site interaction fo r positive octant score (max=18; shown with non transformed data). 12.65 11.95 13.47 14.26 12.44 13.57 0 2 4 6 8 10 12 14 16 PretestFollow-up Assessment TimePositive Octant Score USF (n=78) SDSU (n=19) UCSD (n=61) In summary, the three groupsÂ’ expectancy octant scores appeared unaffected by any of the interventions. As such, results did not support the hypothesi s that participants in the high-level engagement group woul d display greater changes in alcohol expectancies than the low-level engagement and control groups. Alcohol Consumption The final hypothesis proposed in this study was that level of drinking would decrease for the high-level engagement group relative to th e low-level engagement and control groups by one-month follow-up. Because the expectancy challenge intervention focuses more on decreasing quantity than fre quency of drinking, the methods chosen for
50 analyzing drinking data focused on this outcome More specifically, drinking data were conceptualized and analyzed in three ways : (1) mean drinks consumed per day, (2) quantity/frequency (total numbe r of drinks consumed/number of drinking days), and (3) proportion of binge days (where a binge is defi ned as 5 or more drinks in one day and the proportion of binge days equaled the number of binge days/numbe r of drinking days). Del Boca et al. (2004) reporte d that college student drin king is contingency driven such that it tends to increase around holiday s and decrease when academic requirements intensify. Thus, in an effort to minimize s ite differences both in terms of local school holidays and time of year, drinking data we re excluded from analysis when they coincided with holidays in which students were off from classes. However, drinking data falling on midterm and final exam days were included in analyses because of the variability in individual st udentsÂ’ exam schedules, particularly across sites and during summer classes. In addition, not all students have midterm exams or exams every day of finals, and some have exams the week before finals week. Thus, as a rule, any day when classe s were cancelled due to a holiday was removed from analysis. In addition, because students would likely begin drinking more heavily the night before a free day, the day be fore a free day was removed from analysis. For major holidays such as spring break and winter or summer recess, the weekends preceding and following such holidays were removed from analysis as well. Stated more specifically, at USF, spri ng break, summer break, Memorial Day, and Independence Day were removed from student sÂ’ drinking data. At SDSU, Thanksgiving, winter recess, Martin Luther King Jr. holiday, and spring break were removed from
51 studentsÂ’ drinking data. At UCSD, PresidentÂ’s Day, spring break, Memorial Day, and a local drinking holiday called the sungod fe stival (which, although classes are not cancelled, is widely accepted as the heaviest drinking day of the academic year) were removed from studentsÂ’ drinking data. Mean drinks consumed per day. To obtain this measure, the total number of drinks each participant consumed was divided by the total number of days in his particular Timeline Follow-Back (i.e., after removing holidays) during pretest and again dur ing follow-up. The mean drinks variable was normalized by square root transformati on and then entered into a 2 (time) X 3 (group) X 3 (site) mixed m odel analysis of variance. The results for the analysis indicated a si gnificant main effect for time such that all three groups decreased in mean drinks consumed by follow-up, WilksÂ’ lambda = .768, F (1,149) = 45.11, p < .001, multivariate 2 = .232. The interacti on of time with group was not significant, WilksÂ’ lambda = .989, F (2,149) = 0.81, p = .445, multivariate 2 = .011, nor was the interaction of time with site, WilksÂ’ lambda = .971, F (2,149) = 2.23, p = .112, multivariate 2 = .029, or the three-way interaction of time by group by site, WilksÂ’ lambda = .940, F (4,149) = 2.36, p = .056, multivariate 2 = .060. Thus, participants in all three study gr oups decreased their mean drinks by follow-up. There were no differences among gr oups or among sites with regard to the level of change over time. As a result, no in tervention appeared more favorable that the others with regard to the m ean number of drinks consumed per day. Figure 10 displays the change in mean drinks acro ss groups from pretest to follow-up.
52 Figure 10. Mean drinks consumed per day at pretest and follow-up for each group (shown with non transformed data). 1.37 1.77 1.34 1.82 1.69 2.31 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 PretestFollow-up Assessment TimeMean Drinks Per Day Control Low level High level Quantity/Frequency. To obtain this measure, the total number of drinks each participant consumed was divided by the total number of drinking days in his particular Timeline Follow-Back (i.e., after removing holidays) during pretest and again during follow-up. The Q/F of drinking variable was normalized by taki ng the square root of the s quare root transformation and then entered into a 2 (time) X 3 (group) X 3 (site) mixed model analysis of variance. As with the mean drinks data, results indicated a significant main effect for time such that all three groups decreased in Q/ F of drinking by follow-up, WilksÂ’ lambda = .948, F (1,142) = 7.74, p = .006, multivariate 2 = .052. Non significant results were indicated for the time by group interaction, WilksÂ’ lambda = .997, F (2,142) = 0.22, p = .800, multivariate 2 = .003, for the time by site interaction, WilksÂ’ lambda = .997, F (2,142) = 0.21, p = .812, multivariate 2 = .003, and for the three way interaction of time by group by site, WilksÂ’ lambda = .986, F (4,142) = 0.52, p = .721, multivariate 2 =
53 .014. In addition, there was a between-subjects effect for study site such that, summing across groups and time, USF participants were significantly lower on the Q/F of drinking variable than both SDSU and UCSD participants, F (2,142) = 3.918, p = .022, partial 2 = .052. Just as with mean drinks, participan ts in all three study groups displayed decreases in their Q/F of drinking by fo llow-up. There were no differences among groups or among sites with regard to the level of change over time even though USF participants began and ended at lower levels on the Q/F va riable. Again, these findings did not suggest that one inte rvention led to superior outco mes relative to the others. Figure 11 shows the change in Q/F of drinki ng across groups from pretest to follow-up. Figure 11. Quantity/Frequency of drinking at pretest and follow-up for each group (shown with non transformed data). 6.09 6.48 4.95 5.9 4.8 5.84 0 1 2 3 4 5 6 7 PretestFollow-up Assessment TimeQ/F Drinking Control Low level High level Proportion of binge days. As previously stated, participants were considered to have binged on any day in which they consumed five or more drinks dur ing that day. This variable was created by summing the number of binge days (after re moving holidays) and dividing that sum by
54 the number of drinking days during pretest a nd again during follow-up. The variable was then entered into a 2 (time) X 3 (group) X 3 (site) mixed model analysis of variance. The results of the analysis indicated no significant main effect for time (WilksÂ’ lambda = .981, F (1,142) = 2.78, p = .098, multivariate 2 = .019). There were also no significant interactions (tim e by group WilksÂ’ lambda = .998, F (2,142) = 0.14, p = .871, multivariate 2 = .002, time by site WilksÂ’ lambda = .992, F (2,142) = 0.55, p = .579, multivariate 2 = .008, time by group by site WilksÂ’ lambda = .981, F (4,142) = 0.69, p = .600, multivariate 2 = .019). Thus, as displayed in Figure 12, there was virtually no change in the proportion of binge drinking days among groups from pretest to follow-up; there were also no differences over time among the study sites. Figure 12. Proportion of binge days at pretest and follow-up for each group. 0.57 0.65 0.5 0.56 0.49 0.53 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 PretestFollow-up Assessment TimeMean Proportion of Binge Days Control Low level High level In summary, results of the analyses did not support the thir d and final hypothesis that the high-level engageme nt group would decrease their le vel of drinking more than the low-level engagement and control groups Instead, all three groups appeared to decrease in both their mean drinks per da y and quantity/frequency of drinking by one-
55 month follow-up. While the overall decrease in drinking levels for the three groups is a positive outcome, these results did not provide support for the efficacy of this studyÂ’s expectancy challenge interven tions over and above that of the control groupÂ’s assessment and defensive driving intervention. Exploratory Analyses Of the three study hypotheses, only the fi rst (that the high-level engagement group would report being more engaged by thei r intervention) was supported by analysis. However, there were a number of sources of additional variance that were not initially incorporated into the study a nd that may have contributed to a lack of findings with regard to the second and third hypotheses. One source of additional vari ance was that data was collected at three different sites rather than one. The sites not only di ffered in important demographic characteristics such as age and ethnicity but also differed w ith regard to which holidays were recognized by the universities (possibly affecting drinking levels). Another source of variance was created by the long period of time over whic h the study was run. Un like the Darkes and Goldman (1993, 1998) studies, which took pl ace over approximately one and three months (respectively), data co llection for this project pro ceeded over the period of an entire calendar year. Based on the findings of Del Boca et al. (2004), it is not hard to imagine that contingencies affecting stude nt drinking over winter recess may differ significantly from those contingencies in e ffect during finals w eek, spring break, or summer session.
56 As reported earlier, steps were taken to reduce the impact of these additional sources of variance such as including site as a between subjects factor and removing holidays from each participantÂ’s drinking da ta. However, it may be that real group differences were obscured in the process of removing the additional sources of variability. Therefore this section addresses the attempt to cull cleaner, more homogenous samples from the database to explore, very tentatively, whether any conditions exist in which future, similar interventions might be more successful. During the process of conducting the explorat ory analyses, data were divided into twelve different subsamples reflecting the atte mpt to separate participants by site, by time of year when they participated in the study, and by level of drinking where the split between light and heavy drinking was based on the heavy drinking sample shown to benefit most from the expectancy challeng e in Darkes and Goldman (1998). Finally, Caucasians in general and t hose between the ages of 18 a nd 25 were also examined as groups because they comprised a large propor tion of the whole sample and because, as discussed earlier, they tend to have the highe st drinking trajectories during typical college age years. Table 3 lists the specific ways in which the data were grouped. Asterisks in the table indicate subsamples in which significant results or interesting trends were found. These results will be further described below.
57 Table 3 Data Groupings for Exploratory Analyses USF SDSU Site *UCSD USF with spring break in pretest *UCSD with spring break in pretest USF and UCSD with spring break in pretest *UCSD with spring break in follow-up Time of Year *USF summer 2003 and summer 2004 Caucasians *Heavier drinkers with mean drinks greater than or equal to 2.16 per day (comparable to heavier drinking sample in Darkes and Goldman, 1998) *Caucasian heavier drinkers (mean of 2.16 or more drinks per day) ages 18 to 25 Drinker Level *Lighter drinkers (mean less than 2.16 drinks per day) Note. denotes significant results or interesting trends meriting elaboration. In the process of conducting the explorat ory analyses, two interesting findings began to materialize. The first was that, when there were trends where the challenge groups showed decreases in expectancies re lative to the control group, the positive octant tended to be the one most often affecte d. The second finding was that analyses conducted with lighter drinking subsamples te nded to be the ones that showed more favorable results for the challenge groups with regard to changes in expectancies and drinking. Interesting trends and even significant results were observed in a number of seemingly different subsamples that, upon fu rther inspection, had in common that their participants on average drank less than th e 2.16 drinks per day during pretest adopted from the Darkes and Goldman (1998) study. Before discussing particular subsamples, however, it is important to note that, while th e full sample from which these subsamples were drawn (i.e., with holidays intact) had a daily pretest mean of 2.29 drinks, the sample
58 upon which the main study analyses were conduc ted (i.e., with holidays removed) had a daily mean of only 1.97 drinks during pretest. This latter mean is also below the 2.16 drinks per day mark. Nevertheless, the subsam ples appear to yield more favorable results Â– perhaps due to different proportions of pa rticular ethnic gro ups, participants of particular ages, time of year during which data was collected, ch ance, or some other factors. UCSD. The first subsample was composed of par ticipants from the UCSD site. It was comprised of 61 participants who on average drank 1.90 drinks per day during pretest. The subsample was analyzed in a 2 (time) X 3 (group) mixed model analysis of variance. As can be seen in the top line graph of Figure 13, the high-level engagement and lowlevel engagement group appeared to have a sh arper rate of decrease in positive octant expectancy scores than the control group. The group by time interaction, however, was not significant, WilksÂ’ lambda = .950, F (2,58) = 1.54, p = .223, multivariate 2 = .050. As shown in the bottom line graph of Figure 13, a similar pattern held for the mean drinks per day. Again, however, the gr oup by time interaction was not significant, WilksÂ’ lambda = .956, F (2,58) = 1.34, p = .271, multivariate 2 = .044. Thus, this subsample fits both the patter n of favorable trends for the challenge groups in lower drinking samples and positive oc tant expectancies being most affected. In addition, a decreasing trend in some alcohol expectancies coincided with a decreasing trend in drinking levels for the challenge groups.
59 Figure 13. Positive octant scores (max=18) and mean drinks per day for UCSD subsample. 13.76 12 13.95 13.45 11.5 13.3 0 2 4 6 8 10 12 14 16 PretestFollow-up Assessment TimePositive Octant Score Control Low level High level 1.68 1.44 1.29 1.91 1.92 1.86 0 0.5 1 1.5 2 2.5 PretestFollow-up Assessment TimeMean Drinks Per Day Control Low level High level UCSD with spring break in pretest. The next subsample was composed of pa rticipants from the UCSD site who received their interventions shortly after spring break such that their pretest drinking data included spring break. This subsample held 36 participants with a mean pretest number of daily drinks equaling 1.89. The subsample wa s also analyzed in a 2 (time) X 3 (group) mixed model analysis of variance. As can be seen in the top line graph of Figure 14, the two challenge groups appeared to have a sli ghtly sharper rate of decrease in positive
60 octant expectancy scores than the contro l group. The group by time interaction was not significant, however, WilksÂ’ lambda = .933, F (2,33) = 1.18, p = .319, multivariate 2 = .067. Figure 14. Positive octant scores (max=18) and positive/ arousing octant scores (max=18) for UCSD subsample with spring break in pretest. 13.92 14.38 13.45 10.91 11.75 13.83 0 2 4 6 8 10 12 14 16 PretestFollow-up Assessment TimePositive Octant Score Control Low level High level 10.85 11.08 8.64 8.55 11.42 8.83 0 2 4 6 8 10 12 14 PretestFollow-up Assessment TimePositive/Arousing Octant Score Control Low level High level As seen in the bottom line graph of Figure 14, the high-level engagement group had a much larger rate of decrease in positive/ arousing octant expectancy scores than the control group, while the low-le vel engagement group experienced a slight increase in
61 their slope relative to th e control group. The group by time interaction was not significant, WilksÂ’ lambda = .874, F (2,33) = 2.37, p = .109, multivariate 2 = .126. While the trends for changes in expectanci es were not consistent decreases for both challenge groups and there were no correspondin g decreasing trends in drinking levels, it is nonetheless interestin g to note that the trends that we re found occurred in a lighter drinking subsample and positive octant expectancies were again affected. UCSD with spring break in follow-up. This subsample was comprised of UCSD participants whose follow-up assessment period overlapped with spring brea k. There were 24 participants in this subsample, and they consumed an average of 1.91 drinks per day during pretest. As before, the subsample was analyzed in a 2 ( time) X 3 (group) mixed model analysis of variance. For this subsample, ther e were three interesting results. The first was a statis tically significant effect for the time by group interaction for the mean drinks per day, WilksÂ’ lambda = .674, F (2,21) = 5.09, p = .016, multivariate 2 = .326. Follow-up comparisons revealed signifi cant differences in the change over time between the control and low-level en gagement groups, WilksÂ’ lambda = .714, F (1,14) = 5.60, p = .033, multivariate 2 = .286, and between the control and high-level engagement groups, WilksÂ’ lambda = .651, F (1,13) = 6.98, p = .020, multivariate 2 = .349. These findings are illustrated in the top line graph of Figure 15 where the slope of the control group remains fairly constant ove r time while the slopes of the challenge groups decrease over time.
62 Figure 15. Mean drinks per day and arousing octant scores (max=18) for UCSD subsample w ith spring break in follow-up. 1.43 1.49 1.45 2.27 1.03 1.87 0 0.5 1 1.5 2 2.5 3 PretestFollow-up Assessment TimeMean Drinks Per Day Control Low level High level 5.78 6.86 7.57 7.89 6.88 4.75 0 1 2 3 4 5 6 7 8 9 PretestFollow-up Assessment TimeArousing Octant Score Control Low level High level The second interesting result was another significant effect for the time by group interaction for the arousing octant exp ectancy score, WilksÂ’ lambda = .715, F (2,21) = 4.18, p = .030, multivariate 2 = .285. In this case, however, follow-up comparisons revealed significant differences in the change over time between the low-level engagement group and the high level e ngagement group, WilksÂ’ lambda = .641, F (1,15) = 8.40, p = .011, multivariate 2 = .359. There was unfortunately a reliable difference in
63 the two group slopes where the low-level group decreased and the high-level group increased in arousing octant scores (s ee bottom line graph of Figure 15). The third interesting finding for this s ubsample was a trend in which the control group increased in positive octant expectan cy scores over time while the low-level engagement group decreased and the high-le vel engagement group remained about the same (see Figure 16). The group by time in teraction was not si gnificant, however, WilksÂ’ lambda = .937, F (2,21) = .708, p = .504, multivariate 2 = .063. Figure 16. Positive octant scores (max=18) for UCSD subsample with spring break in follow-up. 13.43 13 12.22 13.44 12.38 12.5 0 2 4 6 8 10 12 14 PretestFollow-up Assessment TimePositive Octant Score Control Low level High level In summary, then, this lighter drinking subsample actually produced statistically significant findings where the mean number of drinks per day decreased for the experimental groups relative to the control group. Contrary to expectation, however, these findings were accompanied by a statistica lly reliable increase in the slope of the high-level engagement groupÂ’s arousing octant score and a trend for the positive octant score to remain unchanged over time. Neverthe less, in line with e xpectation, the reliable
64 decrease in the low-level engagement groupÂ’ s mean number of drinks per day was also accompanied by a reliable decrease in that gr oupÂ’s arousing octant score and a decreasing trend in its positive octant score over time (while the control groupÂ’s mean number of drinks remained unchanged and the groupÂ’s pos itive octant score di splayed an increasing trend over time). USF summer 2003 and summer 2004. This subsample was comprised of USF part icipants who participated in the study during one of the two summers in which data was collected. There were 23 participants in this subsample, and they consumed an av erage of 1.71 drinks per day during pretest. As before, the subsample was analyzed in a 2 (time) X 3 (group) mixed model analysis of variance. The first interesting finding was a statis tically significant effect for the time by group interaction for the quantity/freque ncy of drinking, WilksÂ’ lambda = .689, F (2,20) = 4.52, p = .024, multivariate 2 = .311. Follow-up comparisons revealed significant differences in the change over time betw een the control and low-level engagement groups, WilksÂ’ lambda = .661, F (1,13) = 6.66, p = .023, multivariate 2 = .339, and between the low-level and high-level e ngagement groups, WilksÂ’ lambda = .726, F (1,13) = 4.90, p = .045, multivariate 2 = .274. These findings are il lustrated in the top line graph of Figure 17 where the slope of the low-level engagement group decreases while the slope of the control group increases and the slope of the high-level engagement group remains fairly constant over time.
65 The next interesting finding for this s ubsample was a trend in which the control group increased in proportion of binge drinking days over time while the low-level engagement group decreased and the high-le vel engagement group remained about the same (see bottom line graph of Figure 17). The group by time interaction was not significant, WilksÂ’ lambda = .862, F (2,20) = 1.61, p = .226, multivariate 2 = .138. Figure 17. Quantity/Frequency of drinking and proportion of binge days for USF subsample for summer 2003 and summer 2004. 4.03 3.52 4.88 3.43 4.61 4.95 0 1 2 3 4 5 6 PretestFollow-up Assessment TimeQ/F Drinking Control Low level High level 0.39 0.3 0.46 0.26 0.37 0.47 0 0.2 0.4 0.6 0.8 1 PretestFollow-up Assessment TimeMean Proportion of Binge Days Control Low level High level
66 Again, statistically significant results and trends favored the low-level engagement group (but not the high-level e ngagement group) over the control group for two drinking variables in this lighter dri nking subsample. There were no favorable findings, however, for expectancy scor es, positive octant or otherwise. Heavier drinkers. There were two subsamples of heavier dr inkers. The first was one where heavier drinkers were defined as c onsuming greater than or equa l to 2.16 drinks per day on average. The second heavier drinking subsample had the same drinking criterion but was further restricted to Caucasians aged 18 to 25 years. The first heavier drinking subsample held 66 participants who drank an average of 3.68 drinks per day. Analysis in a 2 (time) X 3 (group) X 3 (site) mixed model analysis of variance revealed a trend for the positive octant expectancy score where the control group increased over time while the two cha llenge groups decreased. The group by time interaction was not signifi cant, WilksÂ’ lambda = .968, F (2,57) = .93, p = .399, multivariate 2 = .032. See Figure 18. Figure 18. Positive octant expectancy score (max=18) for heavier drinker subsample (mean daily drinks = 2.16 or higher). 13.83 13.3 12.97 13.84 13.12 12.52 0 2 4 6 8 10 12 14 PretestFollow-up Assessment TimePositive Octant Score Control Low level High level
67 The second heavier drinking subsample (c onsisting of Caucasians aged 18-25) held 47 participants who drank an average of 3.72 drinks per day. As with the other heavier drinking subsample, analysis in a 2 (time) X 3 (group) X 3 (site) mixed model analysis of variance revealed a trend for the positive octant expectancy score. In this case, the control group increased slightly while the low-level engagement group decreased slightly and the high-level engagement group decreased more substantially over time. The group by time interaction was again not significant, WilksÂ’ lambda = .941, F (2,38) = 1.19, p = .314, multivariate 2 = .059. See Figure 19. Figure 19. Positive octant expectancy score (max=18) for heavier drinker subsample (mean daily drinks = 2.16 or higher) comprised of Caucasians aged 18-25 years. 14.56 13.64 11.92 14.02 14.03 13.23 0 2 4 6 8 10 12 14 16 PretestFollow-up Assessment TimePositive Octant Score Control Low level High level These subsamples were comprised of h eavier drinkers relative to the other subsamples. However, in both cases there we re trends where positive octant expectancy scores decreased over time for both challe nge groups relative to the control group.
68 Lighter drinkers. This last subsample was created to cont rast the heavier drinking subsamples and to test further the theory that more favorable trends for the challeng e groups occurred in lighter drinking subsamples. As such, th e final subsample was comprised of all individuals who consumed less than 2.16 dr inks per day on average. It held 92 participants who actually consumed 1.30 drinks per day on average. This subsample was also analyzed in a 2 (time) X 3 (group) X 3 (site) mixed model analysis of variance and resulted in an interesting trend for the positive octant expectancy score. In this case, the control group increased slightly while the high-level engagement group decreased slightly and the low-level engagement group decreased more sharply over time. The group by time interaction was not significant, WilksÂ’ lambda = .961, F (2,83) = 1.67, p = .196, multivariate 2 = .039. (See Figure 20.) Figure 20. Positive octant expectancy score (max=18) for lighter drinker subsample (mean daily drinks < 2.16). 14.23 13.82 11.41 13.28 11.78 12.03 0 2 4 6 8 10 12 14 16 PretestFollow-up Assessment TimePositive Octant Score Control Low level High level This trend did fit the patter n of more favorable trends for the challenge groups in lighter drinking subsamples, but, in essence, it displayed the same pattern of only
69 favorable trends in the positive octant scores found in the two heavier drinking subsamples. Furthermore, no favorable trends were found for drinking variables in this subsample. In summary, given the explor atory nature of these analyses, it is impossible to draw any firm conclusions. However, as orig inally noted at the beginning of this section, when more homogenous subsamples were analyzed, favorable trends occasionally resulted. Additionally, there appeared to be some commonality in these findings such that they were detected in lighter drinking subsamples or involved positive octant expectancy scores (or both). While there we re instances where tre nds in both favorable decreases in expectancy scores and drinking outcomes coincided, this occurred far less consistently. In addition to the previous investigat ions, exploratory analyses were also conducted using the main variables of interest (i.e., site, time of year and drinker level) as interactions terms in the hope of gaining mo re statistical power. The results of these analyses were not significantly different from those previously discussed.
70 Discussion The purposes of this investigation were to examine the effect of varying the amount of participant engagement on alcohol expectancies and dr inking outcomes during a social/sexual alcohol expectancy challenge intervention an d to test of the efficacy of administering an expectancy challenge in a co mputerized format rather than in vivo. It was hypothesized that participan ts experiencing a challenge intervention that was more interactive would find it more engaging and would exhibit greater changes in alcohol expectancies and drinking levels, presumably because they paid more attention to and processed the intervention more deeply. Results of this study suggested that, as predicted, participan ts in the high-level engagement group reported finding their interventi on more interesting than participants in either the low-level engagement group or th e control group. Members of the low-level engagement group, in turn, found their interven tion more interesting than members of the control group did. Given that presentation format differed from high-level engagement group to low-level engagement group and that content differed from low-level engagement group to control group, it would s eem that content as well as presentation format contributed to how engaged particip ants were with their interventions. Interestingly, regardless of their self-reported level of engagement, members of all three groups performed equally well when quizze d on the content of their interventions. Because the mean accuracy level was around 70% for each of the groups, a ceiling effect
71 is not implicated. Thus, within this study, higher levels of self-re ported engagement did not lead to greater levels of mastery over the intervention material. If this is the case, it may help explain why the remaining study hypotheses were not supported. With regard to alterations in social/se xual alcohol expectancies targeted by the expectancy challenge interven tion, none of the groups show ed significant changes from pretest to one-month follow-up. Neverthele ss, all three groups displayed significant decreases over time on two of the three alc ohol consumption variables (mean drinks per day and quantity/frequency of drinking). Thes e findings are curious given that alcohol expectancies are the putative mechanism for ch ange in the expectancy challenge, and an examination of potentially contributing as we ll as confounding factors is warranted. One explanation for drinking changes without expectancy changes is that alterations in alcohol expectan cies, while leading to decreas es in drinking in the two experimental groups, may themselves have b een shorter lived and vanished by one-month follow-up. However, such a supposition is not supported by other ex pectancy research and fails to account for the similar level of decrease in drinking found in the control group, which did not experience an expectancy challenge. Another possible explanati on for drinking changes in all three groups that might occur without expectancy changes is that participants became more aware of their drinking habits while completing the Timeline Follow-Back at pretest. This enhanced self-awareness could have contributed to a decrease in their drinking over the follow-up period, assuming they believed there was such a need.
72 A related explanation is that of experime ntal demand. In this case, participants could have become aware that the research goal was to decrease drinking and tried to help the researcher achieve that goal, either by drinking less or by reporting a decrease in drinking. Care was taken to hide the true nature of the study from the control group by adding questionnaires about safe driving practic es and to refrain from sharing with any group members which information would be gather ed at follow-up. In the latter instance, participants were only told th at they would be repeating so me of the same questionnaires at follow-up, not which ones. Nevertheless, the true nature of th e study was still likely implicit inasmuch as participants were in a psychology study that asked numerous questions about alcohol consump tion and beliefs about the effect s of alcohol. If this were the case, though, one might wonder why member s of the two experimental groups did not also endorse decreases in their alcohol expect ancies at follow-up. The answer to this question may lie in a dosing effect. While the notion of counting how many drinks one consumes is not a foreign concept to most drinkers, the idea of noting and challenging alcohol expectan cies likely is. Perhaps members of the experimental groups did not e ndorse lasting changes in alco hol expectancies because the one brief exposure to the concept afforded by this study was insufficient to produce a lasting effect. A more long-term reduction in alcohol expectancies may require multiple exposures to expectancy challenging proce dures as provided in both of Darkes and GoldmanÂ’s 1993 and 1998 studies. In this rega rd, it would have been informative to assess for changes in alcohol expectancies im mediately after the in tervention in addition to the one-month follow-up.
73 Another potential confound for this study wa s that participants were recruited from three separate sites that are disparate with regard to geography and academic as well drinking reputations. Further, the number of participants recruited from each site differed significantly as did important demographic ch aracteristics such as age and ethnicity. Attempts were made when reviewing, cleani ng, and analyzing the da ta to help identify and compensate for site effects, but the study would undoubtedly have been more powerful with a more homogenous sample collected at one site only. A related potential confo und lay in the extended period of time over which data was collected. Unlike the Darkes and Gold man studies (1993, 1998), participants were recruited for this study over the period of a full calendar year. As previously noted, Del Boca et al. (2004) observed significant ch anges in drinking pa tterns throughout the academic year. Even though attempts were made to help control for these contingencydriven fluctuations in drinking that were idiosyncratic to si te and time of data collection, it is very likely that some Â“noiseÂ” remained in the data. Exploratory analyses were conducted to address some of these confounding issues by reducing variability through the creation of smaller, more homogenous subsamples and repeating the main study analyses on them. Although power issues were undoubtedly a problem as well as type I errors, these analyses still s uggested some trends in the data that might inform future challenge studies. One such trend was that positive octant e xpectancies (i.e., that alcohol makes one outgoing, sociable, and social) tended to d ecrease over time for the challenge groups relative to the cont rol group. Such a trend might be e xplained by the fact that the video
74 in the high-level engagement group and the desc ription of it in the low-level engagement group depicted a group of men in a social s ituation and their err oneous attributions regarding the effects of alcohol on each otherÂ’s sociability. While the original Darkes and Goldman (1993, 1998) protocol included such a focus for one of its sessions, that focus was balanced by a second session addr essing more arousing (i.e., that alcohol makes one appealing, attractive, and beautiful ) and positive/arousing expectancies (i.e., that alcohol makes one erotic, horny, and lustful) by having the participants rate slides of females on level of attractiveness. This content was omitted from the current study protocol. It may also be that expectancies did cha nge as a result of the intervention but were not detectable by one-month follow-up either due to their short longevity or because of the assessment measure used to detect them. While the Alcohol Expectancy Circumplex was used in this study and detected no cha nges, the Expectancy Context Questionnaire, which reputedly measures shorter-term change s in alcohol expectancies, was used with more success in Darkes and Goldman (1993, 1998). Another interesting finding of the exploratory analyses was that the challenge groups in lighter drinking subsamples tende d to yield a greater number of favorable outcome trends than their heavier drinking subsample counterparts. This finding is somewhat perplexing given the greater effi cacy of the Darkes and Goldman (1993, 1998) protocol for their heavier drinking participan ts. Perhaps the difference lies in the more social rather than sexual focu s of the current challenge in that the heavier drinkers are more affected by challenges to their sexual expectancies.
75 Another explanation potentially lies in the single-sessi on nature of the current studyÂ’s challenge compared with the thr ee sessions and the expectancy awareness homework assignments of the Darkes and Gold man (1993, 1998) protocol. It may be that a certain threshold is reached after more than one exposure to expectancy challenge material that causes greater or more lasting changes in participants, particularly heavier drinkers. Given these two trends detected duri ng the exploratory analyses, future expectancy challenges may yield better resu lts by addressing both social and sexual expectancies. They may also do better by providing mo re than one exposure to expectancy challenging material, either through more than one session, through homework assignments, or both. This study was a success in translating the expectancy challenge material to computerized formats that were capable of a ffecting participantsÂ’ levels of engagement. As previously mentioned, participant engage ment was one of four identified possible key components to a successful expectancy challe nge, the others being exposure to a salient expectancy-disconfirming experien ce, inclusion of content that is personally relevant to participants, and use of a homogenous sample of relatively heavy-drinking college males. To the degree that exposure to the other three possible key components was achieved and successfully held constant across groups, this study can be judged as useful in determining that varying partic ipant engagement is not suffi cient to effect changes in alcohol expectancies and dr inking levels alone. Given the relative ease and costeffectiveness of administering computerized interventions, however, future research may
76 be informed by the conclusions drawn and que stions raised from this study and should not abandon this methodology as a means for affecting alcohol consumption.
77 References Alterman, A. I. & Baughman, T. G. (1991). Videotape versus computer interactive education in alcohol ic and nonalcoholic controls. Alcoholism: Clinical and Experimental Research, 15(1), 39-44. Brown, S. A. (1985a). Expectancies versus background in the prediction of college drinking patterns. Journal of Consulting and Clinical Psychology, 53, 123-130. Brown, S. A. (1985b). Reinforcement expectancies and alcoholism treatment outcome after a one-year follow-up. Journal of Studies on Alcohol, 46, 304-308. Brown, S. A., Goldman, M. S., & Christ iansen, B. A. (1985). Do alcohol expectancies mediate drinki ng patterns of adults? Journal of Consulting and Clinical Psychology, 53, 512-519. Bryson, R. (1999). Effectiveness of refusal skills software. Journal of Drug Education, 29(4), 359-371. Christiansen, B. A., & Goldman, M. S. (1983). Alcohol-rel ated expectancies versus demographic/background variables in the prediction of a dolescent drinking. Journal of Consulting and Clinical Psychology, 51, 249-257. Christiansen, B. A., Goldman, M. S., & Brown, S. A. (1985). The differential development of adolescent alcohol expect ancies may predict adult alcoholism. Addictive Behaviors, 10, 299-306. Christiansen, B. A., Smith, G. T., Roehling, P. V., & Goldman, M. S. (1989). Using alcohol expectancies to predict adol escent drinking behavior after one year. Journal of Consulting and Clinical Psychology, 57, 93-99. Cohen, J. (1992). A Power Primer. Psychological Bulletin, 112(1), 155-159. Corbin, W. R., McNair, L. D., & Carter, J. A. (2001). Evalua tion of a treatmentappropriate cognitive intervention for chal lenging alcohol outcome expectancies. Addictive Behaviors, 26(4), 475-488. Corbin, W. R., McNair, L. D., French, K. S., & Black, D. (1998). Effects of an explicit alcohol expectancy challenge. Paper presented at the annual meeting of the Research Society on Alcoholism, Hilton Head Island, SC.
78 Darkes, J. & Goldman, M. S. (1993). Expectancy challenge and drinking reduction: Experimental evidence for a mediational process. Journal of Consulting and Clinical Psychology, 61, 344-353. Darkes, J. & Goldman, M. S. (1998). Expectancy challenge and drinking reduction: Process and structure in the alcohol expectancy network. Experimental and Clinical Psychopharmacology, 6(1), 64-76. Darkes, J., Sheffield, F. D., & Goldman, M. S. (2001, November). Replication of the hierarchical structur e of alcohol expectancies. Paper presented at the annual meeting of the Association for the Advancement of Behavior Therapy, Philadelphia, PA. Del Boca, F. K., Darkes, J., Greenbaum, P. E., & Goldman, M. S. (2004). Up close and personal: Temporal variability in the drinking of individual college students during their first year. Journal of Consulting and C linical Psychology, 72(2), 155-164. Duncan, T. E., Duncan, S. C., Beauch amp, N., Wells, J., & Ary, D. V. (2000). Development and evaluation of an interactive CD-ROM refusal skills program to prevent youth substance use: Â“Refuse to UseÂ”. Journal of Behaviora l Medicine, 23(1), 59-72. Dunn, M. E., Lau, H. C., & Cruz, I. Y. (2000). Changes in activation of alcohol expectancies in memory in re lation to changes in alcohol us e after participation in an expectancy challenge program. Experimental and Clinical Psychopharmacology, 8(4) 566-575. Dunn, M. E., & Goldman, M. S. (1996). Empirical modeling of an alcohol expectancy memory network in elementary school children as a function of grade. Experimental and Clinical Psychopharmacology, 4, 209-217. Eisenberger, R. (1992). Learned industriousness. Psychological Review, 99(2), 248-267. Gerstein, D. R., Gray, F., Epst ein, J., & Ghadialy, R. (1994). Mental Health Estimates From the 1991 National Household Survey on Drugs. Rockville, MD: Substance Abuse and Mental Hea lth Services Administration. Ghosh, A. & Marks, I. M. (1987). Self -treatment of agoraphobia by exposure. Behavior Therapy, 18(1), 3-16. Goldman, M. S. (1999). Expectancy operation: Cognitive/neural models and architectures. In I. Kirsch (Ed.), Expectancy, experie nce, and behavior (pp. 41-63). Washington, D.C.: APA Books.
79 Goldman, M.S. & Darkes, J. The alcohol expectancy circumplex: A neurocognitive approach to expectancy measurement. Manuscript submitted for publication. Goldman, M. S., Greenbaum, P. E., & Darkes J. (1997). A confirmatory test of hierarchical expectancy structure and predic tive power: Discriminant validation of the alcohol expectancy questionnaire. Psychological Assessment, 9, 145-157. Henderson, M. J., & Goldman, M. S. (1987, November). The effect of a social manipulation on expectancies and subsequent drinking Paper presented at the annual meeting of the Association for the Advancement of Behavior Therapy, Boston, MA. Jackson, J. S., Williams, D. R., & Gomb erg, E. S. L. (1998). A life-course perspective on aging and alcohol use and abus e among African Americans. In E. S. L. Gomberg, A. M. Hegedus, & R. A. Zucker (Eds.), Alcohol Problems and Aging (NIDA Research Monograph No. 33. NIH Pub. N o. 98-4163). Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism. Keillor, R. M., Perkins, W.B., & Horan, J. J. (1999). Effects of videotaped expectancy challenges on alcohol cons umption of adjudicated students. Journal of Cognitive Psychotherapy: An International Quarterly, 13(3), 179-187. Maddock, J. E., Wood, M. D., Davidoff, O. J., Colby, S. M., & Monti, P. M. (1999). Alcohol expectancy challeng e and alcohol use: Exami nation of a controlled trial. Paper presented at the annual meeting of the Research Society on Alcoholism, Santa Barbara, CA. Massey, R. F., & Goldman, M. S. (1988, August). Manipulating expectancies as a means of altering alcohol consumption. Paper presented at the annual meeting of the American Psychological Association, Atlanta, GA. Mayer, R. E. & Moreno, R. (2002). Aids to computer-based multimedia learning. Learning and Instruction, 12, 107-119. Musher-Eizenman, D. R. & Kulick, A. D. (2003). An alcohol expectancychallenge prevention program for at-risk college women. Psychology of Addictive Behaviors, 17(2), 163-166. National Institute On Alcohol Abuse and Alcoholism. (2000). Tenth special report to the U. S. Congress on alcohol and health : Highlights from current research (NIH Publication No. 00-1583). Retrieved September 10, 2004, from http://www.niaaa.nih.gov/publications/10report/chap01b.pdf
80 Newman, M. G., Kenardy, J., Herman, S., & Taylor, C. B. (1997). Comparison of palmtop-computer-assisted brief cognitive-behavioral treatment to cognitivebehavioral treatment for panic disorder. Journal of Consulting and Clinical Psychology, 65(1), 178-183. Quinn, E. P., Brandon, T. H., & Copeland, A. L. (1996). Is task persistence related to smoking and substance abuse? The application of l earned industriousness theory to addictive behaviors. Experimental and Clinical Psychopharmacology, 4(2), 186-190. Rather, B. C., & Goldman, M. S. (1994). Drinking-related differences in the memory organization of alcohol expectancies. Experimental and Clinical Psychopharmacology, 2(2), 167-183. Rather, B. C., Goldman, M. S., Roehric h, L., & Brannick, M. (1992). Empirical modeling of an alcohol expectancy memory network using multidimensional scaling. Journal of Abnormal Psychology, 101(1), 174-183. Rickert, V. I., Graham, C. J., Fisher, R., Gottlieb, A., Trosclair, A., & Jay, M. S. (1993). A comparison of met hods for alcohol and marijuan a anticipatory guidance with adolescents. Journal of Adolescent Health, 14(3), 225-230. Seidner, A. L., Burling, T. A., & Mars hall, G. D. (1996). Using interactive multimedia to educate high-risk patients about AIDS and sexually transmitted diseases. Computers in Human Services, 13(4), 1-15. Selmi, P. M., Klein, M. H., Greist, J. H., Sorrell, S. P., & Erdman, H. P. (1990). Computer-administered cognitive-beh avioral therapy for depression. American Journal of Psychiatry, 147(1), 51-56. Sheffield, F.D., Darkes, J., Del Boca, F.K., & Goldman, M.S. (2001). Confirming the structure and va lidity of the Alcohol Expectan cy Inventory Â– Short Form. Alcoholism: Clinical and Experi mental Research, 25 (Supplement), 38A. Sher, K. J., Wood, M. D., Wood, P. K., & Raskin, G. (1996). Alcohol outcome expectancies and alcohol use: A latent variable cross-lagged panel study. Journal of Abnormal Psychology, 105, 561-574. Smith, G. T., Goldman, M. S., Greenbaum, P. E., & Christiansen, B. A. (1995). Expectancy for social facilitation from dri nking: The divergent paths of high-expectancy and low-expectancy adolescents. Journal of Abnormal Psychology, 104, 32-40.
81 Sobell, L.C. & Sobell, M.B. (1992). Timeline follow-back: A technique for assessing self-reported alcohol consumption. In R. Z. Litten,.& J. P. Allen (Eds), Measuring alcohol consumption: Ps ychosocial and biochemical methods (pp. 41-72). Totwa, NJ: Humana Press. Sobell, L.C., Sobell, M.B., Leo, G.I., & Ca ncilla, A. (1988). Reliability of a timeline method: Assessing normal drinke rsÂ’ reports of recent drinking and a comparative evaluation across several populations. British Journal of Addiction, 83, 393402. Stoney, S. & Oliver, R. (1998). Interac tive multimedia for adult learners: Can learning be fun? Journal of Interactive Learning Research, 9(1), 55-81. Taylor, C. B., Agras, W. S., Losch, M., Plante, T. G., & Burnett, K. (1991). Improving the effectiveness of computer-assisted weight loss. Behavior Therapy, 22, 229-236. Tolman, E. C. (1932). Purposive behavior in animals and men. New York: Century Company. U.S. Department of Health and Human Services. (2000). Tenth special report to the U.S. Congress on alcohol and health from the Secretary of Health and Human Services. Rockville, MD: National Institutes of Health, National Institute on Alcohol Abuse and Alcoholism. WebsterÂ’s ninth new collegiate dictionary. (1990). Springfield, MA: MerriamWebster. Wechsler, H., Lee, J. E., Kuo, M., & L ee, H. (2000). College binge drinking in the 1990Â’s: A continuing problem: Results of the Harvard School of Public Health 1999 College Alcohol Study. Journal of American College Health, 48(5), 199-210. Wiers, R. W. & Kummeling, R. H. C. (2004) An experimental test of an alcohol expectancy challenge in mixed gend er groups of young heavy drinkers. Addictive Behaviors, 29, 215-220. Winzelberg, A. J., Taylor, C. B., Sharpe, T., Eldredge, K. L., Dev, P., & Constantinou, P. S. (1998). Evaluation of a computer-media ted eating disorder intervention program. International Journal of Eating Disorders, 24(4), 339-349. Wood, M.D. (1997, January). A critical review of alcohol abuse preventive interventions with college students. Paper presented at th e Center for Alcohol and Addiction Studies, Brown Un iversity, Providence RI.
82 Wooten, B. T. (1995). Challenging alc ohol expectancies: An application to adolescents. Unpublished docto ral dissertation, University of South Florida, Tampa. Zabinski, M. F., Pung, M. A., Wilfley, D. E., Eppstein, D. L., Winzelberg, A. J., Celio, A., & Taylor, C. B. ( 2001). Reducing risk factors for eating disorders: Targeting at-risk women with a computeri zed psychoeducational program. International Journal of Eating Disorders, 29(4), 401-408.
84 Appendix A: Demographics a nd Drinking Styles Questionnaire ____________________________________ _______________ Student name (print neatly) Date Please complete the following so that we may contact you for your follow-up phone appointment: _____________________________ Phone Number ___________________________________ _____________________________ Best day to call Best time to call Please provide answers to the following questions. When were you born? __________/__________/__________ month day year How old are you currently? ___________ (years) What is your GPA? ___________
85 Appendix A (Continued) Please place your responses to the quest ions below on the scantron provided. 1. How would you describe your sexual orientation? (a) Heterosexual (b) Bisexual (c) Homosexual 2. Which ethnicity best describes you? (a) African American, Black (b) Asian (c) Hispanic/Latino/Latina (d) Native American (e) White, Caucasian (f) Other 3. Which of the following best describes you? (a) Never used alcohol (b) Used to drink in the past, but now abstain from alcohol (c) Recovering alcoholic (d) Light drinker (e) Social drinker (f) Moderate drinker (g) Regular drinker (h) Heavy drinker 4. How old were you when you first tried alcohol Â– more than just a few sips? (a) Never used alcohol (b) 10 years old or younger (c) 11 (d) 12 (e) 13 (f ) 14 (g) 15 (h) 16 ( i )17 years or older
86 Appendix A (Continued) 5. During the past year about how frequently did you drink alcohol? Please choose the response which comes closest to describing your drinking pattern. (a) Never; I donÂ’t use alcohol (b) Once or twice during the year (c) 3 to 6 times per year (d) 7 to 10 times per year (e) About once a month (f) 2 or 3 times per month (g) Once or twice a week (h) 3 or 4 times a week ( i ) 5 or more times per week 6. On occasions when you drink, a bout how many drinks do you typically consume? Please estimate the actual number of drinks, where: 1 drink = approximately 1 can of beer, or = 1 glass of wine or wine cooler, or = 1 serving of liquor or a mixed drink. (a) None, I donÂ’t use alcohol (b) One drink (c) 2 drinks (d) 3 drinks (e) 4 drinks (f) 5 drinks (g) 6-8 drinks (h) 9-12 drinks ( i ) 13-16 drinks ( j ) 17 or more drinks 7. During the past month how frequently did you drink enough alcohol to get drunk or Â“high Â”? Please choose the response which comes closest to describing your drinking pattern. (a) Never (b) Once (c) Twice (d) 3 times (e) Once or twice a week (f) 3 or 4 times a week (g) 5 or 6 times per week (h) Daily or almost daily
87 Appendix A (Continued) 8. During the past year how frequently did you drink enough alcohol to get drunk or Â“high Â”? Please select the response which comes closest to describing your drinking pattern. (a) Never (b) Once or twice during the year (c) 3 to 6 times per year (d) 7 to 10 times per year (e) About once a month (f) 2 or 3 times per month (g) Once or twice a week (h) 3 or 4 times a week ( i ) 5 or more times per week
88 Appendix B: Alcohol Expectancy Circumplex This page contains words describing possible effects of alcohol. For each word, imagine it completing the sentence: DRINKING ALCOHOL MAKES ONE___ ." Then, for each word mark the number that indicates how often you think that this effect happens or would happen after drinking several drinks of alcohol "Drinking alcohol" refers to drinking any alcoholic beverage such as beer, wine, wine c oolers, whiskey, scotch, vodka, gin, or mixed drinks. There are no right or wrong answers. Answer each item quickly according to your first impression and according to your own personal belie fs about the effects of alcohol. Please mark your answers on the computer answer sheet. The available responses/numbers and their meaning are indicated below: "DRINKING ALCOHOL MAKES ONE ." 9. Appealing 21. Horny 10. Arrogant 22. Ill 11. Attractive 23. Light-headed 12. Beautiful 24. Lustful 13. Cocky 25. Nauseous 14. Dangerous 26. Outgoing 15. Deadly 27. Sick 16. Dizzy 28. Sleepy 17. Drowsy 29. Sociable 18. Egotistical 30. Social 19. Erotic 31. Tired 20. Hazardous 32. Woozy 0 1 2 3 4 5 6 Never Very Rarely Occasionally Frequently Very Always Rarely Frequently
89 Appendix C: Timeline Follow-Back
90 Appendix D: Driving Practices Please place your responses to the questions below on the SCANTRON 2 provided. 1. How often do you make cell phone calls (without a hands -free device) while operating a motor vehicle? (a) Almost every time I drive (b) Many times when I drive (c) Occasionally when I drive (d) Never when I drive 2. How often do you answer cell phone calls (without a hands-free device) while operating a motor vehicle? (a) Almost every time I drive (b) Many times when I drive (c) Occasionally when I drive (d) Never when I drive 3. How often do you read (e.g., map, newspaper, book, magazine, etc.) while operating a motor vehicle? (a) Almost every time I drive (b) Many times when I drive (c) Occasionally when I drive (d) Never when I drive 4. How often do you eat while operating a motor vehicle? (a) Almost every time I drive (b) Many times when I drive (c) Occasionally when I drive (d) Never when I drive 5. Have you ever taken a safer driving course? (a) Yes (b) No (c) Unsure
91 Appendix E: Driving Beliefs Please place your responses to the questions below on the SCANTRON 2 provided. This page contains a list of behaviors in which people sometimes engage while operating a motor vehicle. For each item, mark the number that indicates how important you think it is to avoid engaging in that behavior while operating a motor vehicle. Answer each item according to your own pe rsonal beliefs about these behaviors. Please mark your answers on the scantron provided. The available responses/numbers and their meaning are indicated below: "WHILE OPERATING A MOTOR VEHICLE ONE SHOULD AVOID ." 6. Eating 14. Talking to others in the vehicle 7. Using A Palm Pilot 15. Making Cell Phone Calls 8. Operating the stereo 16. Receiving Cell Phone Calls 9. Smoking 17. Searching For Objects in the Vehicle 10. Self-grooming (e.g., combing hair, shaving, etc.) 11. Using over-the-counter drugs 12. Reading (e.g., map, newspaper, book, magazine, etc.) 13. Operating Climate Control (e.g. Heat, Air Conditioning, etc.) 0 1 2 3 4 5 6 Not At All Very A Little Somewhat Important Very Ext remely Important Unimportant Unimportant Important Important Important
92 Appendix F: Level of Engagement Questionnaire Please place your responses to the questions below on the SCANTRON 1 provided. For the items below, please indicate your opinion of your time spent on the computer during this study. Please be as honest as you can so that we may take your responses into consideration when ad apting this program for future users. 33. The computer program (a) Attracted my attention throughout (b) Attracted my attention most of the time (c) Attracted my attention once in a while (d) DidnÂ’t really attract my attention at all 34. During my time on the computer (a) I was very involved in the program from start to finish (b) I was mostly involved in the program, but there were moments when my mind wandered (c) I was somewhat involved in the program, but my mind wandered quite a bit (d) I was hardly involved in the program at all. My mind wandered virtually the whole time 35. I found that the computer program (a) Encouraged a lot of active participation on my part (b) Required some active participation on my part (c) Asked me to do only a few things (d) Just required that I sit passively while it ran 36. In general the computer program was (a) Very interesting and didnÂ’t really seem to drag at all (b) Fairly interesting but dragged from time to time (c) A little interesting but dragged quite a bit (d) Pretty boring overall 37. If others were to complete the same program they would likely find it (a) Very engaging (b) Mostly engaging (c) Somewhat engaging (d) Hardly engaging at all
93 Appendix F (Continued) 38. For other individuals completing this computer program (a) It would hold their attention virtually the whole time (b) It would keep their attention most of the time but their minds would wander from time to time (c) Their minds would wander quite a bit (d) They would hardly be able to pay attention at all Please answer the following questions so we may understand how computer experience may affect the way people view ed the computer progr am differently. 39. During the past 30 days (excluding participation in this study) I have used a computer for work/school (a) Never (b) Once (c) Twice (d) 3 times (e) Once or twice a week (f) 3 or 4 times a week (g) 5 or 6 times per week (h) Daily or almost daily 40. How many hours per occasion did you typically use the computer for work/school? (a) None, I havenÂ’t used a computer for work or school in the last 30 days (b) 1 hour or less (c) 2 hours (d) 3 hours (e) 4 hours (f) 5 hours (g) 6 hours (h) 7 hours ( i ) 8 hours ( j ) 9 or more hours 41. During the past 30 days (excluding participation in this st udy) I have played computer/video games (a) Never (b) Once (c) Twice (d) 3 times (e) Once or twice a week (f) 3 or 4 times a week (g) 5 or 6 times per week (h) Daily or almost daily
94 Appendix F (Continued) 42. How many hours per occasion did you typically play computer/video games? (a) None, I havenÂ’t played computer/video games in the last 30 days (b) 1 hour or less (c) 2 hours (d) 3 hours (e) 4 hours (f) 5 hours (g) 6 hours (h) 7 hours ( i ) 8 hours ( j ) 9 or more hours 43. During the past 30 days I have been in internet chat rooms (a) Never (b) Once (c) Twice (d) 3 times (e) Once or twice a week (f) 3 or 4 times a week (g) 5 or 6 times per week (h) Daily or almost daily 44. How many hours per occasion did you typically spend in internet chat rooms? (a) None, I havenÂ’t visited internet chat rooms in the last 30 days (b) 1 hour or less (c) 2 hours (d) 3 hours (e) 4 hours (f) 5 hours (g) 6 hours (h) 7 hours ( i ) 8 hours ( j ) 9 or more hours 45. During the past 30 days I have surfed the internet (a) Never (b) Once (c) Twice (d) 3 times (e) Once or twice a week (f) 3 or 4 times a week (g) 5 or 6 times per week (h) Daily or almost daily
95 Appendix F (Continued) 46. How many hours per occasion did you typically spend surfing the internet? (a) None, I havenÂ’t surfed the internet in the last 30 days (b) 1 hour or less (c) 2 hours (d) 3 hours (e) 4 hours (f) 5 hours (g) 6 hours (h) 7 hours ( i ) 8 hours ( j ) 9 or more hours
96 Appendix G: Defensive Driving Content Questionnaire Please answer the following questions about the material presented by the computer program. 47. Defensive driving includes all the following except for _____________. (a) Making safe turns (b) Managing your space (c) Communicating your intentions (d) Maintaining concentration (e) All of the above (f) None of the above 48. Which of the following behaviors are he lpful defensive driving Â“scanningÂ” techniques? (a) Making sure youÂ’re in the center of your lane when driving (b) Checking your mirrors before changing speed or position in traffic (c) Maintaining a safe following distance (d) Keeping your eyes moving from far to near (e) Â“aÂ” and Â“bÂ”, above (f) Â“bÂ” and Â“dÂ”, above (g) Â“aÂ” and Â“dÂ”, above (h) Â“bÂ” and Â“cÂ”, above 49. When managing your speed, Â“stoppingÂ” time includes which of the following? (a) The time it takes to perceive a hazard (b) The time it takes to react (c) The time it takes to stop once the brakes are applied (d) Â“AÂ” and Â“BÂ” above (e) All of the above (f) None of the above 50. Because honking your horn may spark an aggressi ve response in others, you should only use it to prevent an accident (a) True (b) False 51. There is no benefit to driving with your headlights on during the day. (a) True (b) False
97 Appendix H: Alcohol Program Content Questionnaire Please answer the following questions about the material presented by the computer program. 47. Alcohol expectancies (effects people expect from alcohol that are not really caused by alcohol) include all the following except for _____________. (a) Emotional states like becoming happy or sad (b) Sexual arousal like becoming more sexually aroused (c) Social effects like becoming more extroverted (d) Physiological changes like becoming dizzy (e) All of the above (f) None of the above 48. Which of the following behaviors that people often attribute to the effects of alcohol are inaccurate expectancy effects? (a) Becoming talkative (b) Becoming nauseous (c) Becoming sleepy (d) Becoming sexually aroused (e) Â“aÂ” and Â“bÂ”, above (f) Â“bÂ” and Â“dÂ”, above (g) Â“aÂ” and Â“dÂ”, above (h) Â“bÂ” and Â“cÂ”, above 49. How do people acquire inaccurate expectan cy beliefs about the effects of alcohol? (a) By watching and hearing adults talk about their alcohol expectancy beliefs (b) By seeing adults drink and engage in behaviors based on their alcohol expectancy beliefs (c) By seeing alcohol expectancy belie fs linked with alcohol in advertising (d) Â“AÂ” and Â“BÂ” above (e) All of the above (f) None of the above 50. Alcohol expectancy effects do not really come from drinking alcohol, but people often feel that way when they drink. This is because alcohol causes a general mental slowing and numbing effect and people label what is happening around them based on the situation (e.g., being at a party) and their beliefs a bout the effects of alcohol (e.g., having a good time). (a) True (b) False 51. At higher does, alcohol makes it easier to perform sexually. (a) True (b) False
98 Appendix I: Debriefing Form fo r University of South Florida Thank you for participating in this study! DonÂ’t forget that you will be contacted 30 days from today to complete a brief follow-up phone interview. The goal of this study was to look at how different ways of presenting computerized information can affect how engaged participants become while using a computer program and what type of effect the program has on participan tsÂ’ behaviors. Your participation will aid in psychologistsÂ’ understanding of how these computerized interventions work. Previous research has established principles of software construction that enhance user engagement, motivation, and learning. Th ese suggestions encompass both the modalities of information presentation (e.g., audio, graphic, a nd text-based) and the types of learning activities believed to be the most effective in engaging an d motivating software users. Some examples of techniques believed to improve user engagement and learning include presenting material with words and pictures rather than words alone a nd reducing the number of extraneous words and sounds included in a presentation. If you are interested in learning more about pr inciples of software construction and how these may affect software users, please feel free to contact William Hunt at 974-6963 or at the University of South Florida Psychology Depart ment in PCD 4118G. Additionally, you may find the references below of interest. Related references: Goldman, M. S. (1999). Expectancy operation: Cognitive/neural models and architectures. In I. Kirsch (Ed.), Expectancy, experience, and behavior (pp. 41-63). Washington, D.C.: APA Books. Mayer, R. E. & Moreno, R. (2002). Aids to computer-based multimedia learning. Learning and Instruction, 12, 107-119. Stoney, S. & Oliver, R. (1998). Interactiv e multimedia for adult learners: Can learning be fun? Journal of Interactive Learning Research, 9(1), 55-81.
99 Appendix J: Debriefing Form for San Diego State Univer sity and University of California, San Diego Thank you for participating in this study! DonÂ’t forget that you will be contacted 30 days from today to complete a brief follow-up phone interview. The goal of this study was to look at how different ways of presenting computerized information can affect how engaged participants become while using a computer program and what type of effect the program has on participan tsÂ’ behaviors. Your participation will aid in psychologistsÂ’ understanding of how these computerized interventions work. Previous research has established principles of software construction that enhance user engagement, motivation, and learning. Th ese suggestions encompass both the modalities of information presentation (e.g., audio, graphic, a nd text-based) and the types of learning activities believed to be the most effective in engaging an d motivating software users. Some examples of techniques believed to improve user engagement and learning include presenting material with words and pictures rather than words alone a nd reducing the number of extraneous words and sounds included in a presentation. The information presented in this study con cerned behaviors in wh ich people may engage that have the potential to place them or others at risk of harm or even death. Risky driving behavior may include driving aggressively such as cutting people off on the road, exceeding the speed limit, and making sudden or unexpected la ne changes. Risky drinking behavior may include consuming alcohol in situations that re quire the exercise of good judgment (e.g., sexual encounters or deciding to practice safe sex) or that require coordination (e.g., such as driving). Additionally, as a rule it is not recommended by th e medical profession that adults consume more than one or two drinks per day. It is illega l for individuals under the age of 21 to consume alcohol at all. If you are interested in learning more about pr inciples of software construction and how these may affect software users, in safe driv ing practices, or in the effects of alcohol consumption, please feel free to contact William Hunt at (858) 642-3261 or C/o Sandra A. Brown, Ph.D. 9500 Gilman Dr. McGill Hall (0109) La Jolla, CA 92093. Additionally, you may find the references below of interest: Goldman, M. S. (1999). Expectancy operation: Cognitive/neural models and architectures. In I. Kirsch (Ed.), Expectancy, experience, and behavior (pp. 41-63). Washington, D.C.: APA Books. Mayer, R. E. & Moreno, R. (2002). Aids to computer-based multimedia learning. Learning and Instruction, 12, 107-119. Stoney, S. & Oliver, R. (1998). Interactiv e multimedia for adult learners: Can learning be fun? Journal of Interactive Learning Research, 9(1), 55-81.
About the Author William Michael Hunt received a Bachel orÂ’s of Science Degree from James Madison University, Harrisonburg, Virginia, in 1994 where he majored in psychology and minored in anthropology. He received his first masterÂ’s degree from Hollins University, Roanoke, Virginia, in 1995, st udying general/experim ental psychology, and a second masterÂ’s degree in clinical psychology from the University of South Florida in 2001. Between the two degrees he lived in L ondon and served as a resident advisor for the James Madison University studies abroad program. Upon completion of his second M.A., Mr. Hunt proceeded to doctoral candidacy, earning minors in quantitative methodology a nd in computer programming and data analysis. After fulfilling c ourse requirements for the program, he completed a one-year clinical internship at the Univ ersity of California, San Die go. Mr. Hunt currently resides in San Diego and is working as a National Institute on Alcohol Abuse and Alcoholism postdoctoral fellow at San Diego State University.