Change in VAS sedation expectancies, by condition. Assessments were done a day before the intervention (T1), 24 hours after the intervention (T2), and 6 weeks after the intervention (T3). 

Context in source publication

Context 1

... were recruited by trained peers, who were able to answer any questions that emerged when filling out the preintervention questionnaire. It proved more difficult to recruit participants for the intervention group than for the control group. Participants in the EC group needed to come to a community center for the 1-hour intervention. Many participants indicated they were not willing to come to the intervention location, because it was too far away from the camping site. Moreover, many youngsters who promised to take part in the EC failed to show up at the critical moment (Table 1). In addition, participants did not receive any compensation for the time they spent in the intervention, which may have led many people to decline the offer to participate or fail to show up at the last moment. The holiday situation may be another explanation why difficulties in recruitment were present. Young people on holiday are taking a “time-out” from obligations, which means that appointments are not taken as seriously and no agendas are used. Women were much more difficult to recruit for the study than men, and they were less inclined to drink alcohol for the purpose of an “alcohol experiment.” Feasibility of EC Procedure. The EC sessions were led by a team of two peers. These peers received an 8-hour training on expectancy theory and the EC. Each peer was trained by performing the EC protocol on simulation participants. However, in actual practice, participants were much more noisy and were less willing to listen to the session leaders than in the training. In some cases, some of the participants left before the end. The active deception procedure convinced participants when the protocol was followed. However, in a few cases the drinks were not served ice cold and some participants stated that they did not taste any alcohol. In those cases, identification errors were made, and peers were unable to use discrepancies between predicted and actual drinkers as a starting point for their explanation. This is a point of attention for next studies. The EC was well received among young people, although some participants were a bit disappointed when they learned they had not received alcohol. Costs of the Study. Most of the money was spent on salaries of peers ($2,100) and their expenses for travel and accommodation ($1,500). Because the peers also collected questionnaires, they worked more hours than they would have if no research was attached. Because the purpose of the study was to evaluate the intervention, a professional survey agency was hired for conducting telephone interviews after the holiday was over ($1,400). Other costs were made for the purchase of vodka, tonic, lemons, posters to decorate the EC location, measuring cups, a few knives, and potato crisps ($200). Lessons Learned. There were several points for improvement of implementation. Some of the peers expressed that they should have been trained more thoroughly and had missed guidance when practical problems emerged. There were not enough peers present to be fully in command of the situation, and the locations for performing the EC were too far from the camping sites, which led to high attrition by participants unwilling to show up. For the deception procedure to succeed, it was important that the drinks were served chilled. It can be considered positive that intervention participants were less likely to drop out than control subjects. In addition, young people enjoyed participating in the intervention. VAS expectancies were analyzed in a 3 2 2 (Time Condition Gender) mixed ANOVA, with Time as within-subject variable and Condition and Gender as between- subject variables. The VAS arousal scale showed a statistical trend for the main effect of Time ( F (2 , 83) = 2 . 99 , p = . 06). The Time × Condition × Gender interaction was significant ( F (2 , 83) = 3 . 66 , p = . 03); therefore, we analyzed the interaction between Time and Condition separately in men and women. In men, the interaction between Time and Condition was not significant ( F (1 , 68) = 1 . 43 , p = . 24). The expected decrease in arousal expectancies in the experimental group was not found, although Figure 1 suggests that the EC may have worked to prevent an increase in arousal expectancies observed in the control subjects. In women, the Time × Condition interaction showed a statistical trend ( F (1 , 16) = 2 . 96 , p = . 08), but the difference was only found at pretest ( t (17) = − 2 . 0 , p = . 07), in the absence of differences at the two posttests ( p s > . 25). On the VAS sedation subscale, a statistical trend was found for Time ( F (2 , 83) = 2 . 45 , p = . 09). The Time × Condition interaction was significant ( F (2 , 83) = 3 . 20 , p = . 05). There was no interaction between Time, Condition, and Gender ( p > . 80). Participants in the experimental condition expected significantly more sedating effects from alcohol on the posttests than participants in the control group. However, as shown in Figure 2, sedation expectancies did not increase in the experimental group. The significant effect came from a decrease in sedation expectancies in the control group at the 6-week follow-up. Changes in drinking were explored at two points in time: (1) drinking in the 24 hours after the EC and (2) alcohol consumption on week days and weekend days 6 weeks after the EC (after the vacation was over). To evaluate whether a change in drinking occurred in the 24 hours after the EC, a 2 × 2 × 2 (Time × Condition × Gender) mixed ANOVA was conducted with alcohol consumption as a dependent variable. We compared alcohol consumption at pretest with consumption the day after the EC. Unfortunately, many participants could not be reached for the second measurement or failed to fill out the questions about alcohol consumption, which resulted in high attrition for analyses concerning alcohol consumption (see Methods, Selective Attrition). Alcohol consumption did not change in the first 24 hours after the EC. The effect of Time was not significant ( p > . 50), and no interaction effects were present between Time and Condition ( p > . 50) and Time, Condition, and Gender ( p > . 40). Alcohol consumption 6 weeks after the intervention was analyzed using 2 × 2 × 2 (Time × Condition × Gender) mixed ANOVAs with alcohol consumption during ordinary weekdays and total consumption on a night out as dependent variables. When total alcohol consumption on a regular night out was compared from pre- to posttest, a significant Time effect was found ( F (1 , 88) = 3 . 89 , p = . 05). Either alcohol consumption decreased in both EC and control participants at the posttest or perception of alcohol consumption may have differed at home (lighter) versus on holiday (heavier). No Time × Condition effect was present ( p > . 50), and Gender did not interact with the interaction between Time and Condition ( p 50). Alcohol consumption on weekdays did not show any changes from pre- to posttest. Time and the interaction between Time and Condition were nonsignifi- cant ( p > . 30 and p > . 50, respectively). Gender did not interact with Time × Condition ( p > . 50). We inspected relations between expectancies and alcohol consumption and unexpect- edly found no correlations. Pretest VAS arousal and VAS sedation expectancies did not correlate significantly with pretest alcohol consumption on a night out nor with pretest alcohol consumption on holiday (Pearson’s correlation VAS sedation score, alcohol consumption on a night out r = − . 03; VAS arousal score, alcohol consumption on a night out r = . 13; VAS sedation score, alcohol consumption on holiday r = . 11; VAS arousal score, alcohol consumption on holiday r = . 11). This study examined whether an EC procedure that has been shown effective in (male) heavy drinkers could be transferred from a (bar-) lab setting to a targeted intervention in the field. Performing the EC in a real-life setting proved feasible, and the EC was well accepted by our target group. However, the current study had several limitations: high attrition, high percentage of missing values, and less-than-perfect performance by the peers. Recruitment and attrition problems need to be tackled, for instance, by recruiting youngsters the same day the EC takes place, performing the EC at the actual camping site, or providing transportation to the EC location. The peers performing the EC may need to be trained more extensively, and on-the-spot supervision should be present during the ECs. Presence of a third peer would aid in better performance of the EC. It is important to capture youngsters’ attention by using appealing examples. Further studies should consider conducting the EC in a holiday setting with men only, because males proved easier to recruit and drank much heavier than females and may therefore decrease their drinking more after an intervention. The intervention did not change expectancies exactly as predicted: A decrease in arousal expectancies was not found ( c f . Wiers et al., 2005), and an increase in sedation expectancies was not found (both were found with the same instruments in “heavy drinking” college students in a longer EC in a bar-lab; Wiers et al., 2005). However, it is possible that the intervention countered natural changes in expectancies that may have occurred after the holiday was over. The EC may have countered the decrease in sedation expectancies that took place in the control group but not in the intervention group. In summary, we found some indications of effects of the EC on expectancies in heavy drinking men and women. Although these effects were not exactly as predicted, the EC group may have benefited from the intervention. No effects on alcohol consumption were observed in the 24 hours after the EC, and a decrease was not observed on weekdays or during nights out at the 6-week follow-up. In an attempt to explain our results, we inspected correlations between expectancies and alcohol use in our ...