• Alcohol Drinking;
  • Motivation;
  • Prototypes;
  • Naturalistic Setting;
  • Observations


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Background:  Perceptions about the type of people who drink, also referred to as drinker prototypes, may strengthen young people’s motivation to engage in alcohol use. Previous research has shown that drinker prototypes are related to alcohol consumption in both adolescents and young adults. However, the evidence for the strength of these relationships remains inconclusive. One of the caveats in former studies is that all insights about prototype relations are based on self-reported data from youngsters themselves, mostly gathered in a class situation, which may contain bias due to memory distortions and self-presentation concerns.

Methods:  The present study examined the impact of drinker prototypes on young adults’ drinking patterns by using a less obtrusive measure to assess alcohol consumption, i.e. ad lib drinking among friend groups in the naturalistic setting of a bar lab. Drinker prototypes, self-reported alcohol use in the past, and observed alcohol intake in the bar lab were assessed among 200 college students. Relations between participants’ drinker prototypes and their self-reported and observed drinking behavior were examined by computing correlations and conducting multilevel analyses.

Results:  Drinker prototypes were related to both self-reported and observed alcohol use. However, the drinking patterns of friend group members had a strong impact on participants’ individual drinking rates in the bar lab. After these group effects had been controlled for, only heavy drinker prototypes showed relations with observed alcohol intake in the bar lab.

Conclusions:  These findings further establish the value of drinker prototypes in predicting young adults’ drinking behavior and suggest that people’s motivation to drink alcohol in real-life drinking situations is related to their perceptions about heavy drinkers.

For several decades, scholars have studied prototypes or social images of substance users as possible explanations for adolescent substance use, such as smoking and drinking (Chassin et al., 1981; Gibbons and Gerrard, 1997; Gibbons et al., 2003; McKennell and Bynner, 1969). Most of these studies are based on the idea that young people are conscious about images of products and behaviors, and that their motivation to perform a particular behavior depends on the social image associated with it (Chassin et al., 1985; Denscombe, 2001; Lloyd et al., 1998; MacFadyen et al., 2003). A dominant theoretical model in this line of research is the Prototype-Willingness model, developed by Gibbons and Gerrard (Gibbons and Gerrard, 1997; Gibbons et al., 2003). It postulates that adolescents have clear and salient images of the type of people their age who engage in substance use. Ample research on this theoretical model has shown that positive drinker prototypes are related to both adolescents’ (Blanton et al., 1997; Gerrard et al., 2002; Spijkerman et al., 2007) and young adults’ self-reported alcohol use (Norman et al., 2007; Rivis et al., 2006).

Although previous cross-sectional and prospective studies have consistently demonstrated that drinker prototypes are related to alcohol use, there is no conclusive evidence about the strength of this relationship. One of the caveats in former studies is the lack of unobtrusive measures to assess drinking behavior. All insights about prototype relations are based on self-reported data from youngsters themselves, mostly gathered in a class situation, which may contain bias due to memory distortions and self-presentation concerns. Moreover, drinking alcohol is context-specific and some of the processes that occur during a drinking-conducive situation with friends can not be captured by letting adolescents answer retrospective questions in class about their drinking behavior (Bot et al., 2005). To obtain a more adequate picture of young people’s actual drinking rates, researchers have used observational methods in real-life (Clapp et al., 2008; Lange et al., 2002; Thombs et al., 2008) and naturalistic settings (Aitken, 1985; Bot et al., 2005; Bruun, 1959; Van de Goor et al., 1990). By using observations of people’s drinking practices in drinking situations, they have obtained more objective information on an individual’s actual alcohol use. For example, Bot and colleagues (2005) found that students’ positive and arousal expectancies were related to their alcohol use observed in a naturalistic setting. The paradigm used by Bot and colleagues involved examining students’ drinking behavior among friend groups in an ad lib drinking situation in a bar lab of Radboud University Nijmegen, the Netherlands. This bar lab looked like a normal bar, except that there were cameras to register participants’ drinking behavior. The main purpose of this paradigm was to assess young people’s drinking behavior in an unobtrusive and more systematic way while making the situation as natural as possible. Earlier studies by this research group have demonstrated the validity of this paradigm (Bot, 2007; Bot et al., 2005, 2007a; Van Schoor et al., 2008). Moreover, this type of research method can have several advantages over other possible unobtrusive methods, such as taste-tests (Quigley and Collins, 1999) and observational research in real-life (Clapp et al., 2007). A major problem with taste-tests is that the ecological validity is not very high since tasting beer or wine in a controlled setting will not capture the same social processes as drinking alcohol with some friends in a time-out situation. The other limitation of taste-test paradigms is that people are obliged to drink alcohol. Observations in real-life drinking situations, such as bars and parties (Clapp et al., 2007) will have sufficient, even higher ecological validity compared to the observations in a bar lab. However, this method reveals several limitations which makes it less suitable for systematic examination of predictors of alcohol use (Bot et al., 2005). First, there may be large variations in contextual and individual characteristics of real-life drinking situations. Second, the assessment of predictors may be difficult in real-life drinking settings. Third, the quality of ethical conditions is less easy to control during real-life observations. We propose that doing the same type of research in a naturalistic setting of a bar lab instead of real-life drinking situations may circumvent the aforementioned problems, since many conditions of the drinking situation can be controlled for. In sum, observational assessments of alcohol use in naturalistic settings, such as a bar lab, are more ecologically valid than assessments based on self-reports in nondrinking contexts. Although the ecological validity of observations in a bar lab may be somewhat lower than unobtrusive research in natural drinking settings, research in a bar lab also shows several advantages compared to research in natural drinking contexts. More specifically, it allows for more controlled, intra-individual assessments.

The present study will examine relationships between students’ drinker prototypes and their alcohol intake in a bar lab. We will use the same experimental context and paradigm as Bot and colleagues (2005), except that we will test drinker prototypes, and not alcohol expectancies, as possible predictors of observed alcohol intake. Theoretically, these 2 cognitive constructs bear on different underlying processes of alcohol use. Alcohol expectancies represent people’s motivation to experience psychoactive effects of alcohol, such as mood enhancement, relaxation or arousal (Wiers et al., 2002). In contrast, drinker prototypes represent a person’s openness to the social consequences of alcohol use that are related to general perceptions about the typical characteristics of people who drink (Gibbons et al., 2003). More specifically, the relationship between favorable drinker prototypes and alcohol use is interpreted as an indication that people are more likely to drink alcohol if they find it acceptable to become associated with the social image of the typical drinker (Gerrard et al., 2002). The question whether prototypes predict alcohol intake in a real-life drinking situation, has not been addressed in previous research.

Although much of the prototype literature focuses on adolescents, we focus on college students. The main argument for selecting college students as research participants is that doing this type of research among underaged adolescents is unethical. The choice of studying college students’ drinking behavior has implications for the assessment of drinker prototypes. Whereas most early adolescents have just started to obtain some drinking experience; most college students are already well acquainted with drinking alcohol. Research suggests that a considerable part of the college student population engages in binge drinking and excessive alcohol use (Karam et al., 2007; Maalsté, 2000). Because of these further developed drinking patterns, we have chosen to assess prototypes of specific drinking patterns, such as binge drinking and moderate alcohol use in a social context. This is not entirely in line with previous prototype studies that mainly concentrated on drinker (Blanton et al., 1997; Chassin et al., 1985; Gibbons and Gerrard, 1995; Spijkerman et al., 2007) and nondrinker prototypes (Gerrard et al., 2002). These frequently assessed prototypes solely refer to whether a person drinks or not, and do not take into account how much a person drinks. In 2 previous studies, scholars did examine prototypes that referred to the quantity of alcohol consumed by assessing students’ perceptions of the typical binge drinker (Norman et al., 2007; Rivis et al., 2006). In the present study, we examine 3 prototypes related to different drinking patterns, i.e. abstainer prototypes, social drinker prototypes and heavy drinker prototypes. We define abstainer prototypes as perceptions about the type of person who (almost) never drinks. Social drinker prototypes are defined as perceptions of the type of person who drinks 3 to 5 alcoholic beverages while being in the company of others; and heavy drinker prototypes as perceptions about the type of person who regularly drinks more than 8 alcoholic beverages per occasion.

In sum, the purpose of our study is to test the value of different drinker prototypes in the prediction of students’ actual alcohol use. To this end, we will examine students’ ad-lib drinking behavior among friend groups in a bar lab. The use of this paradigm increases the ecological validity of our study compared to previous studies on drinker prototypes. Moreover, it allows us to draw further conclusions about whether drinker prototypes are important in predicting students’ drinking rates. The impact of drinker prototypes will be tested while controlling for variations in drinking behaviors between friend groups. Earlier findings showed that an individual’s drinking rate was highly dependent on other group members’ drinking levels (Bot, 2007). If our study shows relations between drinker prototypes and students’ actual alcohol use in a naturalistic setting while controlling for friend group drinking levels, then we can conclude that drinker prototypes provide a meaningful contribution to the prediction of young adults’ actual drinking behavior.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References


The present study was conducted in the Netherlands, where the legal drinking age is 16 years for beer and wine, and 18 years for hard liquor. Participants were recruited from the college student population (aged 18 years and older), and only people who consumed alcohol were included. The final sample consisted of 200 college students, of whom 98 were males and 102 females. Participants were in the ages between 18 to 27 years (mean = 21.5 years, SD = 1.98). In total, 31 groups of 5 to 8 friends participated. Seventeen groups consisted of 6 people (51%), 8 groups of 7 people (28%), 4 groups of 8 people (16%), and 2 groups of 5 people (5%). Twenty-two groups included both male and female participants (72.5%), 5 included only males (15.5%) and 4 included only females (12%).


Prototypes.  Three different drinker prototypes were examined, i.e. abstainer, social drinker and heavy drinker prototypes. Students were asked to rate the typical abstainer, social drinker and heavy drinker on a list of adjectives. The adjectives and question-formats for these prototype scales were based on earlier scales (Gibbons and Gerrard, 1995; Spijkerman et al., 2005, 2007) and pretested among 100 college students.1 The final scales consisted of 16 adjectives of which 7 referred to negative characteristics. Examples were: “independent,”“smart,”“cool,”“boring,”“antisocial,”“immature,” etc. For each adjective, students had to indicate whether it matched their perception of the typical abstainer, social drinker, or heavy drinker. Answers could be given on a 7-point scale ranging from 1 = “not at all” to 7 = “very much.” Definitions of the 3 drinker types were provided for each prototype scale. We defined an abstainer as “someone who (almost) never drinks alcohol.” A social drinker was defined as “someone who drinks 3 to 5 alcoholic drinks, while being in the company of others.” A heavy drinker was defined as “someone who regularly drinks more than 8 alcoholic drinks per occasion.” We reversely coded the negative characteristics and than summed all scores of each prototype scale as a measure for students’ general perceptions of the typical abstainer, social drinker, and heavy drinker. High scores indicated a more favorable prototype. Cronbach’s alphas were 0.85 for the abstainer prototype, 0.88 for the social drinker prototype, and 0.91 for the heavy drinker prototype.

Weekly Alcohol Use.  To obtain an estimate of student’s weekly alcohol consumption, we used the Weekly Recall-measure (cf. Hajema and Knibbe, 1998). For this measure, students were asked to report on their alcohol use in the past week. Starting from yesterday, they had to indicate the number alcoholic drinks they had consumed on each day of the past week. The sum score was used as an indication for students’ weekly alcohol use.

Binge Drinking.  Binge drinking was assessed by asking participants how many times they had drunk more than 6 alcoholic drinks per occasion over the past 12 months. Students could answer on a 7-point scale, ranging from 1 = “never,” 2 = “once,” 3 = “1 to 2 times per 6 months,” 4 = “3 to 5 times per 6 months,” 5 = “1 to 3 times per month,” 6 = “1 to 2 times per week,” 7 = “more than twice a week.” A high score implied high frequency of binge drinking in the past 12 months (Knibbe et al., 1991).

Observed Alcohol Use.  Participants received a 45-minute break in-between 2 discussion tasks. During this break, participants could relax and drinks were served. The number of alcoholic drinks that participants consumed during the break was counted by the bartender and recorded on camera. If participants did not finish their drink, the bartender subtracted the remaining volume from the total number of drinks consumed by the participant. The drinks were poured in smaller glasses than standard glasses. The beer we used contained 5% alcohol and was served in glasses of 160 ml. The wine contained 11 to 12% alcohol and was served in glasses of 110 ml. This means that 1 glass of beer contained on average 8 ml pure alcohol and 1 glass of wine contained 12.1 to 13.2 ml pure alcohol. We assessed participants’ observed drinking rates by using the drink counts of the bar tender after having checked whether these drink counts corresponded with the camera recordings.


College students were invited together with 6 to 7 friends to participate in a study on alcohol and group discussions.2 Each group received a 30 Euro (37.63 US $) reward after participation. Before entering the laboratory setting, participants received information about the procedure of the session and signed an informed consent stating that they gave permission for the use of data collected during their participation. As a pretest, each group member individually filled out a questionnaire on a computer containing questions on demographics, drinker prototypes, weekly alcohol use and frequency of binge drinking. The group discussion sessions were held on weekdays between 4.30 pm until 9 pm in the bar lab at our faculty. This bar lab looked like a normal bar except that it had cameras placed to observe participants’ drinking behavior (Bot et al., 2005). At the beginning of the session, the group was presented with a group discussion task which consisted of evaluating 5 non-alcohol-related television commercials. The commercials were shown on television in the bar. When the video stopped, the group was asked to discuss which commercial was most convincing and why. The video tasks were irrelevant for our study, but it covered the actual claim of our research project. After the discussion, participants received a break for approximately 45 minutes during which they could order drinks at the bar. Two cameras registered participants’ ordering and drinking behavior. Available drinks were sodas, orange juice, beer, and wine. To create a similar atmosphere as in a real bar, popular music was played, nuts were offered and participants were allowed to smoke, and to play table-soccer or billiards during the break. After the break, participants had to discuss and evaluate an alcohol prevention campaign which was shown on video. At the end of the whole session, participants’ BAC-levels were measured with a breathalyzer. If one of the members showed a blood alcohol concentration above 0.2 pro mille, all members were offered a taxi. Of the total sample, 64.8% left the experiment with a bac-level higher than 0.2 (M = 0.27, SD = 0.20). After the measurements, participants were debriefed and received the possibility to withdraw their consent for using the observational data in our research. None of the participants withdrew their consent.


Both the questionnaire data and the observations of participants’ drinking behavior during the break were entered in SPSS for Windows (SPSS Inc., Chicago, IL). Relations between participants’ (non) drinker prototypes and their self-reported and observed alcohol use were examined by computing correlations. Associations between drinker prototypes and alcohol use were further analyzed by conducting a series of multilevel analyses. By using a multilevel approach, it is possible to adjust for the effects of nonindependent data (Goldstein et al., 2002). As a consequence of our clustered sampling and data collection by using friend groups, our data should be regarded as nonindependent. The reason is that the behavior and prototypes of group members within a group are more strongly interrelated compared to relations between individual participants who have been recruited via a random sampling procedure. By using a multilevel approach, it is possible to correct for bias due to clustered data and to differentiate between group and individual effects. All analyses were conducted in MLwiN version 2.0 (Rasbash et al., 2005). We first tested 3 basic intercept-only models to test the percentage of variance explained by clustering effects in participants’ self-reported and observed alcohol use. Subsequently, these basic models were used to test the increases in model fit when adding other variables to the model. Students’ alcohol use can be dependent on drinking levels of their friend group members. The impact of these group effects are reflected in the intraclass correlation. For self-reported alcohol use, the intraclass correlations were = 0.38 for weekly alcohol use and = 0.25 for binge drinking. For observed alcohol use, the intraclass correlation was = 0.70, implying that 70% of the variance in students’ observed alcohol use was explained by group effects. Three different models were tested by first controlling for participants’ sex and then adding the 3 drinker prototypes to these basic models. The effect for sex was entered both as a group and individual variable. In addition, effects were either fixed or varied depending on whether it improved the fit of the models.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

We computed descriptive statistics for all study variables (see Table 1). The mean score on the weekly recall measure (M = 17.04, SD = 14.19) suggested that, according to participants’ self-reports, participants had consumed on average 17 alcoholic drinks in the past week. The mean score for binge drinking was 5.07 (SD = 1.43), implying that participants’ average frequency of engaging in binge drinking was 1 to 3 times per month in the past year. Finally, the mean score on observed alcohol use (M = 2.98, SD = 1.96) showed that participants consumed on average 3 alcoholic drinks during the 45-minutes break in the bar lab. The mean scores on drinker prototypes ranged between 4 and 5 suggesting that participants held rather neutral or slightly favorable drinker prototypes. Participants’ prototypes of the abstainer (M = 4.94, SD = 0.68) and social drinker (M = 5.03, SD = 0.64) were more favorable than their prototypes of the heavy drinker (M = 4.19, SD = 0.87). An univariate analysis testing differences in scores on the 3 drinker prototypes showed a significant main effect of prototype, F(2,200) = 78.36, < 0.001. Post hoc analyses with Bonferroni indicated that heavy drinker prototypes differed from the other 2 drinker prototypes, which meant that participants’ abstainer and social drinker prototypes were more favorable than their prototypes of heavy drinkers.

Table 1.   Descriptive Statistics
 Total sample (N = 200)Male (n = 98)Female (n = 102)t-test
  1. Note: Observed alcohol use = number of alcoholic drinks consumed in bar lab; Weekly recall = self-reported alcohol consumption in past week; Binge drinking = frequency of drinking more than 6 alcoholic beverages in the past 12 months; PT = prototype.

  2. *< 0.01, **< 0.001.

PT abstainer4.940.684.770.715.110.62−3.56**
PT social drinker5.030.644.960.615.090.67−1.50
PT heavy drinker4.190.874.350.854.030.862.70*
Weekly recall17.0414.1923.8015.2610.549.277.46**
Binge drinking5.071.435.631.194.531.445.92**
Observed alcohol use2.981.964.171.711.831.4410.47*

Further, when comparing results between male and female participants, our findings showed that males had higher scores both on the self-report drinking measures as well as the observations of their alcohol intake in a naturalistic setting (see Table 1). In addition, compared to their female counterparts, males held less favorable perceptions of the typical abstainer and more favorable perceptions of the typical heavy drinker.

Pearson and Spearman correlations between participants’ sex, age, drinker prototypes and their self-reported and observed alcohol use are presented in Table 2. As shown, students’ abstainer and heavy drinker prototypes were related to both self-reported [weekly recall: r(200) = −0.36, < 0.001; r(200) = 0.40, < 0.001; binge drinking: r(200) = −0.30, < 0.001; r(200) = 0.46, < 0.001] as well as observed alcohol use [r(200) = −0.35, < 0.001; r(200) = 0.23, < 0.001]. These findings indicate that the more students held unfavorable abstainer prototypes and favorable heavy drinker prototypes, the higher their self-reported and observed alcohol use. Our findings also demonstrated that students’ self-reported drinking patterns were related to the observed alcohol use in a naturalistic setting [weekly recall: r(200) = 0.51, < 0.001; binge drinking: r(200) = 0.45, < 0.001] implying that the higher students’ self-reported alcohol use, the higher their observed alcohol use in the bar lab. With regard to participants’ individual characteristics, only students’ sex was related to alcohol use and drinker prototypes. These findings indicated that females showed lower levels of self-reported and observed alcohol use, held more favorable abstainer and heavy drinker prototypes and were younger.

Table 2.   Pearson Correlations (1-tailed) Between Students’ Drinker Prototypes and Observed and Self-Reported Alcohol Use
  1. Note: observed alcohol use = number of alcoholic drinks consumed in bar lab; weekly recall = self-reported alcohol consumption in past week; binge drinking = frequency of drinking more than 6 alcoholic beverages in the past 12 months; PT = prototype.

  2. *< 0.05, **< 0.01, ***< 0.001.

1. Sex−0.35***0.25***0.10−0.18**−0.61***−0.46***−0.42***
2. Age −0.14*−**0.21**0.22**
3. PT abstainer  0.39***−0.14*−0.35***−0.36***−0.30***
4. PT social drinker   0.22**−0.07−0.070.10
5. PT heavy drinker    0.23***0.40***0.46***
6. Observed alcohol use     0.51***0.45***
7. Weekly recall      0.64***
8. Binge drinking       

To control for group effects, we conducted multilevel analyses. The models with the best fit are presented in Table 3. All 3 models revealed a random effect for students’ gender implying that the relationship between students’ gender and their self-reported and observed alcohol use varied across groups. The model with participants’ weekly alcohol use showed fixed effects for participants’ abstainer and heavy drinker prototypes suggesting that associations between these variables and self-reported weekly alcohol use did not differ between the groups. Interpretation of these findings suggests that the more participants held unfavorable abstainer prototypes and favorable heavy drinker prototypes, the higher their self-reported weekly alcohol use, irrespective of group differences. The model on self-reported binge drinking showed similar results, except that participants’ social drinker prototypes also showed a significant fixed effect. These findings imply that, after controlling for group differences, participants’ drinker prototypes were related to participants’ self-reported binge drinking. Most importantly, the model on observed alcohol use showed only a fixed effect for participants’ heavy drinker prototypes, suggesting that students’ heavy drinker prototypes were related to their observed alcohol use, irrespective of friend group differences in alcohol use.

Table 3.   Multilevel Analyses on Students’ Self-Reported and Observed Alcohol Use: Effects of Drinker Prototypes While Controlling for Gender
 Weekly recallBinge drinkingObserved alcohol use
B (SE)B (SE)B (SE)
  1. *< 0.05, **< 0.01, ***< 0.001.

Fixed effects
 Intercept3.18 (7.21)3.80 (0.79)***3.62 (0.74)***
 PT abstainer −3.04 (1.23)**−0.49 (0.13)**−0.12 (0.12)
 PT social drinker−0.45 (1.24)0.33 (0.14)*0.05 (0.12)
 PT heavy drinker4.44 (0.88)***0.59 (0.10)***0.18 (0.09)*
Random effects
 Sex−8.93 (1.73)***−0.79 (0.17)***−1.30 (0.33)***
Deviance intercept-only model1,578.99688.03681.73
Deviance full model1,504.14612.60611.84
χ² (df)74.85 (8)75.43 (8)69.89 (8)

To summarize, correlations and multilevel analyses showed that students’ abstainer and heavy drinker prototypes were related to their self-reported alcohol use. However, only heavy drinker prototypes were related to observed alcohol use after we controlled for group effects.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Previous studies already demonstrated that there is a relationship between young people’s drinking patterns and their prototypes or perceptions of the typical (non) drinker. So far, all evidence has been based on self-reported data (Gerrard et al., 2002; Norman et al., 2007; Spijkerman et al., 2007). In the present study, we used an innovative approach by using observations of students’ drinking practices during a time-out situation in a bar lab. This paradigm allowed us to gain insight into the question whether drinker prototypes explain alcohol use in an actual drinking situation. Our findings showed that besides to self-reported alcohol use, drinker prototypes were also related to observed alcohol intake. However, the friend group members had a strong impact on the number of drinks a participant consumed in this drinking situation. When controlling for this group effect in multilevel analyses, it were the students’ prototypes of the typical heavy drinker that related to drinking behavior in the bar lab. The more students held favorable perceptions of the typical heavy drinker, the more alcohol they consumed in the bar lab over and above the effect of the other group members’ drinking levels. An earlier study by Bot and colleagues (2005) already demonstrated that young adults’ alcohol expectancies predicted drinking behavior in a real-life drinking setting. The more young adults expected that they would experience mood enhancing and arousal effects from alcohol use, the more alcohol they consumed in the bar lab. The present study contributes to these findings by showing that other cognitive-motivational constructs, such as drinker prototypes, are related to observations of young adults’ ad lib drinking behavior in a real-life drinking setting as well.

According to our data, drinker prototypes were related to students’ observed alcohol use in the bar lab, even after taking into account the impact of the friend group. This group effect accounted for 70% of the explained variance in students’ observed alcohol use and was stronger than the group effect found in the study by Bot and colleagues (2005). In their data, the impact of group members explained 46% of students’ observed alcohol use. As described earlier, we followed the same paradigm as was used by Bot and colleagues. This resulted in similar conditions for data collection. However, we recruited friend groups that were generally smaller in size. This overall difference in group size may have lead to diverging group effects suggesting that the relatively small groups induced a stronger impact of the total friend group on student’s observed drinking rates than the larger groups. This can be explained by our qualitative observations indicating that members of small friend groups stayed more closely together, probably increasing group cohesion and the total group effect. In contrast, members of larger friend groups tended to disperse and interact in subgroups with different ordering and drinking rates which may have resulted in a decreased impact of the total group (Bot, 2007).

It is remarkable that only heavy drinker, and not abstainer and social drinker prototypes, were related to observed alcohol use after we controlled for the variance within friend groups. Without taking these group effects into account, students’ abstainer prototypes were related to observed drinking rates as well. The large intraclass correlation indicates however that an individual’s alcohol intake in the bar lab is strongly determined by the overall drinking level of one’s friend group members. Thus, for an adequate interpretation, it is important to take into account this strong group effect. When we controlled for the group effect, only perceptions of the typical heavy drinker contributed to the explanation of participants’ observed alcohol intake. Presumably, these perceptions determine whether participants drink more in comparison to their other group members. The unique contribution of heavy drinker prototypes to the explanation of observed alcohol use among friend groups may be due to the fact that heavy drinkers, in particular, will hold favorable heavy drinker prototypes. These heavy drinkers may be more inclined to differ from the general drinking levels of their friend groups.

An interesting issue that deserves some further elaboration is the question whether we should have controlled for participants’ self-reported drinking practices when examining relations between prototypes and ad lib drinking in a naturalistic setting. In contrast to longitudinal studies that usually include past drinking behavior to detect changes in alcohol use over time, we did not enter self-reported drinking behavior as a control variable in our analyses predicting observed alcohol intake in the bar lab. The reason for this approach was that we were not interested in unique changes in drinking patterns, but in the common variance between different drinking measures. We assumed that self-reported and observed drinking levels would tap into the same behavior (as was confirmed by our data demonstrating positive correlations between participants’ observed alcohol use and their self-reported drinking patterns). Controlling for self-reported alcohol use in analyses predicting observed alcohol intake, would imply that we predicted a unique part in observed alcohol use, which was left unexplained by self-reported drinking measures. This analytic approach would not have provided answers to our research questions. Self-reported drinking behavior was, therefore, left out in the analyses. If we had been interested in unique processes that were present during students’ drinking practices in the bar lab, such as, for example, specific influences of group members, controlling for self-reported alcohol use would have made sense (Bot et al., 2007b).

In contrast to most other prototypes studies, we examined several drinker prototypes, i.e. abstainer, social and heavy drinker prototypes. Both abstainer and heavy drinker prototypes showed relations with students’ drinking rates, but we hardly found any relationship with social drinker prototypes. On first glance, this might be due to the fact that students showed less variance in their perceptions about the typical social drinker. However, when comparing standard deviations of the 3 drinker prototypes we did not find a clear indication that students varied less in their social drinker prototypes compared to the other 2 drinker prototypes. The absence of a relationship between social drinker prototypes and students’ alcohol use could also mean that perceptions about the typical social drinker hardly play a role in students’ drinking practices. One explanation for the low impact of social drinker prototypes may be that impressions of social drinkers are less salient and meaningful to students than impressions of heavy drinkers or abstainers. Since abstention and binge drinking are both at the extreme ends of college students’ drinking practices, these drinking patterns may be more noticeable than moderate alcohol use in social settings. Social psychological research has shown that people have a tendency to focus on the extremity of behaviors and traits when judging or forming an impression of other people (Fiske et al., 2002). In line with this extremity bias, Gibbons and colleagues (2003) proposed that the more extreme a behavior, the more people would hold salient prototypes about the type of people who engage in this behavior. Salient prototypes will probably have a stronger impact on people’s behavioral decisions, than prototypes that are less clear and vivid. Although there is evidence for an extremity bias, specific assumptions about how the extremity of a behavior can affect the formation and impact of prototypes have not been further tested. To gain more insight into this topic, future research could include measures on the extremity of different forms of substance use and the salience of substance-related prototypes. In addition, future studies could include measures asking students what type of drinker they perceive themselves to be. This type of measures will provide information about the personal value participants attach to a specific drinker type, and which drinker types are regarded as less personally relevant.

Another direction for future research could be the use of experimental designs to study the effects of prototype salience on young people’s behavior.

Our study revealed several gender differences in drinking patterns and prototypes. In line with previous studies (Grant et al., 2001; Perkins, 1992), males showed higher alcohol consumption levels than females. In addition, male students held less favorable abstainer prototypes and more favorable heavy drinker prototypes, than their female counterparts. Additional analyses were conducted to test whether gender would moderate the relationship between drinker prototypes and students’ drinking patterns, but this was not the case. This means that males had different drinker prototypes than females, however the impact of these drinker prototypes on participant’s drinking patterns were similar for both sexes. The observed gender differences in drinker prototypes may be explained by the fact that people’s perceptions about drinking are probably related to their own drinking norms and behavior. Since males drink considerably more than females, they will probably be more accepting of heavy drinkers than females.

For the interpretation of our findings, it is important to discuss several limitations of the present study. The first limitation involves our study sample. Since we used college students as our research participants, it can be questioned whether our findings can be generalized to younger samples, such as early and late adolescents. This is especially relevant since conducting a similar study among under-aged youngsters is not feasible due to ethical restrictions. It is clear that alcohol consumption levels differ between adolescents and college students. Still, we expect relations between drinker prototypes and alcohol use to be similar for both populations, as there is evidence that adolescents and college students show large similarities in their vulnerability to peer influences (Poelen et al., 2005). Moreover, the impact of drinker prototypes has been demonstrated in both age groups (Gerrard et al., 2002; Gibbons and Gerrard, 1995). Another selective feature of our sample could be their self-reported alcohol use, which appeared rather high, i.e. 17 glasses of alcohol in the past week. However, the self-reported drinking rates that were found in the present study are comparable with the reported drinking rates in earlier research on alcohol use among Dutch college students, which showed an average drinking rate of 16 alcoholic drinks per week (Maalsté, 2000).

Another limitation of the present study may lie in the credibility of our bar laboratory as natural drinking setting and the question whether participants really behaved as they would have behaved in other drinking situations. In this regard, it is important to bear in mind that students received their drinks for free. This could mean that the drinking situation we created in our research was probably more comparable to drinking at home with friends than to drinking in a normal bar. Observations of participants’ behaviors and social interactions (chatting, telling jokes, playing table soccer, happy faces, drinking, etc.) during the break and the fact that almost all students believed our cover story, suggest that we at least succeeded in creating a relaxed atmosphere were students had a drink. In addition, we found that students’ drinking rates during this break were related to their self-reported drinking behavior. This relation suggests that students’ observed alcohol use partly reflected their self-reported drinking patterns. Students who reported high drinking levels, also showed higher drinking levels during the time-out situation in the bar lab. Importantly, the overlap between self-reported drinking rates, and the observed alcohol intake in the bar lab does not allow further conclusions about the ecological validity of these paradigms. To test the ecological validity of the barlab paradigm future studies should combine unobtrusive assessments of participants’ alcohol use in both natural drinking contexts and a bar lab. This way, researchers can examine to what extent the observed drinking rates in a bar lab correspond with participants’ drinking behavior in their natural drinking environment.

In addition, it should be emphasized that the observational measure in the present study tapped drinking behavior during 1 single drinking occasion within a relatively short period of time. In contrast, the self-report measures we used included a larger time-frame, i.e. past week and past 6 months, and provided therefore a more general impression about participants’ drinking patterns. In future, researchers could test the validity of these measures by conducting multiple assessments over time including both questionnaires and observations in the bar lab.

A fourth limitation that should be addressed is the correlational nature of our data. In the present study, we only conducted observations at 1 point in time and we did not use an experimental design. Therefore, definite conclusions about the direction of the relationship between drinker prototypes and actual alcohol intake could not be drawn. To further gain insight into this issue, future studies could include experimental designs and use a combination of self-report and observational measures.

In sum, with the present study we have taken a first step to further explore the predictive value of drinker prototypes by using a different methodological approach, i.e. the use of a less obtrusive measure for students’ alcohol use. Based on this first study, we can conclude that students’ drinker prototypes are related to their alcohol use, even if we measure students’ actual alcohol use in a naturalistic context.

  • 1

    Details about the construction of the presented prototype scales and/or the pilot study can be retrieved from the first author.

  • 2

    We used a general description of the research project since detailed information on the precise aim of the study could influence the participants’ behavior and seriously affect our results. After the measurements, all participants were fully debriefed. The used protocols were the same as used in earlier studies by Bot and colleagues and were approved by the medical ethical committee (CCMO Arnhem-Nijmegen).


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

This study was supported by the Netherlands Organisation for Scientific Research (NWO). The authors thank Anne Wissink, Inger van de Kamp, Iris van Berkel, Marieke Buursma, and Marion Ooms for their assistance with the data-collection.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  • Aitken PP (1985) An observational study of young adults’ drinking groups-II Drink purchasing procedures, group pressures and alcohol consumption by companions as predictors of alcohol consumption. Alcohol Alcohol 20:445457.
  • Blanton H, Gibbons FX, Gerrard M, Conger KJ, Smith GE (1997) The role of family and peers in the development of prototypes associated with substance use. J Fam Psychol 11:271288.
  • Bot SM (2007) Real life in the pub: an observational study on predictors of young adult social drinking. Radboud University Nijmegen, Nijmegen.
  • Bot SM, Engels RC, Knibbe RA (2005) The effects of alcohol expectancies on drinking behaviour in peer groups: observations in a naturalistic setting. Addiction 100:12701279.
  • Bot SM, Engels RCME, Knibbe RA, Meeus WH (2007a) Pastime in a pub: observations of young adults’ activities and alcohol consumption. Addict Behav 32:491504.
  • Bot SM, Engels RCME, Knibbe RA, Meeus WH (2007b) Sociometric status and social drinking: observations of modelling and persuasion in young adult peer groups. J Abnorm Child Psychol 35:929941.
  • Bruun K (1959) Drinking behavior in small groups. Uudenmaan Kirjapaino, Helsinki.
  • Chassin L, Presson CC, Sherman SJ, Corty E, Olshavsky R (1981) Self-images and cigarette smoking in adolescence. Pers Soc Psychol B 7:670676.
  • Chassin L, Tetzloff C, Hershey M (1985) Self-image and social-image factors in adolescent alcohol use. J Stud Alcohol 46:3947.
  • Clapp JD, Holmes MR, Reed MR, Shillington AM, Freisthler B, Lange JE (2007) Measuring college students’ alcohol consumption in natural drinking environments: field methodologies for bars and parties. Evaluation Rev 31:469489.
  • Clapp JD, Ketchie JM, Reed MB, Shillington AM, Lange JE, Holmes MR (2008) Three exploratory studies of college theme parties. Drug Alcohol Rev 27:509518.
  • Denscombe M (2001) Uncertain identities and health-risking behaviour: the case of young people and smoking in late modernity. Br J Sociol 52:157177.
  • Fiske ST, Cuddy AJ, Glick P, Xu J (2002) A model of (often mixed) stereotype content: competence and warmth respectively follow from perceived status and competition. J Pers Soc Psychol 82:878902.
  • Gerrard M, Gibbons FX, Reis-Bergan M, Trudeau L, Van de Lune LS, Buunk B (2002) Inhibitory effects of drinker and nondrinker prototypes on adolescent alcohol consumption. Health Psychol 21:601609.
  • Gibbons FX, Gerrard M (1995) Predicting young adults’ health risk behavior. J Pers Social Psychol 69:505517.
  • Gibbons FX, Gerrard M (1997) Health images and their effects on health behavior, in Health, Coping and Well-Being (BuunkBP, GibbonsFX eds), pp 6394. Lawrence Erlbaum Associates, Mahwah, New Jersey.
  • Gibbons FX, Gerrard M, Lane DJ (2003) A social reaction model of adolescent health risk, in Social Psychological Foundations of Health and Illness (SulsJ, WallstonK eds), pp 107136. Blackwell Publishing, Malden, MA.
  • Goldstein H, Browne W, Rasbash J (2002) Multilevel modeling of medical data. Stat Med 21:32913315.
  • Grant BF, Stinson FS, Harford T (2001) The 5-year course of alcohol abuse among young adults. J Subst Abuse 13:229238.
  • Hajema KJ, Knibbe RA (1998) Changes in social roles as predictors of changes in drinking behaviour. Addiction 93:17171727.
  • Karam E, Kypri K, Salamoun M (2007) Alcohol use among college students: an international perspective. Curr Opin Psychiatr 20:213221.
  • Knibbe RA, Oostveen T, Van de Goor I (1991) Young people’s alcohol consumption in public drinking places: reasoned behaviour or related to the situation? Br J Addict 86:14251433.
  • Lange JE, Voas RB, Johnson MB (2002) South of the border: a legal haven for underage drinking. Addiction 97:11951203.
  • Lloyd B, Lucas K, Holland J, McGrellis S, Arnold S (1998) Smoking in adolescence: images and identities. Routledge, London.
  • Maalsté N (2000) Ad fundum! Een blik in de gevarieerde drinkcultuur van het Nederlandse studentenleven [Ad Fundum! A gaze in the varied drinking culture of the Dutch student life]. Centrum voor Verslavingsonderzoek, Utrecht.
  • MacFadyen L, Amos A, Hastings G, Parkes E (2003) ‘They look like my kind of people’-perceptions of smoking images in youth magazines. Soc Sci Med 56:491499.
  • McKennell A, Bynner J (1969) Self images and smoking behaviour among school boys. Br J Educ Psychol 39:2739.
  • Norman P, Armitage CJ, Quigley C (2007) The theory of planned behavior and binge drinking: assessing the impact of binge drinker prototypes. Addict Behav 32:17531768.
  • Perkins HW (1992) Gender patterns in consequences of collegiate alcohol abuse: a 10-year study of trends in an undergraduate population. J Stud Alcohol 53:458462.
  • Poelen EAP, Scholte RHJ, Engels RCME, Boomsma DI, Willemsen G (2005) Prevalence and trends of alcohol use and misuse among adolescents and young adults in the Netherlands from 1993 to 2000. Drug Alc Depen 79:413421.
  • Quigley BM, Collins RL (1999) The modeling of alcohol consumption: a meta-analytic review. J Stud Alcohol 60:9098.
  • Rasbash J, Steele F, Browne W, Prosser B (2005) A user’s guide to MLwiN. Version 2. Center for Multilevel Modeling. University of Bristol, Bristol.
  • Rivis A, Sheeran P, Armitage CJ (2006) Augmenting the theory of planned behaviour with the prototype/willingness model: predictive validity of actor versus abstainer prototypes for adolescents’ health-protective and health-risk intentions. Br J Health Psychol 11(Pt 3):483500.
  • Spijkerman R, Van den Eijnden RJJM, Engels RCME (2005) Self-comparison processes, prototypes, and smoking onset among early adolescents. Prev Med 40:785794.
  • Spijkerman R, Van den Eijnden RJJM, Overbeek G, Engels RCME (2007) The impact of peer and parental norms and behavior on adolescent drinking: the role of drinker prototypes. Psychol Health 22:729.
  • Thombs DL, Dodd V, Pokorny SB, Omli MR, O’Mara R, Webb MC, Lacaci DM, Werch C (2008) Drink specials and the intoxication levels of patrons exiting college bars. Am J Health Behav 32:411419.
  • Van de Goor LA, Knibbe RA, Drop MJ (1990) Adolescent drinking behavior: an observational study of the influence of situational factors on adolescent drinking rates. J Stud Alcohol 51:548555.
  • Van Schoor G, Bot SM, Engels RC (2008) Alcohol drinking in young adults: the predictive value of personality when peers come around. Eur Addict Res 14:125133.
  • Wiers RW, Van Woerden N, Smulders FT, De Jong PJ (2002) Implicit and explicit alcohol-related cognitions in heavy and light drinkers. J Abnormal Psychol 111:648658.