Les attentes d’effets positifs (AEP) procurés par l’alcool et le niveau de consommation d’alcool sont souvent considérés comme des variables positivement liées. À partir d’un échantillon de 1 017 jeunes adultes ayant participé à une enquête par questionnaires, cette étude explore à l’aide de la classification TwoStep deux possibilités n’ayant pas été documentées dans la littérature : (1) on peut être un buveur modéré tout en ayant des AEP élevées, (2) on peut être un buveur abusif tout en ayant des AEP faibles. Trois niveaux de consommation ont été déterminés à l’aide d’une version française de l’Alcohol Use Disorders Identification Test et six types d’AEP ont été quantifiés avec une adaptation française de l’Alcohol Expectancy Questionnaire. Les résultats confirment seulement la première possibilité. La classe des buveurs modérés à haut niveau d’AEP est constituée d’hommes à 62 %. Les buveurs catégorisés alcoolo-dépendants sont constitué d’hommes à 71 % et présentent un profil moyen d’AEP élevé.

Aims – The present study investigates the heterogeneity of the population of young adult drinkers with respect to alcohol consumption and Positive Alcohol Expectancies (PAEs). Based on the positive relationship between both kinds of variables, PAE is commonly viewed as a potential motivational factor of alcoholic addiction. Empirical analyses based on the regression of alcohol consumption on PAEs suppose that the observations are statistically homogeneous with respect to the level of alcohol consumption, however. We explored the existence of <sup>[1BROWN SA, CHRISTIANSEN BA, GOLDMAN MS. The Alcohol Expectancy Questionnaire : an instrument for the assessment of adolescent and adult alcohol expectancies. J Stud Alcohol 1987 ; 48 : 483-91.

Cliquez ici pour aller à la section Références]</sup> moderate drinkers with a high PAE profile, and <sup>[2BROWN SA, GOLDMAN MS, INN A et al. Expectations of reinforcement from alcohol : their domain and relation to drinking patterns. J Consult Clin Psychol 1980 ; 48 : 419-26.

Cliquez ici pour aller à la section Références]</sup> abusive drinkers with a low PAE profile. Method – 1 017 young adult drinkers, mean age = 23 ± 2.84, with various educational levels, comprising 506 males and 511 females, were recruited as voluntary participants in a survey by undergraduate psychology students from the University of Toulouse Le Mirail. They completed a French version of the Alcohol Use Disorders Identifiction Test (AUDIT) and a French adaptation of the Alcohol Expectancy Questionnaire (AEQ). Three levels of alcohol consumption were defined using the AUDIT score, and six composite scores were obtained by averaging the relevant item-scores from the AEQ. The AEQ scores were interpreted as measurement of six kinds of PAEs, namely Global positive change, Sexual enhancement, Social and physical pleasure, Social assertiveness, Relaxation, and Arousal/Power. The TwoStep cluster methodology was used to explore the data. This methodology is convenient to deal with a mix of quantitative and qualitative variables, and it provides a classification model which is optimized through the use of an information criterion as Schwarz’s Bayesian Information Criterion (BIC). Results – The automatic clustering suggested five clusters, whose stability was ascertained until 75 % of the sample size. Low drinkers (n = 527) were split into one cluster of low PAEs (I1) and, interestingly, one cluster of high PAEs (I3, 46 %). High drinkers (n = 344) were split into one cluster of intermediate PAEs (II4) and one cluster of high PAEs (II5, 52 %). Interestingly again, abusive drinkers (n = 146) remained a single group (III2), exhibiting high PAEs. Clusters I3 and III3 comprised a significant proportion of males. Constraining the algorithm to find 6 clusters did not affect class III2, but split low drinkers into three clusters. Discussion – Although the present results should be considered cautiously because of the novelty of TwoStep cluster methodology, they suggest a group of moderate drinkers with high PAEs. Also, abusive drinkers express high PAEs (except for 2 cases). Statistical homogeneity of moderate drinkers with respect to PAE variables appears as a dubious assumption.