Objective. - The aim of the study was to evaluate the relative contributions of peers and' own beliefs about use, in the prediction of cannabis use. parental influences and adolescents Method. - Participants were 559 high-school and secondary school students (275 girls, mean age = 15.4 +/- 1; 274 boys, mean age = 15.5 +/- 0.9) who completed questionnaires assessing cannabis use frequency, the number of peers using cannabis, the number of peers opposed to cannabis use, parental attitude toward cannabis use, parental present or past cannabis use and participants' expectations toward use. Parents' opinion of cannabis use was assessed using a ten-point scale ranging from zero (highly opposed to cannabis use) to 10 (highly in favour of cannabis use). The participants' opinion of cannabis use was assessed using a self-report questionnaire which was generated from a preliminary qualitative study on a convenient sample of ten adolescents who agreed to participate in a semistructured interview assessing their perceptions of the effects of cannabis use. Interviewers drew up a List of all the reported perceptions. One rater eliminated redundant responses and combined similar instances into more general terms. Responses were reworded concisely to be appropriate for a close-ended questionnaire. The final questionnaire consisted of 29 items. Items were scored on a 7-point Likert scale, ranging from 1 = disagree strongly to 7 = agree strongly. Results. - In the total sample (n = 559), 22% of girls In = 61) and 28% of boys In = 76) reported having used cannabis once during the last six months (p=0.05); 4% of girls and 9% of boys used cannabis at least 3-4 times per week; water pipe or bong was used by 31% of boys and 28% of girls used cannabis. Cannabis users reported that 49% of their fathers were using or had used cannabis versus 10% of non-users. Cannabis users reported that 39% of their mothers were using or had used cannabis versus 22% of non-users. An exploratory factorial analysis of the cannabis use expectations questionnaire was conducted. The eigenvatue curve suggested either a two-factor solution explaining 46% of the variance. These factors were called 'positive expectancies' (eigenvaiue = 9.0; explained variance = 29%, Cronbach's alpha = 0.86) and 'negative expectancies' (eigenvalue=4.0; explained variance=17%; Cronbach's alpha=0.93). The correlation of factors was negative and moderate (Pearson's r=-0.29). Cannabis users were characterised by a higher number of peers using cannabis, a lower number of peers opposed to use, a lower level of negative opinion of parents, a higher level of positive expectancies and a tower level of negative expectancies. It is to be noted that both users and non-users tended to perceive their parents as highly opposed to use. A logistic regression analysis predicting cannabis use versus non-use was performed entering sex, the number of peers opposed to cannabis use, the number of peers using cannabis, the opinion of parents, parental present or past cannabis use and positive and negative expectations factor scores. A test of the full model with at[ predictors against a constant-only model was statistically reliable: the predictors reliably distinguished between users and non-users (chi(2) (8) 153.9; p < 0.0001). The variance in cannabis use accounted for was high, with McFadden rho(2) = 0.39. Prediction success was satisfactory, with 94% of non-users and 59% of users correctly predicted. The number of peers opposed to cannabis use (B = -0.08; t-ratio = 3.9; p = 0.04), the number of peers using cannabis (B = 0.06; t-ratio = 7.9, p = 0.01), the positive expectations score (B = 0.94; t-ratio = 26.6; p < 0.0001) and negative expectations scores (B = -0.50; t-ratio = 11.8; p = 0.0006) and father's present or past cannabis use (B = 1.17; t-ratio = 8.2; p = 0.004) were significant independent predictors of cannabis use. These results indicated that the higher the number of peers using cannabis and the positive expectations, the higher the risk for initiation of cannabis use. The regression coefficient of the number of peers opposed to cannabis use and of the negative expectations score were negative. These results indicated that the less the number of peers opposed to cannabis use and the lower the negative expectations, the higher the risk for initiation of use. Parental attitudes toward use and mother's present or past cannabis use were not significant independent predictors of use. Discussion. - As our sample was non-ctinical, a first limitation of our findings is that they may not be transposable to patient populations. Another limitation of our study is linked to its cross-sectional design, which prevents the attribution of causal explanations for the associations found. One of the study's strengths is that it assesses potentially important variables not evaluated in previous studies, such as the number of peers opposed to cannabis use and positive and negative expectations of use. The results of the present study suggested that the number of peers using cannabis, father's present or past cannabis use and participants' positive expectations of cannabis use were risk factors for use, whereas the number of peers opposed to cannabis use and the negative expectations of use were protective factors. Parental attitudes toward use did not appear to influence adolescents' cannabis use. In conclusion, our results may have some implications for prevention interventions. They add weight to the view that normalisation of non-use by peers facilitates abstinence. The absence of influence of parental attitudes toward use suggests that parental disapproval of use is not effective in preventing use, whereas the example of father's use or non-use influences adolescent use. The quite low correlation between positive and negative expectancies suggests that prevention interventions presenting information concerning the effects of cannabis use should focus on both reducing positive expectancies and enhancing negative expectancies. (c) L'Encephale, Paris, 2008.
