FRANÇAIS :
Au cours d'une étude longitudinale auprès d'adolescents de Belfast, le taux de rétraction est évalué par le nombre de réponse positive vis-à-vis d'usage de drogues au cours de la vie, qui est dénié une année plus tard. Les taux de rétraction observés sont de l'ordre de 7% pour un usage passé d'alcool à 87% pour un usage passé de champignons hallucinogènes. Plus la stigmatisation sociale de l'usage du produit est grande, plus la rétraction est élevée. Ces taux importants de rétraction mettent en doute la fiabilité des témoignages des jeunes adolescents.

ENGLISH:
Aims To assess the reliability of drug use reports by young respondents, this study examined the extent of recanting previous drug use reports within an ongoing longitudinal survey of adolescent drug use. Here, recanting was defined as a positive report of life-time drug use that was subsequently denied 1 year later. The covariates of recanting were also studied. Design An ongoing longitudinal survey of young adolescents (Belfast Youth Development Study) in Northern Ireland. Setting Pencil and paper questionnaires were administered to pupils within participating schools. Measurements Measures analysed included (a) recanting rates across 13 substances, (b) educational characteristics, (c) offending behaviour and (d) socio-economic status. Findings High levels of drug use recanting were identified, ranging from 7% of past alcohol use to 87% of past magic mushroom use. Recanting increased with the social stigma of the substance used. Denying past alcohol use was associated with being male, attending a catholic school, having positive attitudes towards school, having negative education expectations and not reporting any offending behaviour. Recanting alcohol intoxication was associated with being male and not reporting serious offending behaviour. Cannabis recanting was associated with having negative education expectations, receiving drugs education and not reporting serious offending behaviour. Conclusions The high levels of recanting uncovered cast doubts on the reliability of drug use reports from young adolescents. Failure to address this response error may lead to biased prevalence estimates, particularly within school surveys and drug education evaluation trials. (Author' s abstract)