Anonymity and confidentiality in school surveys on alcohol, tobacco, and cannabis use

School surveys are currently the most important method of collecting data o n alcohol and drug use among youth. Although methodological studies suggest that school surveys yield reliable and valid estimates of substance use am ong youth, it has been argued that results will be affected by the level of perceived anonymity. Longitudinal research designs raise an important ques tion in this respect since follow-up makes complete anonymity impossible. I n fact responses to such surveys should be regarded as confidential rather than anonymous. This study compares the reported use of cigarettes alcohol and cannabis between an anonymous, cross-sectional survey and a confidentia l, longitudinal survey. Both females and males in the confidential survey h ave a slightly higher rate of nonresponse for lifetime cannabis use. Furthe rmore, females fend to be slightly less likely to admit to any use of alcoh ol and cannabis in the confidential survey, and those who do admit to havin g used cigarettes and alcohol, report slightly fewer occasions. The correla tions between use of cigarettes, alcohol, and cannabis are not affected by this bias. These results add further support to earlier research that has f ound school surveys to be a robust method of data collection, and suggests that the bias introduced by identification numbers in longitudinal research has limited practical significance.