School and family participation in a longitudinal study of tobacco use: some methodological notes

School-based recruitment of youth for a research purpose is problematic, as one has to deal with multiple ethical and practical constraints. This study reports on predictors of schools' and families' participation in a longitudinal study of tobacco use among 11-year-old children. School size, but not tobacco policy, was linked to participation. High parental education and average social status in the residence area were associated with parental consent. Given the social background, prevalence of parental tobacco use was close to the expected. Compliance of responsible adults in youth research cannot be predicted on the basis of simple behavioural models.