using computer assisted interview which yields higher reported prevalence of risk-taking behaviour than regular interviews (Resnick et al., 1997). Sensitive questions, including those on pubertal development and substance use were given via headphones. Audio presentation of items relating to risk-taking behaviour has also been associated with a reduction in an under-reporting bias (Turner et al., 1998). Detailed information about the design, sampling and data instruments is available in Bearman, Jones and Udry (1998).
