| Household survey data on age at first use of alcohol can be biased from both sampling and response errors. The paper presents a procedure for obtaining a more accurate distribution of age at first use by viewing the NHSDA as a cohort study. One potential concern is attrition, whereby individuals who start alcohol use at an early age become increasingly less likely to be surveyed in successive years. This paper suggests that such attrition causes substantially less bias than inaccurate responses. This contention was partially confirmed through a comparison of NHSDA cohort data with longitudinal data obtained from a panel study of New Jersey youths The paper presents findings for the 1968-73 birth cohort, which provide extensive bias presumably caused by forward telescoping. By the time of the interview at age 25, the average age at first use of alcohol reported was 16.1 as opposed to the presumably more accurate 14.3 estimated. Additionally, the proportion of 25 year-olds who reported first use of alcohol before age 12 was only a third of the proportion reported at age 12. |
Updated 05/20/2006