The purpose of this appendix is to provide an account of the National Child Development Study (NCDS) and British Cohort Study (BCS70) populations and samples as they change over time, covering the major sweeps of the two cohorts until the year 2000. In the following the response rates in the two cohorts, the analytic sample and sample attrition, as well as the approaches used for handling the possible effects of missingness and non-response will be described. Another issue considered here concerns the reporting of effect sizes of observed associations.

Participation in the 1958 National Child Development Study (NCDS)

The target sample for NCDS is defined as all babies born in Great Britain during the reference week 3–9 March 1958, until they die or permanently emigrate from Great Britain. In NCDS the original target sample included altogether 17,634 babies. For the first three follow-up sweeps at ages seven, eleven and sixteen, cohort members were traced through records and the National Health Service Central Register. Throughout adulthood, tracing of the survey members was based on existing address records and was continuously updated by sending an annual birthday card.

As Table B1 shows, the overall level of response has remained high, with 10,979 cohort members taking part in the 2000 survey at forty-two. The amount of sample attrition up to the age of eleven years was quite small. Even after age eleven the attrition rate was smaller than in most panel studies (Kerckhoff, 1993; Shepherd, 1993; Shepherd, 1995).