Aims: Adolescents' reports of parents' education are sometimes used as indicators of socioeconomic status in surveys of health behaviour. The quality of such measurements is questionable. We hypothesized that consistent reporting of parents' education across measurement occasions in prospective panel studies indicates a higher quality of data than single or inconsistent reports.
Methods: A multi-site, prospective panel study (three measurement occasions) was carried out among adolescents in Cape Town and Mankweng (South Africa), and Dar es Salaam (Tanzania). Analyses were based on data from students participating at baseline and with a valid code for school number (n = 15,684).
Results: For Cape Town and Dar es Salaam students, the associations between parents' education and an alternative indicator of socioeconomic status (both measured at baseline) increased with increasing consistency of reports about parents' education across measurement occasions. For Cape Town, the associations of father's education with a range of behavioural and social cognition variables were significantly stronger among ;;consistent'' than among other students. The pattern was the same for mother's education, but with fewer significant interaction effects.
Conclusions: Requiring consistency of reports across data-collection occasions may, under the right combination of circumstances, make a difference. Insignificant and "close to zero'' associations may turn out to be at least moderately strong and statistically significant. When applying indicators of socioeconomic status, such as parents' highest level of completed education, it is most advantageous to use data from prospective panel studies, and to check for consistency of answers across measurement occasions.