Abstract
Longitudinal research is a useful and unique approach for providing insights into oral health, but longitudinal studies are expensive, labour intensive, and time-consuming. It is crucial that investigators be able to locate participants for follow-up interviews or examinations, often on numerous occasions. Study members may move location, lose interest, or simply 'disappear'. In the current study, data from the Dunedin Multidisciplinary Health and Development Study are presented to demonstrate how non-participation bias may affect the validity and generalisability of the findings of longitudinal studies. The Dunedin Study is a study of a complete birth cohort, and it is unique among such studies in that participation rates of greater than 90% have been maintained for over 30 years. A large amount of effort and funding was used to ensure that Study members who were difficult to locate and recruit were found and included in the Study. The 'difficulty to recruit' each Study member was identified after the age 26 assessments, and the social and dental characteristics of those Study members who were considered to be 'most difficult to recruit' were compared and contrasted against the characteristics of the remainder of the cohort. The same was done at age 32 years, and these data are also presented. Important differences between the findings of these comparisonsat age 26 and age 32 years will be considered; finally, techniques used in the Dunedin Study to ensure high participation rates and minimise non-participation bias are presented and discussed.