The choice between a panel and cohort study design

Abstract

The UK has a long and rich history of longitudinal studies, including cohort studies and longitudinal panels, and we summarise the main studies and their key features. We go on to identify the key differences between cohort and panel approaches, in the population coverage and whether refreshment samples are expected; what types of questions can be used; the sample design and sample size; and the choice of spacing of waves. Over a long period panels and cohorts collect similar amounts of information, but in a single cohort there is no way to measure the cohort effect separately from time effects. Early in the studies, the differences are greater. Cohort studies can have questionnaires much more tailored to the age and expected life course events of the respondents than a panel study, which must cater for many ages. Accelerated cohort designs offer an intermediate solution, where the collection can be tailored, but where cohort effects can be estimated, as long as there is some overlap between the ages in the cohorts. There are several additional factors to consider in an accelerated cohort design. Making choices about which features are important in a new cohort study depends on the answers to many questions about what the intended analytical outcomes are, and how long the study will run