Refreshment sampling for longitudinal surveys
Jul 25, 13:45
As a longitudinal survey matures, there may be a need to consider refreshing the sample. This may be due to the impacts of attrition on the sample size or representativeness of the sample, or a desire to increase the size of the overall sample, target (over-represent) particular segments of population, or provide coverage of new population entrants. For example, household panel studies or cohort studies may add an immigrant sample top-up to include those who arrived in the country after the study began. Or a ‘boost’ sample restricted to a sub-population of particular policy interest, such as a geographical region or a minority group, may be added to better address policy needs.
In this chapter, we discuss the different types of refreshment samples drawing on the experiences of longitudinal surveys around the globe, including household panels (such as PSID, SOEP, SLID, SHP, HILDA Survey, and Understanding Society), birth cohorts (such as NCDS), youth cohorts (such as NSLY), and older-age cohorts (such as ELSA, HRS and SHARE). We provide considerations for refreshment frequency, sampling frames, sample design, questionnaire design, integration of the sample into the fieldwork, weighting, and the impact these refreshment samples have on analysing the data.