Conferences

The paper reports work conducted in the first stage of a project directed at reducing attrition in longitudinal surveys. The first stage comprises an investigation of the correlates of non-response in longitudinal surveys through statistical modelling of attrition processes using data relating to attrition in established UK longitudinal surveys: British Household Panel Study (BHPS) 1958 and 1970 British birth cohort studies, Family and Children’s Survey (FACS). The second stage is to formulate, in the light of the results of the modelling, optimum strategy for encouraging continued participation as a basis for improved field strategy to reduce attrition. The third stage is to design and implement an exemplar field experiment to test the new strategy using a survey from a series of repeated cross-sectional ('Omnibus’) surveys to supply the first wave of a longitudinal survey in the second wave of which respondents are allocated at random to an experimental and control condition. The results of the secondary analysis identify amongst other factors interviewer attributes and experience as factors in attrition and provide the basis for the manipulation of standard interviewer allocation procedures in the Omnibus experiment