Exploring Survey Non-Response: The Effect of Attrition on a Follow-Up of the 1984-85 Health and Life Style Survey

SUMMARY Two rounds of a survey on health and life styles among adults in Great Britain have been conducted-in 1984-85 and 1991-92. Each round of the survey consisted of an interview followed by a physical examination and then a selfcompletion questionnaire. This multistage design meant that several levels of panel non-response were incurred. This paper explores the differential characteristics of responders and non-responders with the goal of expanding general knowledge about survey non-response, seeing to what extent non-response may have been connected with the topic of the survey and providing valuable information for users of the data. Non-response was examined at the various stages of the survey: firstly through demographic and socioeconomic characteristics of the sample and secondly through a selection of key health-related and psychological variables. The effects of the various components of attrition (i.e. refusals, non-contacts, those not traced and deaths) were also considered. Finally, the effect of attrition on substantive analyses was examined. The general conclusions drawn are that different components of non-response are associated with different patterns of biases, and that the effect of these biases over time tend to be cumulative rather than offsetting. Implications for survey research practice are discussed.