SUMMARY
Longitudinal surveys typically devote considerable resources to tracking procedures designed to minimise attrition through failure to locate sample member who move. Although these tracking procedures are often very successful, there is relatively little methodological evidence about the relative success, and cost-effectiveness, of different tracking procedures (Couper and Ofstedal, 2009). This paper extends the existing literature by exploring the relative effectiveness of office tracking and field tracking and by examining the role of respondent characteristics as a determinant of tracking success rates. These issues are explored using the Millennium Cohort Study, a large-scale birth cohort study in the UK. The existing research on tracking procedures has been based on household panel surveys, but in the context of a birth cohort study with relatively high mobility rates among the study population and longer intervals between waves, the effectiveness of office tracking procedures is particularly important. Our main finding, that respondent characteristics are associated with the overall tracking success rate but do not have a significant association with office tracking success, implies that survey practitioners should consider ways of improving their tracking procedures certain groups of respondents.