Panel studies and migration

Attrition is a critical concern in panel data sets, and migration processes are an important influence affecting attrition. In this paper, we examine patterns of attrition in a household panel in a setting, Northeast Thailand, where migration effects are visible because refusing to participate is essentially non-existent. We develop a model of locating panel members, and then examine different stages of the follow-up process separately. Not only is the pattern of attrition influenced by a variety of individual, household, and community-level factors, but the types of variables that are important and the direction of their effects vary from stage to stage in the follow-up process. The findings are consistent with hypotheses derived from the migration literature. Our findings also hint that social network factors are important in locating those who had migrated, suggesting the importance of incorporating social network components in follow-up strategies.