Abstract: In panel studies, respondents are lost when they die, decide to drop out of the study, or when they cannot be traced during later rounds of data collection. Much has been written about attrition, the loss of individuals over time, in developed countries, but literature from developing countries is scarce. The experiences from developed countries may not be relevant for developing countries as the main reasons for attrition differ. While attrition in developed country contexts largely arises from refusals to participate, attrition in developing country contexts is mostly due to respondents moving.

Attrition can cause bias if it is selective, and efforts should be made to track respondents. Such efforts can be costly and difficult as populations in developing countries are often highly mobile, infrastructure is poor, structures frequently change, and formal population and address records rarely exist.

In this paper, the author reviews the experiences from panel studies in developing countries to consider the importance of attrition and of tracking respondents to reduce attrition, and makes recommendations for establishing systems to track respondents in developing countries. Tracking can reduce attrition by up to 45 per cent, and is feasible if procedures are locally appropriate, well planned, and involve the community; if data is collected locally as much as possible; if criteria are explicit; if tracking is conducted regularly; and if interviewers are well trained, supervised, and motivated.

Keywords: panel studies, attrition, tracking