ISSUES IN LONGITUDINAL SURVEYS

Abstract

Longitudinal survey is the lifeblood of the study of development and change over time. In many questions concerning changes in both characteristics and composition in survey populations, a longitudinal survey can provide far greater insight than the traditional cross sectional survey. Studies in sociology, psychology, public health, medicine, income and many other fields have greatly benefited from the rich source of individual database provided by longitudinal surveys. However, no one should conduct a longitudinal survey without first understanding the many important issues concerning this type of survey. In this thesis, we shall first look at the nature of longitudinal survey and various research dimensions that demand the use of longitudinal survey. Next, we examine three important issues of longitudinal surveys: attrition, panel conditioning, and population change and sampling unit. These are issues that are either absent or less acute in cross sectional surveys. We shall touch on the causes, problems and rectification of attrition. Next, a discussion on the nature of panel conditioning, its merits and demerits, and several case studies that help illuminate the significance of panel conditioning is presented. We proceed to look at how population change can be accounted for in longitudinal surveys and the implications and considerations in choosing an appropriate sampling unit.

Finally, we shall look at the Singapore scene for the application of longitudinal survey in ongoing national surveys and new studies in various fields. We consider how a longitudinal design in national surveys like the Household Expenditure Survey and the Labour Force Survey can provide more in-depth information and present a more complete picture of the survey population. This will aid policy makers and researchers in the implementation of new policies. We also look at how the issues discussed above relate to Singapore and why it is feasible and necessary to conduct longitudinal surveys in Singapore.

We hope that this thesis will provide the reader with much insight, thereby assisting them in conducting a longitudinal survey.