Citation

Polak J. Transp. Res. Rec. 1999; 1660: 164-171.

Copyright

(Copyright © 1999, Transportation Research Board, National Research Council, National Academy of Sciences USA, Publisher SAGE Publishing)

DOI

10.3141/1660-21

PMID

unavailable

Abstract

Panel data, in which the same observational units are surveyed at several points in time, offer a number of important advantages compared with more conventional cross-sectional or time-series data. However, panel data are also susceptible to specific forms of bias that do not affect cross-sectional or time-series data, for example, panel attrition, fatigue, and conditioning. If these sources of bias are not properly taken into account in the preanalysis of panel data, they may lead to serious distortion of the conclusions of substantive analyses. It is therefore important to seek a better understanding of the nature of these biases, how they arise, and how to deal with them. The objective of this paper is to explore these issues by presenting the results of the preanalysis of two different panel datasets collected in different countries at different times and using different survey approaches. The patterns of wave-to-wave attrition in the two datasets are compared, and various approaches to dealing with this problem are discussed.

Language: en