Citation

Sandefur GD, Tuma NB. Soc. Sci. Res. 1987; 16(4): 301-328.

Copyright

(Copyright © 1987, Elsevier Publishing)

DOI

10.1016/0049-089X(87)90006-8

PMID

unavailable

Abstract

This paper investigates the implications of using different types of panel data to study individual mobility by means of empirical analyses of data from the Panel Study of Income Dynamics on intercounty migration of white male heads of household in 1975-1980. While event history data are clearly superior to panel data for studying individual mobility, often event history data are unavailable. We examine the consequences of different panel designs (varying number and spacing of waves) and of different timings of the measurements of explanatory variables. The results indicate that: (1) measuring the covariates at the end of an interval can be misleading, especially if these covariates are likely to change during the interval as a consequence of mobility; (2) a study of the mobility of a group of individuals during a single short interval can lead to different conclusions from a study of the mobility of the same group of individuals during a single long interval; and (3) multiwave panel data have clear advantages over data from a single short interval or a single long interval.