Citation

Schoenberg R. Soc. Sci. Res. 1977; 6(2): 133-144.

Copyright

(Copyright © 1977, Elsevier Publishing)

DOI

10.1016/0049-089X(77)90004-7

PMID

unavailable

Abstract

Most social variables are inherently imbedded in time and, therefore, are generated by a dynamic process of some kind. This paper explores the consequences of applying a static model to such variables. We find that unbiased estimates of the underlying dynamic parameters through the application of a static model to a cross-section of data are possible only if the underlying dynamic process is "nonergodic," that is, that the process is a function of the initial conditions. A wide-sense stationary dynamic model that is not a function of the initial state of the system is proposed here, and we find that the application of static models to such processes is not successful, in particular, we find that the static estimates are attenuated or inflated versions of their dynamic counterparts.