ABSTRACT

The long-standing interest of sociologists in causal interpretation of statistical relationships has been quickened by discussions focusing on linear causal models. The basic work of bringing such models to the attention of the discipline was done by Blalock [2], drawing upon the writings of Simon [23, Chapter ii] and Wold and Juréen [31] in particular. The rationale of this approach was strengthened when Costner and Leik [8] showed that “asymmetric causal models” of the kind proposed by Blalock afford a natural and operational explication of the notion of “axiomatic deductive theory,” which had been developed primarily by sociologists working with verbal formulations. Most recently, Boudon [6] pointed out that the SimonBlalock type of model is a “special case” or “weak form” of path analysis (or “dependence analysis,” as Boudon prefers to call it). At the same time, he noted that “convincing empirical illustrations are missing,” since “moderately complicated causal structures with corresponding data are rather scarce in the sociological literature.” This chapter presents some examples (in the form of reanalyses of published work) that may be interesting, if not “convincing.” It includes an exposition of some aspects of path technique, 56developing it in a way that may make it a little more accessible than some of the previous writings.