ABSTRACT
This chapter considers rival hypotheses abstractly by representing the basic features of an experiment and of potential rival hypotheses in the form of a causal model. Experiments are typically designed, not to document the effects of specific manipulations, but to serve as a test of a more abstract causal theory. Originally developed by Sewall Wright for working with problems in genetics, path analysis consists in making assumptions about causal structure explicit and, on the basis of those assumptions, estimating the effect of one variable on another by appropriate regression procedures. The applications and discussions of “path analysis,” “structural equation models,” or “causal models” in sociology have, until recently, concentrated very heavily on the estimation of effects for systems in which all variables are measured variables. In summary, the procedures sketched allow a test of the tenability of a set of assumptions about the causal structure by which experimental manipulations have observed effects.
