Causal Models, Unobserved Variables, and Experimental Data *

This chapter is concerned with possible applications of causal modeling techniques in the analysis of experimental data. It examines the treatment of unobserved variables in path analysis to the analysis of experimental data. The chapter discusses ways in which causal models may be fruitfully used in the analysis of experimental data and to illustrate their uses. One of the early concerns of causal analysts working with nonexperimental data was with the treatment of unobserved or unobservable variables in structural equation models representing causal systems. Blalock and Costner have emphasized the importance of viewing experiments in terms of unobserved causal processes. An important contribution Costner is the suggestion that one can examine the presence of “demand characteristics” and/or “experimenter” effects within the framework of the causal modeling approach. A causal model that includes both observed and unobserved variables is set forth to represent the causal process underlying the experiment.