As useful as LISREL may be in model estimation and testing, its most significant contribution to date is the encouragement and example it gives for right thinking about research and right planning of research. The encouragement to hypothesize the best possible model for the process that is the object of study, and to plan measurements that adequately specify the model, is the aspect of LISREL which is transforming educational research. The estimated model encourages researchers to believe that criterion achievement was heavily influenced by input achievement and modestly influenced by input attitude, but not by instructional process. If the major question of instructional process research is whether the measurement of instructional process variables yields information which contributes to the explanation of learning variance, the answer given for this example by the FaM model squares with the answer given by the LISREL model. The newest version of LISREL redresses previous grievances indicating that it is going to be easier to compute and interpret LISREL models. (Author/PN)