The nature and use of LISREL (LInear Structural RELationships) analysis are considered, including an examination of college students' commitment to a university. LISREL is a fairly new causal analysis technique that has broad application in the social sciences and that employs structural equation estimation. The application examined in this paper involves an aspect of Tinto's model of college student attrition: students' commitment to an institution. Data were obtained from a persistence study of first-year students at a major public university. During the first week of classes, students provided information on their background, their motivation for enrolling in school, and their commitment to the institution. Two months later, students were surveyed again concerning their commitment to the school and their activities on campus. The analysis involves a measurement model and structural equations model. The measurement variables: "peer relationships,""residency," and "social activities" are each expressed as some multiplier times the latent construct "social integration." The structural equations model specifies the causal relationships among exogenous variables (e.g., age and ethnicity) and endogenous variables (e.g., initial institutional commitment). (SW)