Detecting serial correlation in the error structure of a cross-lagged panel model

Cross-lagged panel studies are statistical studies in which two or more variables are measured for a large number of subjects at each of several waves or points in time. The variables divide naturally into two sets and the primary purpose of analysis is to estimate and test the cross-effects between the sets. Such studies are found in the mainstreams of social, behavioral and business research. One approach formulates a multivariate regression model in which the cross-effects are parameters. We contribute to this approach by considering the problem of testing whether the regression model should allow for serial correlation in the error structure. We demonstrate the tests developed by considering a panel study of the attitudes of patients toward the health maintenance organization in which they are enrolled.