A research project developed a computer program for analyzing time series quasi-experimental data. The program generates a nonstationary, integrated moving average time series model; it is used to estimate a parameter which indexes the instantaneous change in level of the time series due to a predesignated treatment. The entire method is based on a Bayesian model for which a uniform rectangular prior to and following the treatment carry primary weights. Computer Print-out Appendix deleted from document due to irreproducibility of original. (Author)