Abstract
The effectiveness was assessed of selected demographic and psychological variables in distinguishing between those male alcoholics who completed an 8 wk inpatient treatment program at a Veterans Administration facility and those who left the program prematurely. Employing a stepwise multiple linear regression procedure resulted in nine variables of sufficient predictability to account for 27% of the variability in the criterion of program completion. Further application of the derived equation resulted in correct classification of 78% of the cases. This represented a 7% improvement over and above knowledge of the observed Stay-Split ratio. These findings were discussed in the context of a multidimensional conceptualization of alcoholism.