Date of Award


Publication Type

Doctoral Thesis

Degree Name





Psychology, Clinical.




The present study used univariate and multivariate data analyses to investigate the combination of variables that could be used to differentiate between and predict drinking habits of male alcoholics after treatment. Alcoholic subjects were Abstainers (N = 60), Controlled drinkers (N = 60) and Relapsed drinkers (N = 60). Forty variables were assessed including measures of demographics and social stability, problem drinking, treatment characteristics, and psychopathology. The ANOVA revealed that alcoholics who relapse after treatment had lower incomes, higher alcohol consumption, and higher Michigan Alcoholism Screening Test and Behavior Impairment Index scores than the controlled drinkers. Relapsed drinkers had more frequent and longer previous inpatient treatment and attended more A.A. meetings than the controlled drinkers. The relapsed drinkers also had less ongoing treatment and poorer attendance in treatment than the abstainers. The relapsed drinkers had significantly higher scores than the abstainers or controlled drinkers on the Frequency (F), Hypochondriasis (Hs), Depression (D), Psychopathic deviate (Pd), Paranoia (Pa), Psychasthenia (Pt), Schizophrenia (Sc), Social Introversion (Si), and Anxiety (A) scales of the MMPI. A stepwise multiple regression technique that maximized the multiple correlation coefficient (R('2)) produced a 28 variable model which accounted for 25 percent of the variance in drinking behavior. Discriminant analysis differentiated between the three groups such that the relapsed drinkers were very high on psychopathology (Function 1) and near the mid range of problem drinking (Function 2). The abstainers were low on psychopathology and moderately high on problem drinking while the controlled drinkers were low on psychopathology and moderately low on problem drinking. Classification coefficients were generated for use in predicting or classifying drinking behavior of new subjects. Unfortunately, the discriminant analysis revealed that, although better than at a chance level (33 percent), the model was able to correctly classify only 63 percent of the abstainers, 60 percent of the controlled drinkers, and 65 percent of the relapsed drinkers. The implications and problems of the present study are discussed.Dept. of Psychology. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis1985 .S658. Source: Dissertation Abstracts International, Volume: 46-09, Section: B, page: 3231. Thesis (Ph.D.)--University of Windsor (Canada), 1985.