First, bivariate differences between abstainers and non-abstainers were examined using Chi-square and t-tests. Next, variables that differed significantly between abstainers and non-abstainers were entered into multivariable logistic regressions in a stepwise manner to predict non-abstinent recovery: in Step 1, demographic variables were entered; in Step 2, treatment history was entered; and in Step 3, DSM-IV alcohol dependence severity was entered. For the aim regarding quality of life, a linear stepwise regression was used with quality of life as the dependent variable: in Step 1, demographic variables were entered; in Step 2, treatment history was entered; in Step 3, DSM-IV alcohol dependence severity was entered, and in Step 4, an indicator for non-abstinence was entered. All analyses were conducted in SPSS (v18).
