Stepwise regressions within the WIR sample allowed us to examine how non-abstinence related to quality of life above and beyond demographic, help-seeking, and problem severity variables. We found that compared to total abstinence, non-abstinence was associated with 0.3 to 0.45 point-reductions (on a 4-point scale) in QOL, even when controlling for length of time in recovery (Table 3). All control variables were significantly related to quality of life except education, having attended formal treatment and problem severity. Furthermore, the effects of Hispanic ethnicity and 12-step group attendance dropped out when controlling for length of time in recovery (Model 4 vs. Model 5). Once controlling for all demographic, help-seeking, and severity variables (Table 3, Model 5), the significant correlates of quality of life were being female (B=.11), older age (B=.06), White race (B=.19), being employed (B=.23), being married (B=.26), non-abstinence (B=-.36) and length of time in recovery (B=.06). According to standardized coefficients for the full Model 5 (not shown), the strongest correlates of higher QOL were: longer length of time in recovery (Beta = 0.19), being married (Beta = 0.19) and abstinence (Beta = 0.16).
