Linear mixed models were used to determine the effect of treatment over time on the AA Involvement and attendance variables (log transformed), on the five IPA subscales, and on Number of Abstinent Friends in the Social Network. Because we had specific hypotheses as to how the treatments should influence the IPA and AA variables, planned contrasts were included in these analyses. One contrast compared the NS treatment with Case Management, and the other compared the two NS treatments. It was expected that NS would yield greater abstinence orientation in the social network than would CaseM, and that NS+CM would yield more network change than would NS. Similar analyses were used to evaluate the effects of treatment on self-efficacy and on coping, which we believed would be significant mediators of treatment effects based on our past findings (e.g., Litt et al., 2003).
