Chunk #24 — 3. Results — 3.2. Relation between AA, network support/activities, and alcohol use over time — 3.2.4. Social networks and activities as mediators of AA's effect on alcohol use
We conducted mediation analyses for the cases where all paths in the lagged models were statistically significant using the MacKinnon method noted previously. As shown in Table 4, there was significant consistent mediation observed for pro-abstinent and pro-drinking network ties across the outpatient and aftercare samples on PDA. This ranged from a low of 7% of the direct effect between AA and PDA (mediated by pro-abstinent network ties for aftercare patients) to a high of 16% of the direct effect (mediated by pro-abstinent network ties for outpatients). AA-influenced reductions in pro-drinking network ties explained 13% of the effect on PDA for both patient samples. In contrast, there was less consistency in the meditational effects of the social network variables for intensity of drinking (DDD). Having more pro-abstinent network ties did not help explain lower DDD in relation to AA attendance for either outpatient or aftercare patients. However, reductions in pro-drinking network ties explained a portion of the effect of AA on DDD among outpatient (14% of the direct effect) and aftercare patients (12% of the direct effect).
