An Agent-Based Model of the Influence of Lifetime Attendance On the Stability of Self-Help Groups

Background and purpose: Within the literature on Alcoholics Anonymous, lifetime attendance has been controversial (Bean, 1975; Walters, 2002). Empirical evidence for the benefits of lifetime attendance in maintaining sobriety is mixed (Kaskutas, Ammon, DeLucchi, Bond & Weisner, 2002; Lloyd, 2002). However, in addition to delivering treatment, self-help organizations face the challenge of maintaining growth in the absence of professional staff. No researcher has systematically examined the possibility that lifetime attendance may benefit the growth of the organization.

Methods: In this study, two agent-based models of self-help group growth were developed using NetLogo 4.0.2. Each agent represents an individual alcoholic searching for a meeting. Each agent has a sobriety score that translates into a chance of returning to drinking. Once an agent does return to drinking that agent drops out of the model. An agent affiliates with a self-help organization by affiliating with a mentor who is attending a group that is part of that organization. Groups add members through this process of individual affiliation. Once a group reaches sufficient size it splits into two different groups. The model tracks both the number of groups and the number of members in each organization. Each model was run 100 times for 2000 iterations each.

Results: The study first compared two self-help organizations in which agents practice lifetime attendance. In this comparison, following 2000 iterations the number of groups in each organization was negatively correlated (r = -.815) as was the number of members (r = -.842). There was no statistically significant consistency in the difference between the number of groups (t = -1.39, df = 99, p = .169) or members (t = -1.35, df = 99, p = .181). Thus, neither organization had a consistent advantage in growth. Rather, outcomes were multistable; from an identical starting point with identical model settings, either organization could end up as the larger. This is typical of path-dependent growth driven by positive feedback loops (Arthur, 1994). The second model compared an organization with lifetime attendance to one in which members graduate when their sobriety scores reach 200. The lifetime attendance organization had a strong tendency to have more groups, more current members, and more members overall (i.e., when we compared current membership of the lifetime attendance organization to current membership plus graduates of the organization that encouraged graduation) (t = 8.80, 9.23, and 6.43 respectively, all p-values <0.001). The groups eventually returned to parity as the sobriety score required for graduation increased.

Conclusions and implications: This study suggests that self-help organizations show path-dependent growth and that those with lifetime individual membership may be more likely to succeed. Self-help group researchers may have looked at lifetime membership through too narrow a lens, and should consider its relationship to organizational stability as well as individual sobriety. Social workers who work with self-help groups may want to encourage members to continue to participate for some period of time after presenting problems appear resolved.