Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Sep 16.
Published before final editing as: J Subst Use Addict Treat. 2026 Jul 28;191:210070. doi: 10.1016/j.josat.2026.210070

Longitudinal Engagement and Drinking Outcomes Associated with Mode of 12-Step Attendance

J Scott Tonigan 1, Matthew R Pearson 1
PMCID: PMC13577259  NIHMSID: NIHMS2202174  PMID: 42521159

Abstract

Introduction.

12-step programs have demonstrated efficacy, and evidence suggests that online 12-step meeting attendance has become an established pathway to the 12-step program, both as a sole point of access and in combination with in-person attendance. Because the relative effectiveness of these two modes of 12-step attendance have yet to be compared clinicians have little empirical guidance in making referrals to either in-person or online 12-step meetings. This study contrasts online versus both online and in-person attendance and experiences among early 12-step affiliates. Three domains central to the 12-step program were compared: engagement in prescribed behaviors, impressions of 12-step group dynamics, and drinking outcome.

Method.

This was an observational longitudinal study (N = 192) examining the change trajectories of early 12-step members. Participants were recruited via online platforms and selection criteria restricted enrollment to individuals with limited AA histories. Assessments for the entire sample included online interviews at baseline, 3, and 6-months and a subset of the sample also completed twice daily EMA online surveys for 6-months (n =133). The EMA sample constituted the sample for this study and, of the 133 EMA participants, 110 were classified by 12-step attendance mode.

Results.

Demographic profiles did not differ between adults attending only online meetings (n = 40) versus attending both online and in-person meetings (n = 70), with the exception that attending both 12-step modes was associated with pre-baseline treatment episodes. Rates of 12-step attendance were equivalent over 3-months between 12-step modes although online only participants reported less engagement in prescribed activities, and they reported significantly lower agreement on the tasks and objectives of the 12-step program. Rates of abstinence (baseline to 3-months) were significantly higher when attending both 12-step modes (94.51%; SD=17.24) relative to only online meetings (81.44%; SD=17.38). No between-group differences were found on measures of drinking intensity and negative consequences.

Conclusion.

Online 12-step meeting attendance is a popular method to access the 12-step program. While online meetings offer enhanced accessibility the exclusive use of online meetings was associated with less engagement in prescribed 12-step activities and lower rates of abstinence relative to attending both online and in-person meetings. Awaiting replication, findings suggest that clinicians should recommend attending both 12-step meeting modes during early recovery.

Keywords: Alcoholics Anonymous, 12-step, recovery, virtual meetings

1. Introduction

A substantial proportion of people go to Alcoholics Anonymous (AA) rather than to formal treatment for help with alcohol problems (Bergman et al., 2024; NSDUH, 2023), and most treatment providers in the United States encourage AA attendance during treatment and as aftercare (Kelly, Yeterian, & Myers, 2008). Meta-analytic work (e.g., Kelly & Humphreys, 2020; Tonigan et al., 1996; Tonigan et al., 2018) offers strong support for this practice with reported positive and significant correlations between frequency of AA attendance and later increases in abstinence. AA participation has also been linked with improvement in quality of life and well-being (Wnuk, 2022), reduced health care costs (Humphreys & Moos, 2001; Kelly et al., 2020), and improvement in recovery self-efficacy (Adelman-Mullally & Sullivan-Kerber, 2021).

Online 12-step meeting attendance is becoming a popular and an established pathway to access the 12-step program, fueled in part from the social distancing mandates during the pandemic. In a national sample of adults with a lifetime alcohol use disorder (AUD), for example, Timko et al., (2024) reported that among participants reporting 12-step as their primary mutual-help group (MHG), only 9% of the individuals exclusively attended in-person meetings. In contrast, about 65% of the 12-step participants attended both in-person and online meetings and 27% attended only online meetings. Similar rates of online 12-step attendance have been reported in the 2022 AA membership survey (Alcoholics Anonymous World Services, Inc, 2024). Here, of the 6,000 respondents, about 75% of the individuals reported attending both in-person and online meetings.

Various factors may account for the increase of online 12-step attendance. In a qualitative study of 50 individuals recruited mostly from AA, Hassett-Walker (2025) reported that participants felt virtual meetings offered greater and easier access, especially in rural communities, and that virtual meetings were also more conveniently scheduled to accommodate daily activities. In a second qualitative study (Sousa et al., 2025), the authors recruited a nationally representative sample of MHG attendees (N = 20, 75% AA) and conducted semi-structured interviews to identify the motivations and experiences of online users. Four motivational profiles were identified: Maximizers (online to supplement in-person meetings), Enthusiast (desired enhanced anonymity and less anxiety), Compliance seekers (ease of complying with court orders), and Strategists (desired new ideas and perspectives). Clearly, much remains to be understood about the reasons and motivations for the migration to online AA meetings. It should be noted, however, that in-person AA attendance was valued by participants in both qualitative studies and that a majority of the AA members in the 2022 annual membership survey preferred in-person AA meetings.

A critical issue facing AUD treatment providers when making AA referrals is whether to encourage in-person, online or both modes of 12-step attendance. Indirect evidence suggests that mode of AA attendance may moderate 12-step fidelity, involvement, and outcomes (Timko et al. 2022; Timko et al., 2024; Zemore et al., 2018; Zemore et al., 2025; Zemore et al., 2026). Interpretative caution of this literature is warranted, however, because, to date, MHG mode comparisons by this research group have only been conducted after first aggregating participant responses from different types of MHG programs to form in-person versus online groups (i.e., 12-step, Women for Sobriety, LifeRing, and SMART Recovery). This practice may mask within 12-step mode comparisons given that MHG programs espouse and embrace different and, at times, oppositional theoretical orientations, prescriptions, and task demands.

With this caveat, Timko et al, (2022) found that relative to no online attendance, any online MHG attendance was associated with less involvement in MHG prescriptions and, at a trend level, less perceived cohesion with other group members. Notably, the finding of less involvement with any online exposure was replicated when contrasting only online attendance with only in-person MHG attendance (Timko et al., 2024). This replication is notable because the authors expanded the original 2022 composite measure of involvement to include additional mode-specific items (e.g., have you texted or emailed another member) to more fully capture the construct of interest. More recently, Zemore et al. (2025) reported that people who only attended virtual meetings (relative to only in-person meetings) had poorer outcomes at both 6 and 12 - months on three measures of past 30-days alcohol use: abstinence, alcohol problems, and heavy drinking days. Interestingly, contrasting drinking outcomes of people who attended only virtual meetings with individuals attending both MHG modes yielded similar results on two of the three drinking measures at 6 months, but this difference faded at the 12-month follow up. Collectively, it appears that virtual MHG meetings (broadly defined by combining participants from multiple MHG programs) mobilizes less MHG involvement which, in turn, results in poor drinking outcomes. This interpretation is consistent with the extant AA research of the importance of the main effect of AA involvement activities, above and beyond attendance, in predicting drinking reductions (e.g., Kelly et al., 2015; Tonigan & Rice, 2013; Zemore et al., 2013). Noteworthy, the importance of involvement or commitment to prescribed MHG activities in predicting positive outcomes has recently been extended and replicated across multiple non-AA MHG participants (Zemore et al., 2026).

To our knowledge, no investigation has compared 12-step engagement and outcomes of individuals who attended both in-person and online 12-step meetings with AA members attending only virtual online meetings. For several reasons, this within MHG comparison represents a critical gap in knowledge, made even more important given the growing popularity of virtual 12-step meetings. Foremost, while alternative MHG has increasingly been identified as conferring benefit (e.g., Zemore et al., 2018; Zemore et al., 2025; Zemore et al., 2026), AA is, by far, the most popular MHG among the general U.S. population and treatment providers (Bergman et al., 2024). As such, gains in understanding the strengths and limitations of online 12-step meeting attendance have significant and pragmatic clinical implications when seeking to initially engage clients in 12-step programs. Second, several 12-step prescriptions have demonstrated associations with later drinking reductions (e.g., Pagano et al., 2011; Tonigan & Rice, 2010). It is unclear, however, if these prescriptions (e.g., getting a sponsor and service work) are initiated and sustained by individuals attending only virtual meetings. And third, social support to reduce alcohol use is a demonstrated mechanism accounting for 12-step-related benefit (e.g., Kelly et al., 2011). The nature and extent that 12-step members perceive the diverse dimensions of social support via online meetings is thus important.

This study has three aims. Based on a nationwide sample of early 12-step members without extensive 12-step histories, we contrast the extent of engagement in 12-step prescribed practices of individuals attending both in-person and virtual meetings with individuals attending only online 12-step meetings. Second, we compare the two 12-step modes on the perceived social dynamics and support of 12-step meetings, with close attention paid to the multidimensional nature of this construct. Third, we contrast the drinking outcomes of these two 12-step groups using multiple indicators of drinking frequency, intensity, and consequences. Our effort offers several refinements relative to past investigations of the benefits and limitations of online MHG participation. Specifically, (1) perceptions of 12-step group social dynamics were assessed comprehensively with a 10 scale self-report measure with demonstrated reliability, (2) a continuous daily account of drinking behavior for 3-months was used instead of sampling the most recent 30-day period in a 6-month window to infer overall drinking status and, (3) early 12-step members were recruited to best inform practitioners about the engagement behaviors and outcomes associated with different modes of 12-step attendance during initial 12-step exposure. This sampling decision contrasts with earlier reports that recruited individuals who had achieved sustained recovery (26.5% of the sample had 1–5 years of abstinence) or stable recovery (26.1% of the sample had 5+ years of abstinence) according to the NIAAA definition of alcohol recovery (Timko et al., 2022; Timko et al., 2024; Zemore et al., 2018; Zemore et al., 2025; Zemore et al., 2026).

2. Method and Measures

2.1. Study Design.

No intervention was offered in this observational study. Recruited participants were assigned to one of two groups: (1) a fixed assessment group that included comprehensive interviews at baseline, three and six-months (n = 59), and (2) a group that had twice daily ecological momentary assessment reports plus the three fixed assessments at baseline, three and six-months (n = 133). As the first intensive EMA study of AA, the fixed assessment group was designed to provide a comparison or control group to detect potential assessment reactivity associated with the twice daily EMA protocol. Evidence for assessment reactivity was negligible in the present sample (Hatwan et al., 2026).

2.2. Eligibility criteria.

Eligibility criteria included 1) 12-step attendance in the past 3 months, 2) lifetime 12-step attendance less than 9 months, 3) an Alcohol Dependence Scale (Skinner & Allen, 1982) score of 8 or higher to indicate the likely presence of an alcohol use disorder, 4) no pending legal convictions to minimize attrition, and 5) having a smartphone (in the event participants were randomized to the ecological momentary assessment group).

2.3. Recruitment and Screening.

Printed flyers and videos were posted on Facebook and InTheRooms.com to recruit study participants between March 2021 and March 2023. Figure 1 displays the Consort diagram of participant flow through the study. As shown, interested prospects were directed to complete an online screening survey (N = 29,725) and incomplete (n = 3,335) and duplicate screens (n = 5,586) were discarded. Of the remaining 20,404 entries, 16,605 were deemed ineligible for the following reasons: (1) not attending 12-step meetings in the past 90 days (n = 2,755), (2) having too much 12-step exposure (n = 12,732), (3) having a pending legal conviction (n = 554), and (4) having an Alcohol Dependence Scale (ADS; Skinner & Allen, 1982) score < 8 (n = 564). A total of 954 prospects were contacted (25.1%) from the remaining 3,799 prospects to achieve the study recruitment goal, at which time contact efforts ceased. Prospect reasons for not participating during our contact efforts included: (1) not interested (n = 76), (2) no-shows (n = 56), (3) repeated cancelled appointment (n = 13), (4) unresponsive to contact efforts (n = 593), and (5) deemed ineligible (n = 18).

Figure 1.

Figure 1.

Open in a new tab

Participant Flow (CONSORT Diagram)

Note. ADS=Alcohol Dependence Scale; EMA = Ecological momentary assessment group; FA = Fixed Assessment group

During these initial contacts, research assistants confirmed study eligibility, answered questions about the study and, if still interested, scheduled a baseline interview. A scripted protocol was used to describe the purpose of the study. Specifically, prospects were informed that we were interested in (1) how people experience AA, (2) the types of activities they may engage as AA members, (3) healthcare utilization, and (4) drinking and drug use status. Prospects were told that they may be assigned to a group that included only three interviews (baseline, 3-month, and 6-month) or a group that included the same three interviews plus twice daily EMA assessments for 6-months. 199 prospects (20.9%) agreed to participate and were consented and assigned to a group. 7 participants withdrew, leaving 192 participants who were virtually interviewed for the baseline assessment.

2.4. Minimization & Group Assignment.

We used the QMinim platform (Saghaei & Saghaei, 2011), which is a free service for minimizing participants to different arms of a controlled trial. Minimization is an alternative to randomization, with the advantage of balancing arms of the trial with respect to preselected covariates (Altman & Bland, 2005). The benefits of minimization (compared to simple randomization) are enhanced in the case of small sample sizes (Chen & Lee, 2011) and when there are several strong prognostic factors, or covariates that strongly relate to outcomes (Ye et al., 2022). In the proposed study, we selected three covariates to be used in the minimization procedure: 1) number of 12-step meetings attended in the past 90 days (0–8 vs. 9+), 2), drinking days in the past 90 days (0–40 vs. 41+), and 3) total score on the Alcohol Dependence Scale (Skinner & Allen, 1982) (0–24 vs. 25+). These factors were selected to minimize differences across groups on variables expected to predict alcohol-related outcomes.

2.5. Determination of 12-step Mode.

Both the fixed assessment (n = 59) and EMA (n = 133) groups were asked about frequency of AA attendance, but only the EMA group was asked whether such daily AA attendance was virtual or in-person. The EMA protocol involved completing two surveys per day (a random survey between 8 am-12 pm and an evening survey at 8 pm). In the evening survey, participants were asked if they attended any 12-step meetings. If they did attend a meeting, they were asked whether any meeting they attended was virtual. Aggregating across the first 3 months of data collection, we used this evening survey (70.6% median compliance; M=57.42 days, SD=26.81; Median=63.5) to characterize the proportion of meetings attended that were virtual/online and in-person to derive the groups used for subsequent analyses. All other data in this study were obtained from retrospective and comprehensive assessments obtained at baseline and the 3-month follow-up interviews. Ethical approval for these study procedures was obtained from the IRB at the participating university (protocol # 22219).

2.6. Measures

2.6.1. Demographic interview.

The 10-item CASAA demographic interview was used to collect information about participant age, gender, year’s education, employment status, occupation, income, ethnicity, residence status and contact information.

2.6.2. Form 90.

The Form 90 (Miller, 1996) is a semi-structured interview that includes calendars and memory prompts designed to help participants remember their day-to-day alcohol consumption. Research staff conducted virtual Form 90 interviews at each assessment point to assess participants’ healthcare utilization, frequency of 12-step attendance (but not mode) and alcohol use for the prior 90-day period. Previous studies have demonstrated that the Form 90 has satisfactory reliability for measuring number of days on which participants abstained from alcohol use (r = .79 for outpatients and r = .97 for aftercare patients) and days on which they engaged in heavy alcohol use, r = .96 for outpatients, and r = .97 for aftercare patients (Tonigan, Miller, & Brown, 1997). Form 90 interviewers were trained, certified, and monitored by a staff member, Ms. Roberta Chavez at CASAA, who routinely trains research staff across the country in the administration of the Form 90. Form 90 data were used to calculate two measures of alcohol use. We calculated proportion of days abstinent from alcohol (PDA) by dividing the number of days the participant reported being alcohol abstinent in the assessment period by the total number of days in that period. Drinks per drinking day (DPDD) were calculated by dividing the number of drinks consumed per drinking day by the number of drinking days in a period. Finally, proportion of 12-step attendance days was calculated by dividing the number of days of attending at least one 12-step meeting by the number of days in the assessment period.

2.6.3. DrInC.

The Drinker Inventory of Consequences (Miller, Tonigan, & Longabaugh, 1995) is a 50-item self-report measure that captures a wide range of alcohol-related consequences including the domains of impulse control (12 items), interpersonal (10 items), intrapersonal (8 items), physical (8 items), and social responsibility (7 items) consequences. All items use a four-point response scale that ranges from “Never/Not at all” to “Daily or almost daily/Very Much”. The response timeframe used in this study was, “In the past 90 days, about how often has this happened to you?” The DrInC has been normed using a national sample and has strong test-retest reliability and psychometric properties. We computed DrInC total scores by averaging all items (M=1.90, SD=.780, Min = 1.00, Max = 3.87, α=.979).

2.6.4. AUDIT.

The Alcohol Use Disorders Identification Test (AUDIT; Saunders, et al., 1993) is a 10-item screening measure developed by the World Health Organization to assess hazardous alcohol use, dependence symptoms, and harmful alcohol use. All items are scored from 0 (never/no) to 4 (daily to almost daily/yes, during the past year), and items were summed to create a total score (α=.919).

2.6.5. The Group Environment Scale Form R

(GES: Moos, 2002) This measure has 10 scales to assess perceptions of social group interactions (90-items total; 9 items per scale). The GES has been normed extensively and each scale has good psychometric properties. The GES scales include Cohesion (α=.798), Task Orientation (α=.678), Leader Support (α=.762), Expressiveness (α=.504), Innovation (α=.659), Anger and Aggression (α=.755), Order and Organization (α=.743), Independence (α=.538), Self-Discovery (α=.395), and Leadership Control (α=.523). Instructions for completing the GES items were as follows:

“We want you to think about the 12-step meeting you attend most often. You are to decide which of these statements are true of this meeting and which are not. If you think the statement is true or mostly true of your group, select “true”. If you think the statement is false or mostly false of your group, select “false”.

Scale scores were summed with true coded as 1 and false coded as 0, having a range of 0 to 9.

2.6.6. Twelve Step Participation Questionnaire.

The TSPQ was adapted from the Alcoholics Anonymous Inventory (Tonigan et al, 1996) which has demonstrated reliability. The adapted and shorter TSPQ version has 14 items, with the first 10 items asking about prescribed 12-step behaviors (True or False; coded 1 and 0 respectively). Sample items are: Have you had a sponsor, Have you celebrated a 12-step birthday, Have you considered yourself to be a member of a 12-step group, Have you read 12-step literature, and Have you called a 12-step member for help. Item 11 asked respondents to indicate the steps, if any, they had completed, and item 12 inquired if respondents had a treatment experience and, if so, whether 12-step attendance was mandated/encouraged in the treatment program. Item 13 asked if 12-step attendance was required by a court or judge (yes, no). Item 14 asked about frequency of meeting attendance at Narcotics Anonymous, Al-Anon, Other 12-step meetings, SMART, and LifeRing. At baseline, frequency of attendance was collected for three time periods, last 90-days, past year, and lifetime and, at follow-up, only the last 90-day period was queried.

3. Results

Of the 192 participants in the parent study, 133 were assigned to the EMA condition which formed the sample for this study. Of these 133 participants, 1 participant never initiated the EMA protocol, and 12 participants did not report attending a 12-step meeting during the 3 months of EMA data collection and therefore could not be classified as attending a 12-step attendance mode. Of the remaining120 EMA participants who attended a 12-step meeting, 40 (33%) reported only attending online/virtual 12-step meetings (i.e., 100% virtual), 10 participants (8%) reported only attending in-person 12-step meetings (i.e., 0% virtual), and 70 (58%) reported attending both online and in-person meetings (M=51.5%; SD = 31.1%; range: 2.9%-97.5%),. Given study aims, we excluded the in-person only mode from further analyses which reduced the final EMA baseline sample to 110.

3.1. Attrition.

At 3 months, 96 of the 110 classified participants were interviewed (87%; n = 62 for both modes, n = 34 for online only mode). One participant interviewed at 3 months provided Form 90 drinking data but did not complete the GES and TSPQ surveys. Thus, the 3-month EMA follow up rate differs from the CONSORT figure because the diagram includes unclassified and/or EMA noncompliant participants who completed the 3-month zoom interview. We examined whether there was differential attrition at 3-months by 12-step mode and, collapsing groups, if baseline alcohol severity measures predicted attrition. A 2 × 2 Chi square with 12-step meeting mode on one dimension and attrition at 3-months (completing the interview, yes/no) on the second dimension indicated no association between 12-step mode and attrition, χ2(1) =.29, p =.59. Logistic regression was then used to determine if four baseline measures of alcohol severity (PDA, DPDD, total DrInC score, and total AUDIT score) jointly predicted 3-month attrition after collapsing 12-step mode. The omnibus tests were not significant, χ2(4) =7.77, p=.10, although the total DrInC score was significant after controlling for the other three measures of alcohol severity (Wald = 6.42, p =.011). Here, 3-month completers reported, on average, 1.83 (SD =.75) adverse consequences while non-completers reported an average of 2.45 (SD =.93) consequences.

3.2. Sample Characteristics.

Baseline (N = 110) characteristics are reported in Table 1. As shown, the two groups did not differ in age and educational level. Further, most of the participants were female and were employed full or part time. Likewise, a majority in both groups reported some or completion of college degrees, and about half the members in the groups indicated that they were living with a spouse. The two modes did significantly differ in reported treatment experiences outside of this study in the 90-day period before recruitment such that 83% of the individuals attending both 12-step modes reported some treatment for emotional and/or substance problems in contrast to 58% of the individuals who only attended online 12-step meetings. Of the 23 participants reporting a pre-baseline treatment experience but who reported only online 12-step attendance after recruitment into this study, 16 participants reported that they were required to attend 12-step meetings during their pre-treatment episode, but they elected only online AA attendance after treatment.

Table 1.

Baseline Characteristics of Sample (n=110) by AA Mode

Both In-Person & Online (n=70) Online Only (n=40)
Continuous Variables M (SD) M (SD) p Hedges g
Age 43.96 (10.02) 42.58 (12.30) .52 .126
Level of Education 4.99 (1.38) 5.49 (2.00) .13 .298
Total AUDIT Score 20.59 (10.62) 19.53 (10.73) .62 .100
Total DrInC Score 1.98 (.86) 1.77 (.66) .17 .269
Proportion Days Abstinent 85.73 (22.70) 71.81 (38.71) .02 .469
Drinks per Drinking Day 6.35 (7.84) 4.90 (5.10) .30 .205
Proportion 12-Step Days (Form 90) .49 (.34) .43 (.36) .38 .171
Proportion 12-Step Days (TSPQ) .45 (.35) .43 (.36) .74 .064
Categorical Variables p Hedges g
Gender
 Male 25 11 .37 .169
 Female 45 29
Marital Status
 Single 17 9 .71 .226
 Married/Cohabitating 34 20
 Divorced 17 8
 Widow 1 2
Living Status
 Alone 13 11 .18 .492
 With Spouse 33 21
 With Parent/Friend 15 7
 Homeless 9 1
Educational Status
 High Schol and Less 15 7 .13 .527
 Vocational 7 2
 Some College 22 12
 BA and Advanced Degree 26 14
 Other 0 3
Employment Status
 Full & Part Time 41 19 .13 .198
 Home/Retired 6 9
 Unemployed 22 11
Treatment Episode
 Yes 58 23 .004 .564
 No 12 17
Open in a new tab

Note. TSPQ = Twelve Step Participation Questionnaire

Participants in the two 12-step modes did not differ on three of four measures of drinking severity, the exception being that the group attending both 12-step modes reported significantly higher proportion days abstinent in the 90-day period before recruitment. On average, however, members of both groups reported high to a very high risk of an alcohol use disorder (AUDIT), typically consumed 5–6 drinks per occasion, and they had experienced modest consequences associated with drinking. The groups did not differ at intake on the proportion of days 12-step attendance in the 90-days before recruitment, with estimates indicating that early 12-step members attended a 12-step meeting about every other day.

3.3. Follow-up 12-Step Attendance.

Early 12-step attenders reported stable rates of 12-step attendance between baseline and the 3-month follow-up. Mean proportion days 12-step meeting attendance for the combined 12-step mode was .48 (SD = .34) and .55 (SD = .38) for the online only group. The mean ratio of online meetings to the in-person number of 12-step meetings for the combined 12-step group was .52 (SD = .31) indicating that about 50% of the meetings for this mode were attended online. Controlling for baseline 12-step attendance, a Generalized linear model (GLM) indicated that the two 12-step modes did not differ, on average, in proportion days 12-step attendance at 3-months, χ2(1) = 2.899, p = .089. Overall, then, participants in both modes attended 12-step meetings about every other day during the 90-day follow-up period.

3.4. 12-Step Engagement.

Seven hierarchical logistic regressions were conducted to assess 12-step engagement patterns of the two 12-step groups. TSPQ items 3, 4, 5, 6, 8, 9 and 10 were the single-item dependent measures (collected at 3-months), with corresponding baseline items serving as covariates (entered in block one). A dummy coded variable representing 12-step mode was entered in block two. No adjustment was made for the multiple comparison problem, with primary interest on the pattern of obtained effect sizes across contrasts. Results of these logistic regressions are displayed in Table 2. None of the seven tests contrasting the two 12-step modes on measures of engagement were significant. Descriptively, the participants who attended both modes of 12-step meetings reported higher endorsement of all seven prescribed behaviors relative to the online only group, especially celebrating a 12-step birthday, considering themselves active 12-step member, and reading approved 12-step literature.

Table 2.

Comparison of AA meeting Mode and engagement in 12-step prescribed behaviors and activities in the TSPQ

TSPQ Item Both In-Person & Online (n = 61)
% “Yes”		 Only Online (n = 34)
% “Yes”		 p Hedges g
Considered oneself an AA member 85% (n=52) 62% (n=21) .08 .358
5+ AA Meetings in Week 56% (n=34) 50% (n=17) .63 .098
Celebrated 12-step Birthday 55% (n=33) 29% (n=10) .054 .404
Have a 12-step Sponsor 72% (n=44) 59% (n=20) .72 .073
Read 12-step Literature 90% (n=55) 82% (n=28) .06 .387
Called 12-step member for help 71% (n=43) 52% (n=16) .098 .346
Had a spiritual awakening 59% (n=36) 47% (n=16) .76 .063
Open in a new tab

3.5. 12-Step Social Dynamics and Support.

Ten GLM’s with corresponding baseline values of the dependent measure indicated that at 3-months, the two 12-step groups significantly differed on only one GES scale depicting impressions of the social dynamics of 12-step meetings, Task Orientation. Again, no adjustment was made to the multiple comparison problem. Table 3 displays Hedges g effect sizes for the 10 contrasts, with positive effect sizes indicating higher endorsement of a scale for the combined 12-step mode. Six effect sizes favored the combined 12-step mode at the small to moderate range in magnitude while the remaining 4 effect sizes were less than .10.

Table 3.

Comparison of AA Social Dynamics by Mode of AA Group Attended: Adjusted Means (SE) for Group Environment Scale

GES Scale Both In-Person & Online Mean (SE) n=61 Only Online Mean (SE) n=34 Main Effect Slope Wald p Hedges g
Cohesion 7.20 (.27) 6.53 (.36) .667 2.26 .13 .39
Expressiveness 5.82 (.22) 5.65 (.30) .173 .214 .64 .10
Independence 6.08 (.23) 6.08 (.31) −.006 .000 .99 .00
Task Orientation 6.87 (.24) 5.71 (.32) 1.16 8.55 .003 .62
Self-Discovery 6.30 (.19) 6.21 (.26) .089 .078 .78 .06
Anger & Aggression 2.03 (.26) 2.18 (.35) −.144 .110 .74 .07
Order 7.72 (.24) 7.06 (.32) .62 2.74 .10 .35
Leadership Control 5.16 (.23) 5.44 (.31) −.277 .500 .48 .15
Innovation 2.52 (.24) 2.50 (.32) .025 .004 .95 .01
Leader Support 7.20 (.27) 6.53 (.36) .667 2.26 .13 .34
Open in a new tab

3.6. 12-Step Mode Drinking Outcomes.

Three GLM’s were employed to contrast the two 12-step modes on 3-month PDA, DPDD and the total DrInC score, with baseline values of the dependent measure entered as covariates. At 3-months, the omnibus test for PDA was significant, χ2(2) =75.03, p<.001, and the main effect of 12-step mode was also significant, χ2(1) =12.16, p<.001. Here, the adjusted mean for the 12-step group attending both modes was 94.51% (SD=17.24; n = 62) abstinent days in the 90-window between baseline and 3-month follow-up while the online only group adjusted mean was 81.44% (SD=17.38; n = 34) abstinent days (Hedges g=.75). A follow-up GLM was done to determine if the higher abstinence rate for the combined 12-step mode was an artifact of this group reporting significantly higher rates of formal treatment for substance use and emotional problems at baseline. With proportion days treatment (90-days before baseline) added as a covariate in the model (grand mean = .147, SD = .187) the main effect of 12-step mode remained, χ2(1)=10.70, p<.001. In contrast, while the omnibus χ2 was significant for DPDD at 3-months, χ2(2)=31.08, p<.001, the main effect of 12-step mode on 3-month DPDD was not significant, χ2(1)=2.24, p=.13 (Hedges g =.32). Third, the omnibus test for the DrInC consequences measure was nonsignificant, χ2(2)=5.22, p=.07, and the main effect of 12-step mode was also nonsignificant, χ2(1)=1.149, p=.28 (Hedges g =.23).

4. Discussion

Virtual online 12-step meetings are a popular avenue for early 12-step affiliates to access the 12-step program. Thirty-six percent (40 of 110) of the participants reported attending only online meetings and, on average, participants who attended in-person 12-step meetings also attended online meetings about 50% of the time. Results highlighted a counter-intuitive pattern that differentiated how early 12-step members may engage in prescribed 12-step behaviors contingent upon type of 12-step mode. Specifically, at the trend level (all p’s < .10), the in-person plus online attendance mode reported higher endorsement on a majority of the prescribed AA behaviors, e.g., reading 12-step literature, celebrating a 12-step birthday, considering themselves a 12-step member, and calling a 12-step member for help. Except for calling another 12-step member, higher rates of endorsement on the remaining three practices does not appear to be a function of greater ease of contact afforded by in person 12-step attendance. In contrast, the two AA modes did not differ on rates of having an AA sponsor (effect size = .07) at three months, a practice that would favor in person interactions. Collectively, these contrasts of 12-step mode-related practices are perplexing and underscore the need to reconceptualize and expand our notions about the relative advantages of in person versus online 12-step attendance.

Social support for abstinence is an important mechanism accounting for AA-related benefit (e.g., Kelly et al., 2011) and the use of a multidimensional measure of social dynamics in this study offers a unique perspective to document the nature of social support in the two 12-step modes. First, the rank order of the ten GES scale endorsements was moderately consistent between modes. In the two modes, Order, Cohesion, and Leadership support were highly and identically ranked 1, 2, and 3, respectively. Illustrative items from these scales are: Order, “The rules of the group are clearly understood by members”, Cohesion, “There is a feeling of unity and cohesion in this group”, and Leadership support, “The leader takes a personal interest in the members”. The order of the four lowest ranked endorsements on the GES scales was also the same for the two modes: Expressiveness (7), Leadership control (8), Innovation, (9), and Anger/Aggression (10). Here, a few items include: Expressiveness, “When members disagree with each other, they usually say so.”, and Anger/Aggression, “Some members are quite hostile to other members”. This preliminary evidence suggests that the nature of the social support “signal” may be relatively consistent between the two 12-step modes. Second, on some dimensions, the strength of the signal may differ between the two 12-step modes. Most pronounced, participants who attended both 12-step modes also reported significantly higher task orientation within meetings compared to people who only attended virtual 12-step meetings. Here, task orientation referred to the extent that group members agreed upon organizational objectives and pragmatic ways to achieve these objectives, here alcohol abstinence. Likewise, at a trend level (effect sizes range from .34 to .39), participants attending both modes reported experiencing group Cohesion, Order, and Leadership Support more than members attending only online meetings. If replicated, we encourage investigation of how, if at all, the experiencing of 12-step mode social dynamics changes with longevity in the 12-step program.

To our knowledge, the present study offers the first view of the relative effectiveness of two 12-step modes to reduce alcohol use. Findings suggest that attending both in-person and online 12-step meetings is associated with, on average, significantly higher rates of abstinence between baseline and the three-month interview, even after controlling for formal treatment episodes. The actual magnitude of this difference was considerable (effect size .75) with the combined 12-step mode group reporting 94% abstinent days whereas the online only 12-step mode group reported, on average, 81% alcohol-free days. This difference may be a function of higher involvement in 12-step practices and/or higher endorsement of the tasks and objectives of the 12-step program, but requires future investigation. No between-group differences were observed on measures of drinking intensity when drinking occurred or alcohol-related consequences, however. It appears, then, that neither 12-step mode offers different protective benefit if alcohol use occurs.

Our findings are mixed in replicating earlier MHG mode comparisons that aggregated participants from different MHG programs with longer periods of abstinence. The task orientation finding suggests that individuals accessing only online 12-step meetings during early 12-step affiliation may be more ambivalent/less receptive about the goal of alcohol abstinence relative to participants attending both online and in-person meetings. In contrast, Timko et al (2022) found no difference in abstinence goals between participants attending any online MHG meeting and participants attending just in-person MHG meetings, a finding that descriptively was replicated in a companion study when contrasting both MHG modes with only online MHG meetings, but excluding participants attending SMART (Timko et al., 2024). Next, with the in-person only MHG mode as the reference group, Timko et al. (2024) reported that participants who attended only online MHG meetings were significantly less involved in prescribed MHG program activities as depicted by a composite measure of involvement. Under the condition that a majority of the 7 comparisons in this study can be considered implying greater involvement, our findings are consistent with earlier work (Timko et al., 2022; Zemore et al., 2025). At the level of specific prescribed beliefs and practices, however, findings suggest the need for a more nuanced characterization of 12-step involvement when comparing 12-step modes. The question posed is, is the simple tally of prescribed 12-step behaviors sufficient to infer less or more involvement or, alternatively, should extent of involvement be defined by the level of effort required to do a prescribed behavior and the probability that such behavior predicts positive outcome? Accepting the latter perspective, and recalling that there was no between mode difference in acquiring a sponsor (which often involves intensive and time-consuming social interactions) we feel it may be premature to make the blanket statement that online 12-step attendance is associated with lower rates of involvement in the AA program.

Continuing, our findings indicated that attending both 12-step modes was associated with a higher proportion of abstinent days during early recovery, Contrasting in-person only with online only modes in a 12-month cohort study, however, Zemore et al., (2025) reported that the in-person mode was associated with significantly higher abstinence and significantly reduced heavy drinking and consequences, accounted for by lower MHG involvement among online participants. Reconciling this difference in findings is most likely attributable to sampling, design, and measurement issues discussed below. And, finally, studies have reported that being female (e.g., Timko et al. 2022), marital status, educational attainment, income level, and recovery goal (among SMART members) (Timko et al., 2024) was associated with selection of MHG mode. We did not observe any demographic mode profiles. Rather, we found that participants who had a recent treatment episode for substance use and/or emotional problems (90-days prior to baseline) were significantly more likely to report attending both 12-step meeting modes (83%) relative to participants only attending virtual 12-step meetings (58%). Because many treatment providers in the U.S. encourage clients to attend in-person 12-step meetings (Kelly, Yeterian, & Myers, 2008) this finding was not surprising. More interesting was the finding that 28% of the people who reported a treatment episode attended only online 12-step meetings after treatment and in the early stage of recovery (months 1–3). A plethora of explanations can be offered for this transition, each having plausibility: convenience, comorbidity leading in-person interactions to be uncomfortable, lack of transportation, use of virtual 12-step meetings in treatment, lack of in-person 12-step meetings in rural areas, to name only a few.

Differences in study design, sampling, and measurement may reconcile how our findings differ with previous investigations. Our study design sought to contrast the use of different 12-step modes during early recovery, and we therefore focused on 12-step behaviors months 1–3. Timko et al., (2022; 2024) and Zemore et al., (2025; 2026) cast a broader net in their reports and followed MHG attenders for 12-months. The absence of demographic predictors of 12-step mode in our study may simply reflect that individual preferences develop over time and are not manifest during early recovery. On the flip side, findings that baseline treatment experiences predicted 12-step mode use during initial efforts to reduce drinking may not be the case at 6 and 12-months. Sampling criteria were also quite different between studies. Ambivalence about change and the determination of the extent of change necessary (recovery goal) is most pronounced during early stages of reducing problematic drinking. Among individuals with stable and sustained recovery, however, recovery goal is more likely established, and 12-step mode selection may become more influenced by factors such as convenience and accessibility.

4.1. Limitations.

Several notable limitations of the present study should be mentioned when interpreting our findings. First, this study was not designed to test the relative effectiveness of different 12-step attendance modes. As such, these exploratory findings should be regarded as such and warrant replication in an independent sample. Second, given that we recruited participants from an online recovery website, it is highly likely that we oversampled individuals who attend online 12-step meetings, reflected in the fact that we only had 10 participants who reported exclusively attending in-person 12-step meetings. To more definitely compare outcomes based on the modality of 12-step meetings being attended, future research is needed that prioritizes the recruitment of individuals from both online and in-person sources. Related, study recruitment (March 2021) began during mandated social distancing although recruitment efforts continued for one year (March 2023) after most states had lifted such mandates (early 2022). In addition to amplifying concern about oversampling online participants, latent post pandemic misgivings about social interactions may have influenced participant responses, especially for the construct of AA meeting social dynamics. If true, participant responses about AA meeting social dynamics may have been attenuated for those attending in person meetings. And, finally, some subscales of the Group Environment Scale had poor internal consistency in our sample, which reduced our ability to detect differences in these domains even if they were present (e.g., Self-Discovery).

4.2. Conclusions.

More people go to 12-step programs than to formal treatment for alcohol problems in the U.S. and how people access community-based 12-step programs has shifted dramatically to online meeting attendance. Awaiting replication, our preliminary findings indicate that exclusive use of online 12-step attendance is associated with good but less favorable abstinence outcomes relative to attending both in-person and online 12-step meetings. Factors that may account for this differential benefit include lower rates of engagement in the number of prescribed 12-step activities and less agreement with the objectives and goals of 12-step programs. Several refinements were included in this investigation, and they are encouraged for future research on MHG. Foremost, we encourage within MHG investigations when assessing the relative value of different modes of attendance. Second, heterogeneity in MHG histories and longevity in recovery may obscure temporally important factors and processes. For this reason, we encourage narrowing and tailoring study eligibility criteria to best address MHG mode research questions of interest.

Highlights.

Alcoholics Anonymous (AA) meetings are popular among people trying to reduce drinking.

Accessing online AA meetings only is associated with less AA involvement.

Attending in-person and online AA meetings is related to higher abstinence rates.

Footnotes

Declaration of Interest Statement

The authors report no conflict of interest. Research reported in this publication was supported by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) of the National Institutes of Health (NIH) under Award Number R01AA027508 (PI: Tonigan). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Adelman-Mullally TA & Sullivan-Kerber C, (2021). Alcohol abstinence self-efficacy and recovery using Alcoholics Anonymous ® An integrative review of the literature, Journal of Psychosocial Nursing and Mental Health Services, 59(12):1–7, DOI: 10.3928/02793695-20210324-05 [DOI] [PubMed] [Google Scholar]
  2. Alcoholics Anonymous World Services Inc. (2024). A.A. 2022 Membership Survey. Alcoholics Anonymous. https://www.aa.org/alcoholics-anonymous-2022-membership-survey. [Google Scholar]
  3. Altman DG, & Bland JM (2005). Treatment allocation by minimisation. Bmj, 330(7495), 843. 10.1136/bmj.330.7495.843 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bergman BG, Greene MC, Zemore SE, & Kelly JF (2024). Prevalence and correlates of 12-step and second-wave mutual-help attendance in a nationally representative US sample. Alcohol: Clinical and Experimental Research, 48, 545–555. doi: 10.1111/acer.15268. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Chen LH, & Lee WC (2011). Two-way minimization: a novel treatment allocation method for small trials. PLoS One, 6(12), e28604. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Hassett-Walker C (2025). How individuals seeking help for substance use disorder adjusted to virtual 12-step meetings during the COVID-19 pandemic. INQUIRY: The Journal of Health Care Organization, Provision, and Financing, 62. 10.1177/00469580251320767 [1, 2]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Hatwan ML, Pearson MR, & Tonigan JS (in press). An experimental evaluation of assessment reactivity in ecological momentary assessment of Alcoholics Anonymous involvement and alcohol-related outcomes: One day at a time. Psychology of Addictive Behaviors. [DOI] [PubMed] [Google Scholar]
  8. Humphreys K, & Moos R (2001). Can encouraging substance abuse patients to participate in self-help groups reduce demand for health care? A quasi-experimental study. Alcoholism: Clinical and Experimental Research, 25(5), 711–716. 10.1111/j.1530-0277.2001.tb02271.x [DOI] [PubMed] [Google Scholar]
  9. Kelly JF, Green MC, & Bergman BG (2016). Recovery benefits of the “therapeutic alliance” among 12-step mutual-help organization attendees and their sponsors, Drug and Alcohol Dependence, 162, 1, p. 64–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Kelly JF, Humphreys K, & Ferri M (2020). Alcoholics anonymous and other 12-step facilitation programs for alcohol use disorder. Cochrane Database of Systematic Reviews, 3, 012880. 10.1002/14651858.CD012880.pub2.www.cochranelibrary.com [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Kelly JF, Stout RL, Magill M, & Tonigan JS (2011). The role of Alcoholics Anonymous in mobilizing adaptive social network changes: A prospective lagged mediational analysis. Drug and Alcohol Dependence, 114, 119–126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Kelly JF, Yeterian JD, & Myers MG (2008). Treatment staff referrals, participation expectations, and perceived benefits and barriers to adolescent involvement in twelve-step groups. Alcoholism Treatment Quarterly, 26(4), 427–449. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Miller WR (1996). Form 90: A structured assessment interview for drinking and related behaviors (NIAAA Project MATCH Monograph Series, Vol. 5, NIH Publication No. 96–4004). National Institute on Alcohol Abuse and Alcoholism, Rockville MD. [Google Scholar]
  14. Miller WR, Tonigan JS, & Longabaugh R (1995). The Drinker Inventory of Consequences (DrInC). National Institute on Alcohol Abuse and Alcoholism. [Google Scholar]
  15. Moos RH (2002). Group Environment Scale: Instrument (Form R, I, E) and Scoring Guide. Mind Garden, Inc. [Google Scholar]
  16. National Survey on Drug Use and Health (NSDUH), Substance Abuse and Mental Health Services Administration (SAMSHA), 2023. [Google Scholar]
  17. Pagano ME, Post SG, & Johnson SM (2010). Alcoholics Anonymous-related helping and the helper therapy principle. Alcoholism Treatment Quarterly, 29 (1), 23–34. doi: 10.1080/07347324.2011.538320 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Saghaei M, & Saghaei S (2011). Implementation of an open-source customizable minimization program for allocation of patients to parallel groups in clinical trials. Journal of Biomedical Science and Engineering, 4(11), 734–739. 10.4236/jbise.2011.411090 [DOI] [Google Scholar]
  19. Saunders JB, Aasland OG, Babor TF, De La Fuente JR, & Grant M (1993). Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption-II, Addiction, 88 (6) p 791–804. [DOI] [PubMed] [Google Scholar]
  20. Skinner HA, & Allen BA (1982). Alcohol dependence syndrome: Measurement and validation. Journal of Abnormal Psychology, 91(3), 199–209. 10.1037/0021-843X.91.3.199 [DOI] [PubMed] [Google Scholar]
  21. Sousa JL, Huskamp HA, Mehrotra A, Busch AB, & Uscher-Pines L (2025). Experiences with online mutual support groups for problematic drinking, Psychiatric Services, 76, 7, 10.1176/appi.ps.20240551. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Timko C, Mericle A, Kaskutas LA, Martinez P, & Zemore SE (2022). Predictors and outcomes of online mutual-help group attendance in a national survey study. Journal of Substance Abuse Treatment, 138, 108732. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Timko C, Mericle A, Vest N, Delk J, & Zemore SE (2024). Mode of mutual-help group attendance: Predictors and outcomes in a US national longitudinal survey of adults with lifetime alcohol use disorder. Journal of Substance Use and Addiction Treatment, 164, 209395. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Tonigan JS, & Rice SL (2010). Is it beneficial to have an Alcoholics Anonymous sponsor? Psychology of Addictive Behaviors, 24(3), 397–403. 10.1037/a0019013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Tonigan JS, Miller WR, & Brown JM (1997). The reliability of Form 90: An instrument for assessing alcohol treatment outcome. Journal of Studies on Alcohol, 58(4), 358–364. [DOI] [PubMed] [Google Scholar]
  26. Tonigan JS, Connors G, & Miller WR (1996). The Alcoholics Anonymous Involvement scale (AAI): Reliability and norms. Psychology of Addictive Behaviors, 10(2), 75–80. [Google Scholar]
  27. Tonigan JS, & Rice SL (2013). Is it beneficial to have an Alcoholics Anonymous sponsor? Psychology of Addictive Behaviors, 24(3), 397–403. 10.1037/a0019013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Tonigan JS, Toscova R, & Miller WR (1996). Meta-analysis of the literature on Alcoholics Anonymous: Sample and study characteristics moderate findings. Journal of Studies on Alcohol, 57(1), 65–72. 10.15288/jsa.1996.57.65 [DOI] [PubMed] [Google Scholar]
  29. Tonigan JS, Pearson MR, Magill M, & Hagler KJ (2018). AA attendance and abstinence for dually diagnosed patients: A meta-analytic review. Addiction, 113(11), 1970–1981. 10.1111/add.14268 [DOI] [PubMed] [Google Scholar]
  30. Wnuk M (2022). The role of optimism and abstinence in the mechanism underlying the indirect links of religious and spiritual involvement with the happiness of Alcoholics Anonymous from Poland. Religions, 13(9), 853. [Google Scholar]
  31. Ye T, Yi Y, & Shao J (2022). Inference on the average treatment effect under minimization and other covariate-Adaptive randomization methods. Biometrika, 109(1), 33–47. 10.1093/biomet/asab015 [DOI] [Google Scholar]
  32. Zemore SE, Lui CK, Mericle AA, Li L, Martinez P, & Timko C (2026). Second-wave mutual-help groups: Examining effectiveness for individuals with alcohol use disorders in the longitudinal, US national PAL Study cohorts. International Journal of Drug Policy, 104921. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Zemore SE, Li L, Lui CK, Timko C, Martinez P, & Mericle A, (2025). Effectiveness of online mutual-help group attendance for adults with alcohol use disorders in the pooled, longitudinal, U.S. National PAL Study cohorts. Drug and Alcohol Dependence, 277, 10.1016/j.drugalcdep.2025.112919. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Zemore SE, Lui C, Mericle A, Hemberg J, & Kaskutas LA (2018). A longitudinal study of the comparative efficacy of Women for Sobriety, LifeRing, SMART Recovery, and 12-step groups for those with AUD. Journal of Substance Abuse Treatment, 88, 18–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Zemore SE, & Kaskutas LA (2008). 12-step involvement and peer helping in day hospital and residential programs. Substance Use & Misuse, 43(12–13),1882–1903. [DOI] [PubMed] [Google Scholar]

RESOURCES