Replication Data and R Code for: Mixed-methods systematic review and multilevel meta-analysis reassessing AI conversational agents for mental health across architectures

This repository contains the replication dataset, reproducible R code, documentation, and verified Figure 3 outputs for the study Mixed-methods systematic review and multilevel meta-analysis reassessing AI conversational agents for mental health across architectures (PROSPERO CRD420261285571). Versions 1.1.0–1.1.2 harmonised outcome classifications, distinguished unreliable from non-poolable strata, corrected the Bennion et al. first-dissemination year, clarified the outcome-specific moderator families, and revised Figure 3 directional annotations. Version 1.1.3 records age as unavailable for the two Chen et al. (2025) effect-size rows because the primary report states that demographic data were not collected for privacy reasons. The previously inferred younger-sample category was removed, and the short-term distress age moderator and its 16-test BH-FDR family were rerun. The age omnibus result is praw = 0.744 and pFDR = 0.851; the other 15 tests were unchanged. Engagement analyses estimate aggregate study-level associations and not individual-level dose-response relations.