Session Description
Media and cultural narratives — safety campaigns, cultural diplomacy — are increasingly deployed as deliberate interventions meant to shift societal attitudes and behavior. Yet unlike infrastructure projects that undergo formal impact assessment, there is no shared method for attributing observed societal change to a specific narrative. The challenge is one IA already knows well: changing baselines, cumulative effects, scarce counterfactuals — applied to content rather than construction.
This workshop tackles one defined problem: how can practitioners design a longitudinal or panel-based monitoring framework that credibly attributes change to narrative exposure, given attrition correlated with exposure, self-selection, saturated multi-narrative environments, delayed effects, and an unstable "narrative" construct?
Two short framings (15 min each) set up the problem: methodological failure points in panel and longitudinal designs, and an honest look at what AI tools — narrative classification, synthetic control, causal ML — genuinely solve versus where they manufacture false confidence.
The rest is facilitated small-group problem-solving: participants bring their own attribution challenges and work them against a shared diagnostic framework, rotating between tables. The session closes with a synthesis aimed at a shared, practical checklist for monitoring narrative interventions going forward.
We welcome participants from social impact assessment, M&E, communication research, and computational social science.
Session Summary
A working session on attributing societal change to media and cultural narratives through longitudinal and panel methods. Two short framings — methodological pitfalls and the real (versus overstated) role of AI — followed by hands-on group work to draft a practical attribution checklist for narrative interventions
Session Chair
Yevgeniy Rogov
inDrive
Kazakhstan
Session Co-chair
Maria Lokteva
inDrive
Kazakhstan