Salience as a Mediator: IV-based mediation for event-study natural experiments

Develop an instrumental-variables causal mediation design for event studies where the “treatment” is a shock (e.g., a terrorist attack or riot) and the mediator is public salience or threat perception. Leverage plausibly exogenous shifts in coverage (e.g., news cycle crowd-out from unrelated major events, weekend timing, or broadcast scheduling quirks) as instruments for mediator uptake. Estimate the indirect effect via salience and the residual direct effect using Hogan-Hennessy’s idea of using cost shifters for mediator take-up. This approach addresses the endogeneity of mediators like attention, fear, and identity activation in event studies, extending Hogan-Hennessy (2025) to high-frequency event-study contexts and media-salience instruments. It helps disentangle whether effects (e.g., on polarization) are due to perpetrator type or differential salience/threat amplification. The method generalizes beyond terrorism to other events like anti-refugee riots or policy announcements, informing interventions that target the salience channel rather than the events themselves. The potential impact is a reusable toolkit for mechanism identification in event studies across political behavior, public health communications, and crisis response, clarifying when observed effects stem from actors versus media propagation.

References:

1. Using Natural Experiments to Uncover Effects of Anti-Refugee Riots on Attitudes of Refugees. Nicole Schwitter, Ulf Liebe (2023). Sociology.
2. Terrorism, perpetrators and polarization: evidence from natural experiments. Vincenzo Bove, Riccardo Di Leo, Georgios Efthyvoulou, Harry Pickard (2024). Journal of Politics.
3. Causal Mediation in Natural Experiments. Senan Hogan-Hennessy (2025).