Using Models to Persuade

This article develops a framework for “model persuasion,” in which persuaders influence beliefs by proposing models that organize existing data and shape predictions. The authors show that persuaders face a tradeoff between model fit and persuasive impact: models that fit data well often shift beliefs less. The framework explains how misleading interpretations can arise even when audiences observe the true model, particularly when data are noisy or open to interpretation. The paper applies these ideas to finance, business, politics, and law, highlighting how competing narratives influence decision-making