When and How Do Message Matching Interventions Work? Exploring Principles to Guide the Use of Message Matching Through a Systematic Review and Meta-analysis, and an Experimental Study

Abstract

One of the most common techniques behavioral scientists use to influence people’s attitudes, intentions, and behaviors is message matching—the systematic design and distribution of persuasive messages such that their features (e.g., themes emphasized) are maximally congruent with the characteristics (e.g., motives) of their audience. Despite its popularity, the effectiveness of the technique varies greatly, and there are few established principles on how to best use it. My dissertation addresses this gap through a theoretical review and two empirical projects. The theoretical review summarizes four distinct literatures that use message matching (functional matching, message framing, message tailoring, and context matching), and proposes several principles. For example, I argue that messages vary along a continuum from positive matches (that are congruent with a person’s motivations) to negative matches (that are in opposition to people’s motivations), and that the success of interventions depends on both achieving positive matches and avoiding negative matches. I also discuss how targeting certain types of characteristics (e.g., motivational orientations) should lead to stronger effects than targeting other characteristics (e.g., demographics). Project 1 presents results from an ongoing registered systematic review (PROSPERO CRD42019116688; osf.io/rpjdg) that explores these principles. A three-level meta-analysis of 604 experimental functional matching studies (covering 4,228 effect size estimates) finds that matching messages to motivationally-relevant characteristics leads to effects around r = .20 on attitudes, intentions, and behaviors. The results demonstrate larger effects than works focused on matching to less motivationally focused characteristics (e.g., demographics), and provide evidence that avoiding negative matching while achieving positive matching leads to larger effects than achieving positive matching alone. These comparative inferences, however, depend on correlational differences between matching studies. Project 2 therefore complements the meta-analysis by presenting a registered (osf.io/yqmsd) experiment (N = 1,101) that directly evaluates the relative effects of positively and negatively matched messages (as these compare to neutral messages), in the context of promoting a non-profit organization by targeting people’s political orientation. The results show that both positive matches and negative matches impact persuasion, but that the detrimental effects of negative matches are greater than the beneficial effects of positive matches.