The reconstructability of persuasive message variables affects the variability of experimental effect sizes: evidence and implications

Abstract

Whereas the persuasive impact of message variables such as weaker versus stronger threat appeals, vivid versus pallid messages, and one-sided versus two-sided messages has received much research attention, more abstract properties of such message variables have gone largely unexamined. This article reports an analysis of one such property, reconstructability: the degree to which one of the two messages in an experimental pair can be deduced from the other. Evidence is offered from research on persuasive communication that as message variables become less reconstructable, the variability of the associated effect sizes increases—which creates distinctive challenges for theoretical progress and practical message design. Attention to message-variable properties such as reconstructability promises to shed light on how and why effects differ across message variables.