Business Experiments As Persuasion

Abstract

Much of the prior work on experimentation rests upon the assumption that entrepreneurs and managers use—or should optimally adopt—a "scientific approach" to test possible decisions before making them. This paper offers an alternative view of experimental strategy, introducing the possibility that at least some business experiments privilege persuasion over generating unbiased information. In this view, actors may craft experiments designed to gain support for their ideas, even if doing so reduces the informativeness of the experiment. However, decision-makers are not naïve—they are aware that the results they are reviewing may be the product of a curated information environment. Using a formal model, this paper shows that under a wide range of conditions, actors prefer to enact a less than fully informative experiment designed to persuade—even when a fully-informative experiment is feasible at the same cost.

Keywords: Experimentation, Decision-Making, Entrepreneurship, Strategy, Bayesian Persuasion

JEL Classification: L26, M13, O31