Profiling Players in Dynamic Games: A Mobile Experiment

This article has two distinct objectives. The first objective is to explore what mobile technology has to offer for experimental research. To that end, a team of app developers was recruited to create Blues and Reds, a mobile app (Android/iOS) that operates as an experiment designed to study behavior in dynamic games. In less than six months, the app generated a database consisting of 35,826 observations from 6,463 subjects located in 141 countries. The second objective is to design and test a novel method of profiling players in dynamic games using data from Blues and Reds that satisfies both descriptive and predictive requirements. To describe the reasoning process in dynamic games, a profile consists of two dimensions. The first dimension depicts the quality of the subject’s reasoning (savvy vs naive); the second dimension captures the speed of that reasoning (fast vs slow). Predictive power requires that the subject’s profile indicates her likelihood to behave consistently with backward induction. The proposed hypothesis is that this probability is the highest for savvy-fast profile, followed by savvy-slow, then naive-fast, and naive-slow at the end. This hypothesis is successfully confirmed and replicated in 22 various dynamic games.