| dc.description.abstract | This dissertation includes three experimental studies discussing the role of strategic beliefs in understanding strategic, pro-social and socially complementary behavior.
The first essay, “Testing Stepwise Reasoning”, introduces a test of the underlying assumption in the Cognitive Hierarchy model of strategic behavior: an individual anticipates facing a distribution of opponents that differ in terms of their strategic sophistication. In phase one of the experiment, subjects play games. Different participants in phase two then predict their behavior. We find that a subject often possesses beliefs consistent with the Cognitive Hierarchy model. We conclude that Cognitive Hierarchy is more than “as if” model of behavior – it illustrates, how agents actually think in a strategic environment.
The second essay, “Measuring Trust: A Reinvestigation”, investigates two ways of measuring trust: a survey and an incentivized game. Prior literature established that the two measures are not correlated. However, a common study in this body of research employed a modified version of the original trust game, thus, potentially changing the motives and strategic beliefs that drive trusting behavior. We conduct a replication and a reinvestigation of this study and find that the two measures are correlated, when the original trust game is used. We suggest that trust is a single construct measured by both the survey and the incentivized game.
The final essay, “Norm Misperceptions and Social Network Structure” presents a test of a potential mechanism behind social norm misperceptions. In a laboratory, we exogenously manipulate the structure of the imposed social networks. We observe a positive bias in the socially complementary consumption and perceptions about the average consumption of others in the treatment condition, where the correlation between the induced preferences and the number of individual network links is positive. However, a positive bias also exists in the control condition, where the number of social links is fixed for all subjects. We discuss potential sources of the bias in the control condition and suggest a direction for future work. An improved design may act as a platform for testing methods to battle social norm misperceptions and their negative consequences. | en |
