The theory of global games has shown that converting games with complete information to related games with incomplete information results in a unique equilibrium prediction that typically coincides with risk-dominance. This dissertation experimentally investigates this prediction in three different games: stag hunt, bargaining, and entry games. There are two treatments in each of these games, complete and incomplete information. In the stag hunt games, subjects under incomplete information conditions deviate significantly from the equilibrium prediction in favor of payoff dominance. They play similar strategies to those under complete information conditions. In the bargaining games most subjects conform to the risk-dominant prediction of global games theory, and convergence is stronger in games with incomplete information. In the entry games, in contrast to previous studies, subjects do not over-enter the market. This is because when too many people enter the market, firms' entry decisions become strategic substitutes, and subjects earn more by staying out of the market. There is less entry than the global games prediction. From these three games, I can conclude that subjects follow the comparative static predictions of global games theory, if not the precise predictions. Global games theory predictions are more powerful if there is no payoff dominance as an alternative prediction.
