UC Berkeley

Academic contributions in any discipline are only as convincing as the methods used to establish them. This dissertation highlights two methodological issues in economics---one in experimental economics and one in applied econometrics---and argues for increased caution in both the design and the interpretation of empirical studies. Within experimental economics the Becker-DeGroot-Marschak (BDM) mechanism is widely used to elicit the valuations of experimental subjects. Although it is theoretically incentive compatible, empirical evidence suggests that elicitations are affected by the distribution from which the random price is drawn. The second chapter presents a novel, within-subjects data of sequential BDM rounds with varied distributions to directly investigate and characterize distributional dependence. When analyzing data collected outside of the realm of randomized experiments (in the laboratory or otherwise), fixed effects are frequently used to ``control for'' the potential influence of observed factors on an outcome variable of interest. The third chapter discusses potential pitfalls in the use and interpretation of fixed effects. The goal of each of these chapters is to offer positive suggestions for more careful future research through marginal improvements in empirical design and practice.