Learning from An Experiment: Bayesian Perspectives on Estimation and Testing