In this talk, Vandekerckhove will describe and demonstrate a design strategy useful for replicating empirical effects in psychological science. The strategy involves the indiscriminate randomization of independent experimental variables that may be moderators of a to-be replicated empirical finding, and can be used to test the robustness of an empirical claim to some of the vagaries and idiosyncrasies of experimental protocols. By the same theory of generalizability that licenses inference from a sample of participants to a population, he proposes to generalize from a sample of experiments to a well-defined space of possible experiments. The strategy is made feasible by automation of experiments and by advances in Bayesian inference that allow for the pooling of information across experiments and designs. He will demonstrate the practical feasibility of the strategy with a replication of a study on subliminal priming.

 

© UC Irvine School of Social Sciences - 3151 Social Sciences Plaza, Irvine, CA 92697-5100 - 949.824.2766