What is an informative replication? Duygu Uygun-Tunç
What is an informative replication?
Duygu Uygun Tunç Philosophy and Ethics, Eindhoven University of Technology
Replication is regarded as a benchmark of scientific rigor, but there is vast disagreement as to what makes a study a replication of another and what function replications serve. In this talk I will propose a general account of replication from a novel theoretical perspective on the severity of a test. Assessing the severity of a test involves assessing whether and to what extent a test is a genuine test of a hypothesis. Given that we must make various auxiliary assumptions in formulating and testing hypotheses, one of the challenges in hypothesis testing is to determine how to allocate the impact of negative results between the main hypothesis and the various auxiliary assumptions that are made in devising its test. This problem applies directly to the current dissensus about the success conditions of replications, because the interpretation of replication failure is similarly underdetermined. From the perspective I propose, an informative replication is a test of certain alternative explanations of the findings that are associated with certain auxiliary assumptions of the original test. A replication aims to reveal if the corroboration or refutation of a hypothesis is conditional on certain auxiliary assumptions, thereby to increase severity. We can maximize the informativeness of replications by identifying which particular kind of auxiliary assumptions they can test.
Commentator: Sarah Volz, Psychology, University of Minnesota