This paper is from Session 8B: Inference in Times of Crisis, Part 2
which comes under Topic 8: New approaches to research in statistics education
Paper 8B1 (Tuesday 10th, 14:00-15:30)
- Fiona Fidler (University of Melbourne, Australia)
Students (and researchers) often overestimate the information p-values provide about future replications. Over the last 5 years, large scale reproducibility projects in psychology, biomedicine and other disciplines have exposed just how wrong these interpretations can be. Some have responded with initiatives to teach replication more explicitly and further, to teach statistics through direct replication studies. I will give examples of such programs from psychology and ecology. Benefits include connecting students with real world research (rather than contrived examples), providing context for the comparison of different statistical approaches, and introducing meta-analysis and other data synthesis techniques in an integrated way. They also meaningfully contribute to the progress of science by building an open replication data bank.