Open Module - Towards more credible research: what's wrong and how can we fix it?
Dorothy V. M. Bishop, Professor of Developmental Neuropsychology, University of Oxford
Tuesday, 28 January 2020, 5.15pm to 6.15pm
Hosted by Dr Reena Vohora
Many psychologists were trained in practices that are now recognised as problematic, because they generate non-reproducible results. Particular problems arise if there is no clear distinction made between hypothesis generation and hypothesis testing, so the same data are used to formulate and test a hypothesis. Furthermore, it has not been customary to check statistical power before doing a study, and many resources are wasted in performing studies that give inconclusive or erroneous results. In addition, confirmation bias means we disregard evidence that does not fit a preconceived idea, leading to canonization of false facts. I will discuss a range of solutions, including use of simulated data in statistical training, collaborative working, open data, pre-registration of studies and changes to the incentive structure for scientists.
Spaces are limited, please book your place by emailing Angela Fox (email@example.com).