Exploring published research results without the data by Dr Nick Brown

About the speaker

Dr Nick Brown was recently awarded his PhD at the University of Groningen, Netherlands for his work investigating social claims from positive psychology. He is, however, better known as a ‘data thug’, detecting anomalies in reported statistics along with Dr James Heathers (check out James’ awesome podcast, Everything Hertz). Drs Brown and Heathers won the SIPS Mission Award for their work on SPRITE, GRIM, and their application in error detection in published research. Dr Brown also keeps a regular blog, where he writes about his exploits in error detection and insightful commentaries on statistical analyses and their pitfalls.

About the talk

A goal of the open science movement is for almost every published empirical articles to be accompanied by the relevant data set and analysis code. However, that outcome is not likely to be achieved in the immediate future (and it does not address the issue of data availability for older work). In the meantime, researchers who wish to critically examine the numerical and statistical results of published articles often have difficulty obtaining the data on which the results are based. The original authors might fail to answer the sharing request or even explicitly refuse to share their data. Perhaps the data no longer exist at all (a fate that will, of course, befall the authors too one day). In this talk, Dr Brown will describe a number of techniques that allow the curious reader to identify possible anomalies in reported statistics without having access to the data, and show how these techniques have been applied to reveal some problems in influential research. Only a minimal level of statistical knowledge is assumed.