Speaker: Prof Adrian Barnett
Title: Bad statistics in medical research, Queensland University of Technology

Abstract:

Using the wrong statistical methods can completely invalidate a study's results, potentially wasting years of hard work, misdirecting future research, and misinforming health policy. Despite the importance of statistics many medical researchers do not have adequate training and/or access to a qualified statistician. Medical researchers often attempt complex study designs that require advanced statistical knowledge. For example, a review of 606 published prediction models for COVID-19 found only 7 of the 606 models were potentially useful for practice, with many using statistical methods that were inappropriate and/or badly described. I will make the case for researchers to use simple study designs that only require basic statistical methods. I will discuss the potential for automated statistical reviews to improve research quality and the need for more training in research methods.


Bio of Professor Adrian Barnett:

Adrian is a professor of statistics who has worked for over 29 years in health and medical research. He was the president of the Statistical Society of Australia from 2018 to 2020. His current research concerns improving statistical practice to reduce research waste. He is the president of the Association for Interdisciplinary Meta-Research and Open Science, whose mission is to improve research quality.