Does Statistical Expertise Matter?
Friday 30 August, 3pm
We are very please to welcome Prof Adrian Barnett, Professor of Statistics at Queensland University of Technology, to present our next ISPOR NZ webinar: Does Statistical Expertise Matter?
The webinar will be held via Zoom at 3pm on Friday 30 August. To register, please complete the form on this page.
About the speaker
Prof Adrian Barnett is a Professor of Statistics at Queensland University of Technology. Adrian researches the research process and is interested in how health and medical research can be improved to give a better return on investment and reduce research waste.
Abstract
Statistics is a challenging subject that requires training and expertise to do competently. There are common pitfalls and misconceptions that are frequently repeated in published papers. Many journals struggle to find qualified statistical reviews and so bad practice has taken a foothold in some fields, even becoming the new normal. Software is available to help researchers do their own statistics, but no software warns researchers when they use the wrong method or when their research question is flawed. Most researchers publish results without stress-testing their models or checking important assumptions.
Statistics as a field is being marginalised by a dwindling critical mass and the growth of data science and machine learning - even though these "new" fields rely heavily on statistics. The lack of qualified statistical help means researchers turn to other fields or the internet for advice, with potentially harmful consequences. The field of sports medicine even invented a totally new way of doing statistics that did not involve statisticians, had little theoretical basis, and has created many false positive findings. Many Australian health and medical ethical committees do not have access to statisticians, but some committees are unworried by this as they can call on epidemiologists, psychologists or clinicians with "a working knowledge" of statistics.
I will examine some of the consequences of the waning influence of statistics in research and discuss some potential interventions to raise the profile of statistics. I will discuss some parallels between statistical and economics expertise.
Comments