The reproducibility crisis in health and medical research
During the October's Gold Coast Skeptics in the Pub, Professor Adrian Barnett from School of Public Health and Social Work at QUT talks about how rewarding publication with "sexy" (read statistically significant) results over good science and the truth is costing the public both money and harm.
It's clear from this talk that rewarding positive results and "new" and "innovative" findings only has lead to high levels of wastage in research. This is not to say that new and innovative isn't important, of course it is, but that the quality of the research methods should be rewarded, rather than the findings.
Good quality research actually contribute to public knowledge, whereas low quality studies are not only next to impossible to replicate but contribute to a significant waste of public funds - this includes research with negative or "un-sexy" results that never gets shared/ published.
While this is difficult to solve as members of the public, we can ask a few questions when participating in studies:
Is there a control/ comparison group? Single-armed studies (read pre-post) tend not to add any robust information. This is most relevant for intervention studies. Realistically only evaluating the effect of the new intervention won't tell you much about it's effectiveness. They may claim "things got better" - yes but we should ask, better compared to what and would things have been the same if we did nothing?
Ask to see the results/ publications. This will allow you to see the fruits of your contribution to research and ensure that it was not for nothing.
Listen to Adrian's talk: click here
Slides: click here
About the speaker
Professor Adrian Barnett
Professor Barnett has undergraduate and postgraduate degrees in statistics. He has worked for over 21 years as a statistician, working for drug companies, research councils and universities. He is the current president of the Statistical Society of Australia. He has written popular articles on statistics, including on regression to the mean and predicting football matches.
His current research concerns how statistics are often badly use in research and what can be done to reduce this research waste.
Contact email: a.barnett@qut.edu.au
Twitter: @aidybarnett
