MSPGH Seminar: 19th November 2014 – Statistical significance causing significant problems: Understanding Type S & Type M errors

12.30-1.30pm, Wed 19th November 2014
Seminar Room 515, Level 5, 207 Bouverie Street, Carlton

Professor John Carlin, Professor of Biostatistics, Melbourne School of Population & Global Health & Director, Clinical Epidemiology & Biostatistics Unit and Data Science,
Murdoch Childrens Research Institute

It has been suggested that research faces a “reproducibility crisis”, in that many claims that appear in research papers may be exaggerated or even false.  A not insignificant (shall we say) component of this problem is the heavy reliance in scientific publication on the concept of statistical significance.  A “significant” p-value provides a licence to make a scientific claim or report a “finding”, so it is not surprising that researchers feel a strong compulsion to find and report small p-values.  This presentation will review the dangers associated with over-reliance on statistical significance as the currency of research and discuss concepts such as the probability of a false positive claim, and Type S and Type M errors.

For more details, please view the Event Flyer.

RSVP not required. All other enquiries to Nora Li, Tel: 03 8344 9350.