Comparisons within randomised groups can be very misleading
| Dates: | 15 November 2011 |
|---|
| Times: | 14:00 - 15:00 |
|---|
| What is it: | Seminar |
|---|
|
|
Health Sciences (Primary Care), School of Community Based Medicine present Comparisons within randomised groups can be very misleading
Speaker: Professor Martin Bland
Further details:
'Comparisons within randomised groups can be very misleading' J Martin Bland, Douglas G Altman Department of Health Sciences University of York Heslington York YO10 5DD J Martin Bland, Professor of Health Statistics, email: martin.bland@york.ac.uk
Centre for Statistics in Medicine
University of Oxford
Wolfson College Annexe
Linton Road
Oxford OX2 6UD
Douglas G Altman, professor of statistics in medicine
Rather than comparing the randomised groups in a clinical trial directly, researchers sometimes look at the change in the measurement between baseline and the end of the trial; they test whether there was a significant change from baseline, separately in each randomised group. They report that this difference is significant in one group but not in the other, and conclude that this is evidence that the groups, and hence the treatments, are different. Several examples will be given, including a recent trial which received wide publicity, in which participants were randomised to receive either an “anti-ageing” cream or a placebo. We will show by simulation and theoretically that this approach is fundamentally flawed and capable of giving alpha errors as high as 50%.
(Access to building via Oxford Road, lift to 4th floor, follow signs to 5th flr.)
Travel and Contact Information
Find event
Seminar Room 2, 5th floor
Williamson Building
Manchester
