- Medical and health journals have a bias towards publishing findings which are statistically significant, even when they may not also be clinically relevant. This results in authors describing their non significant findings with creative language, to try and make them seem more interesting. Matthew Hankins (@mc_hankins) has Tweeted many, many examples of these verbal contortions, with his own wry comments. Storified by me @AnnieBruton.
- — Int J Public Health (@IJPH_official)Fri, Mar 01 2013 09:18:19@mc_hankins “Curiously, P values never seem to ‘trend’ away from significance.”) http://blog.amamanualofstyle.com/2012/06/18/bucking-the-trend-and-approaching-approaching-significance/ …fstyle.com/2012/06/18/bucking-the-trend-and-approaching-approaching-significance/
- — Matthew Hankins (@mc_hankins)Sat, Mar 02 2013 13:46:05Interestingly, 'slight non-significance' and 'slight significance' express exactly the same thing, i.e. non-significance #stilnotsignificant