|
- on the ABC Counterpoint
by the Greek epidemiologist who wrote "Why Most Published Research Findings are False" also excellent
John Ioannidis - Winner's Curse
http://www.abc.net.au/rn/counterpoint/stories/2008/2420024.htm
for audio and full transcript
1- The most dramatic paper will get published more easily and then become more cited most will turn out to be proved wrong afterwards. 5-out of 6 for randomised 1/4 for randomised trials.
2- the follow the leader effect : In some circumstances the follow up tests which should independently verify instead suffer from herding , where people try to get the same results in order to access research funding available to dramatic studies and don't want to contradict
3- the Oligopoly Effect - a minority of journals carry a huge weight - such that one report in them would seem to outwiegh 5 reports in others
4- Rush to contradict - opposite of 2 - funding might be available if can find opposite extreme
etc etc
5- examples include no negative publications for antidepressants giving the false impression there are no negative studies
6- improvements - we at least need to be aware of the biases
- Applying his own rules to himself : you can’t more hyperbolic than saying something like Most Published science is wrong... isn't he doing exactly what he accuses other people of ..coming up with extreme alarmist evidence ?
In life we don't get the truth news is about the irregular not the regular .. we don't get offered fruit cos other things are more profitable.
|