Recommended reading
To be, or not to be…
###By Steffen Dettling
As you know, a statistically significant result does not necessarily describe a biologically meaningful effect. The way we are currently using the formal term “statistically significant” has lead to many misinterpretations, implying that a lot of published findings might be incorrect. As UCL pharmacologist & statistician, David Colquhoun pointed out “If you use p-value=0.05 to suggest that you have made a discovery, you will be wrong at least 30 percent of the time”. A reason why some fields of cancer science have been stuck for years?
Maybe a good time to re-read John Ioannidis’s paper on the subject (which, by the way, has become one of the most widely cited papers ever published in PLoS Medicine). And if his paper is too technical for you - you can also go ahead and read this pertinent comic strip…
####Links:
YAAC · WEBSITE
post format