March 5th, 2003, 02:31
|
Administrator
|
|
Join Date: 2001
Posts: 1,085
|
|
"Statistically Not Significant" Does Not Equal "No Difference"
This review of the Cochrane Database found that approximately 20% of reviews concluded that "there was no difference" or "no effect" of the research intervention. Thesee authors conclude that this demonstrates a poor understanding of statistics. Even if the p value is > 0.05, this means that a difference was not identified. It does not mean that there is no difference. Comment: this is a somewhat confusing article in the way it described the statistics involved. A clearer explaination was given 8 years earlier by Altman and Bland, two authors that have published in the BMJ several important articles on the use of statistics in medical research. [ bmj.com Alderson and Chalmers 326 (7387): 475 ]
|