‘Irreproducible’ research is more often due to error than misconduct or fraud, but fighting it is vital.
I’m not so sure about that.
From the dean of the Harvard Medical School in the WSJ:
‘Irreproducible’ research is more often due to error than misconduct or fraud, but fighting it is vital.
I’m not so sure about that.
From the dean of the Harvard Medical School in the WSJ:
Reproducing “groundbreaking” studies doesn’t feed the bulldogs. Producing “groundbreaking” studies does! Scientists succumb very readily to the lure of being “first to print,” just like journalists. The only difference is that if I’m wrong with my “man bites dog” story, it gets cleared up in the next news cycle and nobody is very adversely affected. Be wrong often enough, and I lose credibility. However, if a scientist manages to get a study past peer review on “Human Inflicted Mandible Injuries On Canines and Their Relationship To Anthropogenic Climate Change” – he gains prestige, grants, and defenders whether others can reproduce his data or not.
Estimates of how much science can not be reproduced range from 50-90 %. Much of this is paid for by taxpayers. In contrast, industry publishes relatively little, but what they sell is often based on science and it works.
Editors and funding agencies have much to answer for.
Part of the problem is what the ‘bad scientists’ DON’T put into print – raw data, methods, sources, etc.