My recent chat with John Batchelor broached the important subject of flawed, non-reproducible scientific studies that find their way into journals. It seems that replicating results is as rare as finding a unicorn at a science fair.
p-hacking
In Part 1 of this series, we presented evidence that a not insignificant amount of published science and the “knowledge” resulting from it is wrong. There are many reasons, including experimental carelessness, differences in how researchers approach statistical analyses, journals’ bias against publishing negative results, and outright dishonesty by investigators. Part 2 describes how activist scientists can use meta-analyses to manipulate data to reach a predetermined, but flawed, conclusion.
Much published science and the "knowledge" resulting from it is likely wrong and sends researchers chasing false leads. Without research integrity, we don’t know what we know, so it is incumbent on the scientific community to find solutions.
Two studies look at how you can use words to spin non-significant findings into published studies, and how falsified data spreads unchecked from one meta-analysis to another.
The New York Times recently swallowed whole a study which concluded that those who eat meat die 23% more quickly than those who don't. But the meat study sounded fishy. And it was. ACSH advisor and expert biostatistician Dr. Stan Young turns the meat study into hamburger.