Regular readers know that we here at Irreal have occasionally been hard on researchers in the social sciences. That’s especially true for Psychology where at least one study found that only 39% of published research results could be reproduced.
One wants to believe that things are better in the “hard” sciences—especially in medicine and health where erroneous results can affect us all. Sadly, these fields are not immune either. There are several well-known examples of this such as the fat versus carbs debate where the official orthodoxy of 50 years turned out to be wrong with disastrous results for the health of the population.
Mistakes do happen, of course, but a recent article from the American Council on Science and Health reports that 1 in 4 statisticians say they were asked to commit fraud by other researchers. The particular fraud that 1 in 4 statisticians were asked to commit was altering or removing date, a clear ethical violation that everyone would consider fraud. Other types of fraud were considered, ranging from the trivial such as not showings plots that failed to strongly support the desired result to the very serious such as falsifying data or misreporting \(p\) values. The article has a table summarizing the various frauds and their reported frequency.
These findings are discouraging to those of us who believe that the purpose of science is to seek the truth or even to those of us who want to stay healthy. My guess is that this fraud is mostly the results of the perverse rewards system used by universities and funding agencies. Goodhart’s law predicts that those operating under such a system will find a way to game it. Unfortunately, one such way appears to be fraud.