Thursday, August 6, 2015

Detecting Fraud in Academic Research

The scientific community is often considered to be less filled with fraud than other disciplines. Typically when we think of fraud, we think of it occurring in businesses and think of someone walking away with a lot of money. However, fraud can also easily happen in academic research (see these previous posts for examples) where someone walks away with a lot of published papers. Discover Magazine recently posted an article titled The Perfect Scientific Crime. The article outlines several ways that someone could commit fraud without being detected by creating and using fictitious data. The purpose of the article, however, is not to help fraudsters, but rather to point out a few things that others should be aware of to help deter fraud in the academic community.
If the data appear to be reasonable, it can be nearly impossible to know if someone has fabricated data without comparing their summarized results to the raw data. Because of this difficulty, and assuming the raw data are not published, co-authors are the ones that have the best chance of detecting fabricated data in papers. Additionally, an alternative solution that the article recommends is to require all authors to publish the raw data along with their papers, so that data fraud will be more difficult to commit. As Uri Simonsohn from the University of Pennsylvania said about what to do with data, "Just Post It."

1 comment: