If you have not read the book “Doctored” by Charles Piller (here), you may want to have a look at the last few paragraphs of the Alzforum post (here).  For whatever reason (if someone could explain to us, we would be very happy), most of such publicly known cases of questionable research originate from the neurodegeneration field. 

It may well be a wrong impression based on a biased and incomplete sample. For example, most of the investigations revealed questionable practices by turning the magnifying glass on western blots, immunohistochemistry images, and other types of visual information reported in scientific publications. However, only a fraction of publications includes this type of data sets. There are by far more ‘quantitative’ results reported in form of tables and figures (bar graphs, line curves, etc.). In other words, in most publications, there is nothing to put under the lens and to scrutinize in detail.

If journals would start requiring disclosures of quantitative values for individual data points (yes, this is in the spirit of Open Science!), ‘data detectives’ would have more room to operate and can use a range of statistical methods to detect data and result anomalies. Have you heard about Benford’s Law? If not, here is a nice introduction. 

More broadly, methods used by data detectives were summarized here by Gabriel Crone and Christopher Green.  We agree with these authors that, “… while data fabrication is a rare phenomenon, estimates suggest that it occurs frequently enough to be a concern.”