New analysis in PNAS is used to support several new conclusions:
My faith in the rigor of the authors’ analysis is shaken by their use of 3D, angled pie graphs (Fig. 2)– but I’ll try not to throw the baby out with the bathwater. Speaking of that Fig. 2, it might be interesting to divide by the total scientific output of each country, to see if the proportion of fraudulent papers varies significantly by country.
Also, apparently retractions don’t always staunch the flow of citations…
Interestingly, according to their data, most fraudulent papers from about 1985-2000 were from serial offenders– authors with 5 or more fraudulent papers in total. Even today, these serial offenders account for a lot of the fraud.