According to a survey conducted by the Gallup Organization, a substantial amount of misconduct in government-funded scientific research is going unreported. The survey has been submitted for journal publication, and is not yet available online, but Bloomberg reports:
The data was based on surveys of 2,212 scientists holding research funding from the National Institutes of Health, asking if they believed they had observed fraud in 2002-2005. The scientists surveyed in the report said they observed 201 instances of “likely misconduct” over a three-year period, or about three cases per 100 people per year.
Again, this behooves reporters to respect the weight of evidence when it comes to reporting new research claims which appear novel or shocking.