Misconduct in science is a timely and substantively important problem in the social study of science. But in the absence of comprehensive and reliable data needed for analysis, formidable obstacles stand in the way of its being studied quantitively. Accessible databases, including government data, are flawed, while undertaking new data collection presents its own problems. First, little is known about biases in official government reports. Second, official reports exclude classes of malfeasance other than fabrication, falsification, and plagiarism of evidence (FFP). Third, while drawing on official data is expedient, available official information is thin; it tells little about miscreants and fails to identify potential causes of their actions and the environments in which misconduct occurred. Fourth, it also fails the test of permitting estimates to be made of populations at risk, making it impossible to calculate incidence. A healthy dose of skepticism is in order in evaluating both the findings of current quantitative studies and of proposals for remediation.