As Swazey points out, the earlier study (J. Swazey, M. Anderson and K. Lewis Am. Sci. 81, 542–553; 1993) measured exposure to misconduct, rather than its incidence, when multiple respondents in the same department were reporting on the same case. The only circumstance in which exposure points to incidence is when there is only one observer per unit of observation. We therefore designed our study so that we sampled only one scientist per department.

Swazey's statement, however, implied that our results should be extrapolated only to departments. Her comment assumes that each observer is reporting all incidents in the department, rather than just those that he or she observed. This is unlikely even in a moderately sized department, let alone in a very large one. In a separate analysis (Gallup Organization Final Report: Observing and Reporting Suspected Misconduct in Biomedical Research Washington DC; 2006, table 8, see ORI.hhs.gov/gallup08), we showed that the incidence of reporting was not affected by department size, and so justified an extrapolation based on the number of research personnel supported by the National Institutes of Health.

See also: Integrity: Croatia's standards unusual in much of Europe Integrity: juniors see leaders gain from calculated dishonesty Integrity: misconduct by a few damages credibility for many Integrity: how to measure breaches effectively