Letters Section Editor: Robert M. Golub,
MD, Senior Editor.
In Reply: The additional barriers identified
by Drs Beckman and Cook as influencing quality research in medical education
are important to keep in mind. Observational studies can and should provide
the basis for randomized clinical trials, something that has occurred extensively
in clinical research but not in medical education. These can help us understand
and control for such sources of variation. An unvalidated outcome measure
is a serious flaw in any study, and we agree that untested instruments are
often used in educational research, a practice that must change. We point
to the numerous validity studies on the United States Medical Licensing Examination1,2 to underscore that large-scale well-validated
national data do exist for practicing physicians in the United States, while
acknowledging that predictive validity is lower among subgroups of learners.3 We must develop and rigorously test instruments in
educational research so that we can address their strengths and limitations
in interpreting findings.
Carney PA, Pias CF, Nierenberg DW, Brooks WB, Stukel TA, Keller A. Educational Epidemiology—Reply. JAMA. 2004;292(24):2969-2971. doi:10.1001/jama.292.24.2970-b