To the Editor.—
Recently there has been a dispute among experts concerning the measurement of diagnostic agreement1-3 and the use of the k statistic.4 We share the view of Uebersax5 that the real solution to this problem will require new paradigms for assessing interjudge agreement, and in particular that the more expensive multiple-judge designs or criteriondiagnosis designs are desirable. However, we do not accept that developing correction formulas to equate k values obtained from samples with different base rates is pointless.2,5 Under the conventional models for psychiatric diagnosis that underlie the calculation of k, it is possible to adjust k so as to incorporate base rate differences. This adjustment does not address the more complex issues raised,1,2,5 but may go some way toward reconciling divergent views2,3 about k.Following Spitznagel and Helzer,1 we define the sensitivity of a judge as the probability of
Stewart GW, Rey JM. A Partial Solution to the Base Rate Problem of the k Statistic. Arch Gen Psychiatry. 1988;45(5):504–505. doi:10.1001/archpsyc.1988.01800290126019
Artificial Intelligence Resource Center
Customize your JAMA Network experience by selecting one or more topics from the list below.