Application of AUC Regression for the Jonckheere Trend Test

ABSTRACTA semiparametric regression model for the area under the ROC curve (AUC) is adapted to test of hypotheses for which the Jonckheere trend test (JTS) is appropriate. Since a nonparametric estimate of the AUC and the JTS depend upon the Mann-Whitney statistic, one can exploit this fact to develop a JTS that accounts for discrete covariates. The new method is illustrated with a simulation study and using real and simulated data motivated by three clinical studies.