es.ROC {pwrFDR}

Computes the equivalent Z-test effect size in hypothesis tests involving the ROC. Included here with the intent that it can be used in conjunction with pwrFDR to allow power/sample size calculation for multiple tests of ROC curve based hypothesis. See details.

Description

In hypothesis tests of TPR_1 vs TPR_0 at fixed FPR, or FPR_1 vs FPR_0 at fixed TPR, this computes the equivalent Z-test effect size. This can then be passed the effect.size argument in a call to pwrFDR or controlFDP

Usage

  es.ROC(FPR0, FPR1 = NULL, TPR0, TPR1 = NULL, b = NULL)

Arguments

FPR0

When the TPR is fixed, the FPR under the null. Otherwise the fixed FPR.

FPR1

When the TPR is fixed, the FPR under the alternative. Otherwise left blank.

TPR0

When the FPR is fixed, the TPR under the null. Otherwise the fixed TPR.

TPR1

When the FPR is fixed, the TPR under the alternative. Otherwise left blank.

b

Nominal slope of the ROC at FPR0. Taken to be 1 by default.

Value

The equivalent Z-test effect size in a hypothesis test for difference in TPR at fixed FPR or difference in FPR at fixed TPR

Author(s)

Grant Izmirlian <izmirlian at nih dot gov>

References

Pepe M. S., Feng Z, Janes, H Bossuyt P. M. and Potter J. D. Pivotal evaluation of the accuracy of a biomarker used for classification or prediction. Supplement. J Natl Cancer Inst 2008;100: 1432–1438

See Also

cc.ROC

Examples

    es.ROC(FPR0=0.15, TPR0=0.80, TPR1=0.90)