Fig. 2

Receiver operating characteristic (ROC) curve for the statistical test used in our example. The abscissa in an ROC curve represents the false-positive rate (1 − Specificity) of a diagnostic test; here, for statistical tests, it should be the p significance threshold (α). The ordinate in an ROC curve represents the true-positive rate (Sensitivity) of a diagnostic test; here, for statistical tests, it should be the study power (1 − β). The diagonal dashed gray line is the ROC of uninformative test. The gray point on the curve corresponds to the most appropriate p significance threshold values. The slope of the dashed red line, the tangent line to the curve at the most appropriate p cut-off, is equal to the likelihood ratio at the most appropriate p significance threshold. AUC is the area under the ROC curve