Quantifying performance of a diagnostic test using the sampling distribution of the weight of evidence favouring case over noncase status