The Effect on Test Reliability and Validity of Scoring Aptitude and Achievement Tests With Weights for Every Choice

swer. Such a distribution of responses tends to maximize item reliability (8). To these item writers, scoring procedures seem unduly crude if they do not permit quantitative differentiation among examinees who, for a given item, choose a distracter that is nearly correct and those who choose one that is indicative of ignorance or even gross misinformation about the point tested. Variation among examinees with respect to the merit of the distracters they mark as correct is completely eliminated from test scores when the same -1 credit (either zero or -, as in conventional scoring procedures) k-1 .