Abstract
Using the decision rules and normal distribution assumptions of signal-detection theory as a base, a general strength theory of unidimensional absolute and comparative judgments is described in detail. The componenss of variance in both absolute and comparative judgments are considered, with particular emphasis on criterion variance in an absolute-judgment task and its relation to criterion variance in a comparative-judgment task. Some difficulties are noted in predicting comparative-judgment (forced-choice) probabilities from absolute-judgment (“yes-no”) probabilities. The principal difficulties are concerned with the relative magnitudes of criterion variance in the two tasks, the correlation of distributions, and attention. The question of the equality of variances for different criteria (e.g., yes-no vs confidence criteria) is considered, and two methods are suggested for answering the question (one of which is a new type of operating characteristic). The notion of a random variable being a function of a real variable or being a function of another random variable is used to analyze the effects of noise in an independent variable on the distribution of a dependent random variable.
Submitted Version (
Free)
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call