Equal discriminability scaling by information measures

Includes bibliographical references (page 34)A new method of scaling stimuli for equal discriminability is described and tested against the method of Garner and Hake. The new method uses cumulated transmitted information to form a scale. Garner and Rake's method uses successive intervals scaling. The original set of unscaled stimuli used in the study is unusual in that subjects judging this set of stimuli achieved the greatest information transmission and percent correct responses in the middle categories based on data from this original unscaled set of stimuli, two new stimulus sets were constructed, one scaled according to the equal discriminability method of Garner and Hake and the other scaled according to the cumulated transmitted information method. Subjects then judged the two new stimulus sets. Based on data from these judgments, the two scaling techniques were evaluated by three criteria: the degree to which the standard deviation of information transmitted by each stimulus for a subject was reduced, the degree to which the average deviation of information transmitted by each stimulus for a subject was reduced, and the degree to which overall information for a subject was increased. The stimulus set constructed according to Garner and Hake???s method gave significantly higher information transmission when compared to both of the other two stimulus sets. Other comparisons were not significant, leading to the conclusion that for the situation tested Garner and Hake's method is preferable to the cumulated information transmission method