On a method of multiple-presentation scaling of successive intervals

This thesis discusses a multiple-presentation scaling technique, by a somewhat modified successive interval approach. If a subject gives two or more different responses to the same stimulus in repeated presentations this may be considered as an "uncertainty" or lack of the subject's ability to discriminate between adjacent points on the rating scale. This "interval of uncertainty" is assumed to be a constant, and the scale points are estimated, on the basis of this assumption, by a least-squares technique.

Analysis and explicit computational procedures have been developed for the case of two, three, and four presentations. Numerical illustrations have been added for each case.

The thesis also includes a discussion on the combination of scales for different subjects if their intervals of uncertainty are different. Finally, it proposes methods of testing for bias between presentations, and a randomized procedure to correct for such bias if its presence is indicated by the test.