A METHODOLOGICAL STUDY OF PREFERENCE.
ILLINOIS UNIV AT URBANA DEPT OF PSYCHOLOGY
Pagination or Media Count:
Traditionally measurement of human preferential choice behavior has been predicated upon two rather strong assumptions 1 lack of individual differences in preferential choice and 2 use of a linear unidimensional model of preferentail choice. In this study the appropriateness of these assumptions and their effect upon results of data analyses based upon them are experimentally evaluated. Subjects were asked to indicate their preference for stimuli in three choice experiments one involving rather simple geometric stimuli Color-Form Rectangles, one involving more complex geometric stimuli Hetrosexual Somatic Preference, and one involving rather abstract stimuli Preference for Famous People. These data were analyzed under both the traditional set of assumptions and several less restrictive sets which allowed for individual differences in preference or for a multidimensional model of preferential choice or both. In all cases it was found that rather profound differences among individuals exist with respect to their preferential choices and that multidimensional models of preferential choice do no worse, and usually better, than the corresponding unidimensional models. Further it was shown that already existing techniques such as the Tucker Cluster Analysis Procedure and non-metric multidimensional scaling techniques can be easily employed to analyze choice behavior data with little modification. Author