Integration theories and self-rating : the application of cognitive algebra to schema theory

摘要

An experiment was conducted to investigate theudappropriateness of Anderson's (1981) averaging integration modeludto self-rating. This model would enable simultaneous estimationudof the extremity (or position) and importance (or weight) ofudthose self-schemata involved in a given rating task (Markus,ud1977). It was hypothesised that the averaging model ofudintegration could be described geometrically as predictive of aud"city-block" metric, such that estimates of proximityud(self-descriptiveness of adjective-qualifier pairs) consist ofudsummed estimates of the proximities of each component of theudstimulus item to a constant comparitor point. Geometrically, thisudmodel therefore predicts that movement through the space in whichudrating takes place is horizontal and vertical, but neveruddiagonal. This model is contrasted with a euclidean metric inudwhich ratings are predicted from the square root of the summedudsquared proximities.udStimulus items consisted of pairs of adjectives, eachudqualified by one of six adverbs.There were two Adjective Typeudconditions (Abstract, Concrete) and three Instructional Sets,udyielding a 6 (qualifier of first adjective) X 6 (qualifier ofudsecond adjective) X 2 (adjective type) within-subject factorialuddesign with Instructional Set as a between groups factor.udData were visual-analog ratings from 0 (not at all like me) ududto 100 (exactly like me). These were subject to a range of modeludfitting procedures intended to 'identify convergence with theudaveraging and euclidean models, including Median Polish (Tukey,ud1975), IMSL iteration procedures, and residuals ANOVA (Anderson,ud1982).udResults indicate that neither the averaging nor the euclideanudmodels satifactorally predicted the ratings obtained, as the datauddeparts significantly from each. Of the two models, however, theudeuclidean model shows superior fit to the data.udIt is suggested that the results may be consistent with auddifferential-weight averaging model (Anderson, 1982), althoughudevidence to his conclusion is secondary.