Downloads & Free Reading Options - Results

This study assessed the construct validity of a cognitive structure interpretation of multidimensional scaling solutions of concept similarity data. Using high school subjects, convergent validity was assessed through correspondence of scaling solutions of three similarity rating tasks; word association, similarity judgment, and semantic differential. Further evidence was sought by grouping subjects within each data gathering technique by similarity of response using, separately, Tucker's and Carroll's individual scaling techniques. Subgroup membership comparisons were made within data gathering techniques across scaling techniques, and within scaling techniques across data gathering techniques. Characterization of subgroup members was attempted through several variables, including developmental level, field independence, cognitive complexity, and measures of ability to cope with the similarity tasks. On full group data, fit was excellent between word association and similarity judgment data, and poorer with semantic differential data. Judging by the fit among the interpretability of subgroup solutions, both individual scaling techniques formed groups successfully. There was considerable commonality of subgroup membership across similarity tasks in the Carroll analysis only. Characterization of subgroup members by the chosen variables was unsuccessful. Due to error-full data, only Carroll analysis and not Kruskal analysis produced interpretable dimensions on the subgroup data. (Author/RD)