STIMULUS COMPLEXITY AND SIMILARITY AS FACTORS IN MATRIX PROBLEM SOLVING
Analogies, as represented in figural matrix problems, have long been included on widely used tests of intellectual ability. The appearance of these items, as well as the development of psychometric instruments devoted exclusively to their use, reflects the ongoing belief that analogical reasoning is intimately related to intellectual ability. The present study was concerned with the task analysis of one such test of intellectual ability, the Raven's Standard Progressive Matrices (SPM). The major hypotheses were that figural matrix problems were multidimensional stimuli, where two dimensions were complexity and similarity, and that the matrices which would be difficult to solve would be those scaled highest on the complexity and similarity dimensions.^ Procedure. The study was conducted in two phases. The first phase was concerned with the selection and scaling of the stimulus materials along the dimensions of complexity and similarity. The second phase was concerned with determining the effects of complexity and similarity on solution speed and error rates for the matrix problems.^ Problems 4 through 12 of Set D of the SPM served as prototypes for the construction of 60 randomly generated experimental problems. The four rules utilized by Raven governed the construction of the problems. The matrices were scaled by 80 volunteer undergraduates using a pile sorting technique, and complexity and similarity rating scales.^ A second group of 80 subjects attempted to solve the problems, in the second phase of the study. Data collection was accomplished with an Apple II microcomputer interfaced with a slide projector and a real-time clock. Subject responses and reaction times were automatically recorded. The problems were presented on slides.^ Results. The scaling data were analyzed with the KYST multidimensional scaling program. A three dimension solution was obtained. The complexity and similarity rating scale data were negatively correlated and were predicted by the same dimension. The problem solving data, mean error rates and solution times, were analyzed by regressing them on the complexity data from the first phase and the rule number of the problem. Error rates were predicted by rule while solution times were predicted by both rule and complexity rating. The results were discussed in terms of the correlation between the complexity and similarity ratings, the possibility of a degenerative multidimensional solution, and the relationships between complexity, rule, and problem solving. ^
MCLAUCHLAN, WILLIAM GRANT, "STIMULUS COMPLEXITY AND SIMILARITY AS FACTORS IN MATRIX PROBLEM SOLVING" (1982). ETD Collection for Fordham University. AAI8219254.