ASSOCIATIVE ERRORS IN CHILDREN'S ANALOGICAL REASONING: A COGNITIVE PROCESS ANALYSIS.

摘要

Previous research has suggested that reliance upon association in analogy solution indicates a cognitive style that impairs learning and limits the predictability of achievement by intelligence. Experiment 1 investigated the hypothesis that the achievement-intelligence relation is moderated by associative responding. Experiment 2 adopted the Sternberg componential framework to explore the cognitive processes underlying logical and associative errors on the Children's Associative Responding Test (CART).;In the second study, the cognitive processes underlying the error types were compared and contrasted. Paper-and-pencil tasks were devised to measure encoding, inference, mapping relations, and response evaluation. Also measured were vocabulary, semantic flexibility, and working memory capacity. Factor analysis resulted in four primary factors: vocabulary, encoding processes, inductive reasoning and semantic flexibility, and response evaluation, and a higher order general factor.;Further regression analysis showed that only mapping relations did not significantly predict the two CART scores. Despite considerable criterion overlap, vocabulary and inductive reasoning were more highly related to non-associative errors, and working memory and semantic flexibility were more highly related to associative errors.;These findings can be interpreted to mean that when association is not available or is avoided as a solution strategy, great reliance is placed upon vocabulary and inductive reasoning. However, when association is employed as a strategy, this could be due to limited working memory capacity and perhaps to inflexibility in determining word meanings.;Achievement test scores were obtained on two samples of 5th and 6th graders. The predictors were intelligence, CART associative errors, and their product. In 18 regressions, the associative score explained variance unexplained by intelligence in 11 cases, moderated the effect of intelligence for English and Mathematics, and came close for two additional cases involving Mathematics. It was concluded that only modest evidence exists for the moderator effect but that associative responding and intelligence reflect distinct cognitive processes.