PREDICTIVE ACCURACY OF THIRD AND FIFTH GRADERS ON LINGUISTIC AND NON-LINGUISTIC COMPREHENSION MEASURES: THE RELATIONSHIP BETWEEN PERFORMANCE ON GRADUATED PASSAGE QUESTIONS UNDER LISTENING AND READING CONDITIONS AND THE STANDARD PROGRESSIVE MATRICES

Abstract

This study, as an investigation of predictive accuracy, examined the relationship of subjects' performance on a non-linguistic measure of comprehension, the Raven's Standard Progressive Matrices, to their performance on linguistic measures of comprehension, graduated passages and questions from the Barnell-Loft Specific Skills Drawing Conclusions Series. These linguistic passages and questions were presented under silent reading and listening conditions. All third graders and fifth graders at a suburban Maryland parochial school served as the study population from which twenty subjects at each grade level were randomly selected. This sample was comprised of eighteen males and twenty-two females. All subjects received all treatment levels which consisted of four tasks administered in two sessions. Session I included the administration of the Raven's, with the added verbalization of item C-9, and passages and questions presented in a silent reading mode. In Session II all subjects completed passages and questions presented in a listening mode and the Slosson Intelligence Test. The graduated passages and question were contained in Comprehension Inventory I and II which were administered in counterbalanced format. Half of the subjects received Inventory I as the listening task and II as the silent reading task, while half received Inventory I as the silent reading task and II as the listening task. The first question under research in this study was whether there was a relationship between linguistic and non-linguistic predictive accuracy. A multiple correlation technique was used to ascertain the relationship between the Raven's and the listening and silent reading linguistic comprehension measures. A correlation coefficient significantly different from zero was achieved in the comparison of subjects' performance data for these tasks. The second question addressed by this study was whether the relationship between the non-linguistic and the listening linguistic measure of predictive accuracy would be stronger than the relationship between the non-linguistic and the reading linguistic measure of predictive accuracy. Through the use of a partial correlation technique it was determined that the relationship between the Raven's and the listening task was not statistically stronger than the relationship between the Raven's and the silent reading task. Both partial correlations, however, were determined to be significantly different from zero. The third question investigated by this study was whether there was a significant relationship between the Raven's, as a non-linguistic measure of predictive accuracy, and the Slosson, as a linguistic measure of intellig ence. The coefficient produced by the Pearson Product-Moment technique was assessed to be significantly different from zero. Finally, the study sought to ascertain if there were developmental differences in predictive accuracy. At test for related samples produced significant differences for all outcome measures in the direction of the fifth-grade subjects. These findings confirmed the hypothesis that older students would perform the specified tasks at a higher level of predictive accuracy than younger students. Possible explanations of the outcomes of this investigation, and their relationship to research in the areas of language/thought, analogous reasoning, intelligence/cognition, reading/listening and comprehension/prediction were discussed, along with implications for theory, research, classroom and diagnostic practices.