Predicting Academic Achievement with Models Based on Eye Movement Data

Abstract

This project was designed to test the premise that, because common elementary information processing operations underlie performance on both IQ-test-type tasks and academic tasks, such measures of information processing are potent predictors of academic success. The proposed predictive superiority of measures of information processing over test score data is believed to result because test scores reflect only the product of performance and thus are most closely tied to particular knowledge stores and procedural rules specific to given test items. Two studies are reported in which eye movements are recorded during solution of a set of 3 x 3 figural analogies. Multiple regression analyses indicate that models based on eye movement data account for significant amounts of variation in overall academic achievement, with these information processing models outstripping the model based on test score data. High achieving individuals can be characterized as processing information in large units with a correspondingly small amount of redundancy and as devoting a relatively small percentage of their processing resources to rule application activities. Implications for diagnosis and prescription are discussed.