Long-term retention of semantic knowledge.

Abstract

Two hierarchical regressions were posed to examine the relative contribution of several predictor variables on retention test performance. The retention test encompassed content from a beginning graduate level statistics class. Cross-sectional methodology was employed to include students who had taken the course sometime during a twenty-two year interval. This study had a unique opportunity to examine long term remembering in an ecological setting where the content area and the teaching had been stable. Grade, from the original course, was the strongest predictor in both hierarchical models. Other independent variables which had significant impact on retention test performance were number of continuing classes in statistics and number of classes in research design and methodology. Rehearsal frequency as well as rehearsal recency were significant predictors. The level of original learning and what one does during the retention interval are more important than the length of the interval itself. The effect of spaced vs. mass practice, as defined by the length of the acquisition interval, was examined. Subjects who took the course over a 15 week semester session outperformed subjects who had the 5 week summer session. In this study, the rate of decline was affected by the subject's age at the time of the retention test. This indicates increasingly rapid forgetting during adult development and has implications for the maintenance of marginal knowledge.