Examining the Incremental Validity of Working Memory for Predicting Learning and Task Performance: A Partial Mediation Model

• Main Author: Carter, Devin
• Other Authors: Psychology
• Format: Others
• Language: en
• Published: Virginia Tech 2017
• Subjects: Working Memory; Intelligence; Knowledge; Learning; Performance
• Online Access: http://hdl.handle.net/10919/81312

General intelligence (“g”) has long been used as an effective predictor of both learning and job performance. Further, other more specific cognitive abilities have not been able to consistently predict incremental variance in job knowledge and job performance beyond “g”. However, the processes associated with working memory (WM) are important for these outcomes and are not captured by our traditional tests of “g”. This study tested a partial mediation model in which working memory (WM) incrementally predicts task performance above “g” through task knowledge and through a direct effect. Participants were given measures of “g” and WM in a lab. They were then given a learning opportunity and a task that applies this newly learned knowledge in order to tests the effects of WM. Results indicate that WM explains additional variance in both task knowledge and task performance, and the partial mediation model was supported using one of the two WM tasks used. === Master of Science === General intelligence is widely used in personnel selection because it is consistent in predicting the job performance of future employees. Other cognitive abilities have also been examined to determine whether they are able to predict job performance as well as general intelligence. However, most of these other cognitive abilities have come up short. This study hypothesized that working memory (WM) is a cognitive ability that may be able to predict job performance even after controlling for general intelligence. A sample of undergraduates completed tasks that measured general intelligence and WM, and this study examined how well each measure predicted both learning and performance on a relatively novel task. Results indicated that WM was able to predict both learning and performance after controlling for general intelligence.