Cognitive tasks borrowed from experimental psychology are often used to assess individual differences. A cardinal issue of this transition from experimental to correlational designs is reduced retest reliability of some well-established cognitive effects as well as speed-accuracy trade-off. The present study aimed to address these issues by examining the retest reliability of various methods for speed-accuracy integration and by comparing between two types of task modeling: difference scores and residual scores. Results from three studies on executive functions show that (a) integrated speed-accuracy scoring is generally more reliable as compared with nonintegrated methods: mean response time and accuracy; and (b) task modeling, especially residual scores, reduced reliability. We thus recommend integrating speed and accuracy, at least for measuring executive functions.
Keywords: cognitive tasks; executive functions; individual differences; retest reliability; speed–accuracy integration.