Implementing cognitive component constraints in the development and validation of short-form raven’s advanced progressive matrices based on item response theory

Raven’s Advanced Progressive Matrices (APM) is a widely recognized tool for assessing individual differences in intelligence. However, it often faces challenges related to lengthy administration times, which can diminish its practicality and increase resource costs. Although several short-form APMs have been developed, they typically overlook the differences in cognitive components assessed by various items, potentially compromising the instrument’s content validity. To enhance the consistency of the proportion of cognitive components measured in the short-form instruments compared to the complete APM, this study employed an automated item selection algorithm with cognitive component constraints within the item response theory framework—the revised maximum priority index—to develop a series of short-form APMs with varying item counts. By comparing them with existing short-form APMs, the newly developed instruments demonstrated good reliability while maintaining a nearly consistent proportion of cognitive components measured as in the complete APM. Finally, we recommend two short-form APMs containing 12 and 18 items, respectively, for subsequent use, based on the customary lengths of existing short-form APMs.