Investigating Speed–Accuracy Tradeoff in Multistage Adaptive Language Assessment

The speed–accuracy tradeoff, in which faster decisions are often linked to reduced accuracy and slower responses allow for greater accuracy, has been widely studied in experimental psychology. This study examines how within-person variation in response time relates to accuracy across grade clusters and proficiency tiers in multistage adaptive language testing. Using a flexible modeling approach to identify the speed–accuracy tradeoff patterns, the study analyzes data from the WIDA ACCESS for ELLs Listening and Reading domains in high-stakes settings. Results show a consistently negative relationship (additional time was associated with lower accuracy) in Listening and a curvilinear relationship (accuracy rises with increasing time up to a threshold, then declines) in Reading, reflecting domain-specific processing demands and time-management strategy. This study contributes to speed–accuracy tradeoff research by advancing our understanding of the intraindividual relationship between speed and accuracy in language assessment.