Person-Level Analysis of the Effect of Cognitive Loading by Question Difficulty and Question Time Intensity on Didactic Examination Fluency (Speed-Accuracy Tradeoff).

Abstract

Fluency may be considered as a conjoint measure of work product quality and speed. It is especially useful in educational and medical settings to evaluate expertise and/or competence. In this paper, didactic exams were used to model fluency. Binned propensity matching with question difficulty and time intensity was used to define a 'load' variable and construct fluency (sum correct/ elapsed response time). Response surfaces as speed-accuracy tradeoffs resulted from the analysis. Person by load fluency matrices behaved well in Rasch analysis and warranted the definition of a person fluency variable ('skill'). A path model with skill and load as mediators substantially described the fluency data. The indirect paths through skill and load dominated direct variable effects. This is supportive evidence that skill and load have stand-alone merit. Therefore, it appears that the constructs of skill, load, and fluency could provide psychometrically defensible descriptors when utilized in appropriate contexts.