Methods of identifying individual guessers from item response data

This article investigates several methods of identifying individual guessers from their response data. Both the posterior probability method and the likelihood ratio method are based on the two-state mixture modeling approach to response times. The accuracy method is based on response accuracy data. Results from the simulation study showed that under similar classification criterion, the likelihood ratio method had a higher rate of correctly identifying guessers than the posterior probability method and led to a much higher rate of false-positive misclassification. The accuracy method, complementary to the likelihood ratio method, did reduce the false-positive rate of misclassification while maintaining the higher rate of correct identification of random guessers, given that examinees' response times are inversely related to their abilities.