Development and Calibration of an Item ResponseModel That Incorporates Response Time

This article proposes an item response model that incorporates response time. A parameter estimation procedure using the EM algorithm is developed. The procedure is evaluated with both real and simulated test data. The results suggest that the estimation procedure works well in estimating model parameters. By using response time data, estimation of person ability parameters can be improved. Potential applications of this model are discussed. Directions for further study are suggested. Index terms: item response theory (IRT), response time modeling, parameter estimation, EM algorithm In psychophysics and cognitive psychology, response time has long been a research topic of interest. Psychologists who study response time are mainly interested in inferring the organization of the mental process from the distribution of response time to different components of the mental process (e.g., Luce, 1986). A general result from this research is the well-known speed-accuracy trade-off function (SATF), which describes how a respondent’s accuracy in performing certain tasks changes as the response time changes (see Luce, 1986, p. 81; Roskam, 1997). In educational measurement, however, response time data have traditionally been ignored in practice, probably due to the fact that it is difficult to collect response time data at the individual item level with paper-and-pencil testing. Computer-based testing makes response time data at the item level readily available to measurement professionals. For this reason, there has been an increased interest in response time in recent years. So far, existing research has focused on mod-eling the distribution of response time and its relationship with other variables such as item diffi-culty, examinee ability, and test-taking strategies (e.g., Jansen, 1997; Parshall, Mittelholtz, &amp