Separability of item and person parameters in response time models

Abstract

response times, separate sufficiency, ranking independence, exponential family, scale parameter, location parameter, accelerated life model, proportional hazards model, proportional odds model, pseudolikelihood, double monotonicity,

Pace: An Alternative Measure of Student Question Response Time

“…This article builds on the prior work of others (Bontempo, 2000;Chang et al, 2005;Hornke, 2000;Jansen, 2003;Thissen, 1983;Van Breukelen, 1997;Van der Linden et al, 1999). Our ability to test the hypothesis that persons maintain a constant tempo as they progress through an exam relative to the question duration was possible because of the availability of computer-based testing technology.…”

“…As a direct consequence, significant advances in our understanding of examination response times have been made at the level of the person-question interaction (Bontempo, 2000;Chang, Plake, & Ferdous, 2005;Hornke, 2000;Jansen, 2003;Thissen, 1983;Van Breukelen, 1997;Van der Linden et al, 1999). In recent years, consideration of item response time in selection of items has played an important role in the use of computer-based adaptive testing in which the test-taker's performance on test items determines the selection of test items presented subsequently ( Van der Linden et al, 1999).…”

A flexible latent trait model for response times in tests

“…Such a model was first sketched in the 1950s by Furneaux (1952). Since then, further models have been proposed (Scheiblechner, 1979;Maris, 1993;van Breukelen, 1997;Douglas, Kosorok, & Chewing, 1999;van der Linden, 2006). However, there does not seem to exist a standard model that can routinely be used for response times in tests.…”

Psychometric Modeling of response speed and accuracy with mixed and conditional regression

“…This is a mixed two way ANOVA model without interaction (for an application to RTs, see Van der Linden et al, 1999). The marginals of the person by item data matrix of ln RT are sufficient statistics for the person and item parameters, respectively (Van Breukelen, 1997), as in the Rasch model for binary data. By replicating each person by item combination, interaction can be tested.…”