Modeling Differences Between Response Times of Correct and Incorrect Responses.

Area of Science:

• Psychometrics
• Cognitive Psychology
• Educational Measurement

Background:

• Standard joint models for response time (RT) and accuracy often assume conditional independence, which is frequently violated in practice.
• Ignoring residual dependencies between RT and accuracy can lead to suboptimal model fit and limited understanding of cognitive processes.

Purpose of the Study:

• To propose a novel framework for joint modeling of RT and accuracy data that accommodates distinct processes for correct and incorrect responses.
• To extend existing hierarchical models by allowing item parameters in the speed measurement model to vary based on response correctness.
• To introduce models with two distinct speed latent variables for correct and incorrect responses.

Main Methods:

• Development of a flexible joint modeling framework extending the standard hierarchical model.
• Incorporation of differential item parameters for speed based on response correctness.
• Implementation of models with separate latent speed variables for correct and incorrect responses.
• Evaluation of model selection procedures and parameter recovery through simulation studies.

Main Results:

• The proposed framework demonstrates improved model fit by accounting for residual dependencies.
• Simulation studies validate the proposed model selection procedures and parameter recovery.
• Application to large-scale assessment data highlights the practical relevance of modeling response-specific processes.

Conclusions:

• The proposed joint modeling framework effectively addresses violations of conditional independence in RT-accuracy data.
• Allowing for distinct processes underlying correct and incorrect responses enhances understanding of cognitive mechanisms in assessments.
• This approach offers a more nuanced and accurate analysis of response time and accuracy data in psychometric modeling.