Accounting for individual differences in speed in the discretized signed residual time model

Area of Science:

• Psychometrics
• Educational Measurement
• Cognitive Psychology

Background:

• Computerized tests increasingly record response accuracy and time.
• The signed residual time (SRT) model uses response time to infer ability, assuming faster correct responses indicate higher ability.
• However, individual differences in response speed unrelated to ability (e.g., caution) can confound measurements.

Purpose of the Study:

• To propose an adapted SRT model that accounts for individual differences in overall response speed.
• To develop a two-dimensional model incorporating dichotomized response time to differentiate fast and slow responses.
• To enable correction of ability estimates for confounding speed differences.

Main Methods:

• Development of a two-dimensional signed residual time (SRT) model.
• Inclusion of both ability and speed parameters to model response time variations.
• Evaluation of the model's performance through simulation studies.
• Application to an empirical dataset from formative educational assessment.

Main Results:

• The proposed two-dimensional SRT model effectively models individual differences in overall speed.
• The model allows for the separation of ability and speed components in response time.
• Simulation results demonstrate the utility of the speed parameter in refining ability estimates.
• Empirical analysis highlights the relevance of accounting for speed in formative assessment.

Conclusions:

• The adapted SRT model provides a more nuanced understanding of ability by disentangling speed effects.
• Incorporating a speed parameter enhances the accuracy of ability estimation in computerized testing.
• This approach offers a valuable tool for improving measurement precision in educational and psychological assessments.