Conditional Dependence across Slow and Fast Item Responses: With a Latent Space Item Response Modeling Approach

Area of Science:

• Cognitive Psychology
• Psychometrics
• Educational Measurement

Background:

• Previous research on cognitive tests often assumes uniform patterns of conditional dependence between response accuracy and response times.
• However, individual differences in cognitive abilities and item characteristics suggest this dependence may be heterogeneous.
• Understanding this heterogeneity is crucial for accurate assessment of cognitive processes.

Purpose of the Study:

• To investigate the item and person specificities in the conditional dependence between item responses and response times in cognitive tests.
• To explore the heterogeneity of this relationship across different respondents and items.
• To provide a more nuanced understanding of cognitive test-taking behaviors.

Main Methods:

• Employed a latent space item response theory (LSIRT) approach.
• Utilized an interaction map to visualize conditional dependence in item-respondent interactions.
• Incorporated response time data by applying LSIRT models to slow and fast response groups.

Main Results:

• Confirmed the presence and general patterns of conditional dependence between item responses and response times, consistent with prior studies.
• Demonstrated significant heterogeneity in this conditional dependence across respondents.
• Identified individual-specific patterns in the relationship between accuracy and response time.

Conclusions:

• The conditional dependence between response accuracy and response times in cognitive tests is not uniform but varies across individuals and items.
• LSIRT with interaction maps effectively visualizes and analyzes this heterogeneity.
• Findings offer deeper insights into individual cognitive processes during test-taking and have practical implications for test design and interpretation.