An analysis of an item-response strategy based on knowledge retrieval

Area of Science:

• Educational psychology
• Psychometrics
• Cognitive science

Background:

• Traditional item-response theory (IRT) models often assume consistent ability utilization and response times.
• Unexplained accuracy variations may occur when individuals employ non-standard strategies.
• Assessing these strategies is crucial for a comprehensive understanding of learning and performance.

Purpose of the Study:

• To hypothesize and model a knowledge retrieval-based item-solving strategy.
• To investigate the conditions under which this strategy emerges (e.g., non-utilization of regular ability, relaxed time constraints).
• To explore the relationship between response time, accuracy, and knowledge retrieval in assessments.

Main Methods:

• Utilizing item-response time data within a finite-mixture item-response theory (IRT) modeling framework.
• Analyzing low-stakes assessment data collected without time constraints.
• Employing simulation studies to validate the accuracy and reliability of the proposed modeling approach.

Main Results:

• The finite-mixture IRT model can effectively identify a distinct item-solving strategy characterized by knowledge retrieval.
• This strategy is associated with conditions where typical abilities are not fully engaged and response times are not minimized.
• Empirical data and simulations demonstrate the model's capability to distinguish knowledge retrieval from other response patterns.

Conclusions:

• A knowledge retrieval-based item-solving strategy can be modeled and identified using response time and accuracy data.
• This approach offers a more nuanced understanding of cognitive processes during assessments beyond traditional IRT assumptions.
• The findings have implications for designing more informative assessments and interpreting student performance data.