Exploring the Effects of Item-Specific Factors in Sequential and IRTree Models

Area of Science:

• Psychometrics
• Educational Measurement
• Item Response Theory

Background:

• Sequential and IRTree models are used for complex item scoring.
• Item-specific factors, though unmeasurable, may influence responses across item stages.
• Existing models may not fully account for these persistent item characteristics.

Purpose of the Study:

• To present a conceptual model for item-specific factors in sequential/IRT modeling.
• To demonstrate how these factors impact parameter estimation and interpretation.
• To discuss implications for various item response theory applications.

Main Methods:

• Developed a conceptual model incorporating item-specific factors.
• Analyzed how these factors affect stage-specific discrimination and difficulty parameters.
• Examined empirical data and literature examples for violations of parameter invariance.

Main Results:

• Item-specific factors can create ambiguity in item and person parameter interpretation beyond the initial stage.
• Empirical evidence suggests violations of item parameter invariance across stages.
• The proposed model explains observed patterns in repeated attempts, answer changes, hints, skipping, and Likert scales.

Conclusions:

• Item-specific factors pose a challenge to the accurate interpretation of parameters in sequential and IRTree models.
• Researchers should actively look for evidence of item-specific factors in their data.
• Sensitivity analyses are recommended to assess the impact of these factors on study inferences.