A general diagnostic classification model for rating scales.

Area of Science:

• Psychometrics
• Statistical modeling
• Educational measurement

Background:

• Traditional diagnostic classification models (DCMs) for polytomous items often require large sample sizes.
• Parameter estimation in existing models can be complex and computationally intensive.
• Limited information is typically provided by traditional DCMs regarding general item characteristics.

Purpose of the Study:

• To propose and evaluate a novel general diagnostic classification model (DCM) for rating scales.
• To compare the performance of the proposed DCM against traditional DCMs for polytomous items.
• To assess the parameter recovery and efficiency of the proposed model through simulation.

Main Methods:

• Development of a general diagnostic classification model (DCM).
• Application of the proposed DCM to an existing dataset for performance comparison.
• Conducting a simulation study to evaluate parameter recovery under applied conditions.

Main Results:

• The proposed DCM effectively accommodates smaller sample sizes by reducing the number of parameters.
• Item category response probabilities and individual scores derived from the proposed model closely match those from traditional saturated models.
• The proposed DCM yields general item information not obtainable from traditional DCMs for polytomous items.

Conclusions:

• The proposed general diagnostic classification model (DCM) offers a promising alternative for analyzing rating scale data.
• The model demonstrates advantages in sample size efficiency and data richness.
• This approach enhances the utility of DCMs for polytomous items in various assessment contexts.