Extreme Response Style: A Simulation Study Comparison of Three Multidimensional Item Response Models.

Area of Science:

• Psychometrics
• Statistical modeling
• Survey methodology

Background:

• Response styles, such as extreme responding, can bias survey data.
• Multidimensional item response theory (IRT) models offer potential solutions for analyzing complex survey data.
• Comparing the performance of different IRT models is crucial for accurate data interpretation.

Purpose of the Study:

• To compare the performance of three multidimensional IRT models in accounting for extreme response tendencies.
• To evaluate the effectiveness of the IRTree Model, the multidimensional nominal response model, and the modified generalized partial credit model.
• To identify the most suitable model for analyzing survey data affected by response styles.

Main Methods:

• Simulation study comparing three specific IRT models.
• Evaluation of models across varying sample sizes (500, 1,000), survey lengths (10, 20), and number of response options (4, 6).
• Assessment based on item mean squared error (MSE) and posterior predictive model checking.

Main Results:

• The modified generalized partial credit model demonstrated the lowest item MSE across all simulated conditions.
• The multidimensional nominal response model performed comparably for specific survey designs (10 items, 4 options).
• The IRTree Model was hypothesized to capture additional sources of error, warranting further investigation.

Conclusions:

• The modified generalized partial credit model is a robust choice for analyzing survey data with extreme response tendencies.
• Model selection depends on specific survey characteristics and the primary research objectives.
• Further research is needed to validate the IRTree Model's ability to account for other sources of irrelevant variance.