The Effects of Aberrant Responding on Model-Fit Assuming Different Underlying Response Processes

Area of Science:

• Psychometrics
• Statistical Modeling
• Educational Measurement

Background:

• Aberrant responding (e.g., random or patterned answers) can distort scale psychometric properties.
• Model misfit in statistical analyses can arise from data misspecification or aberrant response patterns.

Purpose of the Study:

• To compare the effects of four aberrant responding types on model fit.
• To evaluate these effects within both cumulative and ideal point model contexts.
• To examine model fit using graded partial credit (GPCM) and generalized graded unfolding (GGUM) models.

Main Methods:

• Simulated data with varying levels of four aberrant response types.
• Analysis using cumulative (GPCM) and ideal point (GGUM) models.
• Assessment of model fit using information criteria (AIC, BIC) and dimensionality.

Main Results:

• Aberrant data severely impacts model fit for both cumulative and ideal point data.
• Longstring responses most strongly affect dimensionality in both data types.
• Random responding most negatively impacts model fit based on information criteria.

Conclusions:

• Ideal point models (GGUM) may offer superior fit for cumulative data compared to GPCM.
• Cumulative models (GPCM) may inadequately fit ideal point data.
• Understanding aberrant responding is crucial for accurate psychometric analysis.