Prediction Models of Personality Item Responding

Area of Science:

• Psychological Assessment
• Cognitive Psychology

Background:

• Models of test item response have expanded from ability to personality testing.
• Understanding response processes is key to accurate personality assessment.

Purpose of the Study:

• To compare the predictive accuracy of four distinct models for personality test item responses.
• To evaluate models using response prediction and item p-value distribution criteria.

Main Methods:

• Utilized three independent samples (N=552) responding to the Personality Research Form (PRF).
• Compared Jackson's threshold model, a content-based threshold model, a cumulative scale score model, and a multidimensional spatial model.
• Assessed predictive accuracy for item responses and item p-value distributions.

Main Results:

• The multidimensional spatial model, cumulative scale score model, and Jackson's threshold model demonstrated statistically significant predictive accuracy.
• The content-based threshold model showed only marginal improvement over chance predictions.
• Model performance was discussed within the framework of cognitive information processing.

Conclusions:

• Certain models, particularly the multidimensional spatial and cumulative scale score models, offer robust frameworks for understanding personality test item responding.
• Cognitive information processing offers a valuable lens for interpreting individual differences in test item response patterns.