Response Styles in the Partial Credit Model.

Area of Science:

• Psychometrics
• Survey Methodology
• Psychological Measurement

Background:

• Ordinal response modeling in psychological and survey research often overlooks specific respondent answering patterns (response styles).
• Ignoring these response styles can lead to poor and biased estimates of item parameters.
• A common response style involves a tendency to use extreme or middle response categories.

Purpose of the Study:

• To propose an extended partial credit model that explicitly incorporates person-specific response style parameters.
• To address the issue of biased item parameter estimates caused by unmodeled response tendencies.
• To provide a practical tool for analyzing response styles in ordinal data.

Main Methods:

• An extension of the partial credit model is developed, introducing explicit person-specific response style parameters.
• The proposed model is estimated using generalized mixed linear models.
• Finite mixture approaches are contrasted with the proposed explicit parameter modeling.

Main Results:

• Ignoring response styles, particularly the tendency towards extreme or middle categories, leads to significantly biased parameter estimates.
• The proposed model effectively accounts for these specific response styles.
• Empirical applications show that the tendency to use extreme or middle categories is prevalent in survey data.

Conclusions:

• Explicitly modeling person-specific response styles, such as the tendency towards extreme or middle categories, is crucial for accurate parameter estimation in psychological measurement.
• The developed model and associated software provide a valuable tool for researchers to address response style biases.
• Ignoring response styles can lead to misleading conclusions in psychological and survey research.