Measuring Response Styles Across the Big Five: A Multiscale Extension of an Approach Using Multinomial Processing...

Area of Science:

• Psychometrics
• Personality Assessment
• Quantitative Psychology

Background:

• Response styles (RS) can bias personality assessments derived from rating scales.
• Existing methods may not adequately differentiate trait-related responses from RS.
• Multidimensional item response theory (MIRT) offers a potential framework for addressing this.

Purpose of the Study:

• To develop and validate a MIRT-based approach for detecting and correcting response styles in personality data.
• To differentiate trait-related responses from non-trait-related response styles.
• To provide a method for scoring personality data adjusted for identified response styles.

Main Methods:

• Utilized a multinomial processing tree model to decompose observed rating data into multiple response processes.
• Applied the approach to a dataset of 2,026 U.S. students using the International Personality Item Pool (IPI) measuring the Big Five.
• Employed item response theory (IRT) principles to assess the reliability and validity of the RS measures.

Main Results:

• The MIRT approach successfully identified and differentiated between trait-related responses and response styles (extreme and midpoint).
• Unidimensional measures for extreme and midpoint RS were obtained after excluding specific items.
• RS measures demonstrated high reliability and correlations, with cultural differences observed in extreme responding.

Conclusions:

• The proposed MIRT framework effectively addresses response styles in personality assessments.
• The method allows for the correction of rating scale data, leading to more accurate personality evaluations.
• Findings highlight the importance of accounting for RS in cross-cultural personality research.