A psychometric model for respondent-level anchoring on self-report rating scale instruments

Area of Science:

• Psychological Measurement
• Psychometrics
• Behavioral Research

Background:

• Self-report rating scales can be influenced by response biases like anchoring.
• Anchoring describes the tendency to choose response options near the previous item's category.
• Understanding these biases is crucial for accurate assessment of respondent engagement.

Purpose of the Study:

• To propose and validate a psychometric model for analyzing response styles and anchoring simultaneously.
• To investigate the relationship between anchoring, midpoint response style, and respondent effort.
• To explore the utility of these biases as indicators of respondent engagement.

Main Methods:

• Developed a multidimensional nominal model incorporating respondent-level anchoring.
• Employed a fully Bayesian estimation procedure for model fitting.
• Applied the model to a real-world dataset measuring extraversion.

Main Results:

• A positive correlation was found between midpoint response style and the anchoring effect.
• Both anchoring and midpoint response styles negatively correlated with response time.
• These findings support the theory that both phenomena indicate reduced respondent effort.

Conclusions:

• Anchoring and midpoint response style are associated with less effortful responding.
• These response biases can serve as valuable indicators of respondent engagement.
• The proposed psychometric model offers a robust method for assessing these biases.