Categorical Judgment Scaling with Ordinal Assumptions

Area of Science:

• Psychology
• Quantitative Psychology
• Psychometrics

Background:

• Likert scales are widely used in psychological research.
• Researchers frequently analyze Likert scale data assuming equal interval properties, which is often a flawed assumption.
• Existing alternatives have limitations, often relying on normality assumptions.

Purpose of the Study:

• To propose a novel method for analyzing Likert scale data using additive conjoint measurement.
• To overcome the limitations of traditional analysis by not assuming equal interval scales.
• To extract interval-level stimulus values and response category boundaries without distributional assumptions.

Main Methods:

• Application of additive conjoint measurement to Likert scale data.
• Assumption that subjects can provide rank-ordered responses.
• No within-subject or between-subject distributional assumptions are made.

Main Results:

• The proposed method successfully extracted interval-level stimulus values and response category boundaries from three attitude datasets.
• Analysis revealed that the equal interval assumption was clearly inappropriate for the data.
• Despite the violation of interval assumptions, arithmetic means effectively represented group attitudes.

Conclusions:

• Additive conjoint measurement offers a robust alternative for analyzing Likert scale data, particularly when interval assumptions are violated.
• The method provides valuable insights into stimulus values and response categories without stringent distributional requirements.
• This approach enhances the validity of Likert scale analysis in psychological research.