A method to analyse measurement invariance under uncertainty in between-subjects design

Area of Science:

• Psychometrics
• Statistical Analysis
• Consumer Behavior

Background:

• Assessing abstract concepts requires reliable measurement scales.
• Existing methods often overlook individual response uncertainty.
• Scale invariance is crucial for valid comparisons in between-group designs.

Purpose of the Study:

• Introduce a new nonparametric test (H-test) for scale invariance.
• Evaluate the adequacy of rating and visual scales when accounting for response uncertainty.
• Analyze perceived service quality using the H-test.

Main Methods:

• Developed the H-test, a nonparametric statistical test.
• Applied the H-test to consumer data on sports service quality.
• Compared scale invariance with and without considering individual response uncertainty.

Main Results:

• Without uncertainty, a 1-7 scale showed invariance.
• With uncertainty, both 1-5 and 1-7 scales became invariant.
• Visual scales demonstrated non-invariance under uncertainty.

Conclusions:

• Individual response uncertainty significantly impacts scale invariance analysis.
• Standard invariance conclusions may be inadequate without considering uncertainty.
• The H-test provides a robust method for evaluating scales in psychometric research.