Testing for Equivalence of Measurement Scales: Simple Structure and Metric Invariance Reconsidered.

Area of Science:

• Psychometrics
• Statistical Modeling
• Quantitative Psychology

Background:

• Quantitative comparisons often rely on multi-item scales.
• Traditional psychometric approaches demand item-level simple structure and metric invariance for scale validity.
• Existing methods can be overly stringent, complicating scale development and analysis.

Purpose of the Study:

• To propose a novel framework for multi-item scale analysis.
• To demonstrate that scale-level invariance is sufficient for valid quantitative comparisons.
• To offer a more parsimonious alternative to traditional measurement models in confirmatory factor analysis (CFA).

Main Methods:

• The study introduces the concept of scale-level simple structure and metric invariance.
• It proposes implementing these concepts via constraints on mean factor loadings and intercepts of item sets within CFA.
• This approach contrasts with the conventional item-level specification.

Main Results:

• Scale-level invariance provides a sufficient condition for valid quantitative comparisons.
• The proposed method simplifies measurement model specification in CFA.
• This offers a practical alternative to current predominant CFA measurement models.

Conclusions:

• Valid quantitative comparisons using multi-item scales necessitate only scale-level, not item-level, simple structure and metric invariance.
• The proposed confirmatory factor analysis (CFA) approach offers a more efficient and conceptually flexible alternative for measurement model specification.
• This research reframes the requirements for robust psychometric scale analysis.