A NEW METHODOLOGICAL REVISION OF EKMAN'S CONTENT MODEL OF MULTIDIMENSIONAL SIMILARITY ANALYSIS

Area of Science:

• Psychology
• Quantitative Psychology
• Multidimensional Scaling

Background:

• Ekman's content model is a standard for multidimensional similarity analysis.
• Ekman's method involves factor analysis of mean similarity estimates.
• Mathematical objections have been raised against Ekman's model.

Purpose of the Study:

• To propose a methodological revision to Ekman's content model.
• To address mathematical objections within Ekman's multidimensional similarity analysis.
• To introduce the Stone-Coles revision for analyzing correlational similarities.

Main Methods:

• The Stone-Coles revision uses factor analysis on correlational similarities.
• Correlational similarities are derived by intercorrelating column vectors of an Ekman mean similarity matrix.
• Reanalyzed 14 previous investigations using the Stone-Coles approach.

Main Results:

• The Stone-Coles method provides interpretable factorial descriptions of judged similarity.
• Results from the revised method are more meaningful than those from Ekman's methodology.
• The revised method yields more parsimonious descriptions of stimulus-object similarity.

Conclusions:

• The Stone-Coles revision offers a mathematically sound improvement over Ekman's model.
• This new methodology enhances the interpretability and parsimony of similarity analysis.
• The revised approach provides a more effective tool for understanding multidimensional similarity.