A cross‐cultural clustering study of similarities and dissimilarities based on concept words cheap, reliable, and high quality and their corresponding color associations. Issue 3 (31st January 2019)

• Record Type: Journal Article
• Authors: Abdullah, Ahsan
• Is Part Of: Color research & application. Volume 44:Issue 3(2019:Jun.)
• Page Start: 433
• Page End: 445
• Publication Date: 2019-01-31
• Subjects: cheap -- colors -- factor analysis -- hierarchical clustering -- high‐quality -- k‐means -- one‐way clustering -- Pakistan -- reliable -- Saudi Arabia -- USA; Color -- Periodicals; 535
• DOI: 10.1002/col.22354
• Languages: English
• ISSNs: 0361-2317

Abstract

Abstract: The world has undergone a techno‐cultural revolution during the last decades with the materialization of the "global village" and more; all having a profound impact on our culture. Quantitative analysis of these profound multidimensional changes needs cross‐cultural data, and scalable tools/techniques that can discover interesting nontrivial hidden relationships in large data sets. Subsequently, presenting these discovered relationships in an intuitive visual form to non‐IT domain experts. In this article, using proven data mining techniques, we perform cross‐cultural clustering of word‐color associations for 344 respondents from three cultures, that is, the United States, Pakistan, and Kingdom of Saudi Arabia. We consider the concept words cheap, reliable, and high quality along with colors liked and disliked for 10 colors. After statistical data validation one‐way cross‐cultural clustering was performed. High similarity of word‐based cluster association was discovered across the three cultures for negative and positive connotation words, and clear dissimilarity in clustering was discovered within each culture for color‐based cluster association; with possible explanations given. The universal consistent similarities in cross‐culture associations and the dissimilarities in local or intra‐culture associations could be useful for global brand decision making. In this article, we also do a technical comparison of the proposed technique with three traditional techniques used for such analysis, that is, hierarchical clustering, k‐means clustering and factor analysis; along with related comparison with principal component analysis and multidimensional scaling.