Cluster-based market segmentation: some further comparisons of alternativeapproaches

Recently, Green & Krieger (1995) reported that the highly popular 'tandem a pproach' to market segmentation, based on a preliminary factor analysis fol lowed by a clustering of rotated, standardised factor scores, does not perf orm as well as a straightforward clustering. Comparative performance is bas ed on how highly associated the resulting respondent clusters are with a se t of background attributes, not used in the clustering. This paper extends the scope of their research from comparisons of conjoint -based importances (reported by Green & Krieger) to more general types of d ata, such as ratings scales for needs-based segmentation.