Experimental research on categorization processes suggests that people respond to environmental stimuli on the base of some typical modalities. Different relationships are usually employed in conceptualizing item sets: when a subject is presented with a triad of stimuli (e.g., 'dog-bone-cat'), he can relate 'dog' to 'bone' showing a complementary preference on classification. On the contrary, if he selects 'cat', he classifies items with respect to their taxonomic relationships. Smiley and Brown (18) found a strong evidence for a shift in conceptual preference: complementary associations are more frequent in preschool children, whereas taxonomic classifications more frequently occur in adult subjects. This developmental trend is questioned by Greenfield and Scott (5), who found a complementary preference across all ages. On this basis, we hypothesized that divergent findings could not depend on the same experimental methods, but on the different measure parameters selected by Smiley and Brown and Greenfield and Scott, respectively. The aim of this study is to test this hypothesis. We replicated the same method of the previous studies, but we utilized both measure parameters: a) overall number of complementary versus taxonomic responses, and b) subject classification with respect to their complementary or taxonomic preference. The results were in accord with Greenfield and Scott's findings: subjects provided more complementary than taxonomic responses across all ages. The subject number who preferred the complementary relationships was significatively greater also when we changed the measure parameter. On the basis of these outcomes the authors question the validity of the conceptual preference as explicative variable of categorization behavior in human development.