Retailers need to decide on the content and structure of their product assortments, and thereby on the degree of variety that they offer to their customers. This paper compares measures of assortment variety and relates them to underlying variety components. We conceptualize assortment variety from a product and an attribute perspective, compare extant measures of assortment variety, and examine the appropriateness of these measures in capturing assortment variety as perceived by consumers. Recently, Hoch, Bradlow and Wansink (1999) introduced a general model of assortment variety based on product dissimilarities. The current study takes an alternative approach and proposes variety measures based on attributes, specifically the dispersion across attribute levels and the association between the attributes of the products in an assortment. Attribute dispersion refers to the diversity of attribute levels in an assortment (e.g. the relative proportion of red, green, blue products), while association between attributes refers to systematic links between attributes (e.g. all red products are large). We show that product-based and attribute-based approaches to assortment variety lead to substantially different measures with different effects on consumers' perceptions of variety. A first, synthetic, data set, shows that measures of attribute dispersion, attribute association and assortment size reflect specific components of assortment variety. The product-based measure proposed by Hoch et al. is sensitive to the size of the assortment, while the attribute-based measures respond only to specific changes in the content of an assortment. A second, consumer, data set shows that the attribute-based approach accounts best for consumers' perceptions of variety, and offers diagnostic power to retailers by explicatingvariety components. Attribute-based measures of variety significantly add to the prediction of consumers' perceptions of variety, over and above the product-based variety measures, while the reverse is not the case. In the final section we discuss how attribute-based measures can be used in assortment management, e.g. when assortments of different size are compared, when the impact of adding or dropping products on assortment variety is to be determined, and when diagnostic information about assortment variety is important
