A Method for Developing Maximum Differential Scaling (MaxDiff) Segmentation Typing Tools

• Activity Number: 467
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Marketing
• Author(s): Shon Magnan*+
• Companies: GfK CRNA
• Keywords: Marketing Research ; Segmentation ; MaxDiff

Abstract

Maximum Differential Scaling or MaxDiff is a popular methodology in marketing research as an alternative to collecting Likert-based ratings for key driver analysis and segmentation. Rating scales, especially when collecting global data, are notorious for scale usage heterogeneity problems. An issue when using MaxDiff utilities in the creating of segments, however, is the creation of a typing tool. Typing tools are short lists of attributes and a model that are used by organizations to quickly predict segment membership for new sets of data. The models used to create typing tools for rating scale segmentations are typically discriminant or multinomial logit method. Obviously, those approaches are much harder to implement with MaxDiff utilities. In a 2009 Marketing Research article, "Typing Tools that Work", Orme & Johnson suggest a methodology for typing tools from a MaxDiff using an iterative search and optimization algorithm. Using their approach as a framework, we illustrate an easier method that produces satisfactory segment classification accuracies by careful and appropriate selection of the items to include in each max-diff set.