Comparing approaches for clustering mixed mode data:an application in marketing research