Abstract
The ability to rank brands or services from 'best'to 'worst' is a well-known marketing need. The current methodology for doing so is at best ad hoc; it involves some arbitrary techniques, and is not generally statistically testable. In this paper some well-known methods from categorical data analysis are combined with some innovative ideas (and a new theoretical result), to derive a methodologically sound and statistically testable new technique to compare brands on an ordinal variable (with special reference to satisfaction). The new technique, based on the notion of stochastic ordering, has the merit of enabling a comparison of brands even under initially arbitrary satisfaction scale. It determines an 'optimal scale' under which the brands can be insightfully compared and ranked. Two examples dealing with satisfaction of car dealers from training of their personnel are analysed to illustrate the advocated methodology.
Original language | English |
---|---|
Pages (from-to) | 249-264 |
Number of pages | 16 |
Journal | International Journal of Market Research |
Volume | 40 |
Issue number | 3 |
State | Published - 1998 |