A new stochastic ultrametric tree unfolding methodology for assessing competitive market structure and deriving market segments

We present a new methodology for simultaneously assessing competitive market structure and deriving market segments. A hierarchical or ultrametric tree representation is estimated in a maximum likelihood framework from collected paired‐comparison choice data. The derived tree portrays both brands an...

Full description

Saved in:
Bibliographic Details
Published in:Applied stochastic models and data analysis 1988-09, Vol.4 (3), p.185-204
Main Authors: Desarbo, W. S., de Soete, G., Carroll, J. D., Ramaswamy, V.
Format: Article
Language:English
Subjects:
Market structure analysis
Maximum likelihood
Paired comparisons
Preference trees
Ultrametric trees
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We present a new methodology for simultaneously assessing competitive market structure and deriving market segments. A hierarchical or ultrametric tree representation is estimated in a maximum likelihood framework from collected paired‐comparison choice data. The derived tree portrays both brands and consumers/households/segments as terminal nodes, where the ‘closer’ a brand is to a particular consumer/household/segment in the tree, the higher the predicted probability of that consumer/household/segment choosing that particular brand. This paper initially presents an introduction to the problem of market structure assessment. We review the extensive marketing literature on market structure and survey several competing methodologies. The proposed stochastic ultrametric tree unfolding methodology is technically described and several program options are indicated. An illustration of the proposed methodology is presented with respect to paired comparison choice data collected from a convenience sample involving the over‐the‐counter analgesics market. Finally, several areas for future research are identified.
ISSN:8755-0024
1099-0747
DOI:10.1002/asm.3150040306