A taxonomy of highly interdependent, supply chain relationships : The use of cluster analysis

Purpose – Cluster analysis provides a statistical method whereby unknown groupings of similar attributes can be identified from a mass of data and is well‐known within marketing and a wide range of other disciplines. This paper seeks to describe the use of cluster analysis in an unusual setting to classify a large sample of dyadic, highly interdependent, supply chain relationships based upon the quality of their interactions. This paper aims to show how careful attention to the detail of research design and the use of combined methods leads to results that both are useful to managers and make a contribution to knowledge. Design/methodology/approach – Data relating to 55 monopolistic relationships in the UK defence procurement sector were collected. Hierarchical cluster analysis using Wards method was undertaken on scores from five dimensions measuring relationship satisfaction. The resulting clusters are described in terms of the scores on the dimensions and also in terms of their relationships with data, quantitative and qualitative, exogenous to the clusters. Findings – The analysis reveals five distinct clusters of relationships. Statistically significant differences are evident in the scores on the five dimensions of satisfaction with respect to these clusters. These scores lead to the labels “Poor 1” “Moderate 2” “Moderate 3” and “Good 4” being assigned to the clusters. The clusters display statistically significant relationships with a number of the exogenous variables including the value of the contract and the age of the technology involved. Relationships with the exogenous qualitative data are indicative of the validity of the clusters. Originality/value – This paper takes a novel approach to gaining an understanding of relationships through the use of hierarchical cluster analysis. This provides an elegant way of exposing the influences on relationship satisfaction at a disaggregate level which are not possible by taking an aggregate approach. This will be of particular interest to researchers who are seeking patterns in large data sets and practitioners who can identify better practice guidelines when working within supply chain relationships. The disaggregate approach using cluster analysis provides extraordinarily detailed insights into relationship patterns.