A metric for the asymmetry in matched-pair data for buyer–supplier dyads
本文提出一种基于马氏距离的多维不对称性度量指标及显著性检验方法，用于识别买方-供应商配对数据中显著不对称的配对，并展示其在实际零售数据集中的应用效果。
Although various difference-based methods are utilized to analyze asymmetry in buyer–supplier matched-pair data within the literature, these approaches are ad hoc and do not always address differences across multiple dimensions. Furthermore, they do not provide a significance test. This paper extends the concept of the paired t-test for dyad-level differences by developing a Mahalanobis distance-based metric in multiple dimensions, along with a significance test. The metric and the significance test can be used in empirical research to identify dyads in a dataset that are significantly asymmetric at any selected confidence level. In practice, the method can identify those suppliers for a buyer that have significantly mismatched expectations relative to other suppliers. The paper utilizes simulated datasets to compare the proposed metric with other distance-based metrics that lack a significance test. Finally, the paper applies a retail dataset to demonstrate (1) the utility of the metric in identifying significantly asymmetric dyads and (2) the use of the same distance concept to consolidate multiple items in any buyer or supplier construct into a single score for the construct, rather than using factor scores. The latter approach is lossless, in contrast to factor analysis. Using distance-based metrics with this retail dataset in a structural equation model suggests that asymmetry can negatively affect relationship-specific operational performance for buyers and suppliers. This study contributes a robust methodological framework, offering a structured basis for future research in the measurement of dyadic asymmetry.