Substitutability assumptions can be tested by building rival aggregation measures.

Evidence

Our plan is to replicate Russett and Oneal’s (2001, 316) logistic regression analysis on the probability of militarized interstate disputes and to compare the performance of the original weakest-link measures to three alternative aggregate measures reflecting different degrees of substitutability. In addition to the weakest-link minimum measure, our analyses encompass the limited substitutability entailed by the geometric and arithmetic means as well as the full substitutability implied by the maximum. If the weakest-link hypothesis is correct we should observe two empirical outcomes. First, the minimum measures of dyadic democracy and trade interdependence should outperform all others by carrying the strongest impact on the probability of disputes. Second, we should observe a discernible pattern of decreasing performance across measures as substitutability increases. — Gary Goertz, Social Science Concepts, 912

Categorization

• ADDENDUM: Directly parallels the published piece on sacrifice as a stabilizer of stratified societies: there, weakest-link logic governs whether a ritual system is a dyad’s binding constraint; here the minimum measure operationalizes the same ‘one weak link governs the pair’ intuition.