Determine whether score-variance reduction improves ranking accuracy

Determine whether the reduction in pooled-score variance predicted by the MCES diversity argument actually produces improved causal-edge ranking accuracy relative to individual methods across data-generating settings.

Background

The theoretical analysis proves that lower cross-method correlation can reduce the variance of the weighted Convergent Evidence Score under stated assumptions. However, lower variance alone does not guarantee that true driver–outcome pairs will be ranked more accurately, because weaker component methods may add noise or dilute signal.

The experiments show that the ensemble does not automatically outperform the best individual method. Thus, the remaining problem is to characterize when the theoretical stability benefit translates into a ranking-accuracy benefit.

References

The argument in this subsection is a conditional, theoretical one: it states when pooling reduces the variance of the score, given low cross-method correlations. Section~\ref{sec:diversity} measures the proposition's own correlation quantity directly, across repeated draws of the data-generating process for fixed pairs, and finds it low ($\bar\rho \approx 0.13$); the step the theory does not supply, and the experiments do not automatically deliver, is from reduced score variance to improved ranking accuracy.

— Multi-Method Causal Evidence Synthesis: Ranking Candidate Drivers by Convergent Cross-Method Evidence from Observational Data  (2608.20187 - Gupta et al., 20 Aug 2026) in Section 4.1, “Why Pooling Could Help: A Score-Stability Argument”; Section 5.3, “Method Diversity”