Convergent Validity
0ConceptWhether metrics that should be related are actually related, and metrics that should not be related are not.
Last updated: 2026-09-01Spec: SigRank Standard v1.0 (proposed)
Definition
Whether metrics that should be related are actually related, and metrics that should not be related are not.
Inputs
Correlation matrix of all metrics across a population of operators.
Derived variables
Inter-metric correlations, correlation structure.
Claim
If metrics have convergent validity, related metrics (e.g., yield and leverage) should correlate, while unrelated metrics should not.
Test
Compute the full correlation matrix. Compare observed correlations to theoretically expected relationships.
Observable
Correlation coefficients, correlation matrix structure.
Falsifier
Related metrics do not correlate, OR unrelated metrics correlate strongly.
Evidence
Field data shows expected correlation structure (yield correlates with leverage and velocity).
Limitations
Correlation does not imply causation. Small samples produce unstable correlations.
Lineage
Psychometric validation methodology, SigRank field data. Architecture: mos2es.com/architecture.