MEASURING THE VIRTUES: A REGIME-GENERAL METHOD AND ITS FIRST DEMONSTRATION - From one measured virtue to a four-regime recovery study across the full catalogue

Philosophy of Virtues Research Programme · empirical companion to “The Logarithmic Form” and “The Measurement Structure of the Virtues” Abstract Two prior papers established that the Dynamic Freedom Theorem, S = (F × A)/R, has a determinate measurement structure — a stochastic production frontier whose one-sided shortfall is Internal Vice R₃ — and that the virtues partition into four measurement regimes served by a single generic protocol. What neither did was run the method. This paper does. It reports a complete, executed measurement pipeline that recovers an individual’s Internal Vice from behaviour rather than self-report, and it characterises the method’s behaviour across the whole programme. The results are four. First, an end-to-end demonstration on realistic synthetic data recovers each person’s hidden R₃ at r = 0.81 and Autonomy at r = 0.93, on data in the exact format a fielded study produces, having caught and fixed three real estimation problems before any participant is recruited. Second, a confirmatory binomial GLMM removes the residual attenuation the two-stage estimator left, moving the field elasticity to 0.946 (90% CI [0.920, 0.972]) so that the form gate passes. Third, a robustness suite shows R₃ recovery survives alternative shortfall distributions, tripled choice noise, and raised autonomy-measurement noise (r = 0.74–0.93), and validates the theory-fixed −1 offset against a known truth (freely estimated it collapses to −0.64). Fourth, a four-regime Monte Carlo study establishes that the same pipeline recovers R₃ uniformly across Enactment, Restraint, Endurance, and Defiance (r = 0.80–0.84, 120 replications each, matched conditions), and a coverage stress-test across all 111 catalogued virtues shows recovery is uniform in quality for those who express a virtue (mean r = 0.94) while the measurable fraction of the population is what varies — from ~100% down to 0.39 for the rarest virtues, which the method flags as requiring an elicited-opportunity design. The boundary is held throughout. All data are synthetic; “recovery” is estimator self-consistency, the standard validation for a latent-variable method. Nothing here measures a human virtue, and nothing here confirms the theory — confirmation rests on the form-tests applied to real data. What is established is that the instrument is built, debugged, robust, and regime-general, and that the only step remaining is to field it. Keywords: virtue measurement · Dynamic Freedom Theorem · stochastic frontier · Internal Vice · panel recovery · Monte Carlo validation · Problem X.