This paper examines whether the top curriculum candidate under uncertainty-aware evaluation also remains the most robust contender once ranking stability is taken into account. Using two evaluation campaigns centred on the 2005 Civil Engineering curriculum at FACET-UNT, the study analyses the relationship between confidence-adjusted return (CAR) and Winner Stability (WS) within the APSV2 evaluation logic. The central question is whether the candidate ranked first in adjusted comparative performance also coincides with the candidate exhibiting the strongest robustness profile under perturbation. Campaign 001 first revealed a substantive divergence between expected comparative performance and ranking robustness. The candidate ranked first by CAR, lowfragmentationₛearch, did not coincide with the candidate showing the strongest stability profile, and the final interpretive outcome remained a conditional leader rather than a robust winner. Campaign 002 then intensified uncertainty through a stronger macro-shock regime in order to test whether this divergence would collapse or reverse. The expected reversal was not supported. lowfragmentationₛearch retained rank 1 under CAR, but clusterfirst remained markedly stronger in Winner Stability (0. 9583 vs. 0. 7500). The divergence between performance leadership and robustness leadership therefore persisted under stronger uncertainty. The contribution of the paper is methodological and evaluative rather than programmatic. It shows that, in curriculum evaluation under uncertainty, expected performance and robustness are analytically distinct and should not be conflated. Even when the same candidate remains first under a harsher uncertainty regime, the strength of the final claim must still be calibrated by the robustness profile of the ranking itself. The paper therefore supports a more disciplined interpretation of curriculum superiority in simulation-based comparative evaluation.
Hugo Roger Paz (Sat,) studied this question.