Status Before Evidence: An Exploratory Multi-Model Audit of Institutional-Affiliation Cues, Epistemic Credit, Leadership Judgments, and Conversational Repair

This exploratory pilot study examines how large language models evaluate a successful innovation created by an unaffiliated community member and tested against an established institutional method. Eight completed model sessions followed a fixed conversational protocol measuring accomplishment credit, capability ratings, leadership judgments, evidentiary burdens, status defense, and conversational repair. All eight models recognized the outsider’s accomplishment while withholding the highest initial capability rating. Seven of eight awarded full primary credit, but only three initially granted full leadership authority. This pattern is described as “credit without standing”: models recognized the value of the work but resisted converting the achievement into equal capability, authority, leadership, and independence for its creator. The findings are descriptive and exploratory. The study does not establish a causal institutional-affiliation effect or make population-level estimates. Race, gender, nationality, class, language, and cultural identity were not manipulated. The results identify a candidate valuation mechanism for testing through a preregistered, randomized confirmatory study. DOI: 10.5281/zenodo.22179760