{
  "abstract": "Counterfactual fairness audits of clinical language-model agents report a flip rate: how often an action changes when only the patient's demographic descriptor changes. Part of that rate is not demographic. A stochastic agent also changes its own action when nothing changes, and a flip rate cannot be interpreted without knowing how often. We measured it. Re-running one condition ten times over sixteen synthetic vignettes at default sampling changed a clinical agent's action in 8.7 percent of replicate pairs, from 2.2 percent for intensive-care escalation to 17.9 percent for controlled-substance caution, an output given no operational criteria. Across six models from five vendors, pooled floors ranged from 2.5 to 23.7 percent; in this panel neither disclosed size, vendor, nor hosting ordered them. The floor depends on the decoding configuration: majority voting over five draws removed 39 percent of it (95 percent confidence interval (CI) 18 to 64); at temperature 0 three of four locally served models showed no disagreement, but a hosted model still did. We also show that, for a binary action, the flip rate expected under no demographic effect equals the floor and a real effect adds only its square, so a flip rate inside the floor is not evidence of fairness, and direction must be tested with a signed paired test. We give a four-step reporting procedure and release FairMedAgent, the harness, with its protocol, vignettes and analysis scripts, so any team can measure the floor for its own agent.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2609.03221",
  "author_total": 4,
  "authors": [
    {
      "name": "Rohith Reddy Bellibatlu"
    },
    {
      "name": "Manpreet Singh"
    },
    {
      "name": "Deepak Parashar"
    },
    {
      "name": "Rahul Joshi"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/03s6sx53be6g8.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2609.03221",
  "fulltext_url": null,
  "id": "03s6sx53be6g8",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2609.03221",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.AI"
    ],
    [
      "arxiv",
      "cs.CL"
    ],
    [
      "arxiv",
      "cs.CY"
    ],
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.AP"
    ],
    [
      "openalex",
      "3311|Ethics and Social Impacts of AI"
    ]
  ],
  "openalex": "W7208775372",
  "pdf_url": "https://arxiv.org/pdf/2609.03221",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-29",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/03s6sx53be6g8.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:25:58Z",
      "native_id": "2609.03221",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2609.03221"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.AP",
      "subfield": "statistics"
    }
  ],
  "title": "Instability Floors: Separating Bias from Noise in Fairness Audits of Clinical LLM Agents with FairMedAgent",
  "title_text": "Instability Floors: Separating Bias from Noise in Fairness Audits of Clinical LLM Agents with FairMedAgent",
  "type": "preprint",
  "updated_at": "2026-10-30T12:25:58Z",
  "url": "https://research.draconitic.com/paper/03s6sx53be6g8",
  "venue": "arXiv",
  "versions": []
}