Practitioner articles on insurance expert bias, published in Advocate Magazine
Two articles trace the framework this site is built on: the 2017 article that named the “expert safe harbor” problem, and the 2024 article that identified Demer’s standards, factors, and presumptions as the answer. Each is reproduced in full on its own page, with the publisher’s permission.
Advocate Magazine · July 2024
The Ninth Circuit’s 2016 decision in Demer v. IBM Corp. LTD Plan supplied the first comprehensive framework for evaluating expert bias in insurance claims — the inference-of-bias standard, a four-factor bias calculus, and a rebuttable presumption that shifts the burden to the insurer. This article argues the framework applies equally to all insurance claims, ERISA and non-ERISA alike, and analyzes Bagramyan v. Gov’t Employees Ins. Co. (2023).
Advocate Magazine · September 2017 · with Evangeline Fisher Grossman
How the genuine-dispute doctrine, extended to factual disputes by Fraley v. Allstate, produced a near-absolute “expert safe harbor” defense — and why Chateau Chamberay’s biased-expert exceptions failed to rein it in. The article maps the three structural reasons insureds lose on bias, and the discovery pathway through other insureds’ claim files.
What Has Evolved
The 2017 article identified the problem. The 2024 article named the framework. Since then, the analysis has continued to develop along two fronts.
The first is the Demer+ extension. The fourth factor in the Demer paradigm — the insurer’s reasonable measures to safeguard expert impartiality and reliability — is expanding from a retrospective inquiry into an affirmative, structural obligation. The question is not only whether the insurer’s expert was neutral in practice, but whether the insurer’s selection, vetting, and oversight process was structurally designed to produce impartial results. That shift reorients the entire theory of liability and is the least-developed area in the current case law.
The second is AI. Automated claims-decision systems are the structural successors to biased human experts. They produce coverage outcomes at scale, without reviewable methodology or cross-examination. The reasonable measures duty applies to AI-driven claims decisions as it applies to human expert retention — and the California legislation now pending before the Assembly Insurance Committee is designed to delete the records that would prove AI-driven bias before litigation can begin.
The 14-chapter e-treatise — drawing from more than 1,000 reported decisions across all 51 jurisdictions — systematizes the full body of doctrine. It is in final preparation and will be available to paid subscribers of Expert Bias Report: Insurance Claims.