FROM LIKELIHOOD TO LEGAL PROOF: A BAYESIAN SEPARATION PRINCIPLE FOR EXPERT EVIDENCE UNDER INDIAN LAW

Statistical and scientific evidence increasingly enters Indian adjudication through DNA profiling, electronic records, algorithmic outputs, medical studies, and other quantitative methods. The principal legal difficulty is not whether such evidence is probabilistic, but how a specialised probabilistic conclusion is translated into legal proof. A source-level DNA conclusion may be treated as an offence conclusion; a likelihood ratio may be mistaken for a posterior probability; an authenticated electronic message may be treated as proof of human authorship; and dependent findings may be counted as independent corroboration. This article develops a proposed Bayesian Separation Principle for expert evidence under Indian law. The Principle is not a new reliability test or a rule of numerical adjudication. It is an interpretive doctrine governing the allocation of inferential responsibility across four stages: proposition specification, comparative evidential assessment, judicial integration, and legal decision. Drawing on the Bharatiya Sakshya Adhiniyam, 2023, Indian Supreme Court jurisprudence, Bayesian evidence scholarship, and European forensic reporting guidance, the article explains what the Principle adds to familiar expert-evidence doctrine and why the label is analytically useful. A compact six-question protocol translates the Principle into judicial practice.