General Causation vs. Specific Causation

Definition

General causation asks a population-level question — is this agent capable of causing this disease or injury at all? — and is typically answered with group-based evidence such as epidemiological studies. Specific causation asks whether the agent actually caused this disease in this particular plaintiff, and requires ruling out the other known, competing causes of that plaintiff's condition. A plaintiff who cannot establish general causation cannot get to specific causation, no matter how compelling their individual story: the two are sequential, not interchangeable, hurdles.

In the Book

The Epidemiology reference guide states this directly: "epidemiology focuses on the question of general causation (i.e., is the agent capable of causing disease?) rather than that of specific causation (i.e., did it cause the disease in a particular individual?)." The manual's preface, written by the committee co-chairs, flags this as one of two issues "germane to the interpretation of all scientific evidence," noting that judges are asked to rule on both general causation and specific causation and that "concepts of causation familiar to scientists... may not resonate with judges" who must decide without the option of waiting for more data. The book grounds the distinction in case law: courts have permitted a toxicologist to testify to general causation but not specific causation, while an epidemiologist testifying to both was allowed in Landrigan v. Celotex Corp. The manual is explicit that expert testimony resting on a differential-etiology method (see the companion concept) is worthless for specific causation unless general causation has independently been established first — an expert must "rule in" the suspected cause via general-causation evidence, not just "rule out" alternatives.

Why It Matters

This distinction separates a question about mechanisms-in-general from a question about this-instance-in-particular, and conflating them is a common reasoning error whenever aggregate evidence gets applied to an individual case: knowing that a policy works "on average," that a drug is effective "in trials," or that a factor is a "risk" tells you nothing yet about whether it explains one specific outcome without a second, separate argument. Any domain that moves between population statistics and individual diagnosis — medicine, insurance underwriting, algorithmic risk scoring, root-cause analysis of a single incident — needs this two-step structure to avoid smuggling a population claim into an individual conclusion.