counterfactual-reasoning

Define the target variable (the act or omission) and the outcome variable, lock all background conditions, then construct a counterfactual world where only the target variable changes. Compare outcomes across both worlds to determine whether the variable caused the result, and annotate each inference step with a confidence level.

The But-for test asks whether the outcome would have occurred without the act, and works for single-cause scenarios. The NESS test checks whether the act is a necessary element of a sufficient condition set, and is used for overdetermined multi-cause cases where But-for fails.

It constructs a counterfactual world where proper diagnosis and treatment occurred, then uses medical evidence to estimate the probability of avoiding harm. Probabilistic results are handled through the loss-of-chance theory, translating the lost survival or recovery probability into a liability proportion.

It should not be used for pure legal interpretation questions with no factual causation component, when the target variable cannot be meaningfully altered, when no reliable causal knowledge supports the inference, or when the law prescribes an alternative causation framework.

Each inference step receives a rating from CF-HIGH to CF-LOW based on the reliability of its causal basis, and the overall conclusion takes the lowest rating in the chain. The rating is then checked against the applicable proof standard, such as high probability for civil cases or beyond reasonable doubt for criminal cases.