Inference is the movement from premises to a conclusion. Three kinds are usually distinguished. Deductive inference aims at validity: the conclusion cannot be false if the premises are true, and the formal properties of that relation are well understood and uncontroversial. Inductive inference generalises beyond the evidence, and abductive inference moves to the explanation that would best account for it; neither guarantees its conclusion. What makes a non-deductive inference a good one — what licenses the step, and how to grade the support evidence gives — is precisely the question that remains open, and rival answers belong to different programmes in logic, statistics and epistemology. The formal core is settled; the part that does most of the work in practice is not.
Why this entry carries no source list. This lexicon cites an official primary source wherever one exists. For a concept used across several traditions at once, no such source exists: there is no body with authority over it, and a signed academic reference — the tier this lexicon would otherwise reach for — has no entry under this headword. Citing one school’s formulation would present a position as a definition. So the entry names the senses in play and the distinction between them, which is the part a reader can actually use.
A conclusion drawn by reasoning from evidence rather than direct observation; may be inductive or deductive.
Also defined as
A conclusion reached by reasoning from evidence — deductive inference guarantees its conclusion, inductive makes it probable, abductive picks the best explanation; statistical inference generalises from a sample to a population, using probability to quantify the uncertainty involved.