Inference concept

A concept of inference is a statistical concept that justifies the conclusion of observations on hypotheses. Well-known inference concepts are classical inference, likelihood inference, Bayesian inference and statistical decision theory (decision-theoretic inference). Fidical inference, structure inference and pivot inference are less important.

Concepts of inference can be characterized by the purpose of the conclusion, the elements of the model used, and the quality of the conclusion.

• In a cognitivist Inferenzkonzept is the purpose of the conclusion to draw lessons (classical inference, Likelihoodinferenz, Bayesian inference), while the conclusion in a decisionistic Inferenzkonzept making decisions is (decision-theoretic inference).
• If conclusions from observations are based on a model that contains only objective elements, one speaks of an objectivistic concept of inference (classical inference, likelihood inference). In contrast to this, subjective premises may also be included in a subjectivistic inference concept, e.g. B. the so-called a priori knowledge (Bayesian inference, decision-theoretic inference).
• If the quality assessment of a conclusion is based on how often it leads to a correct statement on average, if it is applied to many different observations, it is a frequentistic inference concept (classic inference). In a non-frequencyistic inference concept, on the other hand, a conclusion is judged according to how plausible it is with regard to a given observation (likelihood inference, Bayesian inference, decision-theoretic inference).

literature

• B. Rüger: Test and estimation theory. Volume 1: Basics. Oldenbourg, Munich 1999, ISBN 3-486-23650-4 .