Chapter 6 Classical Inference

(Berger 1985, 16) and (Paulino, Turkman, and Murteira 2003, 6) indicate that procedures based on the classical paradigm are based on some principles, such as maximum likelihood, non-bias21, minimum variance, least squares, consistency, sufficiency and efficiency. The classics consider that there is an unknown parameter \(\theta\) for which no probabilities can be assigned. The sample is obtained randomly from a universe of interest, being one of many - if not infinite - possible samples. (Berger 1985, 26) points out that this principle places the classics as unconditionalists, as it considers all possible data and is not conditioned on what was observed. Therefore, it is important to clarify the concept of conditional independence, discussed in Section 3.4.2.

There are three basic forms of classical approach, detailed below.
- Point estimation
- Confidence Interval (CI/CoI)
- Hypothesis Test (HT)


Berger, James O. 1985. Statistical Decision Theory and Bayesian Analysis. 2nd ed. Springer Science & Business Media.
Paulino, Carlos Daniel Mimoso, Maria Antónia Amaral Turkman, and Bento Murteira. 2003. Estatı́stica Bayesiana. Fundação Calouste Gulbenkian, Lisboa.

  1. Definition 6.3.↩︎