Diagnostic in normal linear regression:
principles and applications

 Artur José LEMONTE[1]

§    ABSTRACT: Statistical models are very used to extract and knowing the characteristics of a data set. Models, however, are always approximate descriptions of a process more complicated. So considerations of quality of a model are extremely important. We present in this article some techniques of diagnostic in normal linear regression model, which can be seen as methods to study the variation in the analysis of the results when the model in the study is subject to some kind of disturbance. Additionally, we present in detail the method of local influence developed by Cook (1986, 1987).

§    KEYWORDS: Local influence; measures of diagnostic; normal linear model.


[1] Departamento de Estatística, Universidade de São Paulo – USP, Rua do Matão, 1010, CEP: 05508-090, São Paulo, SP,  Brasil. E-mail: arturlemonte@gmail.com