L'analyse implicative bayésienne : une méthode pour l'étude des dépendances orientées. 2. Modèle logique sur un tableau de contingence
In Bernard & Charron (1996) we proposed a new method, the Bayesian Implicative Analysis, for the study of oriented dependencies between two binary variables, method which enables to conclude in terms of quasi-implication between modalities of the variables. In this article, we extend the method...
Uloženo v:
| Vydáno v: | URI:https://journals.openedition.org/msh, |
|---|---|
| Hlavní autoři: | , |
| Médium: | Article ou chapitre numérique |
| Jazyk: | Français |
| Vydáno: |
Mathématiques et sciences humaines
2006
|
| On-line přístup: | Accès Université d'Orléans et IFPM Accès Université d'Orléans et IFPM |
| Shrnutí: | In Bernard & Charron (1996) we proposed a new method, the Bayesian Implicative Analysis, for the study of oriented dependencies between two binary variables, method which enables to conclude in terms of quasi-implication between modalities of the variables. In this article, we extend the method to the general case of an A < B contingency table, whth the problem of measuring the degree of quasi-conformity of the data to a given logical model. At the descriptive level, the method is based on the "Del" index, proposed by Hildebrand et al. (1977), which generalises Loevinger's index H. Here again, the inductive step is envisaged within the Bayesian framework. |
|---|