Treatments for undefined log ratios in matching analyses [r-libre/3206]

Caron, Pier-Olivier (2024). Treatments for undefined log ratios in matching analyses. Journal of the Experimental Analysis of Behavior, 122 (1), 52-61. https://doi.org/10.1002/jeab.925

Fichier(s) associé(s) à ce document :

PDF - Caron(2024)_JEAB.pdf
Contenu du fichier : Version de l'éditeur

Télécharger

Catégorie de document :	Articles de revues
Évaluation par un comité de lecture :	Oui
Étape de publication :	Publié
Résumé :	A challenge in carrying out matching analyses is to deal with undefined log ratios. If any reinforcer or response rate equals zero, the logarithm of the ratio is undefined: data are unsuitable for analyses. There have been some tentative solutions, but they had not been thoroughly investigated. The purpose of this paper is to assess the adequacy of five treatments: omit undefined ratios (OMIT), full information maximum likelihood (FIML), mean divided by 100 (M/100), constant 1/10 (1/10) and constant .50 (CONS). Based on simulations, the treatments are compared on their estimations of variance accounted for, sensitivity and bias. The results show that FIML and OMIT had the best overall performance with negligibly biased and more accurate estimates than M/100, 1/10 and CONS. The study suggests that M/100, 1/10 and CONS should be avoided and recommends FIML to deal with undefined log ratios in matching analyses.
Adresse de la version officielle :	https://onlinelibrary.wiley.com/doi/full/10.1002/j...
Déposant:	Caron, Pier-Olivier
Responsable :	Pier-Olivier Caron
Dépôt :	06 juin 2024 13:09
Dernière modification :	02 juill. 2024 15:46

Actions (connexion requise)

RÉVISER