Validação das versões reduzidas do inventário de externalização em uma amostra universitária e clínica

Abstract

Studies in Psychophysiology, psychometry and behavioral genetics indicate that uninhibited personality traits and psychiatric disorders such as substance use and antisocial behavior are explained by a common latent factor called Externalization. This construct has been empirically investigated by means of the Externalizing Inventory originally composed of 415 items and, later, in its two reduced versions, 57 and 160 items respectively. The aim of this study is to assess the psychometric qualities of those Inventory reduced versions. The method used for assessing internal consistency was Cronbach´s Alpha and, for the validation study, the Exploratory and Confirmatory Factor Analysis. A satisfactory level of reliability was detected for the two versions. Yet, as regards the factor structure, the model with a better adjustment for the 415-item Inventory, unlike the three-factor original model, was a fourfactor solution. The 160-item Inventory produced indexes of model adequacy below that which is recommended in the specialized literature, no possible reliable factor having been extracted from it. A few variables observed (items and scales) correlated with more than one factor, thus making the model confusing and not so parsimonious. Results point to a need to reformulate the items in both reduced versions, in order to make the underlying trait more evident.