Psicothema, 2012. Vol. Vol. 24 (nº 1). 127-132

Antonio del Pino Pérez¹, Ignacio Ibáñez Fernández¹, Francisco Bosa Ojeda², Ruth Dorta González¹y María Teresa Gaos Miezoso¹

El objetivo del estudio era validar en una muestra de 205 enfermos coronarios un modelo factorial para el BDI-II, especialmente un modelo que permitiera modelar los síntomas depresivos tras eliminar explícitamente el sesgo asociado a los síntomas somáticos que pueden confundirse con síntomas de la enfermedad. Se realizaron análisis factoriales exploratorios y confirmatorios para datos ordinales. Se analizan un modelo monofactorial, seis modelos con dos factores correlacionados y, derivados de éstos, siete modelos con un factor general y dos factores no correlacionados. El análisis exploratorio aísla dos factores, somático-afectivo y cognitivo. En los análisis confirmatorios, el modelo monofactorial obtiene el peor ajuste. Los modelos bifactoriales son superados en bondad de ajuste por los modelos de factor general y de grupo. Entre éstos destaca el modelo General, Somático-afectivo y Cognitivo (G-Sa-C) de Beck con estudiantes. El peor el General, Somático, Cognitivo (G-S-C) reducido de Ward. Nuestro modelo supera los puntos de corte de todos los índices de ajuste. Se concluye que la inclusión de factores generales y de grupo en todos los modelos supera los resultados del modelo G-S-C y, por tanto, lo cuestiona. El modelo G-Sa-C resulta fortalecido.

Factor models of the Beck Depression Inventory-II. Validation with coronary patients and a critique of Ward’s model. The objective of this study was to validate in a sample of 205 coronary patients a factor model for the BDI-II, especially a model that would allow for modeling of depressive symptoms after explicitly removing bias related to somatic symptoms of depression that would overlap those of heart disease. Exploratory and confirmatory factor analyses for ordinal data were conducted. A one-factor model, six correlated two-factor models and, derivatives thereof, seven models with a single General Depression factor and two uncorrelated factors, were analyzed. Exploratory analysis extracted two factors, Somatic-affective and Cognitive. Confirmatory factor analyses showed the worst fit for the one-factor model. Two-factor models were surpassed in goodness of fit by the models of general-factor and group factors. Among these, the General, Somatic-affective and Cognitive (G-Sa-C) model of Beck with students is noteworthy. The reduced General, Somatic and Cognitive (G-S-C) model of Ward showed the worst goodness of fit. Our model surpasses the cutoff criteria of all fit indexes. We conclude that the inclusion of a general-factor and group factors in all the models surpasses the results of G-S-C model and, therefore, questions it. The G-Sa-C model is strengthened.