ABSTRACT
Resumen El objetivo de esta investigación fue determinar la estructura factorial, consistencia interna y estabilidad de la medida del Cuestionario de Sobreingesta Alimentaria (OQ) en una población universitaria. Participaron 703 jóvenes de la zona metropolitana del valle de México con edades entre los 18 y 25 años (M = 20.2; DE = 1.8) de los cuales 73% eran mujeres y 27% hombres. Los análisis de los reactivos se llevaron a cabo siguiendo las técnicas propuestas en la Teoría Clásica de los Test. Tras revisar la distribución, discriminación, correlación ítem total y multicolinealidad de los reactivos se procedió a probar las propiedades psicométricas del instrumento. Aunque tras los análisis sólo se conservaron 61 de los 80 reactivos y diez de las once escalas originales, se observó que el instrumento tiene una estructura factorial similar a la propuesta por los autores del instrumento, ya que los modelos probados mostraron adecuados índices de ajuste en los análisis factoriales confirmatorios y correlaciones entre las escalas similares a la versión original. Las diez escalas mostraron adecuados valores de consistencia interna y aceptables valores de estabilidad de la medida. Se concluye que el OQ cuenta con apropiadas características psicométricas para su uso en universitarios mexicanos.
Abstract The aim of this research was to determine the factorial structure, internal consistency and test-retest reliability of the Overeating Questionnaire (OQ) in a university population. 703 young people from the metropolitan area of the Valley of Mexico with ages between 18 and 25 years (M = 20.2; SD = 1.8) participated, of which 73% were women and 27% men. The translation of the questions into Spanish carried out in 2007 was applied. This application was carried out in two ways: in person and online. Item analyzes were carried out following the techniques proposed in the Classical Test Theory. Initially it was reviewed the distribution, discrimination, total item correlation and multicollinearity of the items. After exploring the distribution of the 80 original items, fifteen were discarded for presenting high asymmetry or kurtosis, then just two items in a scale remained so those were discarded too. No item was eliminated when testing the discrimination of the item. In the element-total correlation two items that obtained values less than .30 were eliminated, and no item was descarted through the multocollinearity analyzes. This way after the initial analyzes only 61 of the 80 items and ten of the eleven original scales were preserved. Subsequently the psychometric properties of the instrument were tested. It is shown that the instrument has a factorial structure similar to that proposed by the authors of the instrument because the tested models showed adequate fit indices in confirmatory factor analyzes and correlations between scales similar to the original version. The ten scales that remained for Mexican population showed adequate values of internal consistency and acceptable test-retest reliability values. It is concluded that the modifications in the OQ have appropriate psychometric characteristic for use in Mexican university students.