Psychometric evaluation of the big five questionnaire for children (BFQ-C): A rasch model approach

Abstract

The Big Five Questionnaire for Children, (BFQ-C) is an instrument for personality assessment in children and adolescents widely used worldwide. The aim of this work was to study the psychometric properties of the instrument scores from the item response theory (IRT) perspective. We worked with a Partial Credit Rasch Model to analyze an Argentinean sample to validate the scale for its use in this population. We opted for an instrumental design, and for each factor we applied an item calibration plan consisting of different analysis: unidimensionality, classification of response categories, fit levels of items and persons, specific objectivity, and differential item functioning as regards sex. We worked with a sample of 1162 high school students aged 12–17 years. The five original subscales did not show satisfactory fit, so modifications were made to improve their properties. As a result, we could demonstrate that each subscale measures a single latent trait, meets the invariance assumption regarding the sample and the assumption of local independence, showing no sex differential item functioning (DIF). Finally, the ordinal scores were converted to an interval scale, which allows more accurate analysis and better confidence in outcomes. Our results showed that the five subscales corresponding to each factor were in line with the IRT key parameters, although we suggest further studies on both the test capacity to assess extreme scores and the relevance of using a five-response category scoring.