Abstract

Measuring temperament is an important, yet challenging matter. The Children Behavior Questionnaire (CBQ) is a widely used measure, yet its psychometric properties have not been established in Latin America, and few studies have analyzed its structure at the item level. We analyzed the factor structure and reliability of the CBQ-short form in 998 Chilean children (M-age = 5.95 years) in central Chile. Confirmatory factor analysis revealed that the 15 factors proposed by the theory unsatisfactorily fit the data (CFI = .58, TLI = .56, RMSEA = .046, SRMR = .080), and reliability was lacking (range of alpha = .30-.74). We extracted the 36 items of the CBQ-vsf and it performed better (CFI = .82, TLI = .80, RMSEA = .078, SRMR = .074). Exploratory analyses suggested that the surgency factor was composed of two latent variables, and separating them generated a model with better validity and reliability (CFI = .77, TLI = .75, RMSEA = .076, SRMR = .077, range of alpha = .68-.77). We suggest the CBQ-vsf provides more validity, reliability and parsimony than the CBQ-sf. Finally, we discuss the functioning of the CBQ in Chilean culture and child-rearing patterns, and issues with the wording of questions in the Spanish translation.