Confirmatory Factor Analysis (CFA) of the CBCL/6-18 Syndrome Structure in Turkish Samples

The correlated 8-factor model on which the 2001 CBCL syndromes are based was derived from English language data on children from Australia, the United Kingdom, and the United States.

Although research with the CBCL has been published from 62 cultures, there have been few tests of the 2001 correlated 8-factor model with large representative samples of children from non-English-speaking cultures. To test the 8-factor model with CBCL data from a very different cultural/linguistic group, a team of Turkish and American researchers performed confirmatory factor analyses (CFA) on CBCL data from the following three Turkish samples (Dumenci, Erol, Achenbach, & Simsek, 2004): (a) A nationally representative sample of 5,195 Turkish 6- to 18-year-olds whose parents completed the CBCL in a home interview survey using the same data collection procedures as were used to obtain the U.S. national probability sample; (b) a clinical sample of 963 6- to 18-year-olds referred to mental health services in Ankara, Turkey; and (c) a combined sample that included 3,498 children from the national and clinical samples whose Total Problems scores were at or above the age/gender-specific medians of the Turkish national sample.

CFA were performed on the tetrachoric correlations between items in each of the three samples. As a measure of how well the Turkish data fit the CBCL 8-factor model, the Root Mean Square Error of Approximation (RMSEA) was .041 for the Turkish national sample, .057 for the clinical sample, and .054 for the combined sample. All three RMSEAs were well within the range that indicates good fit. For the sake of comparison with the simplest possible model, RMSEAs were also computed for a single-factor model for each sample.

All three RMSEAs were larger than the ones obtained for the 8-factor model. In addition, 99% of the items in the Turkish sample had significant (p<.01), positive, and substantial loadings on the same syndrome as in the 8-factor model. Thus, despite major cultural, linguistic, and religious differences, problems reported by Turkish parents on the CBCL formed syndromes like those embodied in the 8-factor model derived from CBCLs completed for children in Australia, the United Kingdom, and the United States.

Reference: Dumenci, L., Erol, N., Achenbach, T.M., & Simsek, Z. (2004). Measurement structure of the Turkish translation of the Child Behavior Checklist using confirmatory factor analytic approaches to validation of syndromal constructs. Journal of Abnormal Child Psychology, 32, 337-342.