Background: The Beck Depression Inventory (BDI) has demonstrated excellent psychometric properties and good performance as a screening measure in different contexts and languages. However, comparison of its structure across countries and languages remains understudied. Measurement invariance is a prerequisite for considering the BDI equivalent across versions, and for using it to make valid and interpretable comparisons of the severity of depression among different groups.
Methods: As part of a five-country (UK, Ireland, Spain, Norway, and Finland), two-stage epidemiological study of depressive disorder, 7934 persons were screened using the BDI. The item equivalence and measurement invariance of the BDI across the samples of the five countries was tested using Item Response Theory (IRT) and Multiple Indicators Multiple Causes (MIMIC) models.
Results: Overall results support the factorial validity of the BDI, with a unidimensional structure. Item 19 (weight loss) presented a clear misfit in the five countries. IRT models, as well as MIMIC models, suggest that complete measurement invariance cannot be assumed across the five countries. The Spanish sample accounted for the majority of the differences, with a moderate to low Differential Item Functioning for the other countries on the different items.
Conclusions: The BDI could be used cross-culturally in Europe, with particular cautions regarding the Spanish sample, and with the constraints and limits pointed out for the present results. The methodology used for the present work is suggested as a soundness approach for testing the cross-cultural validity of severity rating scales.