Taiwanese depression questionnaire revisit: Factor structure and measurement invariance across genders

• Main Authors: Yi-Chun Sun, Chien-Lin Chen, Shu-Hui Wen
• Format: Article
• Language: English
• Published: Elsevier 2019-09-01
• Series: Journal of the Formosan Medical Association
• Online Access: http://www.sciencedirect.com/science/article/pii/S0929664617304862

Background/purpose: Taiwanese Depression Questionnaire (TDQ) was widely used in Taiwan to measure individuals' proneness to depression. In light of well-established gender differences in the manifestation of depression, TDQ's structure invariance and measurement invariance across genders were examined in this study. Methods: Data from 2604 adults was analyzed. Exploratory factor analysis was first performed to explore the factor structure. Possible factor structures and the original 2-factor structure were then compared by confirmatory factor analysis (CFA). The best fitting model was tested for measurement invariance across genders using multi-group CFA. Results: EFA revealed 2 possible factor structures and suggested cross-loaded items deletion needed. This resulted in 4 models along with the original 2-factor structure to be tested in the follow-up CFA. Among the 5 models tested, the best-fitting model was the 3-factor solution with 3 cross-loaded items removed. The multi-group CFA performed on this 15-item TDQ revealed that the configural invariance, metrix invariance, and scalar invariance were all supported; partial strict invariance was supported when residual variances of items 1, 9, and 18 were set to vary. Conclusion: The 15-item TDQ with a 3-factor structure reflects the manifestations of depression of Taiwanese general population well. Although the results of measurement invariance suggest no gender difference in factor structure, the partial strict invariance suggests that there is gender difference in the amount of variance captured by TDQ. Hence, examining gender differences in the predictive value of different factors regarding health behaviors or medical conditions may bring fruitful results. Keywords: TDQ, Factor structure, Measurement invariance, Gender differences