A cross-sectional study examining the reliability and invariance of the Depression, Anxiety, and Stress Scale–21 (DASS-21) across ten countries/regions
Xiu-Mei Chen, Yang Li, Iqbal Pramukti, Daniel Kwasi Ahorsu, Shikha Kukreti, Ji-Kang Chen, Wan Ying Gan, Kamolthip Ruckwongpatr, Rashmi Gupta, Cheng-Kuan Lin, Dang Tran, Pham le An, Mark D. Griffiths, I-Hua Chen, Chung-Ying Lin, Olga MalasBackground
The Depression, Anxiety, and Stress Scale-21 (DASS-21) is widely used to assess psychological symptoms among university students. However, its factor structure remains inconsistent, and few studies have employed advanced modeling approaches such as exploratory structural equation modeling (ESEM), particularly in cross-cultural contexts.
Objectives
The present study evaluated the factor structure of the DASS-21 by comparing five models: unidimensional, oblique three-factor, bifactor, three-factor ESEM, and bifactor ESEM. For the best-fitting model, internal consistency and measurement invariance across cultural, linguistic, and demographic groups were further examined.
Methods
Using secondary data from an international mental health project, this cross-sectional study analyzed responses from 12,059 university students ( M = 20.37 years, SD = 1.98; age range: 18–34; 36.6% men and 63.4% women) across ten distinct countries/regions: Mainland China, Ghana, Hong Kong, India, Indonesia, Malaysia, Spain, Taiwan, Thailand, and Vietnam.
Results
The three-factor ESEM model showed the best fit across the ten countries/regions with acceptable fit indices, satisfactory internal consistency (McDonald’s ω > 0.70), and acceptable internal structural validity evidence. Measurement invariance testing supported configural and metric invariance across all ten countries/regions, whereas scalar invariance was not achieved, limiting direct cross-cultural comparisons. This pattern was also observed across the seven language groups. In contrast, scalar invariance was established across age, gender, data collection mode (online versus offline), and pandemic timing (onset versus post-COVID-19), further highlighting potential cultural and linguistic influences on DASS-21 responses.
Conclusions
These findings support the DASS-21’s robust internal structure and reliability across diverse student populations, while emphasizing caution in cross-cultural comparisons.