The Difficulties in Emotion Regulation Scale (DERS, Gratz & Roemer, 2004) is the most widely used measurement tool and consists of 36 questions with a 5-point scoring system based on a 4-factor model of emotion regulation. It comprises 6 dimensions: emotional response unacceptance, difficulty in engaging in target behaviors, difficulty in impulse control, difficulty in emotion detection, lack of emotional clarity, and limited use of emotion regulation strategies (Gratz & Roemer, 2004). However, researchers are full of concerns about whether the dimension of difficulty in emotion detection exists independently, so some researchers merged it with the dimension of lack of emotional clarity, thus becoming five factors (Bardeen et al., 2012; Brown & Melas, 2024; Nguyen et al., 2024). Moreover, the five factors model of the scale has problems of high similarity of dimensions and poor differentiation (Todorov et al., 2023). Additionally, there is redundancy among questions within almost every dimension, where positive and negative measurements are repeated up to six times. Therefore, one current task is to improve the measurement efficiency and simplify the questionnaire appropriately.

In light of this, developed The Difficulties in Emotion Regulation Scale-8 (DERS-8; Penner et al., 2023) could be an improvement. Notably, the DERS-8 has language improvement with each item beginning with “when I am upset.” and also has an even representation of positive and negative items for clarity. Difficulty in emotion regulation has been modified to mean an individual’s insufficient ability to engage in adaptive emotion-regulating behaviors when experiencing negative emotions.

At present, the cross-cultural applicability of the DERS-8 has yet to be verified. This study aimed to test DERS-8 reliability, validity, and measurement invariance of scores in Chinese adolescents.

A total of 1114 Chinese adolescents were participants (females = 593, mean age = 15.27 SD = 1.86). They attended high school in Hebei Province, China. Among these students, 190 are enrolled in Grade 7, 200 are enrolled in Grade 8, 185 are enrolled in Grade 9, are enrolled in Grade 7, 142 are enrolled in Grade 10, 208 are enrolled in Grade 11 and 189 are enrolled in Grade 12. There are 436 students from urban areas and 678 students from rural areas.

This research received ethical approval by the Ethics Committee of Hebei Normal University and obtained informed consent from all participating students. At the time of data collection, we declare to participants that they can withdraw at any time during the filling process and that the results of filling will be kept confidential. Three samples were involved as followed: Sample 1 (N = 424 for item analysis and exploratory factor analysis), Sample 2 (N = 586 for confirmatory factor analysis), and Sample 3 (N = 104 for test-retest reliability).

SPSS24.0 was used to analyze the scale’s differentiation, Cronbach’s α coefficient, criterion validity and factor structure. AMOS 24.0 was used for confirmatory factor analysis and measurement invariance tests. The fitting degree of confirmatory factor analysis is mainly evaluated by looking at Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), Incremental Fit Index (IFI), Normed Fit Index (NFI), Goodness of Fit Index (GFI) and Root Mean Square Error of Approximation (RMSEA). Among them, CFI, TLI, IFI, NFI, GFI are considered good if they exceed 0.90, and RMSEA is considered good if they are less than 0.08 (Jekauc et al., 2013; Sun et al., 2021).

The item-total correlation coefficient of the DERS-8 was calculated. The results showed that the discrimination of each item ranged from 0.64 to 0.74, all above 0.4, indicating that the item discrimination was good, and all items were retained for subsequent analysis (Table 1).

Principal component analysis and promax oblique rotation were used to conduct exploratory factor analysis on the data of sample 1. The results show that the KMO value is 0.90, indicating that it is suitable for factor analysis. The scree plot shows only one factor with an eigenvalue greater than 1 (Figure 1), and the load of this factor in each problem is more than 0.7 (p < 0.01), as shown in Table 1. This factor explained 60.65% of the total variation.

Confirmatory factor analysis was performed on the data of sample 2, and the model operation results were well fitted, with fitting indices meeting the measurement standards (χ²/df = 4.05, RMSEA = 0.07, CFI = 0.98, TLI = 0.97, IFI = 0.98, NFI = 0.97, GFI = 0.97). This indicates that the DERS-8 has a single-factor structure (Figure 2).

The total DERS-8 score was positively correlated with both the anxiety scale (r = 0.43, p < 0.01) and the depression scale (r = 0.41, p < 0.01). There was a significant negative association with expression inhibition (r = −0.58, p < 0.01) and cognitive reappraisal (r = −0.61, p < 0.01) on the emotional regulation scale. This indicates that individuals with more difficulty in emotion regulation have less use of emotion regulation strategies and more obvious negative emotions. The results show that the ERSP accords with reality and has good external validity.

The Cronbach’s α coefficient of the DERS-8 is 0.90. The split-half reliability coefficient of the DERS-8 for parity was 0.87. Based on sample 3, the test-retest reliability of the DERS-8 was calculated to be 0.66.

Based on sample 2, the equivalence of DER-8 cross-gender measurement was tested. Firstly, the single-dimensional structure model was well fitted for both male (χ²/df = 2.75, RMSEA = 0.07, CFI = 0.98, TLI = 0.97) and female (χ²/df = 2.95, RMSEA = 0.07, CFI = 0.98, TLI = 0.95). Then, the model of configural invariance, metric invariance, scalar invariance and strict invariance are established and compared in turn. If ΔCFI ≤ 0.01 and ΔRMSEA ≤ 0.015, the measurement equivalence of the model is acceptable (Jekauc et al., 2013). The results showed that the baseline model was well fitted (χ2/df = 4.38, RMSEA = 0.96, CFI = 0.96, TLI = 0.93), indicating that the scale had morphological invariance between different genders and was a single factor structure. On this basis, the DERS-8 factor loadings were then restricted to be equal between different genders. It was found that compared with Model 1, the ΔCFI and ΔRMSEA of Model 4 was less than 0.01 and 0.015 respectively, indicating that the DERS-8 factor loadings were within Same for different genders. On this basis, we further restricted the measurement intercepts to be equal in different genders. It was found that compared with Model 2, the ΔCFI and ΔRMSEA of Model 4 was less than 0.01 and 0.015, respectively, indicating that the DERS-8 measurement intercepts were equal among different genders. On this basis, we further limited the error variance to be equal for different genders. The results showed that the ΔCFI and ΔRMSEA of Model 4 was less than 0.01 and 0.015, respectively, compared with Model 3, indicating that the measurement error of DERS-8 is equal for different genders (Table 2). Finally, the independent sample t test found that there was no significant difference (p = 0.12) in the emotion regulation difficulty scores of boys (M = 3.64, SD = 0.83) and girls (M = 3.54, SD = 0.69).

The Chinese version of the DERS-8 has multiple advantages compared with the original emotional regulation difficulty scale. It avoids the variances by positive and negative scoring for a more unified measure. The most important step is to remove many repeated and ineffective measurement questions, significantly reducing the number of test questions without losing accuracy and effectiveness. The adapted more concise and would be incredibly convenient for clinical measurement and screening.

This study revealed that the DERS-8 has measurement invariance among Chinese adolescents. This study has several limitations that need to be considered. The sampling approach adopted was convenience sampling, and all participants were from Hebei Province, which greatly restricts the generalizability of the results. In the future, the sampling range and population can be expanded to test the measurement effect of this tool. Additionally, another limitation is the absence of using Item Response Theory (IRT) technology to analyze the information content of each item. In future research, it is essential to conduct further verification based on the Rasch model to gain more accurate insights into item quality and functionality, thus improving the overall validity and reliability of the study’s measurement approach. In conclusion, the Chinese version of the DERS-8 has various reliability and validity indicators and can be used to measure difficulties in emotion regulation in Chinese adolescents.