Psychometric properties of the Chinese version of the Quiet Quitting Scale

This study was conducted in two phases to ensure the Chinese version of the QQS was conceptually, semantically, and operationally equivalent to the original version. Phase one focused on translation and cultural adaptation, while phase two assessed the reliability and validity of the adapted scale through a cross-sectional study.

The study employed convenience sampling to recruit participants from hospitals in Central China, primarily from Hubei, with some from Henan, Hunan, and other nearby regions, between October and November 2024. Eligible nurses were required to meet the following criteria: (1) current employment, (2) possession of a valid nursing license, and (3) voluntary participation with informed consent. Exclusion criteria included nurses who: (1) were not on duty during the survey period (including those on sick leave, leave of absence, or enrolled in educational programs), or (2) had been absent from their position for over six months. The electronic version of the questionnaire was distributed using the WenJuanXing online platform (https://​www.​wjx.​cn/​).

The QQS consisted of 9 items, and the sample size was determined based on classical measurement theory [19]. For Exploratory Factor Analysis (EFA), the recommended minimum sample size was 5 to 10 times the number of items, requiring at least 56 valid responses, assuming a 20% invalid response rate [19, 20]. For Confirmatory Factor Analysis (CFA) using Structural Equation Modeling (SEM), a minimum sample size of 200 was recommended [20]. To assess test-retest reliability, at least 30 participants were needed, with the retest occurring 2 to 4 weeks after the initial survey [21]. A total of 453 questionnaires were distributed, yielding 420 valid responses, which corresponded to a response rate of 92.720%. The final sample size of 420 participants satisfied the specific requirements for each psychometric assessment, including EFA, CFA, and test-retest reliability, ensuring adequate support for the scale’s psychometric properties. The valid responses were randomly split into two groups using IBM SPSS Statistics version 29.0 (IBM Corp., Armonk, NY, USA). A Random Number Generator was used to randomly allocate 210 responses to the EFA group and the remaining 210 responses to the CFA group [22].

With authorization from Professor Petros Galanis, we translated the QQS into Chinese using the Brislin translation model [23], which involved a forward-backward translation process. In the forward translation phase, two nursing master’s students independently translated the scale from English to Chinese. This version was then reviewed by an experienced bilingual translator, who identified issues with wording and phrasing. Based on their feedback, we refined the Chinese version for clarity and accuracy. In the back-translation phase, the Chinese version was independently translated back into English by a nursing instructor and a medical English instructor. The research team compared the forward and back translations, resolved discrepancies, and made necessary revisions to ensure the final Chinese version’s accuracy and clarity.

To ensure cultural relevance and equivalence for the Chinese context, we conducted a cross-cultural adaptation. An expert panel was formed, comprising three nursing educators and four nursing professionals, all holding master’s degrees. Each panel member independently evaluated the linguistic accuracy of the translation, ensured the use of neutral phrasing, and assessed whether the items effectively measured the intended constructs. Additionally, they considered whether any additional subscales or items were needed to address the specific sociocultural context of China.

The QQS consists of 9 items across 3 domains: detachment (4 items), lack of initiative (3 items), and lack of motivation (2 items) [9]. This concise, reliable, and valid tool effectively measures quiet quitting among employees. It uses a five-point Likert scale, ranging from 1 to 5, with items 7 to 9 being reverse-scored. To calculate the total score for the QQS, the responses to all items are summed and divided by the total number of answers. Similarly, the score for each factor can be calculated separately. The scale demonstrates strong internal consistency, with a Cronbach’s alpha of 0.803. Following translation and cultural adaptation, the Chinese version of the QQS is used to assess its applicability in the Chinese context.

The study was approved by the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (reference number TJ-IRB202406051). All participants voluntarily consented to participate and were free to withdraw at any time. Prior to participation, they signed an electronic informed consent form, demonstrating a full understanding of the study’s objectives, procedures, and potential risks. Anonymity measures were enforced during data collection to protect participants’ privacy and confidentiality.

Not applicable.

This study employed critical rate (CR), correlation coefficient, and internal consistency methods to evaluate the scale design [25]. The Chinese QQS total scores were ranked, with the top 27% classified as high-scoring and the bottom 27% as low-scoring. An independent sample t-test was performed to compare these two groups. Items were considered for removal if differences were non-significant or the CR < 3.00. Additionally, each item’s correlation with the total scale score was calculated, and items with r < 0.30 or P ≥ 0.05 were flagged for potential exclusion [14]. The scale’s Cronbach’s α did not increase after removing any item, supporting the retention of all items.

The validity of the scale was evaluated through content validity and construct validity: (1) Content Validity: Content validity was assessed using both the Content Validity Index (CVI) and the Content Validity Ratio (CVR). Ten experts rated item relevance to the QQS on a 4-point Likert scale. The CVI evaluation included the Scale-level Content Validity Index (S-CVI) and Item-level Content Validity Index (I-CVI), with standards set at I-CVI ≥ 0.78, S-CVI/UA ≥ 0.80, and S-CVI/Ave ≥ 0.90 [26]. S-CVI/UA represents the percentage of items rated “relevant” (scores of 3 or 4) by all experts, while S-CVI/Ave reflects the average I-CVI across all items. Item necessity was rated based on a three-part spectrum from the point of view of necessity (the item is necessary, “the item is useful but not necessary, or “the item is not necessary), and then the CVR was calculated [27]. According to the Lawshe table, CVR ≥ 0.62 was considered acceptable [28]. (2) Construct Validity: Construct validity was evaluated through EFA and CFA. Suitability for CFA was determined with KMO > 0.70 and a significant Bartlett’s test (P < 0.05). Principal Component Analysis (PCA) with Maximum Variance Orthogonal Rotation evaluated the scale’s internal structure. Structural validity was considered acceptable if the extracted factors accounted for more than 50% of the total variance [27]. Model fit criteria required χ2/df ≤ 3, TLI and CFI > 0.90, RMSEA < 0.08, and SRMR ≤ 0.05 [14].

Reliability was evaluated using internal consistency, split-half reliability, and test-retest reliability [27, 29]. (1) Internal Consistency: Reliability within the scale was evaluated using Cronbach’s α and McDonald’s Omega. A Cronbach’s α coefficient and an Omega value greater than 0.70 were considered acceptable [30]. (2) Split-Half Reliability: The Spearman-Brown coefficient was used to assess split-half reliability, with values above 0.70 demonstrating strong reliability. (3) Test-Retest Reliability: The intraclass correlation coefficient (ICC) was a statistical measure used to evaluate the stability and consistency of responses over time, with values ≥ 0.70 considered indicative of satisfactory reliability [27].

The study participants had an average age of 33.17 years (SD = 7.10). Among the participants, 32 were male, and 388 were female. A majority (362 participants) held a bachelor’s degree, and 338 were married. Detailed demographic characteristics are summarized in Table 1.

The expert panel carefully reviewed and discussed each item of the scale, implementing cultural adaptations to improve the clarity and cultural relevance of the Chinese version of the QQS. For example, the phrase “I take as many breaks as I can” was revised to “I will try to secure as much rest time as possible,” reflecting a more natural expression in Chinese. Similarly, the term “work conditions” was replaced with “current situation” to better capture the respondents’ present work environment. These modifications were designed to ensure the QQS remains both effective and reliable within the context of Chinese culture and professional practices. A detailed comparison between the original and adapted versions of the QQS is included in the Additional file 1.

The CR for all items ranged from 9.942 to 23.665, while the correlation coefficients between each item and the total score ranged from 0.561 to 0.792. Following the removal of individual items, the Cronbach’s α values were observed to range from 0.826 to 0.852. Detailed results are presented in Table 2.

The Cronbach’s α for the overall scale was 0.856, indicating good internal consistency, With the values for the three dimensions ranging from 0.814 to 0.838. The split-half reliability coefficient was calculated as 0.921. Test-retest data from 30 participants over two weeks yielded an ICC of 0.851 for the overall scale, with dimension-specific ICCs ranging from 0.705 to 0.803. The McDonald’s Omega for the scale was 0.887, with dimension-specific values ranging from 0.878 to 0.926. These results demonstrate the acceptable reliability of the translated scale (Table 3).

The study’s results confirmed the data’s suitability for factor analysis, with a KMO value of 0.721 and a statistically significant Bartlett’s test of sphericity (χ² = 1115.805, P < 0.001). Three factors with eigenvalues greater than 1 were extracted, collectively explaining 77.93% of the total variance. The scree plot (Fig. 1) further supported a three-factor structure, showing a sharp downward trend after the third factor. Moreover, all items demonstrated factor loadings exceeding 0.4, with no cross-loading observed across multiple factors. These results confirm a strong relationship between the scale items and their respective constructs (Table 5).

CFA was conducted to evaluate whether the questionnaire’s structure aligned with the theoretical framework of the original scale. It also assessed the consistency between each item’s hypothesized relationships and their respective factors. The results indicated a good model fit (Fig. 2), with the following indices: χ2/df = 2.224 (< 3.00, P < 0.05), CFI = 0.963(> 0.90), TLI = 0.945 (> 0.90), RMSEA = 0.077 (< 0.08).

This study has several limitations. First, the convenience sampling method from central China may introduce selection bias and limit external validity. To mitigate response bias, we ensured complete participant anonymity and employed neutral phrasing in survey administration. The gender imbalance in the sample, while reflective of the nursing profession, further restricts generalizability. Future research should aim for more representative, geographically diverse samples and address gender imbalances to enhance applicability across healthcare settings. Second, despite rigorous translation and validation processes, cultural adaptations may have influenced participants’ interpretation of some items. Future studies should refine the scale for diverse populations to ensure its accuracy and effectiveness across cultures. Third, the two-item subscale for ‘lack of motivation,’ while deviating from conventional scale development standards, offers preliminary insights into this dimension within the broader framework. Future research could expand this subscale to better capture its multifaceted nature. Lastly, as a self-reported questionnaire, participants’ responses may have been influenced by social desirability bias. We embedded validity-check items (e.g., reverse-scored questions) and randomized question order to detect and reduce response distortion. Future studies should further minimize bias by incorporating indirect questioning techniques and objective measures. Additionally, adopting mixed-method approaches, such as qualitative interviews and observational studies, could provide deeper insights into participants’ experiences and contextual influences, thereby enhancing data richness and reliability.