Lazarus and Folkman (1984) defined coping as “constantly changing cognitive and behavioral efforts to manage specific external and/or internal demands that are appraised as taxing or exceeding the resources of the person” (p. 141). Classification systems for different coping strategies vary widely, with Skinner et al. (2003) identifying more than 100 different systems that had been used in the field. Amongst these, both Lazarus and Folkman’s scheme (which divides strategies into problem- and emotion-focused approaches), and Endler and Parker’s (1990, 1994) tripartite system (task-oriented, emotion-oriented, and avoidance strategies) have been adopted widely, but various other categorizations have also been published (e.g., Billings and Moos, 1981, 1984; Carver et al., 1989). This section reviews existing instruments that have been widely adopted in the field, only one of which was developed in the Chinese context.

The Ways of Coping Checklist (WCCL) was developed by Folkman and Lazarus (1980) to measure the strategies that people use in specific stressful encounters. The original WCCL comprises 68 items, organized into seven subscales that measure defensive coping; information-seeking; problem-solving; palliation; inhibition of action; direct action; and magical thinking. Participants are asked to think about the most stressful events they experienced in the previous month and then indicate which strategies they used in dealing with this event (e.g., “Made a plan of action and followed it”—yes/no). Although the WCCL has been used in various studies, its factor stability, clinical generalizability and construct validity have attracted critical comment from several authors (Vitaliano et al., 1985; Carver et al., 1989; Amirkhan, 1990; Parker et al., 1993). The WCCL is also relatively lengthy, thereby increasing risk of survey fatigue (O’Reilly-Shah, 2017), particularly if it forms part of a battery of instruments used in an empirical study.

The Ways of Coping Questionnaire (WCQ) is an adaptation of the original WCCL (Lazarus and Folkman, 1984; Folkman and Lazarus, 1988), but uses a four-point rating scale (“not used at all” to “used a great deal”) instead of a yes/no format. Minor revisions to specific items within the WCCL were also made in developing the WCQ, to improve question clarity. The WCQ comprises 66 items, which fall into eight factors: confrontive; distancing; self-controlling; seeking social support; accepting responsibility; escape-avoidance; planful problem-solving; and positive reappraisal (Folkman and Lazarus, 1985; Folkman et al., 1986). The WCQ has been used widely within different samples and across different types of stressors. However, validity evidence has been inconsistent across studies. Senol-Durak et al. (2011) investigated the psychometric properties of the WCQ with two samples of Turkish university students (n₁ = 472; n₂ = 485) as well as a sample of community members (n₃ = 416), and found support for a seven-factor structure with acceptable reliability coefficients. Other studies, however, have been less favorable. For example, Edwards and O’Neill (1998) used confirmatory factor analysis to compare alternative factor structures for the WCQ with a group of business school graduates (n = 654) in the United States, and found poor fits for the proposed structure (e.g., average Tucker-Lewis Indices and Comparative Fit Indices of 0.65 and 0.67, respectively), as well as low internal consistencies (most failing to reach 0.70). The WCQ is also approximately the same length as the original WCCL.

The Coping Orientation to Problems Experienced (COPE) measure was designed to assess different ways in which people respond to stress (Carver et al., 1989), and includes 60 items which are categorized into 13 subscales—active coping; planning; suppression of competing activities; restraint coping; seeking of instrumental social support; emotional social support seeking; positive reinterpretation; acceptance; denial; turning to religion; venting emotions; behavioral disengagement; and mental disengagement. Participants are asked to rate how frequently they use each type of strategy (e.g., “I’ve been turning to work or other activities to take my mind off things”) on a 1 (I haven’t been doing this at all) to 4 (I’ve been doing this a lot) scale. The COPE has been translated into different languages and its psychometric properties have been evaluated extensively in Western contexts. In general, favorable results have been generated (Sica et al., 1997). However, like the WCCL and the WCQ, the COPE is relatively long at 60 items.

Carver (1997) subsequently developed a 28-item version of the COPE (the Brief COPE), which was initially validated with a group of 168 community residents who had been affected by a hurricane. This initial study indicated that the Brief COPE exhibited similar properties to the original version. Validation studies into the brief COPE since, however, have produced mixed results with respect to the factor structure of the instrument across cultures. Both Muller and Spitz (2003) and García et al. (2018) confirmed the 14-factor structure of the Brief COPE after translating the items into French and Spanish, respectively. However, studies in other contexts have reported varying factor structures, including a two-factor structure based on responses from female nurses in the United Arab Emirates (Rahman et al., 2021); a three-factor structure based on responses from British adults (Ingledew et al., 1996); a four-factor structure based on college students in an American university (Cramer, 2019); a six-factor based on responses of people living with HIV in China (Su et al., 2015); and a seven-factor model based on responses from international university students in America (Miyazaki, 2008).

Endler and Parker (1990) developed the 48-item Coping Inventory for Stressful Situations (CISS) to include16 items in each of three subscales (task-oriented, emotion-oriented and avoidance-oriented coping strategies). Respondents rate how often they use each strategy listed in the instrument (e.g., “Talk to someone whose advice I value”) on a five-point scale ranging from 1 = “Not at all” to 5 = “Very much.” Endler and Parker (1994) evaluated the properties of the CISS and reported high internal consistencies ranging from 0.83 to 0.90 across the subscales. The CISS has also been validated in different countries such as Japan (e.g., Furukawa, 1997), and found to exhibit similar internal consistencies in these contexts. Validation studies on the CISS have, however, suggested that the factor structure may vary across cultural contexts (e.g., Choi et al., 2017).

Endler and Parker (1994, 1999) also developed a short form of the CISS (CISS-21), which has since been translated and evaluated in several languages, including Turkish (Boysan, 2012), Urdu (Rodriguez-Diaz et al., 2014), and Chinese (Li et al., 2017). While a number of these evaluations have yielded similar results to those reported for the English version, both the factor structures and the internal consistencies have varied somewhat across studies. For example, in Li et al.’s evaluation based on 972 Chinese university students, Cronbach’s alphas as low as 0.65 were reported for one of the subscales. Furthermore, despite the fact that the CISS-21 has been translated to Chinese, it was not originally developed with Chinese participants, and may as a result exclude strategies that are used by individuals within this context.

The Simplified Coping Style Questionnaire (SCSQ) is the only coping strategies measure the authors could locate that was developed based on Chinese participants. Xie (1998) developed the 20-item SCSQ based on Folkman and Lazarus’s (1980) WCQ. The SCSQ incorporates two dimensions: positive coping styles (12 items—e.g., “Try to find different methods to solve problems”) and negative coping styles (8 items—e.g., “Try to forget about it”). Respondents are required to rate how often they use each strategy based on a four-point scale (“never” to “very often”). Xie (1998) examined the properties of the SCSQ with a sample of 20 Chinese university students and found high test-retest and Cronbach’s α coefficients. Despite its obvious potential advantages in terms of brevity and practicality, no explicit rationale was proffered for the selection of items included within the SCSQ, or for the overall negative/positive classification of the items. For example, it is arguable as to whether the strategy “Try to forget about it” is actually negative in all contexts (e.g., this strategy could actually be positive in situations that are entirely out of the individual’s control). Furthermore, the instrument has never been validated on a large sample of participants.

Given that no widely validated CSSs for use in Chinese contexts could be located by the authors, the current study aimed to develop such an instrument and validate it in a large group of Chinese university students. The instrument was designed to be used in a variety of adverse circumstances (i.e., it asks about the strategies that participants generally use when confronting adverse events). However, it was validated against the backdrop of the global COVID-19 pandemic, having formed part of a larger survey on Chinese university students’ mental health during the pandemic. Therefore, participants in the validation study are likely to have used this event as a recent reference point when they completed the questionnaire.

The Coping Strategies Scale (CSS) was completed by participants as part of a larger (195 item) survey. Participants were Chinese students enrolled in various university institutions across mainland China. Over 1,000 such students entered the survey, but following the data screening to remove partially completed surveys and instances of clearly disengaged responses (e.g., participants who missed five or more items; put the same answer for all questions; or finished the entire survey in less than 5 min), only 734 responses were retained for analysis.

Using a computer-generated list of random numbers, we then divided this sample of 734 responses into two groups for the validation exercise. Group A comprised 367 students, including 168 males (45.80%) and 198 females (54.00%) and one (0.30%) who indicated “other” for gender. Group B also comprised 367 students, including 166 males (45.2%) and 201 females (54.8%). The ages of the respondents ranged from 16 to 48 years, with a mean of 22.20 years (SD = 4.16) from Group A and a mean of 22.08 years (SD = 3.82) from Group B. For Group A, 22.30% of students were married or in a relationship as compared with 26.20% of students from Group B.

The rationale for constructing a scale and operationalizing coping strategies was that different coping strategies used by people have been found related to their mental health (Ito and Matsushima, 2017; Saxon et al., 2017; Tsaras et al., 2018). The CSS scale, therefore, is intended to be used both as an outcome measure in what coping strategies people tend to use in a specific stressful context, and as a predictor variable on people’s mental health. The instrument was developed based on literature that was published in English journals. All items were written in both Chinese and English forms concurrently. The first author is a speaker of both Chinese and English, and thus was able to write the item statements in both forms. The CSS comprised 30 specific strategies/items, with respondents asked to rate how frequently they use each of the strategies listed using a five-point frequency rating scale (Never = 1, Rarely = 2, Sometimes = 3, Often = 4, Always = 5). These were designed to assess 10 types of coping strategies: Opportunities for Growth (Growth); Identifying Alternatives (Alt); Problem Solving (Prob); Prosocial Focus (Prosoc); Seeking Professional Help (Prof); Seeking Emotional Support (Emot); Self-Regulation (Sreg); Disengagement (Diseng); Denial (Denial); and Withdrawal (WithD). A list of all CSS instrument statements can be found in Table 1. Means and standard deviations for all items, based on the two subgroups of n = 367 as described above, are also presented in Table 1.

To provide further evidence related to concurrent validity two further instruments were completed by all participants. The first was a measure of well-being levels. Different coping strategies employed by individuals have been found to relate to well-being levels. Specifically, problem-solving coping strategies, positive reappraisal, support-seeking have been found consistently to relate to higher levels of well-being, while avoidance coping strategies have been found to be associated with lower well-being levels (Gustems-Carnicer and Calderón, 2013; Sagone and De Caroli, 2014; Freire et al., 2016). The World Health Organization—Five Well-Being Index (WHO-5) (World Health Organization [WHO], 1998) was used as the well-being measure in the study. This measure includes five items, each of which requires respondents to rate their wellbeing levels on a six-point Likert scale ranging from 0 (at no time) to 5 (all of the time). A large body of research has affirmed the sound psychometric properties of the WHO-5 as a measure of subjective well-being (Krieger et al., 2013; Downs et al., 2017; White and Van Der Boor, 2020).

The second measure used to assess the concurrent validity of the CSS was a measure of respondents’ self-efficacy for solving challenging problems they encountered in their lives. Self-efficacy is defined as “people’s beliefs about their capabilities to produce designated levels of performance that exercise influence over events that affect their lives” (Bandura, 1994, para. 4). As noted by Bandura (2006), instruments to measure self-efficacy must be constructed to reflect the specific tasks or activities of interest. The brief measure used in the current study was constructed based on Bandura’s (2006) item construction recommendations, using a bi-polar statement format:

Respondents were provided with a 7-point scale for the self-efficacy measure (-3 to +3). As indicated, all items in this scale related specifically to respondents’ confidence in their ability to cope with and address problems within their lives. Previous studies have indicated that high levels of such types of self-efficacy are associated with active coping strategies such as problem-solving (Taiwo, 2015; Manuel Morales-Rodriguez and Manuel Perez-Marmol, 2019), while low levels of such self-efficacy are associated with passive coping strategies such as avoidance (Taiwo, 2015).

Convenience sampling and snowball sampling were used for data collection. After obtaining ethics approval from the authors’ University Ethics Committee, information about the survey was distributed via WeChat. The CSS was completed online through the Qualtrics platform. An anonymous link to the survey was sent to a large group of Chinese university students via WeChat. Those who completed the survey were also encouraged to forward the link to their circles of friends on WeChat to attract more participants. All of the participants and potential participants were informed that it would take them approximately 5–16 min to finish the survey, but that they could stop and resume filling in the survey if necessary. This was done to ensure that participants filled in the survey carefully, rather than in a rush to reach the end. No special incentives or rewards were offered for participation. All participants completed the survey voluntarily and were able to withdraw from the survey at any point prior to submission.

Exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were used to evaluate the internal structure of the CSS. EFA is a statistical technique used to conduct a preliminary analysis of the constructs which are needed to account for patterns of correlation among measures when there is no expectation regarding the factorial structure of an instrument (Fabrigar and Kan, 2018). CFA is a statistical technique used to verify the hypothesized relationships between observed indicators and their underlying latent variables (Randall and Jung, 2018). In this study, EFA was conducted by IBM SPSS V27 and CFAs were conducted by LISREL V8.80.

To develop items for the CSS, a comprehensive search was first conducted to identify all coping strategies questionnaires published previously for use with adults either in Western or Chinese contexts. All elements of these existing instruments were considered for inclusion to ensure that the CSS was as comprehensive as possible. This included all instruments reviewed previously in the current paper (i.e., the WCC and revised WCC; the COPE and Brief COPE; the CISS; and the SCSQ), and those that appeared only in specific papers, such as the instruments published by Stone and Neale (1984), Amirkhan (1990), Jiang (1999), Garnefski et al. (2001), Heppner et al. (2006), Maxwell and Siu (2008), and Aitken (2011). After the CSS was developed by the first two authors, it was sent to two professors in the area of mental health, who commented upon the wording and content of the items. Revisions to the wordings were then made based on this feedback, and the changes were translated to the Chinese version.

After the content of the instrument was deemed to be suitable in light of the expert reviews obtained, cognitive process interviews were then conducted with four Chinese university students on a one-on-one basis, and all the interviews were conducted in Chinese. In these sessions, participants were asked to read each item in the questionnaire carefully and indicate to the interviewer (the first author) any items or questions that confused them. This process was undertaken to ensure that the response processes associated with completing the questionnaire were aligned with the intent of the instrument, and to identify any items that generated confusion or ambiguity. Based on the feedback obtained, some further adjustments were then made to the wording of specific items.

To evaluate the internal structure of the CSS, an exploratory factor analysis (EFA) was conducted on data from Group A (n = 367), and a confirmatory factor analysis (CFA) performed on the data from Group B (n = 367). Cronbach’s alphas were also computed for each group to evaluate the internal consistency of the CSS. IBM SPSS V26 and LISREL V8.80 were used to conduct the EFAs and CFAs, respectively. A principal components analysis (PCA) extraction method was used in the EFA conducted on the Group A data, given that this was intended to be a preliminary analysis to indicate the number of dimensions that were tenable for the CSS. As the components were expected to be relatively independent of one another, these were rotated to approximate simple structure using the Varimax approach.

Preliminary screening analyses indicated no significant violations of EFA assumptions associated with linearity, normality, factorability, and case-to-item ratio. Rotated loadings are presented in Table 2. The scree plot suggested that a seven-component solution would provide the best fit to the data, though the eigenvalue of the seventh component fell marginally below 1. A comparison of the six- and seven-component solutions obtained indicated that the seven-component solution, which accounted for 68.50% of the total item variance, was more conceptually tenable, and better aligned with the theoretical structure. The six-component solution was similar to the seven-component, except that the two of the three self-regulation items (Sreg 1 and Sreg3) loaded with the Prosocial Focus items, while the remaining self-regulation item (Sreg 2) loaded with the Disengagement items. It was thus decided that the seven-component solution would be retained for interpretation, though the superiority of this model over the six-component solution obtained was also tested directly through the CFAs conducted on the Group B data.

Based on the seven-factor solution, the Prosocial Focus, Seeking Emotional Support, Disengagement, and Self-regulation items all fell into their own separate components as predicted (Components 4, 5, 6 and 7, respectively). The other three empirically obtained EFA components incorporated two of the original item clusters. These were Component 1 (which incorporated Opportunities for Growth and Identifying Alternatives items); Component 2 (which incorporated the original Withdrawal and Denial items); and Component 3 (which incorporated the original Seek Professional Help and Problem-Solving items). These three larger components were labeled Positive Adaptation, Withdrawal, and Problem-solving, respectively. The soundness of the seven-component structure was further affirmed by the moderate to high internal consistencies obtained for each of the components using the Group A data, with αs of 0.88, 0.86, 0.86, 0.75, 0.86, 0.77 and 0.79 for the Withdrawal, Positive Adaptation, Problem-solving, Disengagement, Prosocial Focus, Seeking Emotional Support, and Self-regulation subscales, respectively.

The data screening analyses for the Group B data again confirmed that all relevant assumptions for factor analysis had been met. Four CFA models were conducted on these data to cross-validate the results obtained in the EFA based on the Group A data. In the first model, all 30 CSS items were loaded on one factor. The second model was based on the two-factor model (Problem-Focused vs. Emotion-Focused strategies) suggested by Lazarus and Folkman (1984). In the two-factor model, the original Problem-Solving and Seek Professional Help items loaded on one factor (representing the Problem-Focused dimension), and all others loaded on another (representing the Emotion-Focused dimension). The third and fourth models were based on the six- and seven-factor alternative structures suggested by the EFAs on Group A. The change in χ² among the models was used to evaluate whether the model fit statistics for the four models were significantly different. Fit indices for the four models tested are shown in Table 3.

Based on the Δχ² statistics, the seven-factor solution aligned with the EFA results was clearly the best-fitting model. Specifically, there was a significant increase in χ² from the seven-factor to the one-factor model, Δχ²(21) = 3652.43, p < 0.001; from the seven-factor to the two-factor model, Δχ²(20) = 3438.80, p < 0.001; and from the seven-factor to the six-factor model, Δχ²(6) = 377.87, p < 0.001. The sound fit of the seven-factor model was further affirmed by the other fit indices obtained (Kline, 2005, 2016; Hooper et al., 2008). These included the ratio of the chi-square statistic to the model degrees of freedom (χ²/df), which measures the discrepancy between the sample and fitted covariance matrices, taking sample size into account (values ≤ 5 indicating acceptable model fit); the Non-Normed Fit Index (NNFI), which measures the relative fit of the proposed model to the null model (values ≥ 0.90 indicating acceptable model fit); the Comparative Fit Index (CFI), which measures the relative fit of the proposed model to the null model, but is less sensitive to sample size than the NNFI (values ≥ 0.90 indicating acceptable model fit); the Standardized Root Mean Square Residual (SRMR), which measures the difference between the residuals of the sample covariance matrix and the hypothesized model (values ≤ 0.08 indicating acceptable model fit); and the Root Mean Square Error of Approximation (RSMEA), which measures the discrepancy between the hypothesized model and the population covariance matrix (values ≤ 0.08 indicating reasonable model fit).

Moderate to high internal consistencies were also obtained for each component of the instrument, with αs of 0.88, 0.87, 0.84, 0.75, 0.83, 0.73 and 0.76 for the Withdrawal, Positive Adaptation, Problem-solving, Disengagement, Prosocial Focus, Seeking Emotional Support, and Self-regulation subscales, respectively. Based on the Cronbach’s αs, four subscales exhibited high internal consistency (>0.80), while the other three were acceptable (>0.70) within the Group B data. Coefficients for the paths between each of the items and their respective latent factors are shown in Figure 1.

To provide further validity evidence on the instrument, information on two variables that had been found previously to correlate with coping strategies (age and education level) were also collected. Deasy et al. (2014) found that students aged 27 years or older used escape avoidance strategies (which align broadly with the Withdrawal strategies in the CSS) less frequently, and positive reappraisal (which align broadly with the Positive Adaptation strategies in the CSS) more frequently than did younger students. With respect to education levels, Sandover et al. (2015) found that undergraduate medical students favored “distancing” and “escape avoidance” (which aligns with the Withdrawal subscale of the CSS) as coping strategies in times of stress, while graduate medical students favored “planful problem solving” (which aligns with the Problem-Solving subscale of the CSS). Ickes et al. (2015) also found that undergraduate students were more likely than postgraduate students to use different forms of escapism (which again align broadly with the Withdrawal strategies in the CSS) as coping strategies.

In the current study, multivariate analyses of variance (MANOVAs) were performed to determine whether there were any significant differences in the use of coping strategies across different age groups and education levels. Means for the subgroups are shown in Table 4. All evaluations of conformity to MANOVA assumptions produced satisfactory results.

Results from the MANOVAs indicated a significant difference on the linear composite CSS variable across the age groups, λ = 0.91, F(14, 1502) = 4.93, p < 0.001, partial η² = 0.04, and the education levels, λ = 0.93, F(7, 746) = 8.44, p < 0.001, partial η² = 0.07. Follow-up analyses of variance (ANOVAs) were thus performed to explore each of these significant multivariate effects in more detail. The Holm-Bonferroni correction procedure (see Holm, 1979) was applied in evaluating the ANOVA outcomes, to maintain the α level at 0.05 for each MANOVA.

The ANOVAs testing the effect of age group on coping strategies indicated significant differences across groups on four of the seven coping subscales: Problem-Solving, Self-Regulation, Positive Adaptation, and Prosocial Focus, Fs(2, 757) ≥ 3.11, ps ≤ 0.01, partial η² ≥ 0.01. The effect on Withdrawal also approached significance at the adjusted α level, F(2, 757) = 3.08, p = 0.02, partial η² = 0.01, while the effects of age group on Seeking Emotional Support and Disengagement were not significant. Results of the Tukey post-hoc tests for age groups, conducted at the 0.01 level, showed that students under 21 had significantly lower scores than those who were 27 years or older for Positive Adaptation (p = 0.004), Problem-Solving (p = 0.002), and Self-Regulation (p = 0.004). The difference between students under 21 and those between 22 and 26 years approached significance for Prosocial Focus (p = 0.03), as did the difference between students under 21 and those 27 years or older for Withdrawal (p = 0.03). For the latter subscales, students under 21 had higher scores than did older students. The results obtained for Emotional Support and Withdrawal align with the findings of Deasy et al. (2014), and thus support the validity of the new CSS.

For education level, the ANOVAs again indicated significant differences across groups on four of the seven of the coping subscales: Problem-Solving, Self-Regulation, Positive Adaptation, and Withdrawal, Fs(1, 752) ≥ 7.76, ps ≤ 0.01, partial η² ≥ 0.01. The effects on Prosocial Focus, Seeking Emotional Support, and Disengagement were not significant. The significant differences identified indicated that undergraduate students reported using problem-solving, positive adaptation, and self-regulation less frequently, but relied more frequently on withdrawal as a strategy, than did postgraduate students. The latter results (specifically, those obtained for the Problem-Solving, Positive Adaptation, and Withdrawal subscales) align well with the results reported by Ickes et al. (2015) and Sandover et al. (2015), which again leads support to the construct validity of the new CSS.

To provide further evidence related to concurrent validity, relationships between the CSS subscales and those from the WHO-5 and self-efficacy instrument described previously were explored. These analyses indicated significant positive relationships between Positive Adaptation, Prosocial, Seeking Emotional Help, Self-regulation, Problem-Solving and Disengagement with both WHO-5 scores, all rs(725) > 0.15, ps < 0.001, and with self-efficacy scores, rs(725) > 0.29, ps < 0.001. The relationship between Disengagement scores and the WHO-5 was also positive, though the strength of the relationship was more modest, r(725) = 0.08, p = 0.03. Similarly, scores on the self-efficacy measure were positively related to CSS Disengagement scores, though more modestly than for other CSS subscale scores, r(725) = 0.18, p < 0.001. These results are consistent with the previous research cited previously. In contrast, scores on the Withdrawal subscale were not significantly correlated either with WHO-5 scores, r(725) = 0.07, p > 0.05, or with self-efficacy scores, r(725) = 0.03, p > 0.05. This pattern of results is also aligned with findings reported previously in the literature.