Appreciating resilience at work: Psychometric assessment, measurement, and practical implications

Abstract

Limited research has been done on the utility of the resilience at work (RAW) scale in the global south. Furthermore, no study has modelled the general factor of the RAW scale. This is a huge gap given the need for organizational leaders to effectively to constantly monitor and nurture employee resilience if they are to enjoy adaptive advantages and flourish. The reliability, construct validity, dimensionality, and invariance of the RAW Scale were analysed. The levels of resilience at work were also measured. A sample of 213 employees was drawn from employees in government (34%), non-governmental organizations (NGOs, 33%), and the private sector (33%). Statistical software R, and the Bifactor Indices Calculator were used for the analysis. The RAW scale exhibited adequate psychometric properties. Exploratory factor analysis produced a seven-factor structure with 57% total variance explained. The higher model of the seven-factor scale had adequate fit indices. The results of the bifactor model also confirmed the multidimensional structure of the scale, albeit with six latent factors. Gender did not differentiate resilience at work scores. The results suggest that each of the six latent factors captures unique information beyond the full scale. As such, organizational researchers and leaders should use the RAW subscale average scores when measuring, interpreting, and evaluating resilience at work-related remediation actions. The scale was invariant across sectors hence its utility in national workplace surveys.

Keywords:

• Resilience at work
• Dimensionality
• Psychometric Properties
• Bifactor Models
• Levels of Resilience at Work

PUBLIC INTEREST STATEMENT

The investigation provided further evidence that resilience at work, as measured by the RAW scale, is best conceptualized as a multidimensional construct as per the original theoretical specifications for this construct. This implies that organizational researchers and leaders alike can use the sub-test scores in appreciating the levels of resilience among employees to facilitate relevant work-place remediation efforts.

The study showed that there were no significant differences in resilience at work by gender. Traditionally, and hinged on gender-inclined biases, resilience at work has been associated with masculine values of “strength, robustness, boldness, stoutness, bravery and not being womanish” (Cooper, 1995, pp. 146–147). Hence, these findings are of practical, policy and academic importance. The confirmation that differences in resilience at work between male and female employees are due to perceptual bias than actual behavior is indeed a timely and critical reminder for organizational leadership teams.

1. Introduction

The quality of employee resilience is now at the heart of industrial-organizational psychology, driven by the increasingly chaotic and complex environment. Resilience at work is a capability that helps us to understand how employees manage daily adversities, learn from, and rebound, while proactively preparing for future challenges (Malik & Garg, 2018). Embedded within positive psychology, the importance of resilience in fostering and maintaining employee well-being and performance is well documented (Turner et al., 2021). Several measures have been developed to measure resilience at work (RAW). The RAW scale developed by Winwood et al. (2013) is a useful measure to effectively comprehend the strengths related to employee resilience. The RAW scale has a bias towards developable resilience at work strengths. Limited research has been done on the utility of the RAW scale in the global south. Furthermore, to our knowledge, no study has modelled the general factor of the RAW scale.

The underlying theoretical specification of the scale recognizes seven subtests (Winwood et al., 2013). The first subscale, termed “living authentically”, highlights the role of mindset (personal values, deploying personal strengths) and emotional intelligence (through emotional awareness, management, and regulation). The second one, “finding one’s calling”, is hinged on spirituality i.e. work that has a purpose, having a sense of belonging, and alignment of an individual’s core values and beliefs. The third subscale is about “maintaining perspective” and places emphasis on the employee’s capacity to positively reframe adversities, keep on with solutions in the face of adversities, and thus create the momentum to manage any negativity.

The fourth subscale focuses on “managing stress” and is hinged on ensuring positive work-life balance. The fifth subscale, titled “interacting cooperatively,” places emphasis on work styles with a bias on deliberately seeking feedback and work-level specific advice as well as support, while providing the support to others, too. The sixth subscale is about staying healthy, i.e., the importance of being physically and healthy conscious. The last subscale is termed “building networks” and is focused on developing and leveraging personal support networks within and outside the workplace.

Since its conception by Winwood et al. (2013), only one study by Malik & Garg (2018) has interrogated the RAW scale in detail. The study by Winwood et al. (2013) and Malik & Garg (2018) used classical higher-order models coupled with the rigid use of global model fit indices to ascertain dimensionality of the RAW scale. Such models may be inconclusive (Rodriguez et al., 2016a; Schaaps, 2019). Item response models, such as bifactor confirmatory factor analysis, represent a viable alternative (Rogers et al., 2020).

In undertaking the bifactor analysis, the study sought to confirm whether resilience at work can be interpreted as seven separate scores or a single summative score. Use of the RAW scale total scores assumes that each of the 20 observable items on the RAW scale represents a facet of the same general construct. On the other hand, the use of subscale average scores assumes that each of the latent factors captures unique information beyond the full scale (Rogers et al., 2020).

Therefore, confirming the dimensionality of the RAW scale has a bearing on its utility in the workplace. Furthermore, the study by Malik & Garg (2018) produced a six-factor structure viz. the original seven-factor structure conceived by Winwood et al. (2013), hence the need for further investigation to guide research and standard practice.

1.1. Comprehending the resilience at work phenomenon

Studies by Masten (2001), Masten and Reed (2002), and Coutu (2002) are associated with some of the documented early efforts to understand the concept of resilience within the workplace. Building on existing scholarly conceptualizations, Malik and Garg (2018) provided a transformational perspective to work place resilience which recognizes a) employee resilience as an individual ability that utilizes personal resources for successful adaptation and flourishing within the work environment, and (b) employee resilience as a capability (and intervention) beyond mere adaptation, i.e. encompassing continual growth and improvement within both stable and adverse circumstances.

Building on the above premise, Malik and Garg (2018, p.:79) defined resilience at work as the “individual’s capacity to manage adverse situations and daily stress of work and remain healthy, rebound and learn from unexpected setbacks, and prepare for future challenges proactively, demonstrating increased competence, professional growth, and the ability to handle future challenges in the workplace”.

When the operating environment is precarious—as in Zimbabwe since the turn of the new millennium (Kanyenze et al., 2017)—the ability of employees to adapt, be flexible, have purpose, conjure effective coping strategies for positive well-being and doing can be severely tested and at most severely compromised (Hartmann et al., 2020; Näswall et al., 2019).

The tumultuous events since December 2019—as a result of the COVID-19 pandemic—which may yet be the most consequential events of the second decade of the 21^st century (Busby, 2020) further compound an already suboptimal situation for employees. Such catastrophic and extraordinary events require a different level of employee resilience (Veldsman & Johnson, 2016). Therefore, it is timely to investigate the resilience at work phenomenon.

1.2. Appreciating the role of resilience in the workplace

The vital role of resilience in fostering and sustaining employee well-being and doing is well documented (Mak et al., 2011; Malik & Garg, 2018; McCann et al., 2009). Similarly, its negative association with employee turnover intentions is acknowledged (Malik & Garg, 2018). Mallak et al. (2016) posit that higher resilience can reduce work place stress, boost quality decision making, decrease use of sick days, and foster higher job satisfaction.

Similarly, contemporary evidence shows that resilience is positively associated with work happiness, job satisfaction, job performance, and organizational commitment (Mayfield, 2019; Smith et al., 2020; Walpita & Arambepola, 2020).

In addition to being a significant predictor of job performance, resilience at work has also been found to be associated with work engagement (Dai et al., 2019; Kašpárková et al., 2018; Smith et al., 2020). Resilient employees have also been found to be healthier (Salanova et al., 2011).

Lim and Kim (2020) reported that resilience at work improved the professional quality of life of nurses, including significantly reducing the level of secondary trauma stress. In addition, Chadwick and Raver (2020) and Malik and Garg (2020) reported a positive relationship between resilience and innovative work behavior.

Resilience training has been found to be effective in fostering employees’ mental health and well-being (Pipe et al., 2012; Robertson et al., 2015) as well as productivity (Pipe et al., 2012) and observed behavioral performance (Arnetz et al., 2009).

Despite the evidence, limited studies have been conducted to understand resilience at work in the study context. A few qualitative studies have been conducted, see Mugumbate & Gray (2017), Mushonga (2015), and Witter et al. (2017). It is important to complement these studies. This investigation will shed light on the potential utility of a measure that could aid the understanding of the phenomenon and help influence resilience-related processes and outcomes in the workplace.

1.3. Aim of the study

The study evaluated the reliability, construct validity, dimensionality, and invariance of the RAW Scale. Furthermore, the study also reported the observed levels of resilience at work.

1.4. Hypotheses

The present study aimed to test the following hypotheses about the structure of the RAW scale.

Hypothesis 1: The RAW scale exhibits adequate psychometric properties in the study setting.

Hypothesis 2: The RAW scale has an underlying general factor.

Hypothesis 3: RAW scores reliably differentiate employee resilience at work by gender

2. Methods

2.1. Participants and setting

A sample of 213 employees was drawn from employees in the Zimbabwean government (34%), non-governmental organisations (NGOs, 33%), and the private sector (33%). The sample comprised more men (61%) than women (39%), expectedly so given the underlying formal employment profile for the country. The average age was 38.6 years (standard deviation, s, = 8.87 years). There was a similar representation of participants in government, NGOs, and the private sector. By level of employment, senior (37%) and middle (38%) level employees made up the majority of the sample.

2.2. Measuring resilience at work: RAW scale

Quantitative data were collected using the 20-item resilience at work (RAW) scale developed by Winwood et al. (2013). The RAW scale has robust psychometric features, i.e. the Cronbach alpha of 0.81 (Malik & Garg, 2018). Employees self-rated their level of resilience using a 7-point Likert type scale (“1 = strongly disagree”; “7 = strongly agree”). According to Malik and Garg (2018), while research on workplace resilience is generally in its infancy, the RAW scale provides a realistic and practical measure of workplace resilience.

2.3. Ethics

The participants individually consented to the study. They completed the survey online. The purpose of the study was clearly explained on the survey landing page and participation in the study was strictly voluntary (i.e. opt-out). No personally identifiable or organizational information was collected and the study adhered to strict confidentiality and anonymity rules.

2.4. Analytical approach

Two tests were conducted and confirmed the suitability of the data set for structure detection i.e. the Kaiser-Meyer-Olkin sampling adequacy measure (KMO of 0.801; Cerny & Kaiser, 1977) and the Bartlett sphericity test (p < 0.01; Bartlett, 1954).

Data analyzes were performed primarily in R and the Bifactor Indices Calculator, a Microsoft Excel-based tool used to calculate various indices relevant to bifactor CFA models (accessible at http://sites.education.uky.edu/apslab/resources/; Dueber, 2017).

To assess the psychometric properties of the RAW scale in the study sample, the Cronbach Alpha, composite reliability (CR) scores, and the Average Variance Explained (AVE) were calculated.

3. Results

Given the conflicting results regarding the factor structure of the RAW scale (see, Malik & Garg, 2018; Winwood et al., 2013), an exploratory factor analysis was conducted to further determine the underlying structure of the RAW scale. Principal component analysis, using varimax rotation, was employed. Based on the Kaiser-Guttman criterion (Guttman, 1954; Kaiser, 1960), data for the 20-item RAW scale yielded a seven-factor structure for the RAW scale with 57.18% of the total variance explained.

3.1. Reliability of the RAW Scale

The Cronbach alpha is a commonly referenced measure of internal consistency reliability. According to Nunnally (1978), a threshold of 0.70 confirms adequate reliability. The 20-item RAW scale had a Cronbach α of 0.77. The individual subscales had the following alpha values: 0.70 (Finding one’s calling, FOC), 0.71 (Living authentically, LA), Staying healthy, SH), 0.71 (Maintaining perspective, MP), Building networks, BN), 0.72 (Managing stress, MS) and 0.74 (Interacting cooperatively, IC), see, Table 1. The composite reliability (CR) scores exceeded the 0.70 threshold (Fornell & Larcker, 1981), confirming the internal consistency and reliability of the scale items. The reliability of the scale is also affirmed by the low to moderate correlations among the latent factors; see, Table 1.

Table 1. Reliability and validity of the RAW scale

Latent Factor	Cronbach Alpha	CR	AVE	LA	SH	MP	FOC	MS	BN	IC
LA	0.71	0.70	0.53	0.72
SH	0.71	0.71	0.53	0.10	0.73
MP	0.71	0.71	0.54	0.57	0.01	0.74
FOC	0.70	0.74	0.69	0.36	0.34	0.31	0.83
MS	0.72	0.78	0.61	0.52	0.65	0.36	0.28	0.78
BN	0.71	0.72	0.56	0.59	0.43	0.06	0.07	0.49	0.75
IC	0.74	0.73	0.50	0.59	0.12	0.25	0.40	0.29	0.38	0.71

Note: LA = Living authentically; FOC = Finding your calling; MP = Maintaining perspective; MS = Managing stress; IC = Interacting cooperatively; SH = Staying healthy; BN = Building networks.

The Cronbach Alpha assumes that the true score variance is constant across all 20 items on the RAW scale. However, the possibility of a scale resulting in equal sensitivity across all items is unrealistic (Dunn et al., 2014). The study acknowedges the limitation ofthe Cronbach Alpha hence the complementary use of the CR scores.

3.2. Convergent and discriminant validity

The RAW scale had an average variance extracted (AVE) of 0.58, while all seven latent factors had AVEs > 0.5.. This means that 58% of the variation in the resilience at work construct can explained by the 20 items thus satisfying the overarching threshold of 0.5 (Fornell & Larcker, 1981). . Conversely, an AVE less than 0.50 means that the items explain more errors than the observed variance in the constructs. Coupled with the aforementioned CR scores, the results confirm the convergent validity of the scale in the study setting.

Discriminant validity, on the other hand, seeks evidence of low or no correlation among the seven latent factors, i.e. each of the specific items for each latent factor should be uniquely measuring that specific variable. The square root of the AVE of each of the seven latent variables should be much larger than the correlation of the specific latent variable with any of the other latent variables (Malik & Garg, 2018). The bolded values in Table 1 satisfy this claim thus confirming the discriminant validity of the RAW scale. Results in Table support hypothesis 1 i.e., the RAW scale exhibits adequte psychometric properties in the study setting.

3.3. Confirmatory factor analyses

The study evaluated the seven-factor higher order model viz. the bifactor model of the RAW scale.

3.4. Higher-order CFA

For higher-order models, the existing literature references several thresholds depending on the selected global model fit indices. For Comparative Fit Indices (CFI) and (Goodness of Fit) GFI, values > 0.90 are deemed desirable, while for Adjusted Goodness of Fit (AGFI), values.80 or higher are preferable (Chen, 2007). For the root mean square error of approximation (RMSEA), Hu and Bentler (1999) referenced a threshold close to 0.06 (the upper limit of the 90% confidence interval (CI) should be <0.10 (Kline, 2015), while for the standardized root mean square residual (SRMR), values close to 0.08 are deemed acceptable. For the chi-square (CMIN), the CMIN/df (df- degrees of freedom) has to be below 2 to indicate model fit (Chen, 2007). The higher order factor model demonstrated an adequate fit (χ2 = 244.94; df = 146; χ2/df = 1.68; GFI = 0.92; CFI = 0.92; AGFI = 0.88; SRMR = 0.07; and RMSEA = 0.05).

3.5. Bifactor modelling

3.5.1. Factor level

Table 2 presents the factor-level outputs for the RAW scale. According to Rodriguez, Reise and Haviland (2016), at factor level, the Omega Hierarchical (OmegaH) “reflects the percentage of systematic variance in unit weighted (raw) total scores that can be attributed to the individual differences on the general factor” (p.: 224). A low OmegaH (< 0.80) for the general factor indicates that the total scores for the RAW can essentially be considered multidimensional (Rodriguez, Reise, and Haviland, 2016, p.: 224).

Table 2. Factor-level outputs for the RAW scale

	ECV (S&E)	ECV (NEW)	Omega/ OmegaS	OmegaH/ OmegaHS	Relative Omega	H	FD
General Factor	0.2	0.2	0.9	0.5	0.6	0.7	0.8
Living authentically	0.2	0.7	0.9	0.6	0.7	0.9	0.9
Staying healthy	0.1	0.6	0.8	0.6	0.7	0.8	0.9
Maintaining perspective	0.1	0.5	0.8	0.6	0.7	0.8	0.9
Finding one’s calling	0.1	0.8	0.9	0.8	0.8	0.8	0.9
Managing stress	0.1	0.8	0.8	0.7	0.8	0.8	0.9
Building networks	0.1	0.7	0.5	0.1	0.2	0.8	0.9
Interacting cooperatively	0.0	0.9	0.6	0.5	0.9	0.5	0.7

Where: ECV—Explained Common Variance; Omega is a model-based estimate of internal reliability of the multidimensional composite. For the general factor, all items are considered; for specific factors, only items loading on that factor are relevant to OmegaS; Relative Omega is OmegaH divided by Omega. For the general factor, this represents the percent of reliable variance in the multidimensional composite due to the general factor. For specific factors, the relative omega represents the proportion of reliable variance in the subscale composite that is independent of the general factor.

H is a measure of construct replicability and “represent[s] the correlation between a factor and an optimally weighted item composite (Rodriguez, Reise, and Haviland, 2016, p.: 230). H is high (H > 0.70) in six of the seven latent factors on the RAW scale. These six latent factors are deemed well defined by their respective test items (Hancock & Mueller, 2001). This demonstrates their greater stability across studies, i.e. adequate construct replicability. The 'interacting cooperatively' sub-scale had an H value of 0.5, see, Table 2.

Factor determinacy (FD) is the correlation between factor scores and factors (Rodriguez et al., 2016). According to Gorsuch (1983), factor score estimates can be used when the FD ≥ 0.9. Based on this criterion, six of the seven latent factors may be interpreted as their FD values approximate the 0.9 threshold. Again, the general factor did not satisfy the Gorsuch (1983) threshold (FD = 0.8), thus further confirming that the RAW scale does not have a strong general factor, that is, it is multidimensional.

3.5.2. Model-level

Table 3 presents the bifactor model-level outputs for the RAW scale. According to Reise, Schienes, Widaman, and Haviland (2013), 2PUC < 0.80, general ECV > .60, and OmegaH > .70 [of the general factor] suggest that “the presence of some multidimensionality is not severe enough to disqualify the interpretation of the instrument as primarily unidimensional” (p. 22). An OmegaH = 0.516, PUC = 0.789; and general ECV = 0.199 suggest the multidimensionality of the RAW scale.

Table 3. Bifactor model-level outputs for the RAW scale

Percent of Uncontaminated Correlations (PUC)	0.814
Explained Common Variance (of General Factor)	0.199
Average Relative Parameter Bias (ARPB)	0.290

Finally, average relative parameter bias (ARPB) serves as an indication of bias across parameters when items are forced into a unidimensional structure (Rogers et al., 2020). The ARPB of 0.290 further suggests that multidimensionality within the RAW scale is substantial (Rodriguez et al., 2016). Overall, the results of the bifactor analyses do not support Hypothesis 2. The resultsconfirm the multidimensionality of the RAW Scale.

3.5.3. Identifying measurement invariance

Work environments comprise of several subgroups Comparing levels of resilience across employee groups is only plausible if the RAW scale exhibits measurement invariance i.e. the RAW scale should exhibit psychometric equivalence across the groups of interest (Putnick & Bornstein, 2016). To determine the invariance of the measurement, configural, metric, and scalar invariance tests were conducted using gender as the primary group of interest given the male/female dichotomy in the workplace.

Configural invariance was used to test whether the latent factors of the RAW have the same defined pattern of free and fixed loads between genders (Lee, 2018; Putnick & Bornstein, 2016). Table 4 shows weak configural invariance of the RAW scale in the study setting (Hu & Bentler, 1999).

Table 4. Composite scores for resilience of work

Levels of resilience at work	Scale	Proportion of observations
Low	≤ 3,68	8.60%
Moderate	3,69 - 5,35	56.40%
High	5,36 - 7	35%
Mean	4.37
SD	1.232
SE	0.531
Max	7
Min	1

Metric invariance was used primarily to examine whether there was factor loading equivalence between groups inherent in the study sample. Specifically, factor loadings were restricted to be equivalent across genders while item intercepts varied freely (Lee, 2018). Based on the results in Table 4, the Chi-square difference test was not significant i.e. (Δχ² = 91.62; df = 162; and p > 0.05) thus confirming the metric invariance of the RAW scale. Scalar invariance determined whether the factor loadings and item intercepts of the RAW scale are equal across genders. The chi-square difference test was insignificant i.e. Δχ² = 29.13, df = 19, p > .005 thus indicating scalar invariance.

By confirming the RAW scale as largely invariant, the results indicate that any differences in observed factor variances and covariances are not attributable to the underlying properties of the RAW scale thus supporting hypothesis 3. However, there could be significant differences between predefined groups i.e. gender, level of work, age, and sector (Lee, 2018), hence the need to perform a multi-group path analysis.

3.5.4. Multi-group path analysis

Multi-group analysis was used to test if there were significant differences in group-specific parameter estimates by gender, level of work, age, and sector. There were significant differences by sector (χ2 = 79.18 (40); p < 0.05) and level of work (χ2 = 69.71 (40); p < 0.05). The results did not suggest any differences in group effect by age (χ2 = 26.10 (20); p > 0.05) and gender (χ2 = 28.91 (22); p > 0.05).

By sector, government employees exhibited greater effect on “maintaining perspective”, “building networks (BN)”, and “staying healthy (SH)” (p < 0.05). Employees in the NGO sector exhibited greater effect on “finding one’s calling (FOC)” and “managing stress (MS)” (p < 005). Private sector employees had greater effect on “living authentically (LA)” (p < 0.05).

By level of work, senior level employees had greater effect on maintaining perspective and managing stress (p < 0.05). Middle level employees had greater effect on finding one's calling and staying health (p < 0.05). On the other hand, low level employees had greater effect on living authentically and building networks (p < 0.05).

3.5.5. Levels of resilience at work

The study calculated the total average resilience at work scores and tested for differences by sex. Most (56%) of the employees exhibited moderate resilience at work, while 35% scored high on this measure. Almost nine percent of the employees had low resilience at work scores. There were no significant differences in the resilienceat work sub-scores by gender (χ2 = 28.91 (22); p > 0.05).

4. Discussion

The results confirmed the internal consistency of the RAW scale in a global south, providing additional evidence on the adequacy of the psychometric properties of the scale between cultures. Conventional CFA confirmed a higher-order seven-factor structure of the RAW scale with 57.14% of TVE. The results support the findings by Winwood et al. (2013), from which the RAW scale was developed.

Bifactor analysis affirmed the multidimensionality of the RAW scale. However, only six of the seven latent factors met the FD ≥ 0.9 threshold (Gorsuch, 1983). A study by Malik and Garg (2018) also found a six-factor (second-order) structure using an Indian sample. In that study, the sub-factor “interacting cooperatively” decomposed (Malik & Garg, 2018). This study’s findings further confirm the instability of the IC sub-test

Overall, this investigation confirmed that resilience at work, as measured by the RAW scale, is best conceptualized as a multidimensional construct as per the original theoretical specifications for the construct. This implies that organizational researchers and leaders alike can use subtest scores to appreciate the levels of resilience among employees to facilitate relevant remediation efforts.

The model exhibited weak configural invariance, while both the metric and scalar invariance were adequate. By confirming the construct validity and measurement invariance, these findings reinforce the perspective that leadership teams can use the RAW scale consistently and reliably to gauge levels of resilience at work across employee subgroups. Subgroup analysis gives leaders and employees alike the opportunity to tailor their interventions to enhance resilience in the workplace.

The measurement invariance test and multi-group path analysis showed that there were no significant differences in resilience at work by gender. Traditionally, and hinged on gender-inclined biases, resilience at work has been associated with masculine values of “strength, robustness, boldness, stoutness, bravery, and not being womanish” (Cooper, 1995, pp. 146–147). Hence, these findings are of practical, policy, and academic importance given that: i) the majority of females, especially in the developing countries like Zimbabwe, face barriers in being selected for certain jobs, i.e. there are systemic and systematic barriers to their recruitment and ascension to higher levels of work, ii) as females, they bear the brunt of being laid off in instances where organizations take a new strategic direction, and iii) more than ever before, the chaotic/complex environment means that both male and female employees face situations which require them to demonstrate resilience at work on almost a daily basis. The confirmation that differences in resilience at work between male and female employees are due to perceptual bias rather than actual behavior is, in fact, a timely and critical reminder for organizational leadership teams.

5. Practical implications of the study

Organizational leaders should conduct periodic resilience surveys at work. Since the RAW scale is multidimensional, leaders can use the subscale tests to get a pulse of the levels of resilience and individual resilience capabilities. The results can be used to draw specific programs and interventions to promote employee resilience. Through dedicated training and mentoring programs (or broader capacity development), employee resilience can be monitored and nurtured as an organizational leadership excellence initiative.

In addition, known antecedents of resilience at work can be promoted. The desire for a work environment that safeguards talent from harm is at the heart of the global Decent Work Agenda (Tchonda et al., 2020). The Psychology of Working Theory (Duffy et al., 2016) recognizes the centrality of work in the lives of individuals and its effect on the physical and mental well-being of employees. Resilient employees guarantee adaptive advantages that allow organizations to flourish. It is therefore important for organizational leaders to develop and nurture resilience at work as an enterprise-wide capability by promoting a decent working environment.

The conservation of resources theory (Hobfoll, 1989, ;2001;, 2011; Hobfoll et al., 2018) recognizes the importance of work and non-work resources in building resilience. Hence, the need for organizations to avoid depleting these resources through mindful job designs and recognition of resource-preserving employee work and nonwork values. On a personal level, generalized efficacy, self-esteem, emotional stability, and a robust internal locus of control are associated with resilience at work (Haglund et al., 2007). In the same vein, self‐efficacy, work–life balance/self‐care, social support, optimism, and humour have also been associated with resilience (Ford et al. (2020); Franken et al. (2020). Politically skilled individuals have also been shown to be resilient at work. Hence, organizational leaders can deliberately inculcate such attributes through dedicated workplace activities.

6. Conclusions

The RAW scale exhibited adequate psychometric measures as depicted by the appropriate Cronbach alpha, Omega coefficient, composite reliability scores, and average variance extracted. The scale also showed evidence of convergent (AVE>0.5) and discriminant validity. EFA produced a seven-factor structure with 57.17% TVE. The seven-factor higher-order model of the scale had adequate fit indices. Bifactor analyzes also confirmed the multidimensionality of the scale, albeit with six factors. Furthermore, the study confirmed the configural and metric invariance of the measure. Most of the employees reported moderate levels of resilience at work with no significant differences by age, sex, level, and sector of work. The results suggest that the RAW scale is a valid and hence applicable measurement tool to comprehend resilience at work in the study context. The RAW scale can be used in government, private sector, and NGO settings for ssessing specific levels of levels resilience in the workplace as well for evaluating proactive resilience building and development initiatives.

7. Limitations and directions for future research

The research acknowledges the limits of using a cross-sectional survey design in determining causality and changes over time. Future research should adopt longitudinal designs to appreciate individual and group changes in resilience at work in space and time. There may also be value in appreciating resilience at work phenomenon at the group level.

Correction

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Open science badges form

This will be considered in future.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data will be available on reasonable request from the authors.

Additional information

Funding

The authors received no direct funding for this research.

Notes on contributors

Hamfrey Sanhokwe

Hamfrey Sanhokwe is a Doctor of Business Leadership candidate at the Graduate School of Business Leadership at the Midlands State University in Zimbabwe. He holds the following qualifications: MSc Biostatistics and Epidemiology; MSc Population Studies; MA Human Rights, Peace and Development. Hamfrey is interested in promoting workplace-related research to influence policy and practice. The follower-leader choices and actions influence how businesses develop, grow, and remain sustainably future-fit hence the bias towards this area. Hamfrey can be reached via email on [email protected].

Simon Takawira

Simon Takawira recently completed an MSc in Public Health from Africa University in Zimbabwe. He also holds a Masters in Population Studies among other qualifications. Simon is a statistical analysis guru with an inclination towards modeling.