Education plays a pivotal role in developing the South African economy (Jackson et al., 2006). Indeed, a good education system is crucial in creating a well-educated citizenry and maintaining socially responsive economic and political systems (Modisaotsile, 2012). According to the World Economic Forum (2014), quality education is essential for those economies wanting to move up the value chain. Regardless of its importance, South Africa’s educational performance remains a grave social concern (Sosibo & Lomlomo, 2014).

Wolhuter (2014) states that South African citizens experience heightened dissatisfaction with the education system upon the announcement of the annual matriculation examination results. Accordingly, South Africa’s education system has failed to produce competent learners who display the necessary skills and knowledge to complete basic literacy and numeracy-related tasks (Department of Education, 2008). Meier (2011) further highlights the effects of a failed education system in which she states that a large number of South African first-year university students cannot read, write and comprehend to a satisfactory level. Evidently so, the 2006 Progress in International Reading Literacy Study (PIRLS) pointed out that out of 40 countries worldwide, South Africa performed the worst in terms of reading literacy levels (Mullis, Martin, Kennedy & Foy, 2007). Raising additional concern is the fact that in 2014 the quality of South Africa’s education system was ranked to be 140 out of 144 economies worldwide (World Economic Forum, 2014).

Jackson et al. (2006) believe that motivated and vigorous educators are essential to help ameliorate these conditions. Similarly, Kubberud, Helland and Smith (1999) argue that well-functioning teachers are essential for the delivery of quality education. This study proposes that job crafting can be used as an effective strategy for high school teachers to experience increased levels of SWB (work engagement and psychological meaningfulness), which could potentially result in the provision of improved quality education.

The Job Demands-Resources (JD-R) Model postulates that job characteristics can be placed into two distinct categories, namely job demands and job resources (Bakker & Demerouti, 2007; Tims, Bakker & Derks, 2013). Job demands are those characteristics of a job that require sustained effort, and as a result, have certain physiological and psychological costs associated with them (Demerouti, Bakker, Nachreiner & Schaufeli, 2001). The second characteristic making up a job, which stimulates personal development and is used in the process of achieving work-related goals, is referred to as job resources (Demerouti et al., 2001).

An underlying assumption of the JD-R Model is that job strain manifests when job demands out-weigh job resources. More specifically, it occurs when a person perceives their environmental (job) demands or constraints to exceed their resources or capacities (Jackson et al., 2006). Conversely, when employees have a large number of job resources at their disposal, they are subject to more positive work experiences, such as increased levels of motivation and work engagement, and as such, they are better able to combat challenging job demands (Bakker & Demerouti, 2007; Tims et al., 2013). According to Bakker and Demerouti (2014), job resources have the potential to serve as a buffer by reducing the strain caused by excessively high job demands.

In order to understand job demands from a job crafting perspective, a distinction between challenging job demands and hindering job demands needs to be made. Challenging job demands are those aspects of a job that have the potential to promote mastery and future gain for the individual (e.g. adding tasks to their jobs and being involved in new projects) (Bakker, Rodríguez- Muñoz & Vergel, 2015; Crawford, LePine & Rich, 2010). Contrastingly, hindering job demands are those constraining aspects of the job that slow down progress and generally have a negative impact on the individual (e.g. role conflict and role ambiguity) (Bakker et al., 2015; Crawford et al., 2010). Challenging job demands have been shown to increase the positive impact that job resources have on work engagement (Bakker & Demerouti, 2014). The JD-R Model serves as the foundation on which job crafting can be conceptualised.

As a concept, job crafting explicitly focuses on employee job redesign (Wrzesniewski & Dutton, 2001), that is, the active process that employees engage in to shape or mould their own jobs. Unlike traditional top-down approaches, where the organisation sets forth the job structure and subsequently hires the appropriate employee to match the job specifications (Tims & Bakker, 2010), job crafting can be viewed as a bottom-up approach initiated by the employee. In particular, the employee starts to change their job to shape their work identity and to create a better fit against their personal preferences and competencies (Leana et al., 2009; Tims & Bakker, 2010; Wrzesniewski & Dutton, 2001).

Job crafting can be described as the change in the level of job demands and job resources initiated by employees for the purpose of making their jobs more satisfying, engaging and meaningful (Demerouti, 2014). According to Berg, Dutton and Wrzesniewski (2008), job crafting encapsulates the changes that employees actively make when designing their own jobs, in which numerous positive outcomes may present themselves, including job satisfaction, work engagement and thriving at work. More specifically, job crafting pertains to the changes that employees make to their physical, cognitive and relational boundaries (Bakker & Demerouti, 2007; Berg et al., 2008; Tims, Bakker & Derks, 2014).

With regard to physical boundaries, employees may engage in efforts to change the form, scope and number of tasks that they are involved in whilst working (Bakker & Demerouti, 2007; Tims et al., 2014). For example, teachers may make structural changes to their classrooms in terms of how the desks and chairs are positioned, the tasks that are completed and the learning materials that are posted on the classroom walls. Changes in cognitive boundaries pertain to the manner in which employees view or think about their jobs (Bakker & Demerouti, 2007; Berg et al., 2008; Tims et al., 2014). The task itself does not change, but the way in which the employee views the task changes (Tims et al., 2014), usually for the better. For example, the impact and value that a teacher believes their job has on the lives of others, in particular their pupils, may change from a pure negative and inhibiting thought to a more positive and enabling thought. Lastly, employees may craft their interpersonal relationships that they engage in during the process of performing their work (Berg et al., 2008; Tims et al., 2014). For example, mathematics teachers may choose to socialise and dominantly interact with staff belonging to the mathematics department, and may find that through this interaction, they slowly become more confident in their ability to teach mathematics. Through the process of changing any one of these job elements, employees slowly change or redesign their jobs and the social environment in which they operate (Bakker & Demerouti, 2007), allowing employees a more engaging and meaningful work experience (Jackson et al., 2006; Wrzesniewski & Dutton, 2001).

Demerouti (2014) asserts that regardless of the type of organisational environment in which employees finds themselves, and no matter how detailed their job descriptions are, there will always be room for them to modify (craft) their work in order to carry it out more successfully, however, a certain amount of job control is necessary. Petrou, Demerouti, Peeters, Schaufeli and Hetland (2012) view work control as a closely-related aspect to the job crafting phenomenon. Research indicates that individuals who are provided with the opportunity to take control in their workplace, are more likely to engage in job crafting behaviours (Berg et al., 2008; Frese & Fay, 2001), which has shown to greatly influence individual and organisational performance (Berg et al., 2008).

It is important to understand that job crafting is not a single event, but rather a process in which an employee engages over a period of time (Berg et al., 2008). Organisations acknowledge that stimulating the proactive behaviours typically involved in the job crafting process is crucial to improving working conditions (Demerouti, 2014), which, given the stressful nature of their work, may be highly relevant to the teaching profession. Bakker, Demerouti, Hakanen and Xanthopoulou (2007) state that providing teachers with the right resources can help them to counteract the challenges of demanding work conditions.

With regard to employee health, it has been noted that job resources generally have a positive impact on employees’ well-being, whereas prolonged exposure to high job demands can affect employees’ levels of job strain, burnout and work engagement, and as a result can have a negative effect on their well-being (Bakker & Demerouti, 2007; Tims et al., 2014). This study focuses exclusively on the individual, and is interested in exploring the potential impact that job crafting has on the SWB of South African high school teachers. According to Tims and Bakker (2010), a well-designed job can result in increased employee well-being. Thus, it is expected that teachers who craft their work will experience increased levels of SWB (work engagement and psychological meaningfulness).

SWB is associated with various aspects of a person’s subjective experience and an evaluation of the quality of life (Diener, Sapyta & Suh, 1998; Keyes et al., 2000). These evaluations may take the form of both cognitive judgements and affective components (Diener et al., 1998). With regard to the former, cognitive judgements entail the perception of one’s expectations being met, whereas the affective component entails the degree to which emotions experienced by an individual are considered favourable (Diener et al., 1998). Diener (1984) defines SWB as a construct that consists of three distinct components, namely life satisfaction, the frequent experience of pleasant emotions (positive affect), and the relative absence of unpleasant emotions (negative affect). A teacher is said to experience SWB if they experience feelings of life satisfaction, satisfaction with other domains such as work and family, and if they are exposed to frequent positive emotions (Vazi et al., 2013).

Employees who experience SWB appear to be more productive and display more positive behaviours than employees who experience low levels of SWB (Diener, Lucas & Oishi, 2002; Field & Buitendach, 2011). Moreover, organisations that foster their employees’ SWB have been shown to reflect better financial performance (Warr & Clapperton, 2010). According to Dessler (1999), organisations that emphasise the development of their staff’s well-being are likely to encompass a workforce with high levels of commitment. Bajorek, Gulliford and Taskila (2014) contend that well-being plays a major role in quality, productivity and performance, and may therefore have an impact on both business effectiveness and profitability.

A study conducted amongst teachers across the United States of America (USA) reported a statistically significant positive relationship between SWB and organisational performance (Ostroff, 1992). Similarly, Dewberry and Briner (2007) found a statistically significant positive relationship between staff well-being and students’ educational performance. More specifically, 8% of the variance in Statutory Assessment Test (SAT) results could be attributed to teacher well-being (Dewberry & Briner, 2007). Given the above, suffice it to say that teachers with high levels of SWB would be more effective than those with low levels of SWB. Slater, Davies and Burgess (2009) suggest that there may be a positive link between teachers’ effectiveness and students’ educational outcomes. In their study, it was established that pupils who had a ‘good’ teacher (in terms of effectiveness) displayed better examination results in comparison to pupils who had been taught by a ‘poor’ or ‘mediocre’ teacher (Slater et al., 2009).

Psychological meaningfulness refers to the significance and value that a person places on his or her own existence and course of life (Taubman-Ben-Ari & Weintroub, 2008). Hackman and Oldham (1980) refer to psychological meaningfulness as a judgement that an individual makes that compares the value of a work goal in relation to his or her own ideals, standards and values. Psychological meaningfulness is experienced when individuals perceive life to make strong emotional sense, and when life’s challenges are considered worthy of energy investment and commitment (Lips-Wiersma & Morris, 2009). Individuals experience meaningfulness when they perceive their actions to be, at minimum, useful, valuable and worthwhile (Barrick, Mount & Li, 2013). Matuska and Christiansen (2008) state that psychological meaningfulness is essential to maintaining resilience during stressful working conditions. Thus, it is imperative for teachers to find psychological meaningfulness in their work, primarily because of the stressful nature of their work. Wolhuter, Van der Walt, Potgieter, Meyer and Mamiala (2012) found teachers to be inspired by experiences that were psychologically meaningful. Rothmann and Hamukang’andu (2013:2) state that ‘psychological meaningfulness at work is an important topic in the debate about education’.

Research suggests that experienced meaningfulness is most likely one of the strongest mediators between work characteristics (i.e. job demands and job resources) and work outcomes (Barrick et al., 2013; Lips-Wiersma & Morris, 2009). Psychological meaningfulness has shown to contribute significantly to work-related outcomes (Schueller & Seligman, 2010), specifically work engagement (Geldenhuys et al., 2014; Olivier & Rothmann, 2007; Rothmann & Hamukang’andu, 2013). In their study, Geldenhuys et al. (2014) found that psychological meaningfulness predicted work engagement.

Given that people spend a proportionate amount of time at work, it is important that they find meaning in their work. Cameron, Dutton and Quinn (2003) believe that the workplace is an important context for individuals to find this meaning. Finding their work to be meaningful can result in a number of positive outcomes for employees and their employing organisations. For example, Geldenhuys et al. (2014) contend that meaningful work can result in engaged and satisfied employees who display greater productivity, commitment and loyalty to their organisations. Seligman (2002) further highlights the benefits of finding meaning, in which he states that meaning allows individuals to transcend, in which this transcendental experience manifests in the promotion of positive social relationships or connecting to a higher power. Frankl (1984) puts forth that meaningfulness is a contributing factor to psychological well-being, and thus values the importance of finding meaning in a person’s life. In the pursuit of a good life, meaningful work is critical (Beadle & Knight, 2012). According to Lips-Wiersma and Wright (2012), when an individual experiences their work as meaningful, it is said to be a subjective experience of the existential significance of their work.

Furthermore, meaningfulness is an important source of motivation (Frankl, 1984) and thus, it is plausible to argue that psychological meaningfulness may act as a driving motivator for teachers. Self-determination theory proposes that the motivational orientations that drive human behaviour can result in a number of benefits for individuals, including but not limited to healthy behavioural regulation and psychological well-being (Deci & Ryan, 1985). Furthermore, the self-determined behaviour that individuals engage in has resulted in positive outcomes for the individual, such as greater persistence in behaviour and enhanced well-being (Deci & Ryan, 2008; Frankl, 1984). It is important to note that psychological meaningfulness does not solely result from favourable working conditions but also from spontaneous and continuous efforts by the individual to find meaning (Isaksen, 2000). This study argues that job crafting may constitute such efforts in finding meaning.

Work engagement is important for individual performance (Bakker & Demerouti, 2009). As previously mentioned, it is defined as a ‘positive, fulfilling, work-related state of mind’ and ’is characterized by vigor, dedication, and absorption’ (Schaufeli et al., 2002, p. 74). Vigour is characterised by heightened levels of energy and mental resilience, the willingness to invest effort in one’s work and the persistence that an individual displays in the face of difficult situations (Henn & Barkhuizen, 2009; Schaufeli et al., 2002). Dedication encompasses a sense of significance, challenge, enthusiasm, inspiration and pride (Henn & Barkhuizen, 2009; Schaufeli et al., 2002). The third dimension, namely absorption, may be regarded as an individual’s feeling of being completely and happily immersed in their work to the extent that they often find it difficult to detach themselves from it (Henn & Barkhuizen, 2009; Schaufeli et al., 2002). Individuals are considered to be engaged when they experience high levels of vigour and dedication, and elevated levels of absorption (Henn & Barkhuizen, 2009).

Work engagement occurs when employees proactively search for motivating challenges and experiences that elicit effective problem-solving (Schaufeli, Bakker & Salanova, 2006). Work engagement is an important indicator of employee motivation and a potential indicator of successful adaption to change (Petrou et al., 2012). Job characteristics, such as providing skill variety, task identity and feedback on results have been reported as drivers of work engagement (Saks, 2006). Rogers (2001) forwards that engaged employees produce long-term business outcomes such as improved productivity.

In modern work contexts, job crafting is seen as a particularly critical path to meaningfulness (Wrzesniewski, Berg & Dutton, 2010). Although little research has been conducted on the relationship between job crafting and psychological meaningfulness, there has been research that has investigated the relationship between work-role fit and psychological meaningfulness (May et al., 2004). In their study, it was concluded that work-role fit positively influences meaningfulness at work (May et al., 2004). In another study, Janik and Rothmann (2015) reported a non-significant relationship between job resources (i.e. co-worker and supervisor relationships) and psychological meaningfulness.

With regard to the second research construct, research has shown that job crafting positively influences work engagement (Bakker, Tims & Derks, 2012; Laurence, 2010; Petrou et al., 2012; Tims et al., 2014). Work engagement may increase as a result of job crafting because employees have the potential to shape their job demands and job resources to suit their own unique preferences and needs (Tims et al., 2014). A study conducted amongst primary school teachers in the Netherlands found that work engagement amongst employees could be predicted by weekly variations in job resources (Bakker & Bal, 2010). The study confirmed that teachers’ enhanced feelings of work engagement were attributable to a resourceful working environment and led to an increase in weekly job performance. It was further suggested that engaged teachers are better able to create their own job resources (Bakker & Bal, 2010). Bakker and Demerouti (2007) highlight that job resources, such as autonomy and social support, can mitigate the effects that high job demands have on employees, and can ultimately act as motivators in achieving improved work performance and most importantly, work engagement.

In their longitudinal study, Tims et al. (2014) used structural equation modelling (SEM) to determine whether or not employees can affect their own well-being through job crafting behaviours. Results indicated that employees who crafted their work, in particular their job resources, showed increased levels of well-being (work engagement) (Tims et al., 2014). Furthermore, they found a direct link between increasing challenging job demands and increased levels of well-being (work engagement). Bakker et al. (2015) found that individuals who crafted their structural and social job resources and challenging job demands reported increased levels of work engagement. They further found that decreasing hindering job demands was unrelated to work engagement. In a study conducted amongst educators in the North West Province in South Africa, Jackson et al. (2006) used SEM and found that job resources made a significant contribution to high levels of work engagement.

Based on the literature provided, the following two hypotheses are formulated:

Using non-probability sampling, a sample size of 400 high school teachers was targeted. A final sample of 251 was obtained, indicative of a 63% response rate. To be eligible to participate, the following criteria had to be met: Participants needed to possess a qualification in education; they were required to have a minimum of one year’s teaching experience; and they needed to be proficient in the English language. The sample comprised both men (30.3%) and women (69.7%), and consisted of Afrikaans (25.5%), English (55%), IsiZulu (6.4%), and Sepedi (4.8%) speaking participants. The majority of the group identified themselves as belonging to the white population (66.5%), with the remaining participants representing the black population (31.9%), which, according to the Employment Equity Act of 1998 (Republic of South Africa, 2014), consists of African, Indian, and mixed race individuals. With regards to employment, 64% were employed in public high schools and 33% in private high schools. Regarding highest education level obtained, 64.9% possessed a Bachelor’s Degree, 25.1% possessed an Honours Degree and 8.8% held a Master’s level qualification. The mean age and tenure of the participants was 40 and 15 years, respectively. Lastly, the sampled teachers devoted an average of 7 hours per week coaching extra-mural activities.

A biographical questionnaire was used in this study to assess sample-specific demographic information relating to the participants’ gender, age, ethnicity, language and highest qualification obtained.

The Utrecht Work Engagement Scale (UWES-9 item) was used to assess the work engagement levels of the sampled high school teachers (Schaufeli & Bakker, 2004). Consisting of nine items, the UWES was scored on a seven-point frequency scale ranging from 0 (never) to 6 (always). The UWES measures three dimensions of work engagement, namely vigour, dedication and absorption (Schaufeli et al., 2006). Examples of item statements include: ‘I am bursting with energy every day in my work’; and ‘Time flies when I am at work’. De Bruin and Henn (2013) administered the UWES-9 across 10 countries and reported mean reliability coefficients ranging between 0.85 and 0.92, with a medium total reliability score of 0.92. The reliability of the UWES-9 total score for this study was 0.93, meeting the required minimum value of 0.70 (Netemeyer, Bearden & Sharma, 2003), and as such, was adequately reliable.

The Psychological Meaningfulness Scale (PMS) (Spreitzer, 1995) was used to measure teachers’ levels of psychological meaningfulness by averaging six items. A five-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree) was used to capture responses. The scale items measured the degree of meaning that teachers discover in their work-related activities (e.g. ‘My job activities are personally meaningful to me’). Studies have demonstrated good reliability for the PMS. For example, May et al. (2004), and Olivier and Rothmann (2007) reported a Cronbach’s alpha coefficient of 0.90 and 0.92, respectively. The PMS performed reliably well in this study with a reported total mean score of 0.93.

The Job Crafting Scale (JCS) was used to assess teachers’ job crafting behaviour by asking them to indicate how often they engaged in certain behaviours at work (Tims, Bakker & Derks, 2012). The scale consisted of 21 items which measured four aspects of job crafting, namely how often employees, (1) increase their structural job resources (autonomy, variety and opportunity for development), (2) increase their social job resources (feedback, social support and coaching), (3) increase their challenging job demands and (4) decrease their hindering job demands (mental and emotional demands). Using a Likert scale ranging from 1 (never) to 5 (often), respondents indicated how often they engaged in each behaviour. Example items include: ‘I make sure that I use my capacities to the fullest’ and ‘I decide on my own how I do things’. The coefficient alphas reported by Tims et al. (2012) for the four dimensions are increasing structural job resources (a = 0.76), increasing social job resources (a = 0.73), increasing challenging job demands (a = 0.77) and decreasing hindering job demands (a = 0.75). The reliability scores reported in this study are increasing structural job resources (a = 0.83), increasing social job resources (a = 0.83), increasing challenging job demands (a = 0.83) and decreasing hindering job demands (a = 0.84). The total reliability score of the JCS was 0.84.

Principals and administrative staff in South African public and private high schools, within the Gauteng province, were contacted via email to invite their teachers to participate in the study. The email explained the purpose of the research and the voluntary, anonymous and confidential nature of the study. This approach proved inefficient and, as such, a more direct approach was deemed necessary, in which the researcher physically visited the high schools to get in contact with key informants. Once permission was granted to pursue data collection, surveys were physically distributed to all teachers in each high school. The scales were completed in paper-and-pencil format and, prior to participation, informed consent was obtained. A preface accompanied each survey which stipulated the voluntary, confidential and anonymous nature of the study as well as the purpose of the study. The contact details of the researchers were also provided in the event that participants wanted access to a summary of the results or if they had any uncertainties relating to the research study. The study was assessed by an ethics committee and ethical clearance was granted.

Statistical analyses were performed using two statistical programmes, namely R version 0.98.1103 (R Core Team, 2013) and SPSS version 22 (SPSS Inc., 2010). The Psych version 1.5.8 (Revelle, 2015), Lavaan version 0.5–20 (Yves, 2012), and Sem tools version 0.4–9 (SemTools Contributors, 2015) packages, in particular, were used. Descriptive statistics were obtained to calculate the mean, standard deviation, skewness, kurtosis, Guttmann’s lambda and Cronbach’s alpha for the scales utilised in the study. Upon analysis, there were no cases of excessive skewness or kurtosis, and as such, the data was considered normally distributed. SEM was performed to investigate the relationships between job crafting, work engagement and psychological meaningfulness. To investigate model fit, goodness-of-fit indices were evaluated. The weighted least squares measure of variance (WLSMV) was applied as it is a robust estimator that orders variables categorically and which deals with violations of normality (Muthén & Muthén, 2010). It is effective and accurate in estimating confirmatory factor analysis (CFA) model parameters (Millsap & Yun-Tein, 2004). The absolute fit indices for this study included the chi-square (χ²) fit statistic, the weighted root mean residual (WRMR) and the root means square error of approximation (RMSEA). The incremental fit indices included the Tucker-Lewis Index (TLI) and the Comparative Fit Index (CFI). The chi-square difference test was applied to the competing SEM models in support of the most parsimonious model.

Step 1: Confirmatory factor analysis: CFA was used to test three different measurement models. Model 1 consisted of three latent variables, namely, (1) work engagement, (2) job crafting and (3) psychological meaningfulness. Model 1 was then compared to the hypothesised measurement model (Model 2), which consisted of six latent variables, (1) work engagement, (2) psychological meaningfulness, (3) increasing structural job resources, (4) increasing social job resources, (5) increasing challenging job demands and (6) decreasing hindering job demands. Lastly, Model 3 was constructed and termed ‘subjective well-being’. Model 3 consisted of one latent variable or general factor, which comprised 36 observed items.

During the CFA, item 7 of the JCS was removed (‘I decide on my own how I do things’) because of its undesirably low factor loading of 0.14 on the scale. The factor loadings produced by the scale items, for each of their respective latent variables, ranged from 0.50 to 0.90. Furthermore, each latent variable and its observed items indicated a statistically significant relationship (p < 0.05). Table 2 presents the goodness-of-fit indices for the three competing measurement models.

Inspection of the chi-square test as a fit statistic for Model 1 (p < 0.000), Model 2 (p < 0.000) and Model 3 (p < 0.000) indicated that none of the models produced a perfect fit. Fortunately, alternative fit statistics were used. Based on the remaining fit statistics, the hypothesised model (Model 2) fit the data best. The fit statistics of the four fit indices were acceptable: TLI = 0.96, CFI = 0.96, RMSEA = 0.06 and WRMR = 1.16. The chi-square difference test (χ² = –1769.674; df = –7; p < 0.00) between Model 1 and Model 2 revealed that Model 2 has the best fit.

The reliability of the hypothesised measurement model (Model 2) and its individual components were evaluated by examining McDonald’s omega coefficient (McDonald, 1999). The omega coefficient values ranged from 0.76 to 0.97 for each of the variables in the model, indicative of good acceptable reliability. The total reliability of the model was reported as 0.97.

Step 2: Testing the structural model: The CFA indicated that the hypothesised measurement model (Model 2) was the most parsimonious model, and as such, was submitted for further analysis. Five structural models were tested and compared in the second step of the analysis. Given that Model 2 was submitted for further analysis, the five structural models were consequently labelled ‘Model 2a, 2b, 2c, 2d and 2e respectively. Model 2a comprised six latent variables, namely work engagement, psychological meaningfulness, increasing structural job resources, increasing social job resources, increasing challenging job demands and decreasing hindering job demands. To account for the issue of multicollinearity, adjustments were made to Model 2b.

Model 2b consisted of five latent variables namely, work engagement, psychological meaningfulness, increasing structural job resources, increasing social job resources and decreasing hindering job demands. ‘Increasing challenging job demands’ was removed because it produced a substantially high correlation in the early stages of analysis. Model 2c, which was termed the ‘resources’ model, comprised of four latent variables, namely, work engagement, psychological meaningfulness, increasing structural job resources and increasing social job resources. This model excluded the two job demands dimensions (i.e. increasing challenging job demands and decreasing hindering job demands) of the job crafting construct.

In a similar fashion, Model 2d was constructed and labelled the ‘demands’ model. Comprised of four latent variables, Model 2d excluded the job resources dimensions from the job crafting variable, namely, increasing structural job resources and increasing social job resources. Lastly, Model 2e was created and compared to the previous four competing structural models.

Model 2e, the new hypothesised structural model, combined the job crafting dimensions that were highly correlated (i.e. increasing structural job resources and increasing challenging job demands) to form one latent variable, which was termed ‘increasing development-based resources’. Model 2e consisted of five latent variables, namely work engagement, psychological meaningfulness, increasing development-based resources, increasing social job resources, and decreasing hindering job demands. The fit statistics of the various competing structural models are presented in Table 3. The chi-square difference test (χ² = 273.393; df = –160; p < 0.00) between Model 2b and Model 2e revealed that Model 2e has the best fit.