Department of Psychology Master Thesis 5PS22E, 30 ECTS Spring 2020
Replication Study on the Mediating Effect of Work Engagement between Self-efficacy and Job Satisfaction
Authors: Josilda Kola & Sara Gholmi Supervisor: Rickard Carlsson Examiner: Auksė Endriulaitienė
research article choosing constructs that were of interest to the authors. The one chosen was a 2020 article from Orgambídez et al., which investigated a model derived from the job demands and resources model (JD-R) and Quality of Working Life (QWL) in Spain with a sample of nurses. Our study followed the principles of Orgambídez et al. (2020) of a cross- sectional correlational design but with a Swedish sample in the IT, software and technology field, with 101 participants. Correlational analysis, confirmatory factor analysis and path analysis were utilized to test four hypotheses and the mediation effect of work engagement between self-efficacy and job satisfaction. Results confirmed all four hypotheses including the mediation effect. Opposite to the findings of Orgambídez et al. (2020) though, there was no direct effect found of self-efficacy on job satisfaction.
Keywords: JD-R, QWL, self-efficacy, work engagement, job satisfaction, replicability.
Open Science Forum
This study followed the regulations of open data, open materials and pre-registration of the hypotheses as well as the study design according to open science practices. The pre- registration of the study is available at: https://osf.io/5vt3k and the materials and the dataset are available at https://osf.io/mqk9e/.
Replication Crisis ... 2
JD-R Model ... 3
Self-efficacy ... 4
Work Engagement... 5
Quality of Working Life (QWL) ... 5
IT, Software and Technology Companies ... 6
Models and Present Study ... 8
Self-efficacy and Job Satisfaction... 9
Self-efficacy and Work Engagement ... 10
Work Engagement and Job Satisfaction ... 10
Self-Efficacy, Work-engagement and Job Satisfaction ... 11
Method ... 12
Participants ... 12
Procedure and Design... 12
Ethical Considerations... 13
Materials ... 14
Analysis ... 15
Correlational and Descriptive Statistics ... 16
Confirmatory Factor Analysis ... 17
Path Analysis ... 19
Discussion ... 22
Limitations ... 24
Future Recommendations ... 25
Implications ... 25
References ... 27
Theoretical Background
The aim of this study is divided in two levels; firstly, the authors aimed to apply and test the applicability of the replication method on a published article in the area of psychological and/ or occupational field. Secondly, the chosen article for replication was based on constructs that were of interest to the authors. More specifically, constructs of interest were self-efficacy and how that affects job satisfaction. Therefore, after deciding to conduct a replication study, the aim was to select as much as a recent article as possible, that included these two variables in their model. Thus, out of the studies found, the ones that conducted a longitudinal type of research design were excluded, since the time period for this research was limited. The article chosen for the replication was conducted by Orgambídez et al. (2020). Their study is called “Linking Self-efficacy to Quality of Working Life: The Role of Work Engagement”, which was published in the Western Journal of Nursing Research.
The authors examined the effect of self-efficacy and work engagement on job satisfaction and affective organizational commitment. Job satisfaction and affective organizational commitment were taken into consideration as indicators of Quality of Working Life (QWL).
The study revealed that work engagement partially mediated the relationship of self-efficacy and job satisfaction, moreover it was found that the effect of self-efficacy on affective commitment was dependent on work engagement (Orgambídez, Borrego & Vázquez- Aguado, 2020). Therefore, with this replication, the opportunity to test the applicability of their proposed model and findings in a different work field as well as culture was introduced.
This would determine if the main principles, derived from the JD-R model hold true in different occupational fields. According to the main idea of the usability of JD-R, it is supported that the model can be applied to any work field (Demerouti et al., 2001). The research question of this present study was to replicate a variation of the constructs of the model introduced by Orgambídez et al. (2020) and test if indeed work engagement mediates the effect of self-efficacy on job satisfaction.
The study of Orgambídez et al. (2020) as already mentioned, was based upon the JD- R model and how personal resources (individual characteristics) affect QWL, in this case job satisfaction. They based their model on the aforementioned theories on the self-efficacy theory (job resources) and in an extended version of JD-R by Schaufeli (2017) where according to him job demands in combination with adequate job resources lead to work engagement and consequently lead to positive outcomes. The positive outcomes were based
on Nowrouzi’s et al. (2016) concept of QWL. According to Nowrouzi job satisfaction and affective commitment are indicators of quality of working life and wellbeing generally.
The JD-R model explains strain as an outcome when the person experiences an imbalance of job demands and job resources (Bakker & Demerouti, 2007). More specifically, the JD-R model is a theoretical framework that focuses on the aids that help employees reduce stress and accomplish goals. Although some jobs have different work features, the model can still be divided into two conditions: job demands and job resources. All professions face different job demands, and that may entail cognitive, emotional and psychological efforts (Harney, Fu & Freeney, 2018).
The original study was conducted in Spain with nurses in a cross-sectional correlational investigation. This replication study, on the other hand, took place in Sweden in companies where the field and structure of work is focused on IT, software and technology. One of the reasons for selecting this work field for the investigation was that IT, software and technology companies are a growing market in Sweden and an increase of start- ups in this field has been observed as well. Therefore, the need to investigate in depth this type of work field is brought to surface and may be in demand in the near future too. Another reason is the fast paced and changing work structure of such a work field, as well as the high level of competitiveness (internal and external) that describes the work environment of this field. Thus, job demands are presumed to be high, which renders the need for an increase of job resources as well (Chapke, 2011; Passos et al., 2014).
Replication Crisis
Replication investigates if the statistical results remain the same, when studies are repeatedly reproduced. Τhe broadly accepted measurement used in replicated studies is the p-value of significance p ≤ .05 as well as the effect size of that significance. In science replication is of importance, due to the fact that it provides more reliable results for a study and it also strengthens the reliability of already existing theories (Maxwell, Lau & Howard, 2015). In psychology as well industrial psychology (I-O) discussions have been raised on the exceeding amount of false positive results in replicated studies leading to the so-called replication crisis (Maxwell et al., 2015). This became more apparent in the 2015 Open Science Collaboration (OSC), which revealed 97 % of original studies had produced significant results, whereas the OSC indicated significant results for only 36% of them.
Schmidt and Oh (2016) argue that the problem does not lie on a lack of replications but on a variety of reasons such as low power (sample size) that the original studies have, as well as ignoring studies that yield non-significant results. Boos and Stefanski (2011) moreover, criticize the catholic use of the p value as the only basis for the measurement of significance, which oftentimes is not accurate and leads to false positives. Other factors that create and enhance the false-positives besides the p-value, are the flexibility of the types of analysis, the types of data collection as well as the reporting of results (Simmons, Nelson &
Simonsohn, 2011). Simmons et al. (2011) strongly support conducting research with open science principles by providing the data and materials, as it promotes transparency and it allows for increase in the numbers of replications, which can reduce the phenomenon hypothesizing after the result is known ( HARKing) (Kerr, 1998). Mazzola and Deuling (2013) aimed to test Kerr’s (1998) HARKing on I-O psychology by investigating studies published in journals such as i.e. Journal of Applied Psychology, Personnel Psychology. They did a comparison between journal articles (215) and dissertations (127) to see which reported more significant hypotheses (Mazzola & Deuling, 2013, p. 280). They found that journal articles had a significantly higher rate of supported hypotheses than dissertations. Thus, supporting the idea of the publication bias as part of the problem too.
JD-R Model
Job demands are described as aspects that are of a psychological, physical, organizational or social nature that create tension and stress to both body and mind, thus they demand constant and efficient adeptness both physically and psychologically. Examples of job demands can be high workload, organizational competitiveness and intense physical work pressure, consequently leading to stress and physical strain. Job resources on the other hand, even though they may not be directly connected to the job demands, they act as preserves of important work values. Job resources, similar to the job demands can be of an individual nature (physical or psychological), organizational and social. Examples of job resources are high mental and emotional stability, high self-efficacy, good physical health, job security, autonomy and supportive work climate (Bakker & Demerouti, 2007; Harney, Fu & Freeney, 2018).
Xanthopoulou et al. (2007) emphasize the importance of personal resources (i.e. self- efficacy) as coping mechanisms towards job demands (i.e. physical and psychological strain). They are also the means with which employees identify their work environment in an organizational and individual level (Xanthopoulou, Bakker, Demerouti & Schaufeli,
2007). Personal resources are mostly associated with resilience, describing one’s ability to manage and influence the demands of the environment. They are also linked with positive psychological and physical outcomes since they appear to counteract the effects of job demands of emotional and physical strain as well as job dissatisfaction (Bakker & Demerouti, 2007; Xanthopoulou et al., 2007).
Self-efficacy
Self-efficacy is derived from the social cognitive theory (SCT), explored by Albert Bandura (1986). Self-efficacy, which is the personal resource of interest in this study, is the belief one has in her/his abilities to successfully deal with challenging and even stressful, threatening situations and acts as a coping mechanism. Also, it is focused on the actions/
doing perceptions. The definition of self-efficacy specifically is that “ belief in one's capabilities to organize and execute the courses of action required to produce given attainments” (Bandura, 1997, p.3). According to the social cognitive theory, SCT self- efficacy is the most pervasive and central in the perception of control towards external stimuli and a powerful determinant for motivation. Beliefs of self-efficacy function in cognitive processes, motivational processes, affective processes and selection processes (Bandura, 1997). SCT is an extension of the social learning theory (SLT), which views human learning and behavior as part of a social context. It is a causational model (reciprocal causation) on the interaction between the person, the environment and the behavior (a triadic model).
Emphasis is put between the internal and the external social reinforcement (Bandura,1986).
It focuses on the particular ways each individual experiences certain behaviors and how these behaviors are obtained, influenced and maintained by the personal circumstances of the environment they are learned from. Among the triadic factors continuously interacting with each other, there is not any emphasis put on one specific factor. Each factor can be stronger or equal to the other based on particular circumstances. Influential in this theory is also the concept of human agency and how it regulates the nature of behaviors. Agency refers to the ability of the individual to perform certain actions in order to attain goals. More particularly Bandura (1999) perceives human agency as emergent interactive. This means that the aforementioned triad determines the actions and their motivation. Focused on the causal structure, agency functions in different mechanisms such as through self-efficacy beliefs, goal representations, anticipated outcomes, hierarchical duality in the creating and regulating actions and the ability to override feedback control (Bandura, 1999).
Efficacy beliefs are considered to have an essential part of our lives. Self-efficacy may moreover help to cope with unpleasant feelings of inadequacy and high demands in the workplace, thus proving to be an adequate job resource mechanism against job demands.
Bakker and Hueven (2006) suggest that employees with high self-efficacy are capable of handling emotional dissonance in the workplace, as compared to employees with low self- efficacy. This provides the information that people with high self-efficacy have the ability to cope with demanding tasks and solve problems (Bakker & Heuven, 2006; Bandura, 1997).
Work Engagement
Work engagement according to the JD-R model is an outcome of job resources (i.e.
self-efficacy), which can lead to job satisfaction (Schaufeli, 2017). Yalabik et al. (2013) define work engagement as a motivational state of mind, including feelings of independence, fulfillment, and pervasiveness. Work engagement is divided in three groups which are the following: vigor, dedication and absorption. Vigor refers to being energetic, enthusiastic and having beliefs of achievement. Dedication refers to a psychological identification and captures employees’ feelings of pride, inspirations, and commitment to their work. Lastly, absorption is described by being attached to one’s work, and being deeply concentrated on the tasks. As a general rule, engaged employees are characterized as having energy, being enthusiastic and feeling attached to their work (Yalabik, Popaitoon, Chowne & Rayton, 2013).
Quality of Working Life (QWL)
According to Davis and Cherns (1975), QWL is defined by four distinct aspects that focus on the individual: integrity of the self (how much one is appreciated and esteemed and it is related to the individual identity), integrity of the body (physical health and the immediate connection to the physical work environment), social growth and development (it involves the growth of the society, usually in a democratic context, that enhances individual wellbeing) and lastly integrity of life roles (the level of consideration put on the individual’s personal and work roles and responsibilities). QWL as a concept has been indivisible with the concept of job satisfaction, meaning that one cannot consider and investigate quality of working life without measuring job satisfaction as a part of it.
Huzzard (2003) in a report on the Swedish market and on companies that cater internationally, focused on organization competitiveness and QWL. He supports that, even though working conditions and working life have improved in the Swedish work
environment, the constant need to increase competitiveness and improve performance in as low a cost as possible has a serious effect on QWL. Abraham (2015) while investigating workplace happiness of employees in software companies, found that there is a mix between individual characteristics and organizational characteristics. Elements such as resilience, vitality as well as satisfaction with personal life are labeled as individual factors affecting workplace happiness and consequently satisfaction. Whereas, good team management and perceived trust from the manager to the employees, are labeled as organizational elements that enhance workplace happiness (Abraham, 2015).
Martel and Dupuis (2006) note that contrary to the various theories and models through the decades, it is almost impossible to make a distinction between QWL and quality of life (QOL) since the one affects the other. The triadic components of the individual (employee), the environment (organization) and the social (community) interact with each other making it almost difficult to place these elements in places of hierarchy, of what causes the other. Furthermore, in contrast with the above perceptions they argue that QWL should be defined and related so closely to job satisfaction. This comes to contrast what the view Orgambídez et al. (2020) holds of job satisfaction being an inseparable construct from QWL and a determinant for it.
IT, Software and Technology Companies
According to Harper and Utley (2001), information technology (IT) is widely used and implemented in numerous organizations and it can function as a separate working field.
IT has been expanding within the range of services, applications, processes and equipment and it usually includes three categories: telecommunications, computers and multimedia data (Harper & Utley, 2001). Employees in the IT sector are usually given the opportunity to work independently and use adequate tools to execute their jobs. Teamwork is also used to enhance motivation as well as providing decision-making in environments of IT systems (Harper &
Utley, 2001). The field of both Technology and IT is derived from Software. Mathew et al.
(2012) suggest that this working field (software, IT, and technology) uses almost the same pattern and work structure, independent of the culture it is applied on (i.e. East or West) and emphasis is also put on teamwork (Mathew, Ogbonna & Harris, 2012).
Cordero et al. (1998) describe the environmental aspect of technology companies.
The authors state that employees working in this field are expected to handle time-pressure, make an effort and be dedicated to their work. Technical professions have to incorporate
functional aspects before making a decision, and that can be demanding for the employees.
In this working field, a great deal of attention has been put into working teams, as the profession tries to integrate the members to attain a successful team (Cordero, Farris &
DiTomaso, 1998). Moreover, IT, software and technology work structures are described to be competitive and fast paced work environments. What is sought after and is successful today as a market value may not be so in six months or a year from now. As a result of this, anxiety and high levels of stress are expected to affect employees and in turn it affects the overall group climate that such companies usually rely upon in order to enhance competitiveness (Cordero et al., 1998). However, competitiveness is highly regarded, since research has proven repeatedly that it leads to the generation of new ideas and innovation (Bonaccorsi, 2011). Nevertheless, the need to remain competitive has an effect on the overall quality of working life (QWL) (i.e. job satisfaction) long-term (Huzzard, 2003). In order for competitiveness (job demand) to be effective and not negatively affect e.g. job satisfaction, the organization needs to regulate the pace with which it implants new strategies, rather than focus on the accumulation of certain strategies in one specific time to achieve advantage in the competitive information technology market. This in turn has a positive long-term effect on performance (Henderson & Mitchell, 1997; Huzzard, 2003).
Harper & Utley (2001) suggest that e.g. in the IT field, the most important aspect in an organization is the underlying values that are held by its organizational culture. However, this may intervene with the personal cultural values if not controlled by the management. It is the cultural values of the team members though that determine if the structure is successful.
Passos et al. (2014) support that when team members’ values clash, then not only this does not improve competitiveness and efficiency but it may also increase turnover rates and difficulty to adapt to new software work practices (Passos, Mendonça & Cruzes, 2014;
Harper & Utley, 2001). Chapke (2011) in addition to the above focused on software companies and suggests that human resource (HR) practices are also important to increase creativity and enhance performance. Elements such as providing sufficient training to newly recruited and old employees as well as providing rewards (monetary compensations) and recognizing good performance increases confidence at work and satisfaction with their job.
Moreover, an element that seems to appear in Passos et al. (2014) as well, is that of culture.
Therefore, Chapke (2011) highlights the importance to recognize the different cultural values and construct a diverse work environment. Diversity as is supported here, appears to be a key component for increased competitiveness in the international market for software companies,
supporting that diverse work teams provide higher levels of creativity and innovativeness.
Thus, HR must regulate a work environment that recognizes and values diversity in order to be more competitive (Chapke, 2011 & Passos et al, 2014).
Models and Present Study
Orgambídez et al. (2020) on their proposed research model, in addition to testing the relationship between self-efficacy, work engagement and job satisfaction, they also included as an outcome affective commitment. In addition to proposing that affective commitment was indirectly affected by self-efficacy through work engagement, they further indicated that affective commitment was also caused by job satisfaction. After extensive research, we found unsatisfactory support that affective commitment is caused by job satisfaction or self- efficacy. Furthermore, in the case of their model, affective commitment is presented as an extension of the three other measures. Specifically, it is shown as a result of the mediating relationship between self-efficacy, work engagement and job satisfaction. Their adjusted model can be seen in Figure 1. Moreover, this adjusted model came up from the results of confirmatory factor analysis (CFA) after testing the goodness of fit of their model. Whereas in their proposed hypotheses and initial model they did not predict or hypothesize a direct effect of self-efficacy on job satisfaction.
The reason of choosing to replicate a modified version (excluding affective commitment) of their adjusted model was simply because according in this model they support direct and indirect effect and partial mediation between self-efficacy and job satisfaction, despite the fact that they had no hypothesis expressing such effects (direct or indirect). Choosing the same type of analysis, while remaining true to the principles of replication method, we wanted to test how much verifiable and reliable is to support their findings and conclusions by using correlational designs and path analysis. Furthermore, the path would indicate, direct and indirect effect despite the fact that no such hypothesis was presumed.
Figure 1
Proposed Adjusted Model by Orgambídez et al. (2020)
Based on the aforementioned theoretical models and theories, the focus of the current study’s research question is to investigate the impact of self-efficacy (independent variable) on employees’ job satisfaction (dependent variable) with the mediating effect of work engagement between the independent variable and the dependent variable. This is also in the present study’s model indicated in Figure 2.
Figure 2
The Tested Model based on Orgambídez et al. (2020) and JD-R Model
Self-efficacy and Job Satisfaction
Based on the study of Orgambídez et al. (2020) and supported from the JD-R model self-efficacy seems to play an important role on how job satisfaction is perceived and accepted. This is supported by various studies in various work fields. For example, it has been found that teachers’ self-efficacy perceptions had a direct effect on job satisfaction perceptions (Viel-Ruma, Houchins, Jolivette & Benson, 2010; Karabiyik & Korumaz, 2013).
Lai and Chen (2012) with a sample of sales employees from automobile companies, also support that self-efficacy has a positive effect on job satisfaction. The following hypothesis is based upon the results of the original study as it was not included in their hypotheses.
However, since we investigated the model that they adjusted we proposed:
Hypothesis 1: Self-efficacy is positively related to job satisfaction.
This hypothesis though does not state that a direct effect of self-efficacy on job satisfaction is assumed as is presented in the model. However, if a significant positive relationship is presumed it can enhance the support of the mediation model and the indirect effect that Orgambídez et al. (2020) supported.
Self-efficacy and Work Engagement
Buric and Macuka (2018) measuring the effect of emotions on self-efficacy and work engagement among subject teachers in Croatia, found these three constructs have a significant relationship among each other. More particularly, teachers who experience emotions such as pride, joy and love when working with their students, were generally more vigorous, and were confident enough to deal with barriers which consequently engaged them more in their work. In addition, Federici and Skaalvik (2011) examining the relation between self-efficacy and work engagement among principals working in the elementary school as well as middle school in Norway, discovered that principals with high self-efficacy are more engaged in their work (Buric & Macuka, 2018; Federici & Skaalvik, 2011). According to the above, the original study of Orgambídez et al. (2020) had, “Self-efficacy is positively related to work engagement”. The following hypothesis is the same Orgambídez et al. (2020) used as well:
Hypothesis 2: Self-efficacy is positively related to work engagement.
Work Engagement and Job Satisfaction
Studies have discovered a significant relationship among work engagement and job satisfaction. Job satisfaction refers to a feeling of contentment, motivation and balance in employees' work (Orgambídez-Ramos & de Almeida, 2017). The study of Yalabik et al.
(2013) revealed that employees in the banking sector who are engaged in their work, experienced increased job satisfaction in their workplace. It further demonstrated that engaged employees produced higher work performance (Yalabik et al., 2013; Orgambídez- Ramos & de Almeida, 2017).
Yeh (2013) explored the effect of tourism involvement on work engagement and job satisfaction on hotel frontline employees. It was found that tourism involvement was positively related with work engagement as well as job satisfaction. This implies that employees working in the field of hospitality industry experience work engagement, considering access to more resources and feel more self-motivated towards their working
place. Therefore, these findings suggest that the more engaged the person is in their work, the more likely they are to be satisfied with their jobs (Yeh, 2013).
Kamalanbhan and Sai (2009) investigated the relationship of work engagement and job satisfaction on employees in India within the IT Technology Industry. They found that employees' work engagement in the IT sector leads to an increase of job satisfaction in the workplace (Kamalanbhan & Sai, 2009). These findings support and lead to the following hypothesis “Work engagement is positively related to job satisfaction” (Orgambídez et al., 2020). The following hypothesis too is the same Orgambídez et al. (2020) used:
Hypothesis 3: Work engagement is positively related to job satisfaction.
Hypotheses 1, 2 and 3 were chosen to be investigated in additions to the mediation model as they would provide more information and understanding on the relationships among the three constructs: self-efficacy, work engagement and job satisfaction. Despite the fact that enough evidence was provided from the literature related to the relationships among the above constructs, lack of investigation concerning the field of IT, software and technology made it necessary to test H1-H3.
Self-Efficacy, Work-engagement and Job Satisfaction
Yakin and Erdil (2012) conducted a study with the aim of investigating the relationship between self-efficacy, work-engagement and job satisfaction, indicating a significant relationship between the three variables. Likewise work-engagement was shown to have an impact on intrinsic job satisfaction. In another study Skaalvik and Skaalvik (2014) with a sample of teachers working in elementary as well as middle schools in Norway found that self-efficacy and teachers’ autonomy were significantly associated with work engagement and job satisfaction (Yakin & Erdil, 2012; Skaalvik & Skaalvik, 2014). Caesens and Stinglhamber (2014) also found that employees with high self-efficacy develop feelings of dedication, absorption and become enthusiastic about their work (Caesens &
Stinglhamber, 2014), thus they become more engaged at their work.
According to the above studies and in the relationship among self-efficacy, work engagement and job satisfaction the following hypothesis originally from the replicated study
“The positive relation between self- efficacy and job satisfaction is mediated by work engagement” (Orgambídez et al., 2020) was presumed:
Hypothesis 4: The positive relation between self-efficacy and job satisfaction is mediated by work engagement.
Method Participants
The survey was distributed to 13 IT, software and technology companies from different regions of Sweden. These companies were selected from an extensive list of companies in Sweden and after agreement the survey was distributed to them. The overall number of employees in those companies was N = 839. The overall number of participants in this study was 101, with a response rate of 12,04%. Referring to the demographic characteristics of the participants, out of 101 participants 47 were women (46.5%) and 54 were men (53.5%). The mean age score was 36.09, and the mean score for years of working experience (tenure) was 13.34 years.
Procedure and Design
A cross-sectional correlational survey study was conducted, where employees from different companies and regions in Sweden were asked to voluntarily participate in the study.
The type of study design was based on the original study design of Orgambídez et al. (2020) in order to maintain accuracy for this replication. The survey was distributed in a web-based form to the employees of the companies that participated in the study through their head managers or through the HR departments we contacted. The survey was available to be filled via an email link or an App download . The planning and the preparation for this study started in late January 2020 and it was expected to conclude late May 2020. The study followed open science practices, where all the design and materials were pre-registered in the Open Science Forum (OSF), prior to the data collection.
The data collection started on February 27th, 2020 and in the beginning of the study the data collection aimed for companies operating in the South of Sweden. However, due to the fact that limited numbers of employees could be found working in such types of companies (IT, software and technology) the allocation was expanded to take place in the broader context of Sweden and its regions. It was aimed that the data collection would continue until the end of March or April 2020. Alas, the worldwide pandemic Covid-19 intervened and affected the data collection while it was in its early stages. More specifically, until March 11th, 2020 while the data was being collected and responses had a good flow we were simultaneously approaching and inviting new companies to participate in our study.
On March 12th and 13th, 2020 though, the responses suddenly dropped quite substantially.
In those two days, even though the Swedish government had not declared an official countrywide quarantine, many companies in Sweden decided for their employees to work from home. In addition, the following days a few of the companies that had agreed to participate were not responding anymore after we tried to get in touch with them. Others stated that the distribution would be an additional burden for employees due to Covid-19.
Therefore, on March 16th, 2020 when universities closed and students worked from home and more and more companies did the same, we made the conscious decision not to contact anymore companies besides the 13 companies that had already given us a positive answer.
Although, this decision may have affected the power of study, if we kept approaching companies in the midst of the pandemic crisis then the responses and results would be prone to biases that could act as confounders and not related to our study objectives (self-efficacy, work engagement and job satisfaction). More specifically, responses could have been affected by the sudden crisis that could have caused an attribution error effect. Only measuring the relationships between the three variables without taking into consideration the external circumstances that could confound those relationships (i.e. increase of job insecurity, fear of unemployment, increase of stress levels, social isolation) would lead to unreliable and biased results. The data collection was kept open only for the companies that we had already contacted and received positive responses and was finally closed at the end of April 2020.
Ethical Considerations
Prior to the start of the study and the data collection, several ethical considerations and regulations were taken into account and followed the regional Ethical Review Board (EPN) guidelines. The survey included an introduction section prior to the questionnaire, which informed respondents about the purpose of the study and its aim. They were also informed that participation was voluntary and that they could withdraw at any given time while they were filling the survey. In accordance with the general data protection regulation (GDPR) EU 2016/679, respondents were assured that any personal information given was going to be protected to ensure the confidentiality and anonymity of their protection. To ensure the personal data protection of all individuals, all demographic data was filed and protected with an encrypted password and was not made public. In the open science database only the rest of the data was provided, thus excluding the demographic information. The data collection was aimed for statistical use and the fact that the demographical data was not made
public allows this study to be used for both academic and possibly publishing purposes in the future.
Materials
In this section the scales used in the survey are indicated given the fact that the survey was measuring three different constructs: self-efficacy, work engagement and job satisfaction.
Self- efficacy was measured with The Swedish General Self-Efficacy Scale (S-GSE) which is originally from Jerusalem and Schwarzer (1995) and translated in Swedish by Koskinen et al. (1999). The S-GSE was also validated by Löve et al. (2012). The scale was comprised of 10 items with a 4-point Likert scale ranging from 1 (Not at all true) to 4 (Exactly true). An example of the items was “I can always manage to solve difficult problems if I try hard enough” (Löve Moore & Hensing, 2012). The self-efficacy scale after testing with Cronbach's alpha for this study showed a high reliability alpha of .91. Orgambídez et al.
(2020) also reported high alpha score .94.
Work engagement was measured with the short version of the Utrecht Work Engagement Scale (UWES-9) by Schaufeli et al. (2002). The Swedish translation was provided by the same authors. The scale consists of 9 items with a 7-point Likert scale ranging from 0 (Never) to 6 (Every day). An example used for this scale was “When I am working, I forget everything else around me” (Schaufeli et al., 2002). The Work-engagement scale for this study revealed a Cronbach's alpha of . 94, similar to Orgambidez et al. (2020) where the alpha score was .90.
Job Satisfaction was measured with the short form of the Minnesota Satisfaction Questionnaire (MSQ) created by Weiss et al. (1977). The translation into Swedish was made by Nystedt (1993; 1994 ). The scale was validated by Nystedt et al. (1999). The scale consists of 20 items with a 5-point Likert scale ranging from 1 (Very satisfied) to 5 (Very dissatisfied).
An example of the items for this scale was “The chance to do something that makes of my abilities” (Nystedt, Sjöberg & Hägglund, 1999). Job satisfaction revealed Cronbach's alpha of .91. The job satisfaction scale was the only scale that was not the same as the original study. This was due to the fact that Orgambidez et al. (2020) used the Job Satisfaction Scale S10/12 by (Meliá & Peiró, 1989), which was only available in Spanish and it is used in Spanish speaking countries. Firstly, we aimed to directly translate the scale from Spanish to English and Swedish, despite the fact that it would be prone to reliability limitations. When
this effort was made though, several items in the scale, even after the translation, did not have an applied meaning in the Swedish context and organizational culture as they had in the Spanish culture. Therefore, this scale could not be used for the Swedish sample. The main reason for choosing the MSQ as a replacement for the S10/12 was because it is widely used in Swedish studies, and the overall meaning of 20 items was similar to what Orgambídez et al.(2020) were testing with their chosen scale.
Three demographic items were also included measuring gender, age and years of work experience. Age and years of work experience were asked to be answered with the exact value, whereas gender was in a multiple-choice form (female, male and other) (Hughes, Camden, Yangchen & College, 2016).
Analysis
In the pre-registration was stated that SPSS Amos 26 and JASP software were going to be used for the analysis of our model. However, Amos software was not used in the analysis after the data was concluded. This was due to a technical issue during the design of the model in the confirmatory factor analysis. As well as time saving decision while CFA was performed. It was decided to only use SPSS 26 from IBM. Moreover JASP ver. 0.10.2.0 as submitted in OSF with the addition of PROCESS Macro ver. 3.4.1 by Hayes (2017). A confirmatory factor analysis (CFA) was used as it was necessary to follow the methodological rationale and do a replica analysis of the original study as a means of evaluating the goodness of fit of the model based on the variance left after the latent factors are taken into consideration (self-efficacy, work engagement and job satisfaction). CFA was conducted in JASP to test the relationship between the three latent factors and how well the indicators (39 items) loaded in each latent factor. More specifically, CFA is a method to confirm the factor design of observed variables, it is a means to investigate that the relationship between observed variables and the underlying latent factors exists (Hayes, 2013). To confirm the fit of the model different indices were used such as the root mean square error of approximation (RMSEA), the standardized root mean square residual (SRMR), the Tucker – Lewis Index (TLI) and the comparative fit index (CFI).
Mediation analysis (Path analysis) was conducted with PROCESS Macro to test the relationship among the three constructs and if the path (chain) of the model holds true. More specifically, how and why self-efficacy (IV) affects work engagement (M), which in turn affects job satisfaction (DV). The path analysis followed the principles of linear regression
and specifically the ordinary least squares (OLS) regression (Hayes, 2013). OLS regression type of path analysis, as reported by Hayes (2013) is only appropriate when the outcome (or dependent) variables are continuous as was the case for this study (job satisfaction).
In order to test if there is a relationship between the variables (self-efficacy and work engagement, self-efficacy and job satisfaction, and work engagement and job satisfaction) Pearson's correlation coefficient (r) was used to examine the associations between the three variables and providing support for the hypotheses H1, H2 and H3. It is a statistic that can be applied in linear regression and it tests the linear dependence between two variables (Field, 2009). Path analysis results moreover were referred to specify the path and causal effect of the correlations, as an additional method of support. Despite the support that suggests that correlation does not imply causation states that by using a variety of different designs to test a research question and hypothesis, it only strengthens the causal conclusions, thus is presumed to increase reliability (Rohrer, 2018).
Results Correlational and Descriptive Statistics
Pearson’s correlation coefficient r revealed a significant and positive correlation between self-efficacy and job satisfaction r = .48, p < .001, thus supporting hypothesis 1, that self-efficacy is positively related to job satisfaction. A significant and positive correlation was also found between self-efficacy and work engagement r =.64, p < .001, thus supporting hypothesis 2, that self-efficacy is positively related to work engagement. Finally, a positive and significant correlation between work engagement and job satisfaction r = .65, p < .001, supports hypothesis 3, that work engagement is positively related to job satisfaction. Table 1 indicates the means and standard deviation of the employees for all the three variables tested.
All the mean scores were revealed to be moderate to high.
Table 1
Total Scale Value Means, Standard Deviations and Correlations.
N M S.D. 1 2 3
1. Self-efficacy 101 30.2 5.01 1.00
2. Work engagement 101 47.67 8.87 0.64 1.00
3. Job satisfaction 101 78.32 10.47 0.48 0.65 1.00
Confirmatory Factor Analysis
Confirmatory factor analysis (CFA) was conducted using the software JASP ver.
0.10.2.0 to test the goodness of fit of the model aimed for testing. This model included the three latent factor of self-efficacy, work engagement and job satisfaction based on data from 101 employees, with no missing data from 13 IT, software and technology companies.
Starting with CFA’s assumptions, which are the multivariate normality, an adequate sample size of 200 and above participants, the data should include random samples, and a correct priori model specification (Statistical Solutions, 2013).
For the assumption of multivariate normality, using the Mahalanobis distance as well as a P-P plot, which can be viewed in Figure 3, it was assumed that the variables were distributed normally on a univariate level, despite a slight variation from the diagonal line.
The assumption that a required sample size though of at least N ≥ 200 in order to be considered as sufficient for the CFA is rendered a limitation, since the overall sample size of this study is N = 101, therefore not met adequately. Regarding the assumption of a random sample, since the selection of the companies was done from a random list, we assume that this assumption was met. The assumption of having a correct priori model specification, is controlled by the fact that this study is a replication, which tested an already existing model formed by Orgambídez et al. (2020). Therefore, any omission of latent and confounding variables that may have not been included in the original model may be a limitation for the study of Orgambídez et al. (2020) and based on the principles of the JD-R model.
The CFA revealed the following fit indices values with Chi-square χ2 (699)= 1195.6, p < .001, with Tucker-Lewis fit index (TLI) = .78,comparative fit index (CFI) = .80, the standardized root mean square residual SRMR = .86 and root mean square error of approximation RMSEA= .08 (CI=.07 - .09). These values indicated an acceptable enough fit between the model and the observed data, which were adequate support to proceed with the path analysis (Hooper, Coughlan & Mullen 2008; Kline, 2011). Figure 3 shows the standardized parameter estimates of the CFA model analysis among the three latent factors (self-efficacy, work engagement and job satisfaction) and Table 2 indicates the covariances between these factors.
Figure 3
Confirmatory analysis parameter estimates
Table 2
Latent factor covariances
The moderate fit indices and the low sample size though render the results of this model to be viewed with caution as the it is below the value of N < 200.
Estimate SE z p Stand. Estimate
Self-efficacy Self-efficacy 1.000 ᵃ
Work Engagement 0.684 0.0606 11.28 < .001 0.684
Job Satisfaction 0.556 0.0775 7.17 < .001 0.556
Work Engagement Work Engagement 1.000 ᵃ
Job Satisfaction 0.747 0.0516 14.48 < .001 0.747
Job Satisfaction Job Satisfaction 1.000 ᵃ
ᵃ fixed parameter
Path Analysis
Path analysis was used to test the mediation model, and it presents how X (self- efficacy) affects Y (job satisfaction), considered as an outcome, via the variable M (work- engagement) (Hayes, 2013). Assumptions of the path analysis based on the ordinary least squares (OLS) regression, include linearity, with which it is presumed that the relationship of the predictor variable and the outcome variable is linear or approximately linear. Other assumptions are the normal distribution of errors, homoscedasticity of errors and independence of errors (Hayes, 2013). After running a simple linear regression analysis to test for the linearity of the slope of the standardized residuals, a P-P plot revealed that the distribution followed a generally normal line, with a slight deviation to the left as standardized residuals varied from -3.388 to + 3.066 when standardized residuals should not deviate from -3 to + 3 (Field, 2009). See Figure 4 to observe this assumption.
Figure 4
P-P Plot on the normal distribution of errors of the three latent factors
For the assumption of homoscedasticity, a scatterplot revealed that the majority of the distribution of the residuals in the middle of the plot is between -2 to +2 however there is a slight variation to the left. For the assumption of independence of residuals, the Durbin-
Watson statistic was calculated and revealed a value of 1.953 (almost 2 value). According to this statistic, which ranges from 0-4, in order to assume independence of residuals the Durbin- Watson must be approximately 2, which is also the case here. Therefore, we assumed that there was independence of residuals. See Table 3.
Table 3
Model Summary of Independence of Residuals
Model Summaryb
Model R
R Square
Adjusted R Square
Std.
Error of the Estimate
Change Statistics
Durbin- Watson R
Square Change
F
Change df1 df2
Sig.
F Change
1 .658a .433 .421 7.965 .433 37.427 2 98 .000 1,953
a. Predictors: (Constant), Work Engagement Total, Self-efficacy Total
b. Dependent Variable: Job Satisfaction Total
Testing for the assumption of the normal distribution of errors, a histogram revealed that this assumption was met, as the distribution despite a slight curvature to the left can be considered as a normal distribution (Field, 2009; Hayes, 2013). This can be observed from Figure 5.
Figure 5 Histogram
A simple mediation analysis was conducted using the OLS form of path analysis as described in (Hayes, 2013) and it revealed that work engagement mediated the effect of self- efficacy on job satisfaction (indirect effect), F (2,98)= 37.43, p = .000, R2= .43, thus supporting hypothesis 4 that the positive relation between self-efficacy and job satisfaction is mediated by work engagement. From this finding, considering a as the coefficient for self- efficacy and b the coefficient for work engagement, yields the indirect effect that self-efficacy causes job satisfaction through work engagement ab = 1.38 (0.690) = 0.797. Regarding the direct effect of self-efficacy on job satisfaction as reported by Orgambídez et al. (2020), it was not found statistically significant b = 0.22, t (98) = 1.079, p = .280. In their study even though they report a direct effect, they did not include a hypothesis suggesting such an effect.
In our study as well, even though we use their adjusted model to replicate, we only presumed a correlation between self-efficacy and job satisfaction (H1), not a direct effect of self- efficacy on job satisfaction. A bias-corrected bootstrap confidence interval for this indirect effect (0.797) based on a 5.000 bootstrap sample was above zero (0.347 to 1.257).
From the above results we can conclude that regarding the model, hypothesis 4 was confirmed revealing that self-efficacy indirectly affected job satisfaction through work engagement. However, taking into account the moderate model fit as well as the restricted sample size, the reliability of these results is limited. These findings are presented in Table 4.
Table 4
Coefficients, Standard Errors, p Values.
Consequent
M (work engagement) Y (job satisfaction)
Antecedent Coeff. SE p Coeff. SE p
X (self-efficacy) a 1.138 0.137 .000 a 0.224 0.207 .283
M (work
engagement) --- --- --- b 0.690 0.117 .000
Constant i1 13.514 4.145 .001 i2 38.74 5.087 .000
R2 = 0.42, F (1,99) = 69.80, p = .000
R2 = 0.43 F (2,98) = 37.43, p = .000
Discussion
The purpose of this study was to replicate the adjusted model presented in Orgambídez et al. (2020). By utilizing the framework of the JD-R model, the study investigated the relationship between self-efficacy (personal resource) and work engagement (mediator) on job satisfaction in Sweden among employees within IT, technology and Software companies. As a general summary of the results, the statistical testing confirmed all four hypotheses. Regarding the hypotheses H1-H3, which were tested with the Pearson’s correlation coefficient r; the significant results indicated that the three variables were significantly related to each other. These first three hypotheses were pre-registered prior to the beginning of the study and similar to Orgambídez’s et al. (2020), were stated in the form of correlations and not causal effects. The significant results regarding hypothesis 1 that self- efficacy is positively related to job satisfaction, revealed that self-efficacy is positively associated with feelings of being content and satisfied with one's job. From this it is presumed that the more confidence one has on his/ her abilities to handle situations and complete tasks, the more satisfied they will be with their job positions, as also supported by previous studies (Yakin & Erdil, 2012; Skaalvik & Skaalvik, 2014; Caesens & Stinglhamber, 2014).
Hypothesis 2 stating that self-efficacy is positively related to work engagement, also produced significant results regarding this correlation. Individuals with high self-efficacy have higher probabilities to be more attached and engaged with their work, also consistent with previous research (Buric & Macuka, 2017; Federici & Skaalvik, 2011). Hypothesis 3, work
engagement is positively related to job satisfaction was also confirmed, revealing that when one is feeling content and motivated at work, one will also be more engaged with the structure and context of work (Yalabik et al., 2013; Orgambídez-Ramos & de Almeida, 2017; Yeh, 2013).
The mediation model itself was tested as shown in the results with the method of path analysis and it referred to hypothesis 4, stating that the positive relation between self-efficacy and job satisfaction is mediated by work engagement. The model we replicated from Orgambídez’s et al. (2020), was their adjusted one after they tested their model fit and not the one they based their initial hypotheses on. This model as stated in the prior sections, did not hypothesize or have a direct effect of self-efficacy on job satisfaction. In our study also, even though we followed this model with this direct effect we did not hypothesize so. Moreover, the replication results indicated no such direct effect. Hypothesis 4 was confirmed indicating that self-efficacy indirectly affected job satisfaction through work engagement supporting the conceptual model. Orgambídez’s et al. (2020) however, regarding this relationship among the three variables, talk about partial mediation. They mention partial mediation as a support of the fact that self-efficacy, according to their findings, directly and indirectly affected job satisfaction. Therefore, they reported that work engagement only partially mediated the positive relationship between self-efficacy and job satisfaction. This brings forth the dilemma of what is acceptable in terms of assuming the support of a hypothesis and if partial mediation can be accepted as a proof of mediation, instead of a full mediation. In contrast to Orgambídez’s et al. (2020), this replication’s results showed no direct effect of self-efficacy on job satisfaction. Therefore, we suggest that this model revealed a full mediating effect of work engagement between the independent and dependent variable, depending on the rationale of the original study.
In regard to Orgambídez’s et al. (2020) model, job satisfaction is considered as one of the indicators (including affective commitment that was not investigated in this present study) for the QWL and as described in the previous section are usually inseparable from each other. Martel and Dupuis (2006) differentiate from the common ground that job satisfaction is a good and valid indicator for QWL. They suggest that the circumstances which comprise QWL arise from different hierarchy levels, starting from the individual (psychological and physiological state), the group and the organizational level (work structure, performance) and finally extends to the society level. They do not completely deny the level of impact of work wellbeing and satisfaction, as well as contentment with one's job
position. However, they extend QWL to be a bigger and broader concept composed from subjective and objective circumstances.
Limitations
Potential limitations are taken into account according to the findings of this scientific research paper.
Despite the fact that the statistical analysis supported all four hypotheses, we have to take into account the limited sample size (n = 101) that may have affected the power of the findings and consequently the reliability of the results, relating to the findings of the CFA as well as the path analysis. As mentioned in the method section the Covid-19 pandemic which led to the close-up of various companies affected the increase in number and subsequently the power of the study and the overall generalizability of the results. Hoe (2008) suggests that in order to provide an adequate statistical power to either path analysis or structural equation model (SEM) then at least 200 or more participants are mandatory. Furthermore, as supported by Schönbrodt and Perguini (2013), the larger the sample size is, the greater stability the sample size correlations will have. According to them, in order to reach the necessary point of stability (POS) where only minor variations from the true value occur, then the sample must be greater than n=161.
Another limitation is the study design itself, since correlation designs do not prove causal hypothesis but refer to variance hypothesis. The fact that our sample belonged to a specific work field of IT, software and technology companies in Sweden, can be considered a limitation. Although they came from 13 different companies, this can still restrict the applicability of the findings in other work fields. This limitation can also be presumed that it applies as a limitation for the Orgambídez’s et al. (2020) study, as their sample was composed of nurses and nurse assistants.
Other factors that may have contaminated the results are several biases that are linked with survey designs and self-reported information. Such a bias is the social desirability bias, often common in such forms of designs, where the participants will report the more sociably acceptable response rather than the true opinion or belief. Another type of self-reported bias can be the lack of introspective ability, where the participants cannot accurately access his/her own self (Althubaiti, 2016). Except from the self-reported types of biases, it is important to take into account the effect of a common method bias (CMB), which can be caused by the instruments structure instead of the participants’ responses and their predispositions
(Podsakoff et al., 2012). It is necessary to note though, that all three instruments used in this study have been tested for the validity of the scales. Nevertheless, the factor loadings in the job satisfaction scale were not consistent and have may contributed on the low strength of the model according to the CFA.
Future Recommendations
Structural equation modeling (SEM) is a suggestion for a more in depth future investigation of the JD-R model, including confounding constructs that may have affected Orgambídez’s et al. (2020) study findings. Exploratory factor analysis too could help to indicate other underlying factors, except from the already set ones (i.e. the case of job situation can be descried as intrinsic and extrinsic).
Implications
This study provided information on two different levels on the literature, on the one hand the widely acknowledged theoretical models used for the replication (JD-R, self- efficacy, social cognitive theory, QWL) and on the other hand implications about the replication method as a commonly used technique for validating results.
The theoretical models discussed and investigated in this replication study as well as that of self-efficacy theory brought to surface the need to apply them in more diverse work fields rather than the more commonly investigated ones (i.e. medical fields, educational fields) to understand if indeed they are as applicable in work fields such IT, software and technology. Furthermore, the sample including IT and software employees revealed a lack of investigation in terms of I-O psychological studies on aspects concerning the components of the JD-R model and the social cognitive theory on this field. This study for example, based on Orgambídez’s et al. (2020) model, found that self-efficacy affects how much satisfied with their job IT, software and technology professionals are, through work engagement.
These rapidly growing professions though and the ongoing demand of these professions in almost every types of organizations, thus very diverse work structures, expresses the necessity to investigate these professions more in depth. Job resources and job demands are more complex in such diverse work environments, which in turn make it more difficult to determine the quality of working life.
As far as the replication method is concerned, from the investigation of this study, it appears that the problem about the reliability in I-O psychology is not necessarily the lack of replication studies but the practices which research and original studies follow. This study
has shown us the need for universal established regulations in psychological research. These regulations must follow open science practices with promoted transparency of materials, datasets. Pre-registration of hypotheses and designs should be mandatory not optional for researchers. This will provide the opportunity for replications to be more accessible and frequent, which will lessen the phenomenon of replication crisis. Open science practices will aid the findings of original studies to be viewed as more reliable and assist for the different psychological research fields to be more esteemed again.
References
Abraham, S. (2015). Factors influencing workplace happiness among employees in software companies in Bangalore. International Journal of Research in Applied Management Science & Technology, 2 (2), 1-8.
Althubaiti, A. (2016). Information bias in health research: definition, pitfalls, and adjustment methods. J Multidiscip Healthc. 9, 211-217.
https://doi.org/10.2147/JMDH.S104807
Bakker, A. B., & Demerouti, E. (2007). The Job Demands‐Resources model: state of the art. Journal of Managerial Psychology, 22(3), 309–328.
https://doi.org/10.1108/02683940710733115
Bakker, A. B., & Heuven, E. (2006). Emotional dissonance, burnout, and in-role performance among nurses and police officers. International Journal of Stress Management, 13(4), 423–440. https://doi.org/10.1037/1072-5245.13.4.423 Bandura, A. (1997). Self-efficacy: the exercise of control. New York: W.H. Freeman.
Bandura, A. (1999). Social Cognitive Theory: An Agentic Perspective. Asian Journal of Social Psychology, 2(1), 21–41. https://doi.org/10.1111/1467-839X.00024
Bonaccorsi, A. (2011). European competitiveness in information technology and long-term scientific performance. Science and Public Policy, 38(7), 521–540.
https://doi.org/10.3152/030234211X12960315268010
Burić, I., & Macuka, I. (2018). Self-Efficacy, Emotions and Work Engagement Among Teachers: A Two Wave Cross-Lagged Analysis. Journal of Happiness Studies, 19(7), 1917–1933. https://doi.org/10.1007/s10902-017-9903-9
Caesens, G., & Stinglhamber, F. (2014). The relationship between perceived organizational support and work engagement: The role of self-efficacy and its outcomes. European Review of Applied Psychology, 64(5), 259–267.
https://doi.org/10.1016/j.erap.2014.08.002
Chapke, S. G. (2011). HR Practices in IT companies. Indian Streams Research Journal, Vol. 1(10), pp. 89-94.
Cordero, R., Farris, G. F., & DiTomaso, N. (1998). Technical Professionals in Cross- functional Teams: Their Quality of Work Life. Journal of Product Innovation Management, 15(6), 550–563. https://doi.org/10.1111/1540-5885.1560550
Davis, L. E., & Cherns, A. B. (1975). The Quality of Working Life, Volume 2: Cases and Commentary. Retrived from Martel, J.-P., & Dupuis, G. (2006). Quality of Work Life: Theoretical and Methodological Problems, and Presentation of a New Model and Measuring Instrument. Social Indicators Research, 77(2), 333–368.
https://doi.org/10.1007/s11205-004-5368-4.
Demerouti, E., Bakker, A. B., Nachreiner, F., & Schaufeli, W. B. (2001). The job demands- resources model of burnout. Journal of Applied Psychology, 86(3), 499–
512. https://doi.org/10.1037/0021-9010.86.3.499
Federici, R. A., & Skaalvik, E. M. (2011). Principal self-efficacy and work engagement:
assessing a Norwegian Principal Self-Efficacy Scale. Social Psychology of Education, 14(4), 575–600. https://doi.org/10.1007/s11218-011-9160-4
Field, A. P. (2009). Discovering statistics using SPSS: and sex, drugs and rock “n” roll (3rd ed). Los Angeles: SAGE Publications.
Harney, B., Fu, N., & Freeney, Y. (2018). Balancing tensions: Buffering the impact of organisational restructuring and downsizing on employee well-being. Human Resource Management Journal, 28(2), 235–254. https://doi.org/10.1111/1748- 8583.12175
Harper, G. R., & Utley, D. R. (2001). Organizational Culture and Successful Information Technology Implementation. Engineering Management Journal, 13(2), 11–15.
https://doi.org/10.1080/10429247.2001.11415111
Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: a regression-based approach. New York: The Guilford Press.
Henderson, R., & Mitchell, W. (1997). The Interactions of Organizational and Competitive Influences on Strategy and Performance. Strategic Management Journal, 18(S1), 5–
14. https://doi.org/10.1002/(SICI)1097-0266(199707)18:1+<5::AID- SMJ930>3.0.CO;2-I
Hoe, S. L. (2008). Issues and procedures in adopting structural equation modelling technique. Journal of Applied Quantitative Methods, 76–83.
Hooper, D., Coughlan, J., & Mullen, M. R. (2008). Structural equation modelling:
Guidelines for determining model fit. Electronic journal of business research methods, 6(1), 53-60.
Hughes, J. L., Camden, A. A., & Yangchen, T. (2016). Rethinking and Updating
Demographic Questions: Guidance to Improve Descriptions of Research Samples. Psi Chi Journal of Psychological Research, 21(3), 138–151.
https://doi.org/10.24839/2164-8204.JN21.3.138
Huzzard, T., & Arbetslivsinstitutet. (2003). The convergence of the quality of working life and competitiveness: a current Swedish literature review. Stockholm; Katrineholm:
Arbetslivsinstitutet ; Hellman [distributör.
Kamalanabhan, T. J., Sai, L. P., & Mayuri, D. (2009). Employee Engagement and Job Satisfaction in the Information Technology Industry. Psychological Reports, 105(3), 759–770. https://doi.org/10.2466/PR0.105.3.759-770
Karabiyik, B., & Korumaz, M. (2014). Relationship between teachers’ self-efficacy perceptions and job satisfaction level. Procedia-Social and Behavioral Sciences, 116(21), 826-830.
Klassen, R. M., & Chiu, M. M. (2010). Effects on teachers’ self-efficacy and job satisfaction: Teacher gender, years of experience, and job stress. Journal of Educational Psychology, 102(3), 741–756. https://doi.org/10.1037/a0019237
Kline, R. B. (2011). Principles and practice of structural equation modeling (3rd ed). New York: Guilford Press.
Koskinen-Hagman, M., Schwarzer, R., & Jerusalem, M. (1999). Swedish version of the general self-efficacy scale. 1999; Available from: http://userpage.fu-
berlin.de/*health/swedish.htm.
Lai, M. C., & Chen, Y. C. (2012). Self-efficacy, effort, job performance, job satisfaction, and turnover intention: The effect of personal characteristics on organization performance. International Journal of Innovation, Management and
Technology, 3(4), 387-391.
