Predictors of Employee Turnover in the Dutch Fashion Retail:
The Role of Work Overload, Emotional Labour, Emotional
Exhaustion and Multi-faceted Job Satisfaction
Author: Lonneke Schaap
Student ID: 10380965 Supervisor: Daphne Dekker Submission Date: 15-08-2014
Paper Type: Master Thesis – Final version University: University of Amsterdam
Course: Business Studies
ABSTRACT
Purpose – The purpose of this study is to examine the predictors of employee turnover in the retail fashion industry. The research model analyzes the role of work overload, emotional labour, emotional exhaustion and multi-faceted job satisfaction in predicting employee turnover.
Design/methodology/approach – Data was collected by the use of a self-administered questionnaire. 103 employees working in the fashion retail industry participated in the study. Data was analyzed with IBM SPSS statistics 22.
Findings – Both work overload and emotional labour positively influence emotional exhaustion. In turn, emotional exhaustion is negatively related to six dimensions of multi-faceted job satisfaction; job satisfaction with overall job, co-workers, supervision, policy, pay and customers. Only two dimensions are negatively related to employee turnover; job satisfaction with overall job and job satisfaction with promotion. Only job satisfaction with overall job mediates the relation between emotional exhaustion and employee turnover. Practical implications – Both job satisfaction with overall job and job satisfaction with promotion are negatively related to employee turnover. In order to decrease employee turnover, management of fashion retail organizations should consider ways to increase job satisfaction with overall job and job satisfaction with promotion. The model shows that an increase of job satisfaction with overall job can be achieved by lowering emotional exhaustion, work overload and emotional labour.
Originality/value – This study fills the research gap of employee turnover in the fashion retail industry. A comprehensive model is used including industry-specific characteristics such as work overload, emotional labour and emotional exhaustion. The overall model has never been studied before. This is also the first time that the construct of multi-faceted job satisfaction is examined in a fashion retail context specifically.
Keywords - Work Overload, Emotional Labour, Emotional Exhaustion, Multi-faceted Job
TABLE OF CONTENTS
1. Introduction 7
1.1. Introduction 7
1.2. The Dutch Fashion Retail Industry 10
1.3. Theoretical and Practical Relevance 10
2. Literature Review 12
2.1. Work Overload 12
2.2. Emotional Labour 13
2.3. Antecedents of Emotional Exhaustion 15
2.4. Antecedents of Multi-faceted Job Satisfaction 17
2.5. Antecedents of Employee Turnover 21
2.6. Research Model and Hypotheses 23
3. Methodology 24 3.1. Research Design 24 3.2. Questionnaire 24 3.3. Research Sample 25 3.4. Data Collection 25 3.5. Measures 26 3.6. Control Variables 27 3.7. Data Analysis 28 4. Results 31 4.1. Data Cleaning 31
4.1.1. Coding Variables and Recoding Counter-indicative Items 31
4.1.2. Missing Values 31
4.1.3. Detecting Outliers 32
4.1.4. Confirmatory Factor Analysis (CFA) 32
4.1.5. Computing Reliability 32
4.1.6. Computing Scale Means 33
4.1.7. Checking Normality and Distribution of Data 33
4.2. Testing Hypotheses 35
4.2.1. The Influence of Work Overload and Emotional Labour on Emotional 35 Exhaustion
4.2.2. The Relation between Emotional Exhaustion and Multi-faceted Job 36 Satisfaction
4.2.3. Emotional exhaustion as Mediator between Work Overload and 39 Multi-faceted Job Satisfaction
4.2.4. Emotional Exhaustion as Mediator between Emotional Labour and 42 Multi-faceted Job Satisfaction
4.2.5. The Relation between Multi-faceted Job Satisfaction and 45 Turnover Intentions
4.2.6. Multi-faceted Job Satisfaction as Mediator between 46
Emotional Exhaustion and Turnover Intentions
4.3. Research Model with Results 47
5. Discussion 48
5.1. Conclusion 48
5.2. Discussion 49
5.3. Theoretical Implications 51
5.4. Practical Implications 51
5.5. Limitations and Future Research 52
6. References 54
LIST OF FIGURES AND TABLES
Figure 1: Statistics on Employees in the Dutch Retail Industry in 2010 11
Figure 2: Research Model 11
Figure 3: Mediation Model 30
Figure 4: Research Model with Results 47
Table 1: Means, Standard Deviations, Correlations and Reliability Coefficients 34
Table 2: Kurtosis, Skewness and the Kolmogorov-Smirnov Test 35
Table 3: Summary of Multiple Regression Analysis for Emotional Exhaustion 36 Table 4: Summary of Linear Regression Analyses for Multi-faceted Job Satisfaction 38 Table 5: Path Coefficients and Confidence Intervals; Emotional Exhaustion 41
as a Mediator between Work Overload and Multi-faceted Job Satisfaction
Table 6: Path Coefficients and Confidence Intervals; Emotional Exhaustion 44 as a Mediator between Emotional Labour and Multi-faceted Job Satisfaction Table 7: Summary of Multiple Regression Analyses for Turnover Intentions 45 Table 8: Path Coefficients and Confidence Intervals; Multi-faceted Job Satisfaction 46
1. INTRODUCTION 1.1. Introduction
Imagine that you are in charge of a large fashion retail company which is in need for many front-line employees on a regular basis. However, employee turnover rate appears to be high and each time a new employee is hired, this person or another quits within a short period of time. As one might picture, this situation is rather problematic and leads to numerous organizational disadvantages. The overall subject of employee turnover is not a novel one but has been of great interest over the last decades and has proven to be detrimental for companies (Regts & Molleman, 2013). In general, prior literature supports this viewpoint (Argote & Epple, 1990; Shaw, Gupta & Delery, 2005; Siebert & Zubanov, 2009) and numerous organizational disadvantages of employee turnover are found such as an incline in productivity (Batt, 2002; Huselid, 1995), future revenue growth (Baron, Hannan & Burton, 2001) and profitability (Glebbeek & Bax, 2004; Ton & Huckman, 2008). Additional organizational disadvantages of high employee turnover are a decrease in customer service (Kacmar, Andrews, Van Rooy, Steilberg & Cerrone, 2006) and customer satisfaction (Morrow & McElroy, 2007).
Most research explains the negative relation between employee turnover and organizational performance on the basis of the social and human capital theory, as these two theories have received most attention in prior literature (Park & Shaw, 2013). The human
capital theory assumes that employees who have experience are of more value to a company
because they possess the needed skills and knowledge. When these employees decide quit, it is harmful to the organization because it loses employees with experience who contribute to the human capital of the firm. It takes time to substitute these employees and to reach similar levels of knowledge. The replacement of personnel also adds costs in terms of recruiting, selecting and training new employees. The social capital theory argues that turnover has a negative effect on organizational performance because when an employee decides to leave, he or she disturbs the organization’s social capital which has to do with social relations between employees such as trust and collective goal orientation. Moreover, socialization costs are involved when new employees enter an organization.
Originally, labour turnover can be divided into either voluntary or involuntary turnover (Morrell, Loan-Clarke & Wilkinson, 2001; Price, 1977). “An instance of voluntary turnover, or a quit, reflects an employee's decision to leave an organization, whereas an instance of involuntary turnover, or a discharge, reflects an employer's decision to terminate
the employment relationship” (Shaw, Delery, Jenkins & Gupta, 1998, p. 511). Because this study aims to predict the antecedents of why employees decide to quit, this paper refers to the concept of voluntary employee turnover when using the definition of employee turnover. Voluntary turnover is often unexpected and hard to manage for organizations (Shaw et al. 1998).
Much attention has been paid in prior literature to the subject of employee turnover in general but less is known about employee turnover in the retail industry. Rhoads, Swinyard, Geurts and Price (2002) argue that having and retaining a good workforce is a major requirement for being successful as a retail company. By maintaining competent employees, recruitment and training costs can be reduced and more value can be offered to customers. This results in benefits while gaining competitive advantage over other market players. It is therefore important to decrease employee turnover in the retail industry (Kim, Knight & Crutsinger, 2009).
Although a number of studies investigated employee turnover in the retail industry, there is still much to discover (Booth & Hamer, 2007). Even more absent is a comprehensive model on the most important predictors of employee turnover in the fashion retail industry specifically. Research on this subject is lacking in this specific branch. This study tries to gain more insight into the predictors of employee turnover in the fashion retail industry and will thereby contribute to the lack of research in this specific area. Due to this lack of knowledge, this study is mainly based on arguments from research conducted in the overall retail industry since the fashion retail is presented as part of the overall retail industry (Peek & Veghel, 2011). It is therefore assumed that there are many similarities between the fashion retail industry and the overall retail industry. Figure 1 illustrates this in terms of statistics.
Multiple factors influence the decision of employees to leave an organization in the fashion retail industry. Prior literature claims that job satisfaction is one of the most important drivers behind employee turnover in general (Jaramillo, Mulki & Solomon, 2006). This also accounts for the retail industry (Booth & Hamer, 2007; Henrie, 2004). In order to draw an as complete model as possible, this study focuses on the concept of multi-faceted job satisfaction instead of single-faceted job satisfaction as key antecedent of employee turnover. Multi-faceted job satisfaction has mainly been used in business-to-business context and so far, only Chung, Rutherford and Park (2012) explored the concept of multi-faceted job satisfaction in the retail industry. They argue that future research is needed to test the effects of multi-faceted job satisfaction on outcomes such as employee turnover. Therefore this study includes multi-faceted job satisfaction as major predictor of employee turnover.
Multi-faceted job satisfaction in turn, is found to be predicted by emotional exhaustion in both a sales and retail environment (Chung et al., 2012; Rutherford, Boles, Hamwi, Madupalli & Rutherford, 2009). Broadbridge (1999) argues that jobs in retail have become more stressful over time due to organizational changes which are forced by the multiplicity and complexity of retail organizations. These changes can be assigned to technological, environmental and market issues, such as online shopping, customer needs and internationalization. Since jobs in the retail industry are related to long, unsocial hours; psychical endeavor and routine work, one can imagine that emotional exhaustion plays especially in this particular context an important role (Broadbridge, 1999). Therefore emotional exhaustion is assumed to be a major predictor of multi-faceted job satisfaction.
In turn, this study argues that work overload and emotional labour are again major stressors of emotional exhaustion and are specifically present in the retail sector. Broadbridge (2002) found that jobs in retail are considered to be stressful among the majority and that this is partly caused by work overload. Another industry specific characteristic is emotional labour. The display of emotions is particularly important in the service industry in which emotional labour is part of front-line employees’ jobs (Ashforth & Humphrey, 1993). In this study, comparison between front-line employees from the service industry and the retail industry is often made since both can be considered as service jobs which require ‘person to person’ interaction and ‘soft skills’ (Warhurst & Nickson, 2007). According to Warhurst and Nickson (2007), most research on front-line employees from both the retail and service industry is concerned with emotional labour. However, the impact of emotions in the service industry is not a well-understood area yet (Hennig-Thurau, Groth, Paul & Gremler, 2006). Limited research on emotional labour in the workplace has been conducted (Schaubroeck & Jones, 2000).
In sum, this study examines the antecedents of employee turnover in the fashion retail industry. Specific industry characteristics are included in the research model and the role of work overload, emotional labour, emotional exhaustion and multi-faceted job satisfaction in predicting employee turnover is analyzed. A questionnaire will be distributed among a sample of Dutch front-line employees in the fashion retail branch. The following research question will be addressed: “What is the role of work overload, emotional labour, emotional
exhaustion and multi-faceted job satisfaction in predicting employee turnover?” See figure 2
1.2. The Dutch Fashion Retail Industry
The Dutch retail industry is strongly diversified since there are 45 subsectors including fashion (Peek & Veghel, 2011). According to het Hoofdbedrijfschap Detailhandel (hbd, 2011) in 2011, a total number of 680.200 people were employed in the Dutch retail sector. The retail sector is one of the largest commercial sectors and 10% of all jobs in the Netherlands can be found in the retail industry (hbd, 2011). Both absenteeism and labour turnover are found to be high in the retail industry (Broadbridge, 1999; Rhoades & Eisenberger, 2002). According to Eurostat data, in 2006, just over 60% of the Dutch retail workforce stayed in their current job for less than five years and 27% of annual leavers have been counted in the Netherlands.
In 2009, consumers spend 14.6 billion Euros on fashion in the Netherlands (Fashionunited, 2009). A total of 76.800 paid jobs in fashion stores have been counted in the Netherlands in 2013 (hbd, 2013). This number does not only include sales assistants, but covers all positions in a store. Most of the jobs in fashion stores are part-time. 35% contains of jobs less than 16 hours whereas approximately 19% of the fashion stores contains full-time jobs (38 hours or more). The remaining part works somewhere in between part-time and full-time. 86% of all employees in fashion stores are female; this is slightly more than the overall retail industry (hbd, 2010). See figure 1.
1.3. Theoretical and Practical Relevance
As described earlier, there is a lack of research and theories in the area of employee turnover in the fashion retail industry. However, the contribution of this study is not only relevant for theoretical purposes but also for practical ones. The size of the retail industry in relation with the commercial and economic interests, the height of the employee turnover rate and the aforementioned disadvantages of employee turnover indicate the importance of the topic. It is therefore crucial to understand the main drivers behind employee turnover in the fashion retail sector. The results of this study can be helpful for organizations in order to reduce their employee turnover rates and increase performance. For example, if employee turnover appears to be a result of low job satisfaction, management of an organization can make decisions on how to improve job satisfaction among front-line employees. Improved job satisfaction can for example be achieved by giving employees more responsibility and autonomy, stimulating variety in skills and enhancing interpersonal relationships (Harter, Schmidt & Hayes, 2002).
Figure 1: Statistics on Employees in the Dutch Retail Industry in 2010 Fashion Stores Total Retail
Gender Male 14% 37% Female 86% 63% Age till 17 years 5% 11% 17 to 20 years 19% 24% 21 to 25 years 16% 14% 26 to 35 years 17% 16% 36 to 45 years 19% 17% 46 to 55 years 15% 12%
56 years and older 9% 6%
Ethnicity Natives 80% 83% Western Immigrants 9% 7% Non-Western Immigrants 11% 10% HBD en ITS, 2010
Figure 2: Research Model
Work Overload Emotional Labour Emotional Exhaustion Multi-faceted Job Satisfaction - Overall-job - Co-workers - Supervision - Policy - Pay - Promotion - Customers Turnover Intentions
2. LITERATURE REVIEW
This chapter will provide an overview of the relevant literature in the research area of this study. Furthermore, hypotheses will be formed. First, the theoretical background of work overload and emotional labour will be discussed. Second, the literature on emotional exhaustion will be explained in relation to work overload and emotional labour. Third, the concept of multi-faceted job satisfaction will be elaborated in relation to emotional exhaustion. Last, the theoretical background on employee turnover and its predictors will be defined.
2.1. Work Overload
Based on Rizzo, House and Lirtzman (1970), Bolino and Tumely (2005) state that “Role overload describes situations in which employees feel that there are too many responsibilities or activities expected of them in light of the time available, their abilities, and other constraints”, p.741. This definition is used throughout this paper to describe the definition of “work overload”. It should be noticed that “role overload” is replaced by the term “work overload”. Much research on work overload has been conducted in general; especially the impact of work overload on the outcomes of sales people (Jones, Chonko, Rangarajan & Roberts, 2007). However, there is a lack of research on work overload in the (fashion) retail industry specifically.
In general, work (over)load has increased over the past decades. This is partly due to organizational restructuring and a larger focus on productivity improvements, resulting in more responsibilities for employees. Besides, companies focus more and more on maximizing their profits and revenues which means that cost-cutting appears wherever possible, including lay-offs. This in turn leads to fewer employees per organization who remain stuck with too much work (Mulki, Lassk & Jaramillo, 2008). This increase in work overload is found to be harmful to the well-being of employees. Physical and mental health decrease when employees experience more stress (Jones et al., 2007). Also found is that work overload leads to a decrease in organizational commitment and higher levels of absenteeism because of sickness. This leads again to a decline in the overall profitability of an organization (Duxbury & Higgins, 2001). A study by Broadbridge (1999) in the retail sector supports this and claims that outcomes of pressures on the work floor can have a negative impact on an employee’s well-being. Therefore work overload is assumed to be a major job stressor in this study.
2.2. Emotional Labour
Morris and Feldman (1996) state that organizations progressively want to influence the kind of behavior and image their employees transcend to clients. According to Cho, Rutherford and Park (2013) it is important that employees successfully control their emotions, not only for themselves but also for customers. As a result, organizations have increasingly been developing specific guidelines which prescribe desired emotions to be expressed by employees, called display rules. Ekman (1973) specified display rules as standards of behavior which prescribe the appropriate emotions in specific situations and how these emotions should be displayed.
The display of emotions is particularly important in the service industry in which emotional labour is part of front-line employees’ jobs. There are a few reasons for this (Ashforth & Humphrey, 1993). First, front-line employees carry out and promote the organization to customers directly because they operate in boundary spanning roles. Second, front-line employees are often involved in face-to-face interactions with customers in which front-line employees have direct contact with them. Third, the presence of customers can cause uncertain situations in which the quality of the service encounter might fluctuate. Last, it is difficult for customers to judge the level of service quality, because the provided services are often hard to ‘grasp’ and are intangible.
These aspects emphasize the importance of service front-line employees because the behavior heavily influences how customers perceive the overall quality of the organization. Front-line employees have to welcome their customers in a positive manner and during the interaction, positive emotions will be transferred to customers (Cho et al., 2013). Since the behavior of front-line employees plays an important role in the organization’s performance, they are expected to behave according to an organization’s display rules. According to Grandey, Fisk, Mattila, Jansen and Sideman (2005), a widespread display rule is the “service-with-a-smile” rule which means that employees should always express indefectible positive emotions when interacting with customers. Display rules thus emphasize the “superficial” side of emotions; these emotions which are direct observable. Deeper emotions or actual feelings are not taken into account (Ashforth & Humphrey, 1993).
Researchers all agree with the underlying assumption that emotional labour is about the regulation of emotions to ensure that these are in line with the organizational display rules (Goffman, 1959). However, there are different theoretical approaches concerning emotional labour and there is no clear consensus about its specific construct (Glomb & Tews, 2004). According to Morris and Feldman (1996) emotional labour is defined as “…the effort,
planning, and control needed to express organizationally desired emotion during interpersonal transactions”, p. 987. They further argue that as long as the expressed emotions are in line with the organizational display rules, it creates emotional labour. This is supported by Brotheridge and Lee (2003). According to Ashforth and Humphrey (1993), emotional labour is the “act of expressing socially desirable emotions”, p. 88-89. No matter if internal feelings are in line with these desirable emotions. These researchers thus argue that emotional labour involves both genuine and in-genuine feelings.
However, Mann (1999a) describes emotional labour as: “The state that exists when there is a discrepancy between the emotional demeanor that an individual displays because it is considered appropriate, and the emotions that are genuinely felt but that would be inappropriate to display”, p. 353. This definition is closely related to emotional dissonance. Although researchers agree that dissonance is included in the concept of emotional labour, there is no agreement on whether emotional dissonance is a necessary condition for emotional labour (Glomb & Tews, 2004). Nevertheless, this paper follows the reasoning of Mann (1999a). Thereby arguing that emotional dissonance is a requirement for emotional labour to exist and that emotional labour is solely present when employees fake or oppress certain feelings whereby genuinely felt emotions are not taken into account. This definition is chosen, because employees have to make more effort when organizational display rules are incongruent with their genuine feelings (Ashforth & Humphrey, 1993; Morris & Feldman, 1996). This has proven to lead to negative outcomes such as job-related stress (Adelmann, 1995) and emotional exhaustion (Morris & Feldman, 1997) and fits the conceptual model of this study best.
It is thus argued that emotional labour is concerned with the discrepancy between an employee’s true feelings and the expressed feelings which are desired by the organization. “The process of emotional labour itself typically involves two common processes: suppressing the negative emotions that one is feeling and faking positive emotions that one is not feeling”, p. 470 (Glomb & Tews, 2004 in Sliter, Jex, Wolford & McInnerney, 2010). Prior literature has argued that employees in the service industry are often involved in both processes. This means that they welcome customers in a friendly manner when actually not feeling well and they hide their irritation when customers behave impolite (Grandey, Fisk & Steiner, 2005). Engaging in the two processes of emotional labour; suppressing negative emotions and faking positive emotions, has direct implications for an employee’s well-being (Grandey, 2003).
2.3. Antecedents of Emotional Exhaustion
Emotional exhaustion is the most prominent dimension of job burnout, among depersonalization and personal accomplishment (Maslach & Jackson, 1981). Based on these authors, Rutherford et al. (2009) define burnout as “… a psychological syndrome or condition that manifests in reactions to chronic stress experienced by people who provide services”, p.1147). Also based on Maslach and Jackson (1981), Chung et al. (2012) define emotional exhaustion as “…the feeling of emotional overextension and exhaustion attributable to one’s work, p.703. Emotional exhaustion is known to be as a type of stress that is caused by stressors at the work floor (Cropanzano, Rupp & Byrne, 2003) and is the most researched dimension of job burnout in a sales environment (Rutherford et al., 2009). According to Singh, Goolsby and Rhoads (1994), emotional exhaustion is more likely to arise at employees who are involved in boundary spanning roles. Front-line employees engage in boundary-spanning roles because they are always in between the customer and the employer. There are three reasons for this.
First, they are responsible for representing the store and the image of the store. Second, their task is to guarantee service quality and communicate improvements internally in order to meet the demands of the customer better. Last, the level of quality and customer satisfaction relies for a big part on the behavior of front-line employees (Bettencourt & Brown, 2003). Service employees are in a position which requires them to engage in boundary-spanning activities and it is therefore presumed that they are more likely to experience role stress (Kahn, Wolfe, Quinn & Snoek, 1964; Singh, 1993). Cooper and Baglioni (1988) tried to distinguish the level of stress among different occupations and found that jobs in the retail are exposed to an above average level of stress. Broadbridge (1999) found that positions in the retail industry are considered to be stressful. As such, emotional exhaustion is particularly present within the service and retail industry, since employees are often in direct contact with their customers, who expect them to live up to high demands. This makes them more vulnerable to emotional exhaustion (Cordes & Dougherty, 1993; Rutherford et al., 2009).
It is important to understand the antecedents of emotional exhaustion since emotional exhaustion leads to important organizational outcomes, such as job satisfaction, organizational commitment and turnover intentions (Babakus, Cravens, Johnston & Moncrief, 1996). According to Cho et al. (2013) the emotional understanding of employees also affects the way customers perceive levels of service quality, satisfaction and loyalty. Robinson and Griffiths (2005) found that work overload is most often mentioned as the main source of job related stress. This is supported by other research. For example, Janssen, De Jonge and
Bakker (1999) conducted a study among nurses and found that burnout is mainly caused by both workload and limited social support. Furthermore they state that stress is a result of resources which are threatened by demands (e.g. workload or role stress). Another meta-analysis conducted by Lee and Ashforth (1996) claims that there is a relationship between work overload and limited support on one side and emotional exhaustion on the other. However, these outcomes concern organizations in general and are not specifically focused on the retail industry.
Broadbridge (2002) however, did focus on the retail industry and found that work overload is one of the key stressors for employees. Also applicable to the retail environment, Firth, Mellor, Moore and Loquet (2004) found that certain stressors, among which work overload, have a direct influence on stressful feelings and they further argue that work overload has a direct relation with stress feelings and job satisfaction. Since work overload is a potential job stressor, it is argued that there exists a positive relation between stress related feelings and emotional exhaustion. So when an employee experiences more work overload, it is likely that an employee’s emotional exhaustion increases. Therefore the following is hypothesized:
H1. Work overload is positively related to an employee’s emotional exhaustion.
Further argued is that emotional labour is an important antecedent of emotional exhaustion. As stated, front-line employees are more prone to emotional exhaustion due to their participation in boundary spanning roles. Sales persons need to express positive emotions and hide negative emotions in order to accomplish customer satisfaction and customer loyalty (Lam, Kraus & Ahearne, 2010). This refers to the concept of emotional labour: a discrepancy between real feelings and feelings which need to be expressed by the organization. Emotional dissonance, which is in this case closely related to emotional labour, is even larger in organizations where employees have face-to-face contact with customers all the time because they always have to live up to the organizational display rule, even though this is not in line with their genuine feelings (Morris & Feldman, 1996).
Hochschild (1979, 1983) argues that complying with organizational display rules might cause harmful psychological effects among front-line employees. This is supported by other researchers. The engagement in organizational display rules has been found to be linked to stress-related physiological arousal (Gross & Levenson, 1997) and job strain which concerns bad work attitudes and burnout (Brotheridge & Grandey, 2002).
In general, employees have to make more effort when organizational display rules are incongruent with their genuine feelings (Ashforth & Humphrey, 1993; Morris & Feldman, 1996). The negative effects of emotional labour can be explained by the Conservation Of Resources (COR) theory by Hobfoll and Freedy’s (1993). This theory argues that people, when possible, try to preserve their resources. When they engage in fake emotions or when they suppress their emotions, their resources cannot be guarded which leads to emotional exhaustion (Sliter et al., 2010). Grandey, Fisk and Steiner (2005) also argue that faking and oppressing feelings drain personal resources and causes job stress for front-line employees. However, the impact of emotions in the service industry is not a well-understood area yet (Hennig-Thurau et al., 2006). Nevertheless, based on the COR theory, the following is hypothesized:
H2. Emotional labour is positively related to an employee’s emotional exhaustion.
2.4. Antecedents of Multi-faceted Job Satisfaction
Job satisfaction can be described as “a pleasurable or positive emotional state resulting from the appraisal of one’s job or job experiences” (Locke, 1976, p. 1304). This overall understanding of job satisfaction using global measurements has helped in the exploration of the antecedents of job satisfaction (Arnold, Flaherty, Voss & Mowen, 2009; Babin & Boles, 1996). However, there has been critique on global measures of job satisfaction because they leave out measures of individual aspects (Churchill, Ford & Walker, 1974). Widely held literature views job satisfaction as a universal and single-layered dimension, while it is often approached as both a precursor and consequence (Rutherford et al., 2009), as a respond to this, several multi-faceted job satisfaction scales have been designed.
Wood, Chonko and Hunt (1986) use a four dimension scale, including satisfaction on pay, closure, information and variety. Smith, Kendall and Hulin (1969) designed a five dimension scale which covers satisfaction with pay, opportunities for promotion, supervision, type of work and coworkers on the job. Churchill et al. (1974) developed a seven dimension scale with 95 items named INDSALES, covering satisfaction with overall job, co-workers, supervision, policy, pay, promotion and customers. Similar to Chung et al. (2012), this study uses the reduced 28-item INDSALES scale (Comer, Machleit & Lagace, 1989). This scale is chosen because it explains more facets than the other scales and is therefore more likely to obtain richer information. Moreover, the scale has been ultimately developed to be used in a sales environment and is therefore applicable to this study. Besides, the scale simultaneously
measures individual facets of job satisfaction as well as global job satisfaction (satisfaction with overall job), thereby taking advantages from both measures.
The study of Rutherford et al. (2009) is one of the few which focuses on the link between emotional exhaustion and multi-faceted job satisfaction. They found that emotional exhaustion is related to five of the seven dimension of multi-faceted job satisfaction, namely satisfaction with: overall job, supervision, policy, pay and promotion. No significant relationship was found between emotional exhaustion and satisfaction with co-workers and customers. However, they conducted their study in a sales environment and not specifically in a retail context. Another interesting study explored the link between emotional exhaustion and multi-faceted job satisfaction particularly in the retail context (Chung et al., 2012). Results indicate that emotional exhaustion is one of the most prominent predictors of multi-faceted job satisfaction. They found that emotional exhaustion is significantly and negatively related to job satisfaction with: overall job, co-workers, supervision, pay and promotion. They did not find any relation between emotional exhaustion and job satisfaction with policy and customers.
These studies on multi-faceted job satisfaction show different outcomes and in general, not much research has been conducted yet on the effect of emotional exhaustion in relation to multi-faceted job satisfaction, but rather in relation to global job satisfaction (Rutherford et al., 2009). The rest of this paragraph will build theoretical support for the relations between emotional exhaustion and each of the seven dimensions of job satisfaction.
Contrasting results have been discussed in prior research about the relation between emotional exhaustion and global job satisfaction. A number of studies did not find any relationship, assuming that emotional exhaustion has no effect on job satisfaction (Boles, Johnston & Hair, 1997). However, in general, it is widely acknowledged that burnout is negatively related to job satisfaction and as described earlier, emotional exhaustion is a part of burnout (Maslach, 1981; Singh et al., 1994). Singh et al. (1994) argue that in contrast to role stressors, burnout is always destructive and has a negative, linear curve with various job outcomes such as job satisfaction. The negative relation is based on two theoretical arguments. ‘First, because psychological burnout is the outcome of an appraisal process by which an individual evaluates the demands vis-a-vis his or her re-sources, it is posited that the outcome of this appraisal should affect an individual's psychological well-being on the job, including job satisfaction. Second, because both are affective responses, it is hypothesized that burnout feelings should be related to job satisfaction’ (Singh et al., 1994, p.561). The negative link has found support in prior literature in financial firms and service organizations
(Jaramillo et al., 2006), the retail banks sector (Karatepe & Tetinkus, 2006) and in a sales environment (Babakus et al., 1996). A recent study in the retail industry of Cho et al. (2013) also found support for the negative link between emotional exhaustion and job satisfaction. However, their study is conducted in Asia and might therefore not be generalized.
Furthermore, it is expected that emotional exhaustion is negatively related to job satisfaction with co-workers and customers, although this latter construct has not found full support in any of the aforementioned studies on multi-faceted job satisfaction. However, Leiter and Maslach (1988) argue that emotional exhaustion is closely related to interpersonal relationships and that it correlates high with being in contact with other people. When someone is emotionally exhausted, a person’s resources are sooner depleted. It is therefore expected that emotional exhaustion sooner leads to negative experiences with co-workers and customers and in turn, reduces job satisfaction on these dimensions.
Also, a negative link between emotional exhaustion and job satisfaction with supervision and policy is assumed. Employees prefer to work for an organization in which managers respond to their needs (Booth & Hamer, 2007). Following this reasoning, a possible explanation for the negative relation between emotional exhaustion and job satisfaction with supervision and policy could be that an employee, who feels emotionally exhausted, has not many resources left and has more difficulties coping with job-related tasks or activities. This person might be sooner depleted and irritated, requesting for more flexibility. When the policy or supervision of a company is not able to respond to this, an employee might develop a more negative view of the organization and thereby reducing his or her job satisfaction concerning these two facets.
Moreover, emotional exhaustion is expected to have a negative relation with job satisfaction with pay and promotion. This can be explained by the social exchange theory (Hatfield & Sprecher, 1984), assuming that people always strive for equity in an employee-organization ratio. Emotionally exhausted employees have probably put much energy and effort in the organization. This in turn results in a demand for equivalent rewards from the organization in order to maintain a feeling of equity. These rewards might for example be promotion or pay (Van Dierendonck, Schaufeli & Buunk, 1996). When the organization is not able to provide higher rewards, an employee’s job satisfaction with pay and promotion will decrease.
Based on the aforementioned theories, this study argues that emotional exhaustion has a negative impact on all dimensions of multi-faceted job satisfaction. This leads to the following hypothesis:
H3. Emotional exhaustion is negatively related to an employee’s multi-faceted job satisfaction (including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers).
Some researchers have found a direct effect of certain job stressors on global job satisfaction (Wunder & Dougherty, 1982; Currivan, 2000). In a retail context, Firth & al. (2004) found that workload directly relates to feelings of stress and overall job satisfaction. However, this study expects that emotional exhaustion acts as a mediator in the negative relation between work overload on multi-faceted job satisfaction and emotional labour on multi-faceted job satisfaction. Both work overload and emotional labour are expected to result in higher levels of emotional exhaustion. Experiencing emotional exhaustion makes it more likely that the perception of an employee on the different dimensions of job satisfaction changes, rather than work overload or emotional labour on their own. This is supported by Singh et al. (1994), arguing that burnout (among which is emotional exhaustion) is not a job stressor in itself, but rather a result of multiple role stressors. The study recognizes that there is in general a significant and direct effect of role stressors (e.g. conflict, overload) on important organizational outcomes, such as job satisfaction. However, these outcomes do not reflect enough strength and consistency to be fully supported. Further argued is that burnout is a far more prominent predictor of organizational outcomes than role stressor(s). Based on this theory and the overall lack of understanding about the antecedents of multi-faceted job satisfaction in a retail environment (Chung et al., 2012), the following hypotheses are assumed:
H4. The negative relation between work overload and multi-faceted job satisfaction (including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers) is mediated by an employee’s emotional exhaustion.
H5. The negative relation between emotional labour and multi-faceted job satisfaction (including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers) is mediated by an employee’s emotional exhaustion.
2.5. Antecedents of Employee Turnover
Prior research has shown that turnover intention is the best predictor of actual turnover (Regts & Molleman, 2012). “Turnover intentions can be defined as an employees’ state of mind to leave an organization” (Singh, Verbeke & Rhoads, 1996 in Alexandrov, Babakus and Yavas, 2007, p.357). In this study, voluntary employee turnover is measured in terms of turnover intentions. Sager’s (1991) has shown that turnover intentions provide enough validity and makes an effective distinction between those who leave and those who stay.
Multiple factors influence employee turnover in general. For example, a meta-analysis of Hom and Griffeth (1995) found that employees leave an organization when they become dissatisfied and lose their organizational commitment. Hom and Kinicki (2001) build further on this and state that there are three main reasons for employee turnover: the external business environment, the personal element and the job satisfaction element. According to Chang et al. (2013) the main antecedents of turnover intention in previous studies are: job autonomy, fair reward, job satisfaction, organizational commitment, tenure, social support, and demographic variables. Other studies have investigated employee turnover in a retail context specifically. One interesting study is conducted in by Henrie (2004) who found that key arguments for employee turnover are: too little working hours, bad payments, no career opportunities, overwork, unsocial work hours, bad training, poor staff facilities, being afraid of redundancy and staff views were not heard. However, this study took place in the UK and might not be generalized. Alexandrov et al. (2007) explored the effects of psychological climate on turnover in a retail environment and argue that this results in affective responses as job satisfaction and affective organizational commitment and in turn influences employees’ turnover intentions.
As illustrated, there are numerous predictors of employee turnover in the retail context which poses a limitation for identifying a comprehensive model which covers all its antecedents. However, almost all of the aforementioned studies include job satisfaction and as an important predictor of employee turnover. This is supported by other authors, stating that dominant models in literature focus on job satisfaction as one of the main drivers of labour turnover in organizations (Henrie, 2004; Jaramillo et al., 2006; March & Simon, 1958). Based on this, job satisfaction is considered as the major predictor of employee turnover. In order to capture as much information as possible, the concept of multi-faceted job satisfaction is used instead of global job satisfaction.
The relation between job satisfaction and turnover intentions of salespersons has been widely examined (Ladik, Marshall, Lassk & Moncrief, 2002) and several studies (Boles,
Johnston & Hair, 1997; Jaramillo et al., 2006) found a negative relation between general job satisfaction and turnover intentions. However, there is only limited research on the relation between multiple facets of job satisfaction and the intention to leave (Rutherford et al., 2009). Rutherford et al. (2009) only found a significant link between job satisfaction with overall job and job satisfaction with promotion on the propensity to leave, whereas the other dimensions did not strongly enough relate to turnover intentions. However, this study has been conducted in a sales environment whereas findings in a retail context might differ. Based on previous results in the literature and the relative lack of knowledge on the relationship between multi-faceted job satisfaction and turnover intentions, this paper argues that all dimensions of job satisfaction are negatively related to employee turnover intentions. Therefore, the following is proposed:
H6. Multi-faceted job satisfaction (including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers) is negatively related to an employee’s turnover intentions.
Given that emotional exhaustion is expected to predict multi-faceted job satisfaction and multi-faceted job satisfaction predicts in turn employee turnover intentions, it is assumed that multi-faceted job satisfaction mediates the relationship between emotional exhaustion and employee turnover intentions. However, there are mixed findings on the mediating role of job satisfaction in relation to emotional exhaustion and turnover intentions (Rutherford et al., 2009). Boles et al. (1997) argue that emotional exhaustion is directly related to turnover intentions; they did not find support for the mediating role of job satisfaction. O’Driscoll and Beehr (1994) found the opposite; job satisfaction mediates the effects of uncertainty and role stressors on turnover intentions and strain. However, both studies address global job satisfaction instead of faceted job satisfaction. Rutherford et al. (2009) found that multi-faceted job satisfaction mediates the relation between emotional exhaustion and the propensity to leave. Given these mixed findings and the lack of previous research on multi-faceted job satisfaction as a mediator, the following hypotheses will be tested:
H7. The positive relation between emotional exhaustion and employee turnover intentions is mediated by an employee’s multi-faceted job satisfaction (including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers).
2.6. Research Model and Hypotheses Work Overload Emotional Labour Emotional Exhaustion Multi-faceted Job Satisfaction - Overall-job - Co-workers - Supervision - Policy - Pay - Promotion - Customers Turnover Intentions H2 H1 H3 H6 H4 + H5 H7
H1. Work overload is positively related to an employee’s emotional exhaustion. H2. Emotional labour is positively related to an employee’s emotional exhaustion.
H3. Emotional exhaustion is negatively related to an employee’s multi-faceted job satisfaction (including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers).
H4. The negative relation between work overload and multi-faceted job satisfaction (including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers) is mediated by an employee’s emotional exhaustion. H5. The negative relation between emotional labour and multi-faceted job satisfaction
(including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers) is mediated by an employee’s emotional exhaustion. H6. Multi-faceted job satisfaction (including 1.Overall job, 2. Co-workers, 3. Supervision,
4. Policy, 5. Pay, 6. Promotion and 7. Customers) is negatively related to an employee’s turnover intentions.
H7. The positive relation between emotional exhaustion and employee turnover intentions is mediated by an employee’s multi-faceted job satisfaction (including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers).
3. METHODOLOGY 3.1. Research Design
This study enhances a deductive and explanatory approach, because preconceived hypotheses based on existing theory are used while explaining relations between variables (Lewis & Saunders, 2012). In order to collect the data, a quantitative approach was chosen which allows for reaching a high number of responses. This was done through the use of a self-administered, internet-mediated questionnaire. This approach is efficient and simultaneously keeps costs low (Lewis & Saunders, 2012). Moreover, data was collected through the distribution of this self-administered questionnaire by delivery and collection. An advantage of this latter approach is that it is more personal and a decent response is often given (Lewis & Saunders, 2012). The study is cross-sectional which means that data is collected at one point in time instead of a longitudinal study, measuring multiple points in time. Even though results of the latter design appear to be more reliable, a cross-sectional design was chosen in this study due to restrictions of costs and time.
3.2. Questionnaire
The developed questionnaire consists of closed questions in which a likert scale of 1 (e.g. strongly disagree) to 5 (e.g. strongly agree) is used. Closed questions are often recommended in this type of study since this increases reliability (Lewis & Saunders, 2012). Besides, control variables such as age, gender and level of education are adopted in the questionnaire. The questionnaire is based on existing measures in the literature of which the language of instruction is English. Since the questionnaire was distributed in the Netherlands, most respondents were expected to be Dutch speaking natives. Probably, a Dutch questionnaire will therefore be easier accessible to people and increase the response rate. Moreover, a Dutch questionnaire will increase the validity of the answers since familiarity with the language increases understanding (Lewis & Saunders, 2012). Therefore, the questionnaire was translated into Dutch with the “back and forward method”, recommended by Field (2005). Following, the questionnaire was evaluated on clarity and possible errors by two different people, after which improvements were implemented. Last, an introduction was written with clear instructions. The introduction explained the purpose of the research and the time it takes to fill out the survey. It also stated that anonymity was guaranteed; results would be treated in a strictly confidential manner and participation was voluntary. Furthermore, contact information was provided in case respondents had any questions or remarks. Also, a word of
thanks was given. In order to maximize response rates, a gift card of €50, - was allotted among all participants. These last steps all contribute to conducting research in an ethical and responsible manner.
3.3. Research Sample
The sample consists of 103 front-line employees working in fashion retail stores in the Netherlands. 76.7% is female and 23.3% is male. This is congruent with statistics on gender of hbd (figure 2) showing that more females than males are working in this industry. The age of the participants is mostly between 17 and 35 years old (92.4%), representing relative young employees. 31.1% works full-time whereas 68.9% works part-time. This is also in line with statistics from hbd, showing that more employees work part-time than full-time. The tenure in job ranges from 5 days to 15 years, and the average tenure is 3 years. The level of education is respectively 19.4% for secondary or high school, 34% for middle level applied education (MBO), 43.7% higher education or bachelor (HBO) and 1.9% a university or master’s degree (WO). The nationality is predominantly Dutch (87.4%) which is also congruent with statistics of hbd. Since this study is specifically focused on the fashion retail industry, only front-line employees working in fashion stores are included in this research. Data was collected by using non-probability sampling methods. This forms a limitation, because it lowers theoretical value and generalizability (Lewis & Saunders, 2012). However, this method is chosen because there was no sampling frame on front-line employees in the fashion retail industry available.
3.4. Data Collection
The questionnaire was developed and distributed with Qualtrics (www.qualtrics.com). In order to reach the right focus group, a digital link was posted and distributed through social media channels such as Facebook. Everyone in the researchers’ network was requested to further distribute or share the link within their own network. Furthermore, possible respondents working in a fashion store were approached via a private message or email in which the link of the survey was pasted. Besides, an email with the link was distributed among students of the Amsterdam Fashion Institute in order to increase the response rate. According to Lewis and Saunders (2012) it is beneficial to distribute digital questionnaires because one is able to reach a high number of participants in a time and cost efficient manner. Besides, this strategy enhances feelings of being ‘anonymous’ which results in less social desirable answers and interviewer bias. Furthermore, questionnaires were printed and
personally distributed among fashion stores. After, these results were uploaded by hand into the online program of Qualtrics.
3.5. Measures
The independent variable is work overload which is measured according to the modified role overload scale, using 3 items (“I never seem to have enough time to get everything done at work”; α = .82) of Bolino and Turnley (2005). This is measured on a five-point Likert-type scale with anchors 1 = Strongly Disagree and 5 = Strongly Agree. Based on Rizzo et al. (1970), Bolino & Tumely (2005) state that “Role overload describes situations in which employees feel that there are too many responsibilities or activities expected of them in light of the time available, their abilities, and other constraints”, p.741.
Emotional labour is defined as “the state that exists when there is a discrepancy between the emotional demeanor that an individual displays because it is considered appropriate, and the emotions that are genuinely felt but that would be inappropriate to display” (Mann, 1999a, p. 353). This variable is measured by the Response-Focused Emotion Regulation Scale from Grandey, Fisk and Steiner (2005) (“I fake a good mood”). The measurement scale from 1 = Never to 5 = Always. A principal-components analysis showed that it met traditional critera (eigenvalues 1.0). α = .89 for the U.S. sample and α = .83 for the French sample. Items 1, 2, 3, 5 and 7 are from the Surface Acting scale used in Grandey (2003) and items 4 and 6 are taken from the 3 -item Surface Acting scale created by Brotheridge and Lee (2003).
Emotional exhaustion is measured in this study by 4 items from Kreitner and Kinicki (1992) (“I feel emotionally drained from my work”; α = .80). Items are measured on a five-point scale, anchored with 1 = never and 5 = always and measured is how often one feels emotionally overextended and exhausted by one’s work. This construct is an adaptation of the Maslach Burnout Inventory (MBI) which originally consists of 9 items.
Multi-faceted job satisfaction consists of seven dimensions (including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers). It is measured with the reduced 28-item INDSALES scale by Comer, Machleit and Lagace (1989). It measures the level of agreement for each statement (“My job is satisfying”; α = .89) with anchors of 1 = strongly disagree and 5 = strongly agree. In measuring satisfaction with supervision, “supervisor” is replaced by “manager” since this definition is practically more relevant.
Intention to leave is the dependent variable and is based on the 4-item scale of Bluedorn (1982). The scale assesses the chance of quitting the job during the next 3 months, 6 months, next year and next 2 years (“How would you rate your chances of quitting this job in the next 3 months”; α = .84). It is anchored by 1 = very low and 5 = very high.
3.6. Control Variables
Control variables are included in the study to control the relations between predictor and outcome variables, thereby reducing unintended effects and improving results (Field, 2005). In this study, the following control variables are included: age, gender, tenure in job, number of hours working, level of education, nationality and organizational service climate.
Age appears to have influence on the research model of this study. According to Bedeian, Ferris and Kacmar (1992), older employees possess more self-confidence than younger employees and are therefore more likely to indicate a higher degree of job satisfaction. Besides, younger employees generally switch more often between jobs because they are still full of ambition and are less committed to one specific organization (Bedeian, Pizzolatto, Long & Griffeth, 1991). Moreover, one can imagine that older employees in the fashion retail have more difficulty with switching from one organization to another. Therefore, older employees are more likely to stay for a longer period at the same company than younger employees.
Gender is incorporated in the analysis since prior literature has proven that there are fundamental differences between male and female in the way they behave and respond to work-related matters. In retail positions, women might experience more intervention between work and private life than men (Babin & Boles, 1998). It therefore appears that men and women respond differently to stress related issues and job satisfaction (Chung et al., 2012).
Tenure in job is also controlled for. The longer an employee works in an organization, the less likely this person will leave the company because organizational benefits are more likely to arise (e.g. promotion, status) (Hellman, 1997). Also, employees who are new to the company go through a process of socialization first. This might affect their thoughts about their job (Hofstede, 1980).
The number of hours also influences the research model. It appears that employees who work part-time experience lower levels of stress than employees who have a full-time job. Besides, part-time workers are more satisfied with their job and are less prone to quit (Wotruba, 1990). Number of hours is measured by asking respondents if they work part-time or full-time.
Level of education is also included as control variable. It appears that the level of education has an influence on the working experience of employees (Bedeian et al., 1992), which again influences important variables in the research model of this study. One might also imagine that someone who is highly educated might have different ambitions than working in a fashion retail store for which an educational background is not required. This again influences an employees’ motivation and also variables such as job satisfaction and turnover intentions.
Nationality is another control variable. Each nationality has a different cultural background and this background influences perceptions and behaviors (Hofstede, 1980). As one can imagine, these perceptions and behaviors might in turn influence job related stress factors, job satisfaction and turnover intentions.
The last variable which is controlled for is service climate. “Service climate refers to employees’ shared perceptions of the practices, procedures, and behaviors that are rewarded, supported and expected by the organization with regard to customer service and customer service quality” (Schneider, White & Paul, 1998 in Salanova, Agut & Peiro, 2005, p. 1217). Service climate appears to influence employee attitudes, perceptions and organizational citizenship behavior (Walumbwa, Hartnell & Oke, 2010) and thus influences the research model of this study. Service climate is measured with a reduced version (4 items) of the 7-item Global Service Climate Scale (Schneider et al., 1998) (“Employees in our organization have knowledge of the job and the skills to deliver superior quality work and service”; α=.84) Answers range from 1 = strongly disagree to 5 = strongly agree.
3.7. Data Analysis
Data analysis of this study can roughly be divided into two steps. First, data was explored and basic assumptions were checked, also referred to as ‘data cleaning’. These steps are needed to prepare the data for the second step of the data analysis; testing hypotheses. In order to perform the data analysis, IBM SPSS statistics 22 was used. In order to perform the first step, all data was coded and variables were created. Next, data was checked on counter-indicative items and missing values. Furthermore, the overall distribution and normality of the data was explored. After, a Confirmatory Factor Analysis (CFA) was performed. CFA focuses on the latent structure of a measurement instrument. Since this is not possible in SPSS, SmartPLS (Ringle, Wende, & Will 2005) is used as software to conduct CFA. “CFA verifies the number of underlying dimensions of the instrument (factors) and the pattern of item-factor relationships (factor loadings). CFA also assists in the determination of how a test should be
scored” (Brown, 2012, p.3). Last, scale reliability was assessed and scale means were computed.
After this first step, hypotheses were tested. Throughout all tests, a significance level of p = .05 was maintained. Hypotheses 1 and 2 were simultaneously tested with a multiple regression analysis. Multiple regression analysis tests the effect of multiple predictor variables on one outcome variable (Field, 2005). This also accounts for hypothesis 6, which tests the relation of multi-faceted job satisfaction on turnover intentions. Multi-faceted job satisfaction contains of multiple variables and thus multiple predictors. There is one outcome variable only, turnover intentions, and therefore multiple regression analysis was also used to test this hypothesis.
Hypothesis 3 predicts the relation between emotional exhaustion and the different dimensions of multi-faceted job satisfaction. In this case, it is not possible to conduct a multiple regression analysis, because there is only one predictor variable and multiple outcome variables. Therefore simple linear regression analysis was used seven times to test the relations between emotional exhaustion and the seven dimensions of multi-faceted job satisfaction. Although better programs exist instead of SPSS to perform this analysis with (e.g. LISREL 8 by Joreskog & Sorbom, 1993), this study chooses for regression analyses in SPSS over other methods or programs due to the time and scope of this research. Besides, the model is very complex and there is a lack of research-based knowledge on how to perform the analysis from and to the different dimensions of multi-faceted job satisfaction (Rutherford et al., 2009).
Hypotheses 4, 5 and 7 all focus on mediating effects. These mediating effects were tested in SPSS with the process script created by Hayes (2012) in which a bootstrap confidence level of 95% was maintained throughout the analyses. Simple mediation was used by means of process 4, focusing only on one mediating variable. An example of a simple mediation model is showed in figure 3. The relation between the independent (X) variable and the mediator (M) is called a. The path between M and the dependent variable (Y), controlled for X, is named b. The indirect effect is the product of a and b (ab). The direct relation between X and Y is expressed by c’. The total effect is the sum of the direct and indirect effect: c = c’ + ab (Preacher & Hayes, 2008).
Figure 3: Mediation Model
4. RESULTS
This chapter explains the findings of the data analysis. First, data cleaning and the associated results will be discussed. Second, hypotheses will be tested and the outcomes will be described. The variables used in the data analysis are work overload, emotional labour, emotional exhaustion, multi-faceted job satisfaction (including 1.Overall job, 2. Co-workers, 3. Supervision, 4. Policy, 5. Pay, 6. Promotion and 7. Customers) and turnover intentions. The control variables included are age, gender, level of education, tenure in job, hours in job and service climate. Nationality has not been included since this variable contained more missing values than all other variables and this would decrease the reliability of the study (Field, 2005). IBM SPSS statistics 22 is used for all analyses, except for the CFA, which is analyzed with SmartPLS (Ringle, Wende, and Will 2005).
4.1. Data Cleaning
4.1.1. Coding Variables and Recoding Counter-indicative Items
The first step in the analysis was to code all items. After, counter-indicative items were re-coded. Coding scales were reversed by using the button ‘Recode into Same Variables’ in SPSS.
4.1.2. Missing Values
Missing values were replaced by using a HotDeck imputation which is, according to Myers (2011), a valid and simple method to perform. A frequency table was produced in which missing values were shown with the frequencies option. In total, 10 items contained missing values. First it was tested if the missing values were less than 10% compared to the total sample, otherwise it is not possible to run a HotDeck imputation (Myers, 2011). This was the case for all variables which contained missing values.
In order to check corresponding deck variables, a correlation test was done with the
correlate-bivariate option. Important is that deck variables have almost no missing data and
show discrete values. Also, deck variables should be related to the variable in which data are missing but the relation between these should not be too important for the outcomes of the research (Myers, 2011). Considering this, deck variables were chosen on basis of the highest (Pearson) correlation. After running the HotDeck macro, frequencies options was used again to check whether all missing values were corrected.
