1
Master Thesis Leadership & Management Date: 22-6-2018
Job insecurity in the gig-economy: considering turnover as a
way to cope?
Author Jorn van Duijnhoven (10590706)
Qualification MSc. Business Administration – Leadership and Management Track Supervisor Ms. E. Federici
Second Reader Dr. C.T. Boon
University Universiteit van Amsterdam
2
Statement of originality
This document is written by Student Jorn van Duijnhoven who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 Table of Contents Statement of originality ... 2 Abstract ... 4 Introduction ... 5 Literature Review ... 9 Gig-worker ... 9 Job Insecurity ... 12 Turnover Intentions ... 13 Job Satisfaction... 14 Career Adaptability ... 16 Career Exploration ... 20
Data and Method ... 22
Sample and Procedure ... 22
Measures ... 25 Analytical Strategy ... 26 Results ... 29 Correlation ... 29 PROCESS ... 31 Discussion ... 36 Practical Implications ... 40
Strengths, Limitations and Future Research... 43
Conclusion ... 45
References ... 47
Appendix ... 53
Appendix 1: Scales used in survey ... 53
Appendix 2: Control variables ... 55
Appendix 3: Job satisfaction Cronbach’s Alpha ... 56
4
Abstract
The gig-economy is a new phenomenon that challenges the traditional economy by providing extremely flexible labor relations between the worker and the requester for gig-work. In the gig-economy a substantial amount of business risk is transferred from the business to the worker. This is argued to contribute to job insecurity among gig-workers. This research extends the literature by investigating the gig-economy from the perspective of the worker.
It was hypothesized that job insecurity in the gig-economy would be positively related to turnover intentions, and that this relationship would be mediated by job satisfaction. Moderators from career construction theory were introduced to the model. Interaction effects of career construction theory have been largely neglected in previous literature. Career adaptability was hypothesized to moderate the relationship between job insecurity and turnover intentions, and the relationship between job insecurity and job satisfaction. Career exploration was hypothesized to moderate the relationship between job insecurity and turnover intentions, as well as the relationship between job satisfaction and turnover intentions.
This research applied a cross-sectional, survey based, research design (N=172) to investigate the gig-economy through the eyes of the worker. The results supported the hypothesis that job insecurity in the gig-economy is positively related to turnover intentions. It was also supported that this relationship is mediated by job satisfaction. Furthermore, the hypothesis that career adaptability would moderate the relationship between job insecurity and turnover intentions was supported. This relationship was stronger for people with high career adaptability. Career adaptability did not moderate the relationship between job insecurity and job satisfaction. This hypothesis was rejected. Also, this research found no support for the hypotheses of the moderating effects of career exploration. Practical implications of these findings are discussed.
5
Introduction
Technological advancement is at the foundation of a substantial amount of change in society. Among others, it has introduced smart phones, self-driving cars, crypto currencies, and more. Recently, technological advancement has made its entrance into the labor market, and has introduced gig-work. Potentially, ‘jobs’ will cease to exist and will be replaced by ‘gigs’, as the gig-economy begins to take shape (Stewart & Stanford, 2017).
The online platforms in the gig-economy bring together peers to perform a task, service, or ‘gig’ between each other. There is no broker between the peers in the form of a business, instead the online platform matches supply and demand. The gig-economy incorporates a wide variety of gigs, which overlap with normal jobs as well (Huws, Spencer, Syrdal & Holts, 2015). The gig-economy has created employment opportunities, with high(er) pay and autonomy for the worker (Graham, Lehdonvirta, Wood, Bernard, Hjorth, & Simon, 2017).
However, the rise of the gig-economy is not only perceived as a positive development. As stated by Aloisi (2015, p.653) “Uncertainty and insecurity are at the price for extreme flexibility”. Some gig-workers are paid per completed gig. Consequently their income is dependent on the supply and demand for the type of gig they perform. Besides, a substantial amount of other business risk and potential costs are shifted to the workers as well (Aloisi, 2015). Gig-workers have no insurance, or other benefits that come with a traditional employment contract. Because of this, the field of research is dominated by the call for fitting legislation for the gig-economy, in an attempt to protect the gig-worker against being exploited by the platform (Aloisi, 2015; De Stefano, 2015; Steward & Stanford, 2017). Despite that, the dynamics behind working in the gig-economy remain undiscovered in the literature. Consequently, in the literature the gig-economy is largely seen through the eyes of the legislator, neglecting the view of the gig-worker.
6
This research dives into this gap. It is argued that the gig-worker is burdened with the business risk that would otherwise be for an employer, which could cause job insecurity. Job insecurity could have negative consequences for the gig-worker. Job insecurity is a source of strain for the employee, and will cause negative outcomes for the employee and organization (Ashford, Lee & Bobko, 1989; Reisel, Probst, Chia, Maloles & König, 2010). Job insecurity could entail a loss of control over the continuance of a work situation (Mauno, De Cuyper, Tolvanen, Kinnunen & Mäkikangas, 2014). While people strive to achieve control over their environment, withdrawal from the source could be a way of coping with this strain. In other words, gig-workers could consider turnover to withdraw themselves from a situation in which they experience job insecurity, i.e. the platform (Mauno et al., 2014; Staufenbiel & König, 2010). Therefore, the first aim of this research is to explore the relationship between job insecurity and turnover intentions in the gig-economy.
Furthermore, the positive relationship between job insecurity and turnover intentions is expected to be partially mediated by job satisfaction. Job satisfaction is one of the primary affective responses to a job, through economic stability, social contact and self-efficacy. Job insecurity is the perception of the discontinuation of the job and the perception of possible negative task events, through which negative emotions manifest and an individual could become dissatisfied with their job (Ashford, Lee & Bobko, 1989; Reisel et al., 2010). Job satisfaction is among the main predictors of employee behavior (Lambert, Hogan & Barton, 2001). It has been proved to be positively related to outcome variables, such as performance and organizational citizenship behavior, and negatively to turnover (Judge, Thoresen, Bono & Patton, 2001; Whitman, Van Rooy & Viswesvaran, 2010; Tett & Meyer, 1993). The negative consequences of job insecurity are argued to lower job satisfaction first, as this is an immediate attitudinal consequence of job insecurity. In turn this will manifest in turnover intentions, which is a long-term behavioral consequence of job insecurity (Reisel et al., 2010). In other words,
7
the second aim of this research is to investigate job satisfaction as a mediator of the relationship between job insecurity and turnover intentions.
Finally, variables from the career construction model of adaptation by Rudolph, Lavigne & Zacher (2017), are introduced as moderators to the model. The first moderator that is introduced to the model is career adaptability. Career adaptability includes the resources an individual possesses to cope with current and anticipated tasks, transitions and traumas in their job (Savickas & Porfeli, 2012). Career adaptability resources increase mobility of an individual and their willingness to take advantage of opportunities on the labor market, while decreasing their dependence on the organization (Savickas & Porfeli, 2012; Rudolph et al., 2017). This in turn increases their turnover intentions (Ito & Brotheridge, 2005). Career adaptability gives an individual a proactive look into the future, the ability to set goals, and the ability to act on those goals. It gives the power to move on towards more lucrative employment opportunities than their current employment situation (Klehe, Zikic, Van Vianen, Koen & Buyken, 2012). Working in the gig-economy is expected to provide a number of challenges, due to the business risks associated with it. An individual who has high career adaptability is expected to proactively cope with the situation by considering their voluntary turnover, because they are confident in their own mobility on the job market. On the other hand, people with less career adaptability could experience less turnover intentions, because they lack the confidence and mobility to move. Therefore, career adaptability is expected to moderate the positive relationship between job insecurity and turnover intentions, so that this relationship is stronger for people who have high career adaptability.
Fiori, Bollmann and Rossier (2015) have explained how career adaptability relates to job satisfaction. Individuals with high career adaptability see challenges in their career as opportunities and a potential foundation to develop additional skills, thereby increasing positive affection towards the job (Fiori et al. 2015). Additionally, their resources will make them
8
perceive to be more in control over workplace uncertainty and confident to overcome obstacles (Fiori et al., 2015). In the gig-economy, career adaptability could help the individual to cope with the uncertainty and instability of gig-work, and could reduce the negative effect the extra business risk has on job satisfaction. While people with lower career adaptability could have less resources to feel in control of their work environment and could experience lower job satisfaction. Consequently, career adaptability is expected to moderate the negative relationship between job insecurity and job satisfaction, so that this relationship is stronger for people who have low career adaptability.
The second moderator that is introduced in the model is career exploration. Career exploration consists of self- and environment exploration. Through career exploration an individual becomes enabled to review different career options, to the extent they are a fit with their inner values (Klehe, Zikic, Van Vianen & De Pater, 2011; Guan, Wang, Liu, Ji, Jia, Fang, Li, Hua & Li, 2015). It is a way of orienting oneself for available options, and of preparing oneself for a change from a current employment situation (Klehe et al., 2012). A prolonged state of exploration will decrease the dependency of the individual on its employer (Klehe et al., 2011). In the gig-economy, people who experience high job insecurity or low satisfaction, could consider turnover more when they have high career exploration. Individuals with high career exploration could have already considered their options on the labor market and could picture themselves in other employment situations. Potentially they could feel like they are a better fit in another job, than that they are in their job in the gig-economy. At the same time, people with low career exploration could see less options outside their current job, and could feel less tenure to quit. Career exploration is expected to moderate the positive relationship between job insecurity and turnover intentions, so that this relationship is stronger for people with high career exploration. Furthermore, career exploration is expected to moderate the
9
negative relationship between job satisfaction and turnover intentions, so that this relationship is stronger for people with low career exploration.
Therefore, the contribution of this paper is threefold. First, it will unravel previously neglected variables in the gig-economy and explain part of the consequences of the insecure nature of the gig-economy from the perspective of the gig-worker. Secondly, it will extend on the knowledge of the variables that mediate the relationship between job insecurity and turnover intentions. This is a relationship that has received little attention in previous literature (Mauno et al., 2014). Finally, it will explore the conditions under which variables from career construction theory interact with challenges on the job level. The proposed conceptual model is depicted in Figure 1.
Figure 1. Research model
Literature Review
Gig-worker
Similar to the concept of gig-working, the field of research on gig-working is relatively new. There is a broad range of overlapping terms for the concept of gig-working, in which some researchers define the concept more widely than others. Friedman (2014) defines gig-workers
10
as workers that are hired under flexible arrangements as independent contractors or consultants, to complete a task, or for a defined time. In his definition of gig-working he does not mention the online aspect of gig-working. Stanford (2017), does include this online aspect in his definition of gig-working, and defines gig-working as work organized through online matching platforms that facilitate actual production. The definition provided by Stanford (2017) shapes the domain of what could be considered as gig-working. However, for this research the scope is narrowed down further. In line with Kenny and Zysman (2016), this paper defines the gig-economy as part of the platform gig-economy, to which capital allocating platforms belong as well. Like Stewart and Stanford (2017), this paper utilizes the definition provided by De Stefano (2015), who categorizes the matching platforms into crowdwork systems and work-on-demand systems. On the one hand, crowdwork systems involve work that can be completed and delivered online, through open platforms. It can involve microtasks, small, or larger tasks, that cannot be completed by a computer. Linking peers is done by creating and calling a bid. The peers could be in geographically dispersed locations (De Stefano, 2015).
On the other hand, work-on-demand systems are overlapping with traditional jobs. They involve activities such as the transportation of people and goods, cleaning and clerical work. Work-on-demand gigs are assigned through the online platform. Workers are paid per gig, through which their income is dependent on supply and demand (De Stefano, 2015).
This paper focusses on the work-on-demand systems and the people that work through these systems, defining them as gig-workers. Work-on-demand systems are upcoming in several sectors in the traditional economy (Stanford, 2017), are challenging traditional forms of employment (Stewart & Stanford, 2017), and are the source of controversy (Bergvall-Kareborn & Howcroft, 2014).
Gig-work is performed on an on-demand basis. There is no guarantee of an ongoing flow of gigs available. This is dependent on supply and demand. Workers could be compensated
11
per gig performed and are required to supply their own equipment to perform the requested task (Di Stefano, 2015). The platform arranging the work is distinct from both the gig-worker and the entity requesting the work, which implies a triangular arrangement between gig-worker, end-user and platform (Stanford, 2017). The platform is digital and is utilized to commission, supervise, and deliver the work. It is also responsible to facilitate the payment (Stanford, 2017).
Gig-workers are flexible in the sense that they are not connected to one type of task or one platform. Also, in case of sufficient demand for their services, gig-workers can decide on their own the hours that they work per week. They can work full-time or part-time in the gig-economy. Gig-work can also be used to derive extra income next to a traditional job.
The price for this extreme flexibility is insecurity among the gig-workers (Aloisi, 2015). Gig-workers do not have a fixed income; their income is based on the amount of gigs they complete. This amount is dependent on the capacity of the gig-worker to perform gigs, and the supply and demand for gigs in the market. Through shifting a substantial amount of business risk to the gig-workers, by not employing them, the platforms take away benefits that come with a traditional employment contract, such as insurance, vacation, and continued payment during illness.
There are no official statistics on the amount of active gig-workers available (Stewart & Stanford, 2017). Such statistics are also not available for gig-work in The Netherlands. Generally, researchers estimate that a small proportion, of under 1% of the total working population in any country, is currently involved with working through a platform (Stewart & Stanford, 2017). In their research, Huws et al. (2017) reported that approximately 4,9% of the working population in The Netherlands works through a platform on a weekly basis. However, this includes work-on-demand systems as well as crowdwork systems.
12
Job Insecurity
The business environment of today calls for the efficient managing of the organization and its employees. Economic pressures have caused countless organizations to engage in restructurings, mergers, acquisitions and downsizing. Consequently, the demand for flexible labor has increased (Staufenbiel & König, 2010). The rise of the gig-economy thrives on this change in the business environment, because it is highly flexible. However, the development towards more flexibility also increases job insecurity (Reisel, et al., 2010; Aloisi, 2015).
Job insecurity has been defined in several different ways (Sverke, Hellgren & Näswall, 2002). Job insecurity is the subjectively experienced anticipation of an involuntary change in the continuity of a job (Sverke et al., 2002). In this sense, the continuity entails both anticipated job loss and anticipated changes in features of the job to which the individual attaches value (Hellgren, Sverke & Isaksson, 1999). From this definition it leads that job insecurity should be seen as separate from actual job loss. It should also be seen as an individual perception of the situation, as an involuntarily change, and as a stressor which causes various kinds of strain (Sverke et al., 2002). Job insecurity as a stressor is related to several negative outcomes for the individual and the organization, including poor mental and physical wellbeing, impaired job attitudes and reduced performance (Vander Elst, De Witte & De Cuyper, 2014; Sverke et al., 2002; Cheng & Chan, 2008).
The perception of insecurity about the continuity of a job can stem from multiple sources. As said before, working in the gig-economy is extremely insecure, due to the flexible nature of gig-work. The income of a gig-worker could be dependent on the capacity of the gig-worker to perform gigs, the demand for their performance, and the supply from other gig-workers in the market. Therefore, the gig-worker is partially not in control of the amount of gigs they can complete, which affects their income. This business risk is only for the gig-worker. They can expect no support from the platform in case there are not enough gigs available. Furthermore,
13
the gig-worker does not have the benefits of a traditional employment contract, such as payment during illness and vacation. Discrimination, social isolation, opacity, taxation and the disturbed power relationship between worker and platform were identified as additional risks of working in the gig-economy, and could further contribute to the feeling of job insecurity (Graham et al., 2017). Therefore, job insecurity is expected to be inherent to the gig-economy.
Turnover Intentions
Turnover is understood as the termination of the employment of an individual within an organization (Tett & Meyer, 1993). Turnover intention is a conscious and deliberate willfulness of the individual to exit the employment relationship (Tett & Meyer, 1993). Important is the distinction between the intended and actual turnover. Nonetheless, turnover intention is the main predictor of actual turnover (Klehe et al., 2011). Turnover has been a major problem for organizations, since it is costly to replace people that have left the organization (Van Dick, Christ, Stellmacher, Wagner, Ahlswede, Grubba & Tissington, 2004).
In the economy there is no employment contract between the platform and the gig-worker. The gig-worker has the ability to voluntary turnover at any point in time. Therefore, it is likely that turnover intentions will manifest in actual turnover, since mobility is higher for the gig-worker compared to a traditional worker. Consequently, it is important for the platforms to understand turnover intentions among their workers. For a platform to survive, it is important that it can at least match the demand for gigs with the supply by gig-workers. Not being able to deliver up to the demand, will leave customers whose gig was not performed. To prevent this, platforms should uphold a large base of gig-workers, registered to the platform, and should keep turnover intentions low among their gig-workers.
In this study it is expected that job insecurity is positively related turnover intentions. In the gig-economy, job insecurity could spark a feeling of powerlessness, because the individual
14
has lost control over the continuity of the job (De Witte et al., 2014; Mauno et al., 2014; Staufenbiel & König, 2010). It makes it hard for an individual to picture a future within the organization (Hellgren et al., 1999). It is a stressor which is correlated with impaired wellbeing (De Witte et al., 2014; Staufenbiel & König, 2010). While one of the main motivators in life is to achieve control, this is impaired by job insecurity, and might not be achieved in the gig-economy (Mauno et al., 2014). One way of coping with this, could be by withdrawal from the job, to prevent the negative impact of job insecurity and potential job loss. This withdrawal could manifest itself in increased turnover intentions (Mauno et al., 2014; De Witte et al., 2014; Staufenbiel & König, 2010). Therefore, the first hypothesis is:
H1: There is a positive relationship between job insecurity and turnover intentions.
Job Satisfaction
There is a broad range of literature available on the topic of job satisfaction. Job satisfaction is defined as a state of being, resulting from the appraisal of a job or job experiences (Locke, 1976). It is an attitude towards aspects of a certain job or part of the job (Van Dick et al., 2004). In which the job for this research is gig-work. Job dissatisfaction is then a negative state of being, derived from the job or aspects of the job.
Job satisfaction can stem from a variety of sources including context and characteristics of the job, as well as the quality of social relationships (Van Dick et al., 2004). Job satisfaction is the primary affective response to the job and stems from economic stability, social contacts and self-efficacy (Ashford et al., 1989; De Witte, 1999; Reisel et al., 2010).
Job insecurity is the perception of the continuation of the job, so it threatens the continuation of the affective response. In the gig-economy, business risk is transferred to the gig-worker. Consequently, the gig-worker is burdened with the risk, among which the largest
15
is whether there will be enough demand for their services, and whether there is no oversupply. Both could threaten the continuity (feasibility) of the job as a gig-worker.
Anticipating on the potential negative consequences, will disturb the affection of an individual towards the job through negative emotions becoming associated with the job (Ashford et al., 1989; Sverke et al., 2002). The uncertainty regarding the continuation of the job is a stressor, which will result in strain (Sverke et al., 2002). Together these factors could negatively affect job satisfaction, which is in line with the meta-analysis by Sverke et al. (2002), who have tested, and confirmed such a relationship in a meta-analysis. Therefore, hypothesis 2a is:
H2a: There is a negative relationship between job insecurity and job satisfaction.
Job satisfaction frequently appears in research models and has been proven to positively relate to several outcome variables, such as performance and organizational citizenship behavior, and negatively to turnover (Judge et al., 2001; Whitman et al., 2010; Tett & Meyer, 1993).
The positive consequences for the organization of high job satisfaction among employees are expected to be reversed when there is job dissatisfaction (Reisel et al., 2010). It has been proved that an individual voluntary withdraws themselves from a work environment in which they are not satisfied, by avoiding the job or voluntarily terminating the employment contract (Hanisch, Hulin & Roznowski, 1998; Hom & Kinicki, 2001). Furthermore, job satisfaction is the affective attachment to the job. In a situation in which there is job dissatisfaction, this affective attachment disappears and an individual will start thinking about employment elsewhere (Tett & Meyer, 1993; Trevor, 2001). Consequently, hypothesis 2b is:
16
Accordingly, part of the relationship between job insecurity and turnover intentions can be explained by job satisfaction. In other words, job satisfaction is expected to mediate the relationship between job insecurity and turnover intentions. In a situation where there is job insecurity, i.e. the gig-economy, an individual could feel uncomfortable, because job insecurity is a source of strain (Sverke et al., 2002), it impairs well-being (De Witte et al., 2014; Staufenbiel & König, 2010) and entails a loss of control (Mauno et al., 2014). These consequences contribute to a lower job satisfaction as well.
As said, job satisfaction is one of the main predictors of behavior, also for turnover intention (Judge et al., 2001; Whitman et al., 2010; Tett & Meyer, 1993). The consequences of job insecurity can be categorized as immediate and long-term consequences (Sverke et al., 2002; Reisel et al., 2010). Job satisfaction is an example of an immediate consequence of job insecurity, whereas turnover intention is an example of a long-term behavioral effect (Sverke et al., 2002; Reisel et al., 2010). In line with Reisel et al. (2010) it is therefore expected that the effects of job insecurity will manifest first on job satisfaction, which will then act as a mediator of the relationship of job insecurity and turnover intentions. Consequently, it is hypothesized that job insecurity will lower job satisfaction, which will explain partially the behavior that results from job insecurity, in this case turnover intentions. The second hypothesis is:
H2: The positive relationship between job insecurity and turnover intentions is mediated by job satisfaction.
Career Adaptability
Research on career adaptability is relatively new (Maggiori, Johnston, Krings, Massoudi & Rossier, 2013). Career adaptability roots in career construction theory, and explains how an individual constructs a career (Maggiori et al., 2013). Most research utilizes the definition given
17
by Savickas (1997) for career adaptability stating it as a psychological construct that describes the resources of an individual for coping with current and anticipated tasks, transitions, traumas in their occupational roles that, to some degree large or small, alter their social integration (Savickas & Porfeli, 2012, Maggiori et al, 2013; Koen, Klehe, Van Vianen, Zikic & Nauta, 2010; Klehe et al., 2011).
Career theories have become more relevant, since labor relationships between employee and employer have shortened. Restructurings and downsizing have become more frequent with technological advances, global competition, state deregulations and changing market conditions, which in turn contributed to the changing nature of careers (Klehe et al., 2011).
The personal resources that underlie the degree of career adaptability of an individual are called adapt-abilities (Savickas & Porfeli, 2012). They include concern, control, curiosity and confidence. Concern about the future helps an individual anticipate on and prepare for what is coming. Control helps an individual to shape themselves and their surroundings to match what is coming, and roots in self-discipline, effort, and persistence. Curiosity helps an individual to picture themselves in different situations and roles, thereby exploring possible selves. Confidence helps an individual act and take their resources into practice (Savickas & Porfeli, 2012).
The meta-analysis by Rudolph et al. (2017) confirmed that career adaptability positively relates to several career- and non-career related outcomes. For example, career adaptability relates to the quality of life (Johnston, Luciano, Maggiori & Rossier, 2013), personal and professional well-being (Maggiori et al., 2013) and the development of skills (Bimrose, Brown, Barnes & Hughes, 2011). What is lacking in the literature is the knowledge of how career adaptability interacts with job and career challenges (Rudolph et al., 2017). In this research, there are two interaction effects investigated. First, the interaction effect of job insecurity and career adaptability on turnover intentions will be investigated. Then interaction effect of job
18
insecurity and career adaptability on job satisfaction will be investigated. The interaction effect on job satisfaction is particularly interesting, because in the meta-analysis by Rudolph et al. (2017), it was also indicated that how career adaptability relates with job satisfaction requires more investigation.
Career adaptability could decrease dependence of an individual on the organization in which they are employed, because the individual has the resources to be more mobile and the willingness to find employment elsewhere (Ito & Brotheridge, 2005). This has the potential to increase their intentions to turnover. Furthermore, people with high career adaptability should have the power to envision and act upon more lucrative job opportunities, outside their current employment. Moreover, it could give them the ability to adapt their work environment to their needs by considering the possibility of employment elsewhere (Klehe et al., 2012).
In the gig-economy, it is expected that career adaptability will moderate the relationship with turnover intentions. As explained before, the extreme flexibility at the nature of the gig-economy creates risk which could contribute to job insecurity. Career adapt-abilities could help people cope with current and anticipated tasks, transitions and traumas in their jobs (Savickas & Porfeli, 2012). It can be seen as a way of proactive coping with a given situation (Klehe et al., 2012). Career adaptability resources give people the power to set goals and act upon these goals, and the ability to envision a different future state of employment (Klehe et al., 2012). Combined with higher mobility and willingness, that individuals with career adaptability resources possess, people are thought to be more confident in their ability to take control of the situation and could consider turnover as a way of proactive coping (Ito & Brotheridge, 2005; Klehe et al., 2012). A person with high career adaptability is expected to cope with the low job security in the gig-economy by considering turnover. Hence, the third hypothesis is:
19
H3: The positive relationship between job insecurity and turnover intentions is moderated by career adaptability, so that this relationship is stronger for high values of career adaptability.
Job satisfaction was defined as a state of being, resulting from the appraisal of a job or job experiences (Locke, 1976). It is an attitude towards aspects of a certain job or part of the job (Van Dick et al., 2004). It stems from economic stability, social contacts and self-efficacy (Ashford et al., 1989; De Witte, 1999; Reisel et al., 2010).
People with high career adaptability have the possibility to adapt themselves to the demands of their job, and may thus feel more at home in their job (Fiori et al., 2015). They may also see the challenges in their work environment as opportunities to learn from, evoking more positive emotions with their job (Fiori et al., 2015). A similar path was speculated by Rudolph et al. (2017), that people with low job satisfaction may benefit most from career adaptability.
Fiori et al. (2015) found that career adaptabilities have a strong effect on the way an individual copes with negative affect from their job. The job insecurity inherent in the gig-economy might be an example of negative affect and a challenge which could be derived from a job. People with high career adaptability resources could feel more in control of their workspace due to their ability to cope with challenges better than others, and could therefore experience less stress than their peers (Fiori et al., 2015). They could even see these challenges as opportunities to learn (Fiori et al., 2015) The gig-economy is expected to pose a number of challenges, among which job insecurity. Through the career adaptability, an individual will experience less negative affect than their peers and will have a more positive attitude towards their job as a gig-worker. As a result, the fourth hypothesis is:
H4: The negative relationship between job insecurity and job satisfaction is moderated by career adaptability, so that this relationship is stronger for low values of career adaptability.
20
Career Exploration
In line with Hirschi, Hermann and Keller (2015), career exploration is defined as distinct from career adaptability. Career adaptability are adaptability resources, while career exploration is an adapting response (Rudolph et al., 2017). Career exploration entails self-exploration and environmental exploration (Stumpf, Colarelli & Hartman, 1983; Klehe et al., 2012; Guan et al., 2015).
Self-exploration is about reviewing values, goals, interests, skills and experiences, to create a comprehensive overview of one’s career and career ambitions (Stumpf et al., 1983; Klehe et al., 2012). Self-exploration can give direction to the further career of an individual by creating an inner compass (Klehe et al., 2012). Environmental exploration involves collecting and reviewing information on jobs, organizations and industries (Klehe et al., 2012). This process helps individuals to make more informed decisions on their career, and thus on their future employment (Guan et al., 2015).
The exploration activities help an individual to create a wide focus beyond their current job (Koen et al., 2010). A prolonged state of exploration is expected to be related to turnover intentions, because an individual will start to envision themselves in more lucrative and fitting job opportunities elsewhere, that were found during exploration (Klehe et al., 2011). Career exploration also entails an open-mindedness through which an individual could become more curious with other job opportunities (Klehe et al., 2012). Career exploration gives the individual the knowledge on how to move towards other jobs (Ito & Brotheridge, 2005).
Previous research has found career exploration to be positively related to job search intensity (Werbel, 2000), job search strategies (Koen et al., 2010), and reemployment quality (Zikic & Klehe, 2006). Most research performed on career exploration investigated direct- or indirect relationships between career exploration and an outcome variable (e.g. Werbel, 2000; Zikic & Klehe, 2006; Klehe et al 2011; Koen et al, 2010). Consequently, interacting effects
21
have largely been overlooked in current literature on career exploration (Klehe et al., 2011; Rudolph et al., 2017). This research attempts to fill both gaps, by investigating the interacting effects of career exploration and job insecurity and job satisfaction on turnover intentions.
In the gig-economy it is expected that career exploration will moderate the relationship between job insecurity and turnover intentions. An individual in the gig-economy is faced with a substantial amount of business risk, which contributes to job insecurity. An individual with high career exploration is expected to be aware of what employment opportunities are available outside their current job as a gig-worker. They are also thought to know what they want, and to be aware of their own abilities (Koen et al., 2010; Klehe et al., 2011; Klehe et al., 2012). People with high career exploration are expected to see that being in a situation in which they experience job insecurity is not necessary. Instead, they could be aware of employment opportunities in which they are a better ft, either by exploring career opportunities within, or outside the gig-economy. Therefore, the fifth hypothesis is:
H5: The positive relationship between job insecurity and turnover intentions is moderated by career exploration, so that this relationship is stronger for high values of career exploration.
A similar argumentation could be followed for the interaction effect between job satisfaction and career exploration, on turnover intentions. An individual with high career exploration is expected to be aware of employment opportunities outside their current employment (Koen et al., 2010; Klehe et al., 2011; Klehe et al., 2012). In a situation in which they are dissatisfied, potentially they are aware of employment opportunities in which they can picture themselves being more satisfied. People with high career exploration are hypothesized to have the knowledge of what they expect in a job, and a clear view of their needs in a job and how it fits with themselves and their career. They are expected to consider turnover, to withdraw
22
themselves from their current job in which they are dissatisfied. Therefore, the sixth and last hypothesis is:
H6: The negative relationship between job satisfaction and turnover intentions is moderated by career exploration, so that this relationship is stronger for low values of career exploration.
Data and Method
Sample and Procedure
To test the hypotheses a correlational research design was adopted. This allows to create a snapshot of how and when job insecurity is related to turnover intentions, in the context of the gig-economy. Using this design grants the opportunity to make conclusions on the correlation between the variables in the model. Data was collected on a cross-sectional time-horizon, by means of a survey. The data was collected as a joint effort between 4 master students from the Leadership and Management track, at the University of Amsterdam. It was agreed upon that data would be collected from April 19 up until May 18. Data was collected through a non-probability sampling technique. The gig-economy is currently a niche. To get to a decent sized sample, every gig-worker that we had access to was approached to fill in the survey. The goal was to collect 320 responses in total.
This research was originally targeted at the gig-worker and their perception of working in the gig-economy in The Netherlands. We have contacted 19 platforms in The Netherlands, to ask if they were willing to cooperate in our research and spread the survey among the workers that were registered to their platform. The platforms varied in the gigs they host and the way they are organized. There were 6 platforms active in the (car) transportation industry. 4 platforms were active in the food delivery industry. The rest of the platforms performed a variety of gigs, including clerical work and homeschooling. The initial contact was done
23
through e-mail, further contact was done through the phone. Eventually, only 2 platforms agreed to cooperate with us. They either gave us the opportunity to get in touch with their workers, or would spread the survey for us.
In addition to reaching out to the workers through the platforms, we have reached out to gig-workers directly in our network. To collect data, these workers were asked to fill out surveys online, or offline. We would also ask our network to distribute the survey to other gig-workers they knew. Furthermore, we would go in pairs to known hotspots for gig-gig-workers, where they would have to start their shift, for example. A hardcopy survey would take between 10 and 15 minutes to complete. To motivate gig-workers to fill in our survey, we would offer them the chance to win one of the €50,- vouchers.
2 weeks into data collection, the data collection was reviewed. It became apparent that data collection was slower than previously thought. We would not be able to collect 320 responses within the time limit. Qualtrics, the host of the survey, offers data collection services. With Qualtrics we agreed that they would gather responses, in exchange for a fee, to increase the sample size.
As a result, our prior defined scope, The Netherlands, was broadened to the United States of America and the United Kingdom. Fortunately, gig-work is relatively similar in these countries. Gig-work in western countries is rewarded and organized relatively similarly (Graham et al., 2017). Some of the platforms active in The Netherlands, were also active in the USA and UK.
We created a new version of the survey for respondents through Qualtrics. This new version of the survey included minor adjustments. One item was added to request the country in which the respondent would perform gig-work. The control variable for type of employment was adjusted by removing “ZZP” from the answers. ZZP is a term that is unique to The Netherlands. Furthermore, we would not give these respondents the opportunity to win a
24
voucher. Moreover, the introduction text was adjusted to emphasize the fact that the respondent would have to be a gig-worker. Also, 3 quality checks were added to the survey, for example “Please answer strongly agree to this question”. These quality checks allowed us to monitor the quality of the data. Respondents that did not answer these questions correctly, were not included in the final sample.
Consequently, there were 2 separate data sets for respondents that we contacted ourselves, and respondents that were contacted by Qualtrics. In the end, we were able to get 85 respondents ourselves, while Qualtrics got 195 respondents for us. This adds up to a total of 280 respondents when the data sets were merged using SPSS (version 24).
Although research from the perspective of the gig-worker is scarce, research that has previously been performed on the perspective from the gig-worker experienced low response rates. Graham et al. (2017) reported a response rate of 30% and 7% when they asked two platforms to distribute their survey. For this research there were multiple channels through which we reached out to gig-workers. It is not possible to estimate how many people we have reached through all these channels. Therefore, no comment on the response can be made.
The average age of the respondents was 33 years old. 63,5 percent of the respondents was male, 36,5 percent of the respondents was female. On average, respondents worked 24 hours per week through a platform. While the respondents were working for 3,1 years on average through a platform. 47 percent of the respondents were from the United States of America, 36 percent of the respondents were from the Netherlands, and 17 percent was from the United Kingdom. The largest part of the sample is paid per task (53,6%), followed by paid per hour (42%). Only 4,4 percent of the respondents was paid per set amount. 36,4 percent of the respondents was self-employed, 34,3 percent of the respondents had a contract for a set amount of hours, while 29,3 percent of the respondents had a contract for 0 hours. 57,9 percent of the respondents has 1 paid job at the time of filling in the survey, the rest of the respondents
25
had more than 1 job. The largest part of the respondents worked through a taxi platform (44,2%), followed by the 3 food delivery platforms in this sample (24,9%, 13,3% and 8,3%). The other platforms made up for 9,3 percent of the sample.
Measures
The scales used in the survey were all previously validated scales, that were used in other studies and reported adequate Cronbach’s Alphas. To make the survey accessible to the sample, the scales were translated to Dutch. This was done using a back-translation method, if the translation had not been provided in the previous study. No scales required adjusting to the context of the gig-economy.
Job insecurity was measured using the scale by Vander Elst et al. (2014), containing 4 items. An example of an item is: ‘Chances are, I will soon lose my job’. The scale reported a Cronbach’s Alpha (α) of 0,82 in the current research.
Turnover intentions was measured using the scale by Bozeman and Perrewé (2001), containing 5 items. An example of an item is: ‘I will probably look for a new job in the near future’. The current research reported an α of 0,75.
Job satisfaction was measured using the scale by Cammann, Fichman, Jenkins and Klesh (1979), containing 3 items. An example is: ‘All in all I am satisfied with my job’. The current research yielded an α of 0,62.
Career adaptability was measured using the scale by Maggiori, Rossier and Savickas (2017). Respondents were asked how strongly they have developed the abilities in the scale. An example is: ‘Preparing for the future’. The current study reported an α of 0,93.
Career exploration was measured using the scale by Werbel (2000). To shorten the total survey length, only the items related to environmental exploration were used in the survey. Respondents were asked to what extent they have behaved in the ways described in the scale.
26
An example is: ‘I went to various career orientation programs’. The current study reported an α of 0,90.
An overview of the scales used in the survey can be found in appendix 1, along with their translation.
Several control variables were added to the survey. This was done to be able to control for factors that could affect the outcome of the model. A wide variety of control variables was added, including age, gender and hours spend gig-working. An overview of the control questions is presented in appendix 2.
Scales were considered reliable when Cronbach’s Alpha (α) > 0,70. Job satisfaction showed a problematic α of 0,62. Further analysis showed that this was due to the reverse coded item, JobSat2. Deletion of JobSat2 increased the α significantly to 0,87 (appendix 3). The decision was made to exclude JobSat2 from further analysis due to the low reliability. All other scales remained unadjusted.
Analytical Strategy
After merging the 2 data sets, the dataset was cleaned for it to be used for data analysis. The variables that were included to the dataset by default by Qualtrics, but that were of no use to this research, were deleted. An example is IP address of the respondent.
Next, case numbers to each respondent in the data set were computed. The data was sorted by ‘source_international’. The new variable with the case number was called ‘Case_ID’. Cases were numbered 1 up until 280.
Then cases were deleted based on certain quality checks and inclusion criteria. In the survey we have included the inclusion item ‘Do you work through a platform?’ at the start of the survey. 36 cases were filtered out, because the question was empty or was answered ‘No’. Then cases were deleted for which the completion rate of the survey was lower than 44%. These
27
cases would yield a substantial amount of missing data. 57 cases were filtered out due to this criteria. For the data that we gathered ourselves, we distinguished that respondents would not be able to fill out the survey correctly in less than 5 minutes (300 seconds). 1 case was filtered out as a result. Finally, we performed a check on the item ‘Which platform(s) do you work through?’ for the answers ‘Other’, to check whether the platform that was filled in the open field was within our scope. 5 cases were filtered out, because the platform was out of scope.
Afterwards the data was checked for typos. Frequency analysis was performed on the control variables. For example, 2 respondents answered ‘100’ to the item ‘How many paid jobs do you currently have?’. 2 similar cases were found in other variables. Typos like these were treated as missing data.
Dummy variables were created for all variables that were answered through a category (more than 2). Dummy variables were created for 6 platforms through which the respondents worked, for the different pay forms (Task, Hour and Set amount), for the different employment arrangements (0 hour contract, Set hour contract and self-employed) and for the countries in which workers performed gig-work (USA, UK and NL).
Then the reversed items in the scales were reverse coded. The reversed items in the data set were JobSat2, JobIns2, and TUI 3,4 and 5. Each scale used a 5-point Likert Scale, so the items could be reversed in one go, using ‘recode into different variable’ option in SPSS (1=5, 2=4, 3=3, 4=2 and 5=1).
2 separate principal factor analyses (PAF) were performed on the scales, to check which items load to which factors, i.e. if they measure the same factor. The first factor analysis was performed on the scales of career adaptability and career exploration. Kaiser-Meyer-Olkin measure (KMO) = 0,92, indicating no problem with adequacy of the analysis, and Bartlett’s test of sphericity χ² = 2008,77, with p = < 0,01, indicated that the correlations were large enough. There were 2 factors with eigenvalues > 1, which corresponded with the amount of scales in
28
this analysis. Further analysis was done on the pattern matrix, on the criteria, principal loading > 0,30 and principal loading at least twice as big as secondary loading. The results showed a clean factorial structure, with all factors loading onto the related factor. The pattern matrix can be found in appendix 4.
The second PAF was on the scales of job satisfaction, job insecurity and turnover intentions. KMO = 0,75 was substantially lower than the previous PAF. However, it can still be seen as adequate enough when KMO > 0,5 (Williams, Onsman & Brown, 2010). The Bartlett’s test of sphericity χ² = 949,43, with p = < 0,01, showed to be significant. The PAF showed that there were 4 factors with eigenvalues > 1, which is larger than the amount of factors that was expected, considering the 3 scales that were used for the 3 variables. The same criteria as in the previous PAF were considered. It showed that JobIns2 and TUI3, 4 and 5 were not complying to the criteria. TUI 3 and 5 appeared to be loading on more than one factor. Also, it appeared that JobIns2 and TUI4 loaded on the same, fourth factor, that was not included in the model. This could be due to the content of the items, which do overlap. JobIns2 asks whether workers are sure they can keep their job, while TUI4 asks whether workers will actively look for a job in another organization. When a worker is sure they can, and want, to stay in the same job, they will not actively look for another job. Because the scales were reliable in the Cronbach’s Alpha test, and previously validated scales were used, the decision was made not to adjust the scales any further. The related pattern matrix can be found in appendix 4.
Normality of the data was checked by calculating the skewness and kurtosis for each scale. The criteria for normality was skewness and kurtosis between 1 and -1. Job Insecurity and Job Satisfaction did not fall within this range, and were checked for outliers. No outliers were identified as problematic. Outliers were also controlled for when running the model in PROCESS.
29
Results
Correlation
After the described steps, the data was ready for hypothesis testing. The correlation table includes all the variables that are in the model, as well as the control variables.
There is a significant positive correlation between job insecurity and turnover intentions (r=0,36**). There is also significant negative correlation between job insecurity and job satisfaction (r=-0,21**). Finally, there is a significant negative correlation between job satisfaction and turnover intentions (r=-0,42**).
Control variables were added to the correlation table to check whether they have a significant correlation with the variables in the model. If the control variables had a significant correlation, they were included in the analysis, because they could impact the outcome of the model. In case the control variable was multicategorical dummy variables were created and included in further analyses.
On the basis of the criteria of significance, the control variables age, hours gig-work per week, taxi platform, location USA and NL, zero hours contract and set hours contract and paid per task, were added as control variables (covariates) when running the model in PROCESS.
30 Table 1. Means, Standard Deviations, and Correlations
Mean SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1. Job insecurity 3,96 0,89 (,82) 2. Job satisfaction 2,41 1,04 -,21** (,87) 3. Turnover intention 2,80 0,92 ,36** -,42** (,75) 4. Career adaptability 3,86 0,78 -,14 ,39** -,28** (,93) 5. Career exploration 3,21 1,08 ,18* ,15* ,20** ,39** (,90) 6. Gender 1,38 0,49 -,14 ,04 -,20** ,11 ,06 - 7. Age 32,75 10,48 -,19* ,21** -,35** ,22** -,05 ,06 - 8. Hours gig-work per week 24,19 13,69 -,04 ,18* -,21** ,30** ,19* -,02 ,19* - 9. Taxi platform 0,47 0,50 ,10 ,20** -,15* ,34** ,28** ,01 ,24** ,25** - 10. Delivery platform A (D) 0,13 0,33 ,06 -,08 ,02 -,08 -,04 -,02 -,13 -,20** -,36** - 11. Delivery platform B (F) 0,24 0,43 -,02 -,14 ,20** -,31** -,31** -,10 -,20** -,12 -,52** -,21** - 12. Variety platform (T) 0,02 0,15 -,12 -,01 ,21** ,04 ,11 ,04 -,11 -,10 -,14 -,06 -,09 - 13. Delivery platform C (U) 0,08 0,27 -,10 -,03 -,11 ,03 ,06 ,16* -,08 ,10 -,28** -,11 -,17* -,05 - 14. Cleaning platform (H) 0,01 0,11 ,00 ,04 -,06 ,07 ,11 ,03 ,14 -,01 -,10 -,04 -,06 -,02 -,03 - 15. Location USA 0,49 0,50 -,04 ,27** -,35** ,32** ,17* ,15* ,44** ,26** ,57** -,38** -,47** -,15* ,13 ,11 - 16. Location UK 0,17 0,38 -,01 -,02 -,06 ,03 ,12 ,05 -,04 -,02 ,00 ,47** -,26** -,07 -,02 -,05 -,45** - 17. Location NL 0,33 0,47 ,05 -,27** ,42** -,36** -,28** -,20** -,44** -,25** -,61** ,03 ,71** ,22** -,12 -,08 -,70** -,32** - 18. Source of income (primary/secondary) 1,31 0,46 -,01 -,03 ,08 -,13 -,13 -,01 ,03 -,32** ,08 -,07 -,14 -,02 ,03 ,05 ,15 -,14 -,04 - 19. Zero hours contract 0,30 0,46 ,16* -,19* ,12 -,30** -,11 -,02 -,14 -,07 -,18* ,05 ,20** -,10 -,01 -,07 -,22** ,06 ,18* ,08 - 20. Set hours contract 0,34 0,48 -,09 ,19* -,11 ,12 ,00 ,06 ,00 ,13 -,01 -,02 ,17* -,11 -,08 ,15* -,10 ,12 ,01 -,27** -,48** - 21. Self-employed 0,35 0,48 -,06 -,01 ,00 ,16* ,10 -,04 ,13 -,07 ,19* -,03 -,36** ,21** ,09 -,08 ,31** -,18* -,19* ,19* -,49** -,54** -
22. Paid per task 0,54 0,50 ,09 ,02 -,17* ,23** ,19* ,06 ,14 ,01 ,46** ,00 -,58** -,09 ,06 -,01 ,37** ,12 -,49** ,14 -,16* -,17* ,32** -
23. Paid per hour 0,42 0,49 -,06 -,09 ,19* -,25** -,20* -,14 -,17* -,05 -,41** -,01 ,60** ,10 -,12 -,09 -,41** -,11 ,53** -,13 ,19* ,18* -,36** -,92** -
24. Paid per set amount
0,03 0,18 -,09 ,13 -,05 ,04 ,06 ,18* ,10 ,10 -,11 ,02 -,11 -,03 ,18* ,28** ,13 ,00 -,13 ,01 -,06 -,07 ,12 -,21** -,16* - **. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed). c. Listwise N=172
31
PROCESS
Testing hypotheses was done using PROCESS, an extension tool for SPSS, that allows for moderation, mediation and conditional process analysis (Hayes, 2012). PROCESS combines functions from other popular statistical computing programs for testing hypotheses, into one simple-to-use extension tool for SPSS (Hayes, 2012).
With the help of PROCESS a wide range of models can be tested, among which model 29, which corresponds with the model in this research (Hayes, 2017). The use of PROCESS is adequate, because it allows for simultaneous testing of the hypotheses in the model, while controlling for the control variables distinguished in the previous section (Hayes, 2012). Furthermore, it is time-efficient and fail proof, because it does not require extensive learning to use and does not require additional non-automatic regression routines (Hayes, 2012).
Additionally, PROCESS is a robust way of testing hypotheses. It takes a bootstrapping approach to obtaining confidence intervals (Hayes, 2012). Bootstrapping includes resampling from the dataset, and calculating the confidence intervals for each resample (Preacher & Hayes, 2008). This process should be repeated at least 1000 times, but 5000 times is recommended (Hayes, 2012). Bootstrapping will provide a lower- and upper limits to the confidence intervals, and will thus control for a normal distribution of the data. A significant effect can be assumed when 0 is outside the confidence interval.
The model in this research includes moderators and mediators. To work with moderators in PROCESS, the independent variables and moderators must be standardized or mean-centered (Hayes, 2012). While both should yield the same results, standardizing was chosen over mean-centering based on the recommendation by Dawson (2014). This was done in SPPS while doing descriptive statistics, using the ‘save standardized values as variables’ option.
To test the hypotheses, 3 times a different model was ran in PROCESS. First model 4, for simple mediation was ran to test hypothesis 1 and 2. Then model 29 was ran to test the
32
model for moderated mediation. Then model 5 was ran to probe the significant interaction effect. All models were tested in combination with the control variables as defined in the previous section.
Model 4 was ran with turnover intention as Y, job insecurity as X, and job satisfaction as M. This yielded the results as depicted in table 2:
Table 2. Test simple mediation
The results in table 2 show the direct- and indirect relationship between job insecurity and turnover intention, with job satisfaction as mediator. 38% of total variance of turnover intentions is explained by the total model (p<0,01). From the results it shows that there is a significant direct relationship between job insecurity and turnover intention (c’1=0,23. p<0,01). This means that hypothesis 1 is supported.
16% of total variance of job satisfaction is explained by the model (P<0,01). The results show a significant negative direct relationship between job insecurity and job satisfaction
Job Satisfaction (M) Turnover Intention (Y)
Antecedent Coeff. SE p Coeff. SE p
Job Insecurity a1 -,136 ,064 ,034 c'1 ,227 ,057 <,001
Job Satisfaction - - - b1 -,290 ,068 <,001
Age ,004 ,007 ,622 -,010 ,006 ,102
Hours gig-work per week ,005 ,005 ,352 -,005 ,004 ,244
Taxi platform ,111 ,173 ,521 ,347 ,154 ,026
Zero hour contract -,091 ,167 ,585 -,160 ,148 ,282
Set hours contract ,296 ,160 ,066 -,168 ,143 ,243
Location USA ,256 ,193 ,192 -,162 ,172 ,348
Location NL -,151 ,211 ,476 ,699 ,188 ,022
Paid per task -,119 ,152 ,437 -,301 ,135 ,028
Constant i1 3,95 ,379 <,001 i2 4,100 ,377 <,001
R2=,1608 R2=,3838
F=3,576 p<,001 F=10,403 p<,001
Effect SE p LLCI ULCI
Direct effect c'1 ,267 ,059 <,001 ,150 ,384
Total effect c'1 ,227 ,057 <,001 ,114 ,341
BootSE BootLLCI BootULCI
33
0,14. p=0,03). People who experience relatively higher job insecurity, have lower job satisfaction. Therefore, hypothesis 2a is supported.
Furthermore, the results show a significant negative direct relationship between job satisfaction and turnover intentions (b1=-0,29. p<0,01). So, people who experience relatively higher job satisfaction, will have lower tendency to quit. Hence, hypothesis 2b is supported.
Finally, the results show that job satisfaction mediates the relationship between job insecurity and job satisfaction (a1b1=0,039. CI: 0,003 to 0,088). This indicates that people who experience job insecurity, will be more dissatisfied with their job, which translates into turnover intentions. As a result, hypothesis 2 is supported as well.
The results of model 29 are shown in table 3. Using model 29, the moderated mediation model could be tested. In addition to job insecurity, turnover intention, and job satisfaction, career adaptability and career exploration were added as moderators to the analysis.
Model 29 made it possible to comment on the interaction effects in the model. Hypothesis 3 predicted the interaction effect of job insecurity and career adaptability on turnover intentions. The results confirm that there is a significant interaction (c’4=0,19. p<0,01).
The model included 3 other interactions, of which none were significant. This means that H4 (a3=0,06, p=0,341), H5 (c’5=-1,07, p=1,07), and H6 (c’6=-0,73, p=0,351), are all rejected. Consequently there is no moderated mediation as the model predicted. There is only moderation on the direct relationship between job insecurity and turnover intention.
34
Table 3. Test moderated mediation
Job Satisfaction (M) Turnover Intention (Y)
Antecedent Coeff. SE p Coeff. SE p
Job Insecurity a1 -,129 ,066 ,051 c'1 ,154 ,057 ,008
Job Satisfaction - - - b1 -,306 ,067 <,001
Career Adaptability a2 ,267 ,072 <,001 c'2 -,070 ,069 ,311
Career Exploration - - - c'3 ,589 ,332 ,078
Job Insecurity x Career
Adaptability a3 ,064 ,067 ,341 c'4 ,185 ,066 ,006
Job Insecurity x Career
Exploration - - - c'5 -,107 ,066 ,107
Job Satisfaction x Career
Exploration - - - c'6 -,073 ,078 ,351
Age ,004 ,007 ,601 -,003 ,006 ,636
Hours gig-work per week ,000 ,005 ,939 -,006 ,004 ,158
Taxi platform ,043 ,171 ,800 ,276 ,144 ,057
Zero hour contract ,054 ,167 ,749 -,110 ,140 ,434
Set hours contract ,284 ,158 ,075 -,100 ,136 ,464
Location USA ,279 ,190 ,145 -,006 ,162 ,969
Location NL -,118 ,211 ,579 ,699 ,180 ,002
Paid per task -,180 ,153 ,242 -,191 ,129 ,143
Constant i1 3,727 ,341 <,001 i2 4,100 ,377 <,001 R2=,2344 R2=,5143 F=4,4805 p<,001 F=11,0848 p<,001 Career adaptability Career exploration Unstandardized
Boot Effects BootSE BootLLCI BootULCI
Conditional indirect effect at intention to quit for levels of career adaptability and
career exploration -,89 -1,11 ,42 ,04 -,02 ,13 -,89 ,00 ,06 ,04 -,03 ,14 -,89 1,11 ,07 ,06 -,04 ,20 ,19 -1,11 ,03 ,02 -,01 ,08 ,19 ,00 ,04 ,02 -,01 ,08 ,19 1,11 ,05 ,03 -,01 ,11 1,04 -1,11 ,01 ,02 -,02 ,07 1,04 ,00 ,02 ,03 -,03 ,07 1,04 1,11 ,02 ,03 -,04 ,09
Model 29 is limited in describing the conditional effects of a single interaction in the model. Therefore, model 5 was ran, with career adaptability as the only moderator for the direct relationship between job insecurity and turnover intention. This allowed to comment on the conditional effects and probe the interaction. The results are shown in table 4.
The results showed that the relationship between job insecurity and turnover intentions is not significant for low values of career adaptability (effect=0,055. SE=0,088. CI: -0,118 to 0,228). However, this relationship is significant for medium (effect=0,229. SE=0,058. CI: 0,114
35
to 0,345) and high values of career adaptability (effect=0,369. SE=0,077. CI: 0,217 to 0,521). As a result, hypothesis 3 is supported. Figure 2 represents a visualization of the interaction effect.
Table 4. Test interaction effect
Figure 2. Interaction Career Adaptability x Job Insecurity 1 1,5 2 2,5 3 3,5 4 4,5 5
Low Job Insecurity High Job Insecurity
T u rn ove r In te n tion s Low Career Adaptability High Career Adaptability
Job Satisfaction (M) Turnover Intention (Y)
Antecedent Coeff. SE p Coeff. SE p
Job Insecurity a1 -,138 ,066 ,039 c'1 ,198 ,598 ,001
Job Satisfaction - - - b1 -,288 ,705 <,001
Career Adaptability a2 - - - c'2 ,020 ,068 ,772
Job Insecurity x Career
Adaptability a3 - - - c'4 ,163 ,060 ,007
Age ,004 ,007 ,610 -,009 ,006 ,164
Hours gig-work per week ,004 ,005 ,439 -,004 ,005 ,329
Taxi platform ,128 ,175 ,468 ,280 ,154 ,072
Zero hour contract -,072 ,170 ,675 -,168 ,160 ,267
Set hours contract ,301 ,164 ,069 -,171 ,144 ,237
Location USA ,258 ,195 ,188 -,096 ,173 ,580
Location NL -,195 ,213 ,369 ,543 ,190 ,005
Paid per task -,160 ,158 ,314 -,223 ,138 ,110
Constant i1 3,727 ,341 <,001 i2 4,309 ,404 <,001
R2=,2344 R2=,6466
F=4,4805 p<,001 F=9,6385 p<,001
Career adaptability
Unstandardized
Boot Effects BootSE BootLLCI BootULCI
Conditional indirect effect at intention to quit for levels of
career adaptability.
-,877 ,055 ,088 -,118 ,228
,190 ,229 ,058 ,114 ,345
36
Discussion
The purpose of this paper was threefold. The foremost purpose of the study was to extend the knowledge on the gig-economy. The gig-economy is a new phenomenon that is developing every day. Consequently, related literature is also developing. As a result, there are many knowledge gaps that require to be filled. What is currently a hot topic in literature, is how to legislate the gig-economy. Legislators are interested in creating rules and law for the platforms that organize the work in the gig-economy. As a result, this is reflected in the literature, which is dominated by research on legislation for the gig-economy.
The reason why there is need for legislation, is the thought that the platforms do not provide sufficient security for their gig-workers. A substantial amount of business risk, that would originally be for the employer, is transferred to the worker (Aloisi, 2015). The gig-worker does not have the security of a normal employment contract, such as payment during illness, insurance, and stable income. Moreover, their income is sometimes dependent on the supply and demand for the gigs they perform. These are the costs for extreme flexibility for the gig-worker, and the requester of the gig-work (Aloisi, 2015).
In this research, job insecurity was taken as independent variable to explain the effect it has on other parts in the model. Based on the results, it could be said that the thought that platforms do not provide enough security for their gig-workers is justified. The results show a high mean of job insecurity, 3,96 on a 5-point-Likertscale. This means, that on average the gig-worker experiences relatively high values of job insecurity. It could be expected that this is due to the fact that gig-workers are burdened with a substantial amount of business risk, which makes them worry about the continuance of the work.
Hypothesis 1 concerned the direct relationship between job insecurity and turnover intention. It was predicted that there would be a positive relationship between the two variables. This was confirmed by the results. The relationship could be explained by a sense of
