Performing on demand? Digital performance management
systems and turnover intentions the Gig economy
What is the effect of perceived justice of performance management systems on turnover intentions for gig workers active on work-on-demand platforms, and what is the role of organizational commitment and external self-rated employability in this relationship?
Author: Izzy Heemskerk Student number: 10488464 Supervisor: Eloisa Federici Second Reader: Corine T. Boon Date: 21/06/2018
Master Thesis
MSc Business Administration, Leadership & Management Amsterdam Business School, University of Amsterdam
Statement of Originality
This document is written by Student Izzy Heemskerk who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are 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.
Acknowledgements
This thesis reflects a period of 6 months in which I professionally and academically developed, and for this I would like to thank three parties. First I would like to thank my supervisor and the University of Amsterdam in guiding me in writing my thesis with a very new and challenging subject. Mrs. Federici has always provided me with clear feedback, and also encouraged me to be critical towards my reasoning and increase the quality of my writing and research. Next, I would like to thank Qualtrics for providing me, and the other three students, with respondents falling within our scope. Lastly, I would like to thank the other three students of this project in successfully developing a survey together and collecting data.
TABLE OF CONTENT
STATEMENT OF ORIGINALITY _________________________________________________________ 2 ACKNOWLEDGEMENTS ________________________________________________________________ 3 ABSTRACT _____________________________________________________________________________ 5 1. INTRODUCTION ___________________________________________________________________ 6 2. THEORETICAL FRAMEWORK _____________________________________________________ 11 2.1.DEFINITION OF GIG WORK ____________________________________________________________ 11 2.2THE RELATIONSHIP BETWEEN PERCEIVED JUSTICE OF PERFORMANCE MANAGEMENT SYSTEMS ANDTURNOVER INTENTION. __________________________________________________________________ 12 2.3THE MEDIATING ROLE OF ORGANIZATIONAL COMMITMENT____________________________________ 17 2.4MODERATION OF EXTERNAL SELF-RATED EMPLOYABILITY ____________________________________ 20
3. METHOD _________________________________________________________________________ 25 3.1PROCEDURE ________________________________________________________________________ 25 3.2SAMPLE ___________________________________________________________________________ 27 3.5ANALYTICAL STRATEGY ______________________________________________________________ 28 3.4MEASURES ________________________________________________________________________ 29 3.5DATA ANALYSIS ____________________________________________________________________ 33 4. RESULTS _________________________________________________________________________ 37 4.1PRELIMINARY ANALYSIS ______________________________________________________________ 37 4.2HYPOTHESIS TESTING ________________________________________________________________ 38 4.3POST HOC ANALYSES ________________________________________________________________ 43 5. DISCUSSION ______________________________________________________________________ 44 5.1PRACTICAL IMPLICATIONS ____________________________________________________________ 52 5.2LIMITATIONS AND FUTURE RESEARCH ____________________________________________________ 53
6. CONCLUSION _______________________________________________________________________ 56 7. REFERENCES _______________________________________________________________________ 57 8. APPENDIX __________________________________________________________________________ 66
8.1CONTACTED WORK-ON-DEMAND PLATFORMS ______________________________________________ 66 8.2SCALES FOR MEASURING THE CONSTRUCTS ________________________________________________ 67 8.3FACTOR ANALYSIS RESULTS ___________________________________________________________ 69 8.4INTERVIEWS _______________________________________________________________________ 70
8.4.1. Interview Foodora Rider _________________________________________________________ 70 8.4.2 Interview Uber Driver ___________________________________________________________ 81 8.4.3 Interview Temper Worker ________________________________________________________ 101
Abstract
This study examined the effect of perceived justice of performance management systems on turnover intentions for individuals working for working-on-demand platforms in the gig economy. It was investigated whether this relationship was mediated by organizational commitment. Moreover, external self-rated employability was studied as a moderator on the relationship between perceived justice of performance management systems and organizational commitment. Survey was conducted by means of a questionnaire adapted for gig workers working for work-on-demand platforms, and cross-sectional and correlational data was obtained from 181 gig workers working in the Netherlands, USA and UK. It was found that the perceived justice of performance management systems had a relationship with turnover intentions, which was fully mediated by organizational commitment. External self-rated employability did not have an effect on the relationship between perceived justice of performance management systems and organizational commitment. These findings stressed the relevance of taking into account justice perceptions when developing performance management systems in the working-on-demand app or on the website. Further implications and limitations of the current study are discussed as well.
Key words: gig economy, working-on-demand platforms, perceived justice of performance
managements systems, turnover intentions, organizational commitment, self-rated employability.
1. Introduction
A new economy is appearing, the gig economy, and its influence is significant for both consumers and the labour market. The gig economy includes different types of organizations with new forms of employment, offering both employers and employees flexibility (Kalleberg & Dunn, 2016). The number of tasks and new projects offered daily in this new economy has grown with 25% within the past two years (Oxford Internet Institute, 2018). The gig economy can be seen as a digital version of freelance and contingent work arrangements and it is characterized by short-term engagements between workers, customers and employers (Kalleberg & Dunn, 2016). The gig economy includes two forms of work, namely crowd work and work-on-demand via apps. The crowd work platform can be seen as a facilitator and allows demand and supply to meet globally. The work-on-demand apps, on the contrary, link the client and worker online but require a local execution of the work (De Stefano, 2015). Uber, Deliveroo and Helpling are examples of online platforms running work-on-demand apps and offering gigs. Research on gig workers has mainly been focused on the labour conditions and the legal challenges concerning this form of employment (Stewart & Stanford, 2017; De Stefano, 2015; Friedman, 2014). However, since the number of firms offering jobs on online platforms is growing rapidly, it is of interest how these firms can most optimally manage this new form of employment (Oxford Internet Institute, 2018).
This new form of work enables flexibility within a person’s job. They are able to choose their own hours and match their gig work with capabilities and other activities, such as studying or working for another organization (De Stefano, 2015). Even though online platforms offer these advantages, turnover rates are high. More than fifty percent of the participants of labour platforms exit within 12 months and one in every six participants is new in any given month (Farrell & Greig, 2017). Besides this, the percentage of new entrants on labour platforms decreased from 29% in 2014 to only 17% in 2016. This decrease in new entrants and high
turnover rate can indicate a difficulty in the future for acquiring new participants on labour platforms (Farrell & Greig, 2017). Also, this can imply fewer participants in the future, and businesses running work-on-demand apps might not be able to meet the demands of services with their number of workers. Therefore, it was of interest to study the turnover intentions of gig workers using work-on-demand apps.
The first purpose of this study was to investigate the relationship between justice of performance management systems and turnover intentions. Platforms running work-on-demand apps in the gig economy communicate mainly via the app with their employees, and indicators of performance are also received in this app. This is basically a digital form of performance management, also referred to as algorithmic management (Lee, Kusbit, Metsky, & Dabbish, 2015). Studies indicate that the perception of the fairness, mostly referred to as justice, of performance management systems can have an influence on employee performance and turnover intentions (Lee et al., 2015; Lee & Jimenez, 2011; Poon, 2004). The ability of online labour platforms to manage performance by using their app and how this is perceived could, therefore, be of interest for improving the turnover intention of employees. The new form of performance management in the gig economy might alter the feelings of justice towards these systems and have an effect on employee attitudes and behaviour, such as turnover intention. Based on this, the first purpose of the paper was to investigate the relationship between perceived justice of performance management systems and turnover intentions.
Next, it was proposed that perceived justice of performance management systems and its relationship with turnover intention could be explained by organizational commitment. Turnover intention can be a result of different personal and job-related factors, and this has been studied in both permanent organizations and temporary organizations (Nuhn & Wald, 2016; Slattery & Rajan Selvarajan, 2005). Studies have focused on organizational commitment as an antecedent of turnover intention (Lambert, Hogan, & Barton, 2001; Clugston, 2000;
Currivan, 2000; Shore & Martin, 1989; Vandenberghe & Tremblay, 2008; Somers, 1995). Organizational commitment can be of interest for companies running work-on-demand apps since the competition in this industry is rising and the percentage of new entrants decreasing (Farrell & Greig, 2017). Meaning that, for work-on-demand platforms, keeping gig workers attached to the organization is important. Research shows that organizational commitment can diminish turnover intentions (Lambert & Hogan, 2009). Moreover, the study of Joo and Park (2010) found that organizational commitment mediates the relationship between developmental feedback and turnover intention, supporting the role of organizational commitment as a mediator in the relationship between perceived justice of performance managements systems and turnover intentions.
The relationship between perceived justice of performance management systems and organizational commitment might be influenced by external factors. The competition for platforms offering work-on-demand apps is increasing, due to the number of organizations entering this industry (Oxford Internet Institute, 2018). This offers possibilities for gig workers to join other platforms, that might offer better working conditions. The perception of an individual’s possibilities to replicate the same or a better form of employment elsewhere is referred to as external self-rated employability (De Cuyper & De Witte, 2011). Having the perception that better employment elsewhere is an option, could diminish the organizational commitment (De Cuyper, Notelaers, & De Witte, 2009; De Cuyper & De Witte, 2011).
Previous research indicated that workers in the gig economy often accept doing crowd work, only because they cannot find a regular or permanent work arrangement (Huws, Spencer, Syrdal, & Holts, 2017). This stresses the relevance of examining external self-rated employability for gig workers. Moreover, it is indicated that external self-rated employability reduces the insecurity one might experience within a job (De Cuyper, Mäkikangas, Kinnunen, Mauno, & Witte, 2012). These insecurities within a gig worker’s job can result from being
treated unfair by the organization. Combining these insights, it can be implied that external employability can help cope with job insecurities resulting from a lack of justice in performance management systems by providing job security. This coping mechanism will weaken the effect of injustice on organizational commitment. Moreover, it was also indicated that feeling secure within one’s job, as a result of external employability, reduces the importance of justice perceptions (Loi, Lam, & Chan, 2012; Van den Bos & Lind, 2002). Based on the above mentioned studies it was expected that the effect of perceived justice of performance management systems on organizational commitment would be weaker when external employability is high. Perceived external employability of an employee could, in these ways, alter the relationship between perceived justice of performance management systems and organizational commitment for gig workers.
Taken this all together, previous research has focused on the relationship between perceived justice of performance management and organizational commitment on turnover intentions in traditional working arrangements. These relationships, however, might be altered regarding work arrangements in the gig economy, and more specifically, for companies running work-on-demand apps. The new form of performance management might be perceived differently by the workers and this can have an effect on both behaviours and attitudes of gig workers. This study, therefore, examined the relationship of perceived justice of performance management system with turnover intentions, and included the role of organizational commitment and employability in this relationship, for companies running work-on-demand apps. The proposed conceptual model is depicted in Figure 1.
2. Theoretical Framework 2.1. Definition of Gig work
The gig economy holds different kind of work arrangements. The gig economy is argued to be timeless, meaning that workers are hired on the spot regardless of their past employment and with no specific promises for future opportunities, legacy pay or deferred compensation (Friedman, 2014). The two most common work arrangements within the gig economy are ‘crowd work’ and ‘work-on-demand apps/platforms’ (De Stefano, 2015). Crowd work is characterized by work executed through online platforms. These platforms put in contact an indefinite number of businesses, organizations and individuals, and allow clients and workers to connect globally. Work-on-demand apps instead, involve both more traditional working activities, such as transport or cleaning, and clerical work. Work-on-demand apps offer and assign work via the app, and the company running this app often intervenes by setting a minimum quality standard and selecting and managing the workforce.
These forms of work both use online activities to match demand and supply, and are able to do so at a high velocity. Even though they may appear similar, a distinction can be made. Crowd work platforms may not include a relationship between the client and the worker. The platform pays the worker and delivers the result to the client. The crowd work platform can, therefore, be seen as a facilitator and allows demand and supply to meet globally. On the contrary, the work-on-demand platforms link the client and worker online, mostly via an app but also via a website, but require a local execution of the work (De Stefano, 2015). Moreover, crowd work platforms only match the worker with the client. The relationship between the organization and worker could, therefore, be regarded as limited. Companies running a work-on-demand app, on the other hand, hold a certain control over the workforce, and over the prices or standards of the delivered product or service (Stewart & Stanford, 2017). Based on this difference, this study focused on work-on-demand platforms exclusively. The closer interaction
opposed to crowd work companies, was necessary to study the perceived justice of performance management systems, organizational commitment and turnover intentions.
2.2 The relationship between perceived justice of performance management systems and turnover intention.
As outlined before, turnover is increasing in the gig economy. Literature shows that the turnover for online labour platforms is high and participation is decreasing (Farrell & Greig, 2017). The loss of productivity and the likely loss of business opportunities are examples of consequences of high turnover for companies (Chew & Chan, 2008). Regarding employee attitudes and behaviours within an organization, turnover can be measured by the intention of an employee to leave the organization. This is also referred to as turnover intention, or intention to leave, and studies show that this is significantly related to actual turnover in organizations (Sager, Griffeth, & Hom, 1998; Hom, Caranikas-Walker, Prussia, & Griffeth, 1992; Mobley, Griffeth, Hand, & Meglino, 1979). Turnover intention can be defined as the cognitive process of planning on leaving an organization, thinking about quitting or the desire to leave the organization (Lambert & Hogan, 2009).
Antecedents of the intention of employees to quit the organization are examined broadly for traditional organizations and to some extent for temporary work arrangements (Clugston, 2000; Slattery & Rajan Selvarajan, 2005; Nuhn & Wald, 2016; Lambert et al., 2001). However, so far research has neglected to look at turnover intentions in the gig economy, and specifically for companies running work-on-demand apps. Hence, the first purpose of this paper was to identify mechanisms that are related to turnover intention for organizations running work-on-demand apps in the gig economy. More specifically, performance management was examined and proposed as an antecedent of turnover intentions (Huselid, 1995).
Performance management is stressed as an important policy within the HR domain in order to motivate employees and, in turn, enhance employee performance (Jiang, et al., 2012).
Performance management can be defined as all the policies, practices and design features that interact within an organization to enhance employee or group performance. Performance goals, assessing performance and feedback systems are all activities that can be included in a performance management system (Gruman & Saks, 2011; Molleman & Timmerman, 2003). In most systems, the manager controls performance by providing inputs, such as training, and gives feedback based on outputs, such as performance assessments (Molleman & Timmerman, 2003). Performance management systems encourage managers to work together with employees in order to review and measure results, set expectations and reward performance. This is aimed to improve not only employee performance, but also to have a positive effect on organizational performance (Gruman & Saks, 2011; Den Hartog, Boselie, & Paauwe, 2004; Molleman & Timmerman, 2003; DeNisi & Pritchard, 2006).
Performance management systems in organizations running a working-on-demand app have a different approach than other organizations. As mentioned before, the digital form of performance management in organizations running work-on-demand apps is referred to as ‘algorithmic management’. This refers to the use of software that executes management functions, such as optimizing, allocating and evaluating work (Lee et al., 2015). Lee and colleagues (2015) investigated the algorithmic management practices of a work-on-demand platform, solely for taxi services, and focused specifically on the fairness of the performance evaluation system provided by the app. They indicated that the system is sometimes seen as ineffective and perceived as unfair, resulting in negative psychological feelings for workers. This study had a very small sample (i.e. N = 21) and focused solely on two organizations providing a ride-sharing service (Lee et al., 2015). Therefore, addressing the perceived justice of performance management systems further and for other work-on-demand apps in the gig economy was of interest. As the study by Lee (2015) outlines, perceptions of the performer are important to take into account when developing performance management systems, and
fairness can have an influence on the effectiveness of these systems (Buchner, 2007; DeNisi & Pritchard, 2006). Moreover, it is argued that reactions of employees on performance appraisal systems have an influence on their development of a favourable job, organizational attitudes and their motivation to increase performance (Jawahar, 2007).
In literature, fairness of performance management systems is often examined by looking at constructs of organizational justice (Gabris & Ihrke, 2000; Jawahar, 2007; Farndale, Hope-Hailey, & Kelliher, 2011; Shaw et al., 1998). For this study, organizational justice was examined as well in order to capture the perceptions of employees regarding performance management systems. The three constructs of organizational justice are procedural justice, distributive justice and interactional justice. Procedural justice refers to the fairness of the procedures the employees are subject to. Distributive justice refers to the perceived fairness of outcomes, such as rewards and costs. Lastly, interactional justice refers to the perceived fairness of the interpersonal communication and treatment an employee receives (Jawahar, 2007; Farndale et al., 2011).
The study by Jawahar (2007) combined literature regarding justice perceptions and reactions to performance appraisal systems, and, hereby, provided support to examine organizational justice to obtain insight in the fairness of performance management systems. The author argued that the different forms of justice could be studied when examining the justice of a performance appraisal system. The author stressed that performance ratings can be regarded as outcomes and, therefore, distributive justice can indicate the perceived fairness of these outcomes. Moreover, the appraisal system can be argued to be the procedure of distributing the outcomes. Procedural justice can, therefore, give insight into the fairness of the appraisal system (Jawahar, 2007). Jawahar (2007) studied whether satisfaction with a performance management system is influenced by the above-mentioned forms of justice. The study reflected that perceived justice of outcomes (i.e. distributive justice) positively influenced satisfaction with
ratings and procedural justice positively influenced satisfaction with appraisal systems. The study failed to support the prediction that interactional justice influenced satisfaction with the rater. These outcomes stress the importance of perceived justice when studying reactions to performance management systems, and how this can be studied by measuring procedural and distributive justice (Jawahar, 2007).
An example of a study that examined only one construct of organizational justice, in order to measure performance management practices, is the study by Shaw (1998). This study examined the effect of performance management systems on turnover. It measured performance appraisals, procedural justice and electronic monitoring in order to identify performance management practices.This study stresses the relevance of procedural justice in order to study performance management systems in relation to turnover (Shaw et al., 1998). However, only the electric monitoring had a significant effect on quit rates, in that more electronic monitoring resulted in higher quit rates (Shaw et al., 1998). This is still of interest since work-on-demand platforms use their app to obtain information on employee performance. This can be perceived as a form of electric monitoring as well.
A more recent study stressed the importance of measuring two constructs organizational justice in order to study how performance management systems are perceived by employees (Farndale et al., 2011). Distributive justice is argued to be of interest in studying whether an employee perceives a given evaluation as fair. Procedural justice is, as an example, explained as the fairness of the process towards a final grading. It is stated that distributive justice is more related to personal outcomes of a performance management system, and procedural justice more closely related to the evaluation of the organization by the employee (Farndale et al., 2011). Again, this shows that these two forms of justice can be relevant when studying the perceived justice of performance management systems, including both the outcomes as well as the process. Based on the above mentioned studies, it was of interest to measure both procedural
justice and distributive justice when studying the perceived justice of performance management systems of companies running working-on-demand apps. Interactional justice will not be taken into account, as most literature lacks to identify the relationship between interactional justice and performance management systems (Jawahar, 2007; Farndale et al., 2011). Moreover, for organizations running working-on-demand apps, this form of justice appears to be less relevant based on the limited interaction between supervisors and employees.
The perceived justice of performance management systems tends to influence the attitudes and behaviours of employees, as briefly mentioned above (Lee et al., 2015; Farndale et al., 2011; Jawahar, 2007). A study researching the effect of performance-based management practices showed that these practices have a negative relationship with turnover intention in federal agencies (Lee & Jimenez, 2011). Performance-based management practices consisted here of objective performance measurements, performance-based pay systems and supervision supporting performance. The effect of perceived performance appraisal politics on turnover intention was also indicated by Poon (2004). Perceived performance appraisal politics refers here to the impact of employees’ perceptions of performance appraisal systems on their attitudes and behaviours. This study indicated that perceiving unfair performance appraisals, such as using ratings for punishment, was positively related to turnover intentions (Poon, 2004). It was concluded that unfair performance management systems can increase turnover intentions.
Alongside the studies indicating the effect of justice of performance management systems on turnover intention, it is also indicated that procedural and distributive justice, examined separately, have an influence on turnover intentions (Nadiri & Tanova, 2010; Alexander & Ruderman, 1987). Literature shows that both distributive and procedural justice have a negative relationship with turnover intentions (Nadiri & Tanova, 2010). The study by Nadiri (2010) tested the direct relationship between distributive and procedural justice with
turnover intention, and also indicated that distributive justice explains the variation in turnover intention more than procedural justice. Distributive justice was focused on outcomes such as pay and rewards, and procedural justice was measured by examining the decision-making of the supervisor (Nadiri & Tanova, 2010). The decision-making of supervisors and measuring rewards is often part of performance management practices, indicating the relevance of the authors’ findings that both distributive and procedural justice have an effect on turnover intentions (Gruman & Saks, 2011; Nadiri & Tanova, 2010). This study, therefore, supported the expectation that procedural and distributive justice would influence turnover intention.
The first hypothesis was based on the literature regarding the importance of studying justice perceptions of performance management systems, especially distributive and procedural justice, as well as the effect of these perceptions on turnover intentions. It was expected that perceived justice of performance management systems has a negative relationship with the turnover intentions of gig workers.
H1: Perceived justice of performance management systems, consisting of distributive and procedural justice, is expected to have a negative relationship with turnover intention for gig workers, meaning that a higher perceived justice of performance management systems will decrease turnover intention.
2.3 The mediating role of organizational commitment
The relationship between perceived justice of performance management systems and turnover intention could be explained by looking at a less distal outcome of perceived justice. A more proximal outcome of organizational justice is organizational commitment (Farndale et al., 2011; Folger & Konovsky, 1989; Masterson, Lewis, Goldman, & Taylor, 2000). A definition of organizational commitment is given by Porter (1974), stating that organizational
commitment is the strength of an individual’s identification with and involvement in a particular organization. Three factors are identified to characterize organizational commitment. First, there will be a strong belief and acceptance of the values and goals of the organization. Second, a willingness to make a considerable effort on behalf of the organization is shown. Lastly, there is a strong desire to stay a member of the organization (Porter, Steers, Mowday, & Boulian, 1974).
Literature on the relationship between performance management, perceptions of justice and the effect on employee commitment can give insight into the effect of justice of performance management systems on organizational commitment. It is supported that relationship between three dimensions of performance management, in this study referred to as high commitment performance management (HCPM) practices, with employee commitment is fully mediated by organizational justice (Farndale et al., 2011). The mediation of organizational justice in the relationship between performance management practices and organizational commitment, implies that in order for the performance management practices to have an effect on organizational commitment, they need to be perceived as fair. This argument is enhanced by the lack of support for a direct effect of the performance management practices on employee commitment. Besides the mediation of organizational justice, this study also indicated a direct effect of distributive justice of appraisals on employee commitment (Farndale et al., 2011). The direct effect of distributive justice, as well procedural justice, on organizational commitment is also supported by Folger (1989). Folger (1989) explicitly states that procedural justice should be taken into account when aiming to be maximally effective in establishing and maintaining employee commitment towards an organization, as indicated also in other studies (Folger & Konovsky, 1989; Masterson et al., 2000). Both distributive and procedural justice had a significant effect on organizational commitment and were related to performance management practices. Procedural justice was focused on the fairness of the performance appraisals that
employees received, and also on the extent to which performance measurements were present in the process. Distributive justice was focused on the fair distribution of incentives, and whether these incentives were fair given their performance (Folger & Konovsky, 1989). Taken together, this indicated that justice of performance management systems, measured by both procedural and distributive justice, can have a positive effect on organizational commitment.
In the literature, organizational commitment is often studied as a mediator with turnover intention as an outcome. For example, the study of Joo and Park (2010) found that organizational commitment explains the relationship between developmental feedback and turnover intention. This relationship between organizational commitment and turnover intentions has already been studied early on (Cohen, 1993; Tett & Meyer, 1993). Previous research suggests that all components of organizational commitment, as identified by Meyer and Allen (1993), are negatively related to both turnover and turnover intention (Meyer, Stanley, Herscovitch, & Topolnytsky, 2002; Allen & Meyer, 1993). Other studies that did not hold a focus on different components, also indicated that organizational commitment has a direct negative effect on turnover intention (Lambert & Hogan, 2009). This negative effect of organizational commitment on turnover intentions has not yet been supported for working-on-demand platforms and was of interest to investigate.
In conclusion, it was expected that the relationship between perceived justice of performance management systems and turnover intention can be explained by organizational commitment. Meaning, that organizational commitment mediates this relationship. Also, it was expected that there is a positive relationship between perceived justice of performance management systems and organizational commitment and a negative relationship between organizational commitment and turnover intentions.
H2: The negative relationship between perceived justice of performance management systems and turnover intention is mediated by organizational commitment.
H2a. There is a positive relationship between perceived justice of performance management systems and organizational commitment.
H2b. There is a negative relationship between organizational commitment and turnover intentions.
2.4 Moderation of external self-rated employability
To better understand the effect of perceived justice on turnover intention, explained by organizational commitment, it is important to take into account the conditions under which this occurs. The current study focused on perceived employability as a possible moderator on the relationship between perceived justice of performance management systems and organizational commitment. Employability is the ability to keep a job or to get the job that is desired (De Cuyper & De Witte, 2011; Rothwell & Arnold, 2007). Self-rated employability (SRE) holds a focus on the employee perspective and is defined as the belief how easily an individual can find new employment, or the perception of an individual about the possibilities of new employment (De Cuyper & De Witte, 2011; Rothwell & Arnold, 2007).
A distinction can be made between external and internal SRE. Internal SRE is about the employability within an organization. It refers to investments, such as training, that are present within one’s job and how this makes an employee more employable within the internal labour market (De Cuyper & De Witte, 2011). External SRE, on the other hand, refers to the extent an employee believes that the current work arrangement can be easily obtained elsewhere in the labour market (De Cuyper & De Witte, 2011). Referring to external SRE, an employee could not only perceive their current work arrangement easy obtainable elsewhere but also work arrangements with better deals and conditions. Qualitative external SRE refers to this possibility
of better external job opportunities, and could possibly result in withdrawal and dissatisfaction when it is highly perceived (De Cuyper & De Witte, 2011). Moreover, external qualitative SRE is negatively related organizational commitment. Meaning that perceiving better work arrangements elsewhere would diminish the commitment towards an organization (De Cuyper & De Witte, 2011). This stressed the importance of external SRE as a condition for the relationship between perceived justice of performance management systems and organizational commitment. Moreover, it was of relevance for the gig economy since competition for work-on-demand platforms is increasing, resulting in more platforms available that might offer better work conditions (Oxford Internet Institute, 2018).
The conditional effect of external SRE on the relationship between perceived justice of performance management systems and organizational commitment was explained by looking into the role of job uncertainty, often also referred to as job insecurity. Job insecurity can be defined as the employee’s perception or concern about involuntary job loss (De Cuyper et al., 2012). When employees feel insecure in their current job, external SRE might secure them by perceiving employment across organizations (De Cuyper et al., 2012). This can be explained as follows. The perception of being employable can result from investments in skills and competencies, throughout a career. These investments will not only result in a higher employability in the external labour market, it will also increase security within the current job (De Cuyper et al., 2012). In addition to this, the link between employability and job insecurity could also be explained by the underlying feeling of control. Employees that are highly employable in the external labour market have a certain control over their career. This feeling of control will be reflected in their current job, in that they will be able to protect their job, and this, in turn, will result in less job insecurity. The decrease in job insecurity when external SRE is high, has been supported in the literature and could indicate that external SRE can help lowering uncertainty within one’s job. Moreover, studies indicate that employability can help
mitigate the negative effects of job insecurity. Meaning, when employability is high, job insecurity, possibly resulting from other factors, will have reduced negative consequences (Silla, De Cuyper, Gracia, Peiró, & De Witte, 2009; Kalyal, Berntson, Baraldi, Näswall, & Sverke, 2010)
One of the factors contributing to job insecurity, is procedural justice. A study by Loi (2012) states that job insecurity consists of unpredictability and uncontrollability. The latter is argued to result from having no possibility for giving input in a decision-making process and perceiving that the organization holds no norm on fairness (Loi et al., 2012). It is stated that organizations lacking this norm on fairness will increase uncontrollability which, in turn, contributes to feelings of uncertainty. This norm on fairness can be established by procedural justice (Loi et al., 2012). Meaning that when there is a lack of procedural justice this will not be established and job insecurity will increase by means of uncontrollability. This supports a negative relationship between procedural justice and job insecurity (Loi et al., 2012). External SRE can help cope with the negative consequences of this job insecurity resulting from injustice and, in this way, diminish the effect of injustice on organizational commitment (De Cuyper et al., 2012). Therefore, based on the job insecurity underlying both external SRE and justice perceptions, it was argued that the relationship between justice of performance management systems and organizational commitment is moderated by external SRE. It is moderated such that perceived justice of performance management systems has a weaker effect on organizational commitment when external SRE is high.
The uncertainty management theory, as proposed by Loi (2012) also provided insight in the expected moderating effect of employability on the relationship between perceived justice of performance management systems and organizational commitment. This theory states that justice is closely related to uncertainty, and that employees tend to rely more on justice information when they are confronted with uncertainty (Loi et al., 2012; Van den Bos & Lind,
2002). In order to cope and reduce uncertainty, employees will seek out environmental cues, and justice information is one of these cues. Justice information can, when feeling uncertain, reduce concerns about being exploited or excluded by the organization. Fair treatment is argued to make the future more predictable and controllable, and thus reduce uncertainty (Loi et al., 2012). Other studies also indicate that justice perceptions are of great importance when employees try to manage their uncertainty (Lind & Van den Bos, 2002; Wang, Lu, & Siu, 2015). Moreover, they indicate that when employees feel uncertain, the effect of procedural justice on work outcomes is magnified (Brockner, 2011; Lind & Van den Bos, 2002). This also implies that gig workers having a high SRE, will feel more certain and tend to rely less on justice perceptions. This decrease of sensitivity for justice can make the relationship between perceived justice of performance management systems and organizational commitment weaker.
Taken together, external SRE was expected to moderate the effect of perceived justice of performance management systems on organizational commitment. Uncertainty management and the negative relationship between external SRE and job insecurity supported this expectation. First, external SRE provides job security by perceptions of job opportunities elsewhere and can, in this way, help cope with job uncertainties resulting from injustice of performance management systems. Next, a higher external SRE can reduce the need to look at justice perceptions to help diminish one’s job uncertainty. In this way, it was expected that experiencing a high external SRE can mitigate the effects of perceived justice of performance management systems and decrease the relevance of these perceptions. Therefore, hypothesis 3 expected external SRE to weaken the relationship between perceived justice of performance management systems and organizational commitment:
H3: External qualitative self-rated employability moderates the positive relationship between perceived justice of performance management systems and organizational commitment for gig workers, such that this relationship is weaker when external qualitative self-rated employability is higher.
3. Method 3.1 Procedure
This was a quantitative research and was done by means of a cross-sectional and correlational survey design. The research was explorative in that it attempted to explain the hypotheses derived, in a deductive way, from the literature. The cross-sectional survey design was chosen since it provides the possibility to obtain quantitative values that are linked to the variables. Moreover, this cross-sectional design was chosen due to the limited time to obtain data. There was no possibility to obtain data of multiple points in time.
The survey consisted of 49 questions, including questions regarding demographics and questions measuring constructs of this study and four other studies. It was constructed by using Qualtrics research software. The survey took approximately 14 minutes to fill in. In order to participate, respondents were required to work for an on-demand-platform in the UK, USA or Netherlands and fully understand either the Dutch or English language. Translation of original items was done by means of back-translation, in order to create reliable translations. Anonymity was guaranteed by using the anonymous link to the survey, as provided by Qualtrics, as well as outlining the anonymity when participating in the introduction of the survey. For motivating individuals to participate and complete the survey, they had a chance to win a voucher of 50,- for a Dutch online-shop.
The study used non-probability sampling, by means of convenience sampling, due to the lack of a sampling frame. The sampling frame could not be precisely identified because of the large and diverse population. In the first two weeks, the questionnaire was spread under gig workers by directly approaching them, either online or in person. Also, 18 work-on-demand companies were contacted and asked to spread the survey among employees. A list of the contacted platforms can be found in the appendix. Foodora advised how and where to reach as many employees as possible, but did not spread the survey internally. Temper was the only
company to spread the survey internally. All other platforms did not spread the survey internally or provided help in any other way. The volunteer sampling technique, by means of the ‘snowball’ effect, was also applied. Respondents were asked, at the end of the survey, whether they would be willing to help spread the survey. If they were willing to do so, they received an email with further instructions. The sample was expected to be diverse, based on these two sampling methods. However, important to note is that the direct, in person, approached gig workers are mainly working for the same company.
The above outlined sampling techniques gave limited responses, and sampling took longer than expected. Therefore, it was decided to purchase respondents by using a panel service of Qualtrics. This company approached self-employed individuals in the Netherlands, UK and USA and provided 120 respondents falling within the research scope. For this approach, 3 quality checks were added to the survey, as well as the question filtering out all participants outside of the research scope. In this way, the quality of respondents was checked for and in line with the other sampling techniques used.
Regarding the sampling in the three different countries, it was of interest to look at some differences in the gig economy between these countries. A recent study released by the Bureau of Labor Statistics, indicated that 3.8% of the US workforce hold contingent jobs, equal to 5.9 million persons (Contingent and Alternative Employment Arrangements Summary, 2018). More specifically, one study revealed that 1.5% of the adults in the USA received income from a labour platform between 2012 and 2016 (Farrell & Greig, 2017). Looking into these numbers more specifically, indicated that 49% of the participants working for labour platforms (i.e. work-on-demand platforms) are unemployed. Meaning that they solely work for the platform. Regarding platform-work in Europe, in the UK and the Netherlands, it was found that between 9% of the adults has generated some income via an online platform (Huws et al., 2017). Moreover, they indicated that for the Netherlands 1.6% and for the UK 2.7% of the crowd
workers earned more than half of their income via an online platform. In this study, crowdworkers also included individuals working for work-on-demand platforms. Legalization and employee benefits differ per platform and country, and are mostly not received when working for an on-demand-platform. The implications of legalization for gig workers is still an ongoing discussion (Croft, 2018; Nog een rechtszaak tegen maaltijdbezorger Deliveroo, 2018). In the UK and Netherlands 7% and 6%, respectively, indicated to receive some benefits or pensions via crowdwork (Huws et al., 2017).
3.2 Sample
The sample consisted of 280 respondents. Qualtrics provided 195 respondents, and 85 respondents were obtained by directly and indirectly approaching gig workers. However, respondents were filtered out based on the research scope and the requirements for a sufficient quality of responses. All respondents not working for a work-on-demand platform, as asked in the first question, were excluded (36 respondents). Careless responding was corrected for, by the deletion of participants taking less than 5 minutes to fill in the survey (1 respondents), as well as participants filling in 44% or less of the survey (57 respondents). Lastly, 5 participants working for crowdsourcing platforms (i.e., Rev.com, Textbroker, Appen, Bookscouter and Twitch) were excluded, based on the research scope of this study. Regarding the quality checks included for respondents provided by Qualtrics, all respondents met the criteria. The deletion of respondents, based on the above given requirements, resulted in a sample of 181 gig workers.
The average age of the respondents in this survey was 32.56 years old, with a minimum of 16 years old and a maximum of 65 years old (SD = 10.41). In this sample, 36.5% was female and 63.5% was male. Out of the 181 respondents, 84 were American, 36 were from the United Kingdom of Great Britain and Northern Ireland and 37 were Dutch. The remaining 24 respondents had 12 different nationalities. The average time a gig worker was working for the platform was 3.15 years, with a range of 0 to 32 years (SD = 4.80). Gig workers spend on
average 24.02 hours per week working for the platform, ranging from 0.10 to 65 hours per week (SD = 13.69). In the sample, 44.2% worked most of the time for Uber, 24.9% for Foodora, 13.3% for Deliveroo, 8.3% for UberEats, 2.8% for Temper, 1.1% for Helpling and 5.5% for another platform. Regarding the contract form, 29.5% had a zero-hour contract, 34.3% had a contract for a specified amount of hours and 36.5% was self-employed. It was indicated that 47% of the sample is performing gig work in the USA, 35.9% in the Netherlands and 17.1% in the United Kingdom of Great Britain and Northern Ireland. Interesting is, that more than half of the respondents are payed based on a fulfilled task, namely 53.9%. Others are payed based on either hours worked, 42.2%, or based on an agreed upon amount, 3.9%. Lastly, 69.4% of the gig workers in the sample indicated that working for a work-on-demand platform was their primary source of income. For 30.6% it is a secondary source of income.
3.5 Analytical strategy
For this study, the two dataset were exported from Qualtrics and merged in IBM’s Statistical Package for Social Sciences (SPSS) 25.0. This program was used to analyse the data. All missing values were coded 999 and items negatively formulated were recoded. Internal consistency was checked by running reliability analysis, results are given in table 1a and 1b and discussed below. A Cronbach’s alpha level higher than .70 indicates a sufficient reliability, and Corrected Item-Total correlation should be above .30 for all items (Nunnally, 1978). Before testing the hypotheses, an one-way ANOVA was done in order to establish whether any differences occurred between the type of performance management systems. Post Hoc analyses were also done in order to determine whether there was a difference between the forms of justice that were measured.
For testing the hypotheses, PROCESS, an external macro, was used to test the mediation model, as well as the moderated mediation model. To test hypothesis 1, 2a, 2b and 2, model 4 was applied in order to examine the mediation and direct effect. After this, model 7 was applied
to test hypothesis 3 by means of the moderated mediation model. In both models, X was
Perceived justice of performance management systems (PMS), Y was Turnover intention and
M was the mediator Organizational commitment. In addition, for model 7, W was the moderator
Self-rated external employability. By using PROCESS, bootstrapping was done in order to
correct for possible non-normality in the mediation analysis and to estimate properties of the sampling distribution in a non-parametric way (Hayes, 2013). Also, PROCESS gives by default bias-corrected confidence intervals based on 5,000 bootstrap samples, and mean centres all numerical variables (Hayes, 2013). In this way, the assumptions were checked and corrected for in the analyses. Finally, important to note is that all analyses were done with listwise deletion, in order to filter out all respondents that are not subject to any performance management systems and, therefore, did not fill in questions regarding perceived justice.
3.4 Measures
The survey included all before-mentioned variables of this study, as well as demographic and control variables. Respondents were asked to fill in their age, gender, and nationality. Next, they were asked to fill in control variables, including ‘location where work is executed’, ‘company working for’, ‘company working most for’, ‘form of employment’ (self-employed or not), ‘type of contract’, ‘form of payment’, ‘short or long-term intention’, ‘source of income’, ‘hours spend working per week’ and ‘hours spent as gig worker per week’. Since many variables were included, the short version for most scales was preferred over longer ones. All original items that were not provided in Dutch, were translated by means of back translation. All scales, after adaption, can be found in the appendix.
The dependent variable, turnover intention, was measured using the five items of the 5-point Likert scale by Bozeman and Perrewé (2001). An example of an item is ‘I do not intend to quit my job’. The Likert scale ranged from ‘strongly disagree’ (1) to ‘strongly agree’ (5). All reversed items (3) for this scale were recoded. The Cronbach’s alpha of this scale was 0.75.
To measure Perceived justice of PMS, both procedural and distributive justice were measured. Before doing so, participants were asked whether they received any performance evaluations. Participants were given three options, namely ‘no’, ‘Yes, I receive performance ratings from clients in the app or website’ and ‘Yes, my performance is evaluated, but not through clients’ ratings’. In this way, participants that did not receive any performance evaluations could neglect the questions regarding perceived justice. Moreover, this question provides the opportunity to compare the perceived justice of ratings received from clients in the app/website with other performance evaluations.
An adapted version of the validated scale by Tang (1996) was used to measure procedural justice. This resulted in four items of the factor ‘Fairness’ (E.g. How fair do you feel your last performance rating was?) being used, indicated to measure procedural justice (Tang & Sarsfield-Baldwin, 1996). Based on the expectation that most gig workers receive ratings from their clients, the items referring to ‘supervisor’ were adapted to refer to ‘client’ (If you have been evaluating your own performance, how similar would your rating have been to the last one that your client gave to you?). Moreover, two items were deleted, since they were not applicable in the context of gig work. The first item was deleted based on the expectation that gig workers do not receive performance appraisals, but mostly ratings in the app (How fair do you feel your last performance appraisal was?). A second items was deleted based on the assumption that gig workers do not receive any ratings from supervisors (How justified do you feel your supervisor was in his/her last rating of your performance?). The Likert scale ranged from ‘very unfairly’ (1) to ‘very fairly’ (5) for the first two items and ranged from ‘to a small extent’ (1) to ‘to a large extent’ (5) for the last two items. The Cronbach’s alpha of this scale was 0.85.
The five-item 5-point Likert scale by Colquitt (2001) was adapted and three items were used to measure distributive justice (E.g. To what extent is your rating justified, given your
performance?). The adaptation included choosing ‘rating’ as the outcome being measured, as well as deleting one item that was not applicable in the gig work context. This item was deleted based on the assumption that the distal relationship between a gig worker and the organization results in the absence of a direct contribution of a rating to the organization (To what extent does your rating reflect what you have contributed to the organization?). The Likert scale ranged from to a small extent (1) to a large extent (5). The Cronbach’s alpha of this scale was 0.82.
In order to capture the total perceived justice of PMS, the mean of the average score on the two scales of justice was computed. This scale had a Cronbach’s alpha of 0.91. A principal component factor analysis was done for the seven items of the procedural and distributive justice in order to determine whether different constructs were measured. Inspection of the correlation matrix revealed many values exceeding .30. The Kaiser-Meyer-Olkin value was .91, exceeding the recommended value of .60. Bartlett’s Test of Sphericity was significant (p <.001), supporting the factorability of the correlation matrix. The analysis revealed presence of only one factor with an eigenvalue higher than 1, explaining 60.31% of the variance. Based on these results and the clear break after the first component, it was supported that the items load on only one factor. Meaning, that even though the scales were aimed to capture different forms of justice, this cannot be supported by factor analysis. An overview of the results of the factor analysis can be found in the appendix. Based on these above mentioned results, hypotheses were analysed using the scale of the total perceived justice of PMS.
Organizational commitment was measured by an adapted short 9-item version of the
Organizational Commitment Questionnaire (E.g. I am proud to tell others that I am part of this organization) (Porter, Steers, Mowday, & Boulian, 1974; Whitener, 2001). This adapted short version was created by using only the positive phrased items from the original scale (Mowday, Steers, & Porter, 1979). The scale was also adjusted to the gig work context, by replacing
‘organization’ for ‘platform’ for all items. Three items were also deleted to make the scale more suitable for the context. One item was deleted based on the fact that most gig workers are not offered a great variety of tasks when working for a work-on-demand platform (I would accept almost any type of job assignment in order to keep working for this organization) and two items were deleted based on ambiguity and lack of alignment with the context of gig work (‘This organization really inspires the very best in me in the way of job performance.’ and ‘I am extremely glad that I chose this organization to work for over others I was considering at the time I joined.’). The Likert scale for the items chosen ranged from ‘strongly disagree’ (1) to ‘strongly agree’ (5) and the Cronbach’s alpha of this scale was 0.91.
Lastly, self-rated external employability was measured by the five-item 5-point Likert scale of Berntson (2007) (E.g. My experience is in demand on the labour market.). The Likert scale ranged from ‘do not agree at all’ (1) to ‘agree entirely’ (5). The Cronbach’s alpha of this scale was 0.84.
In order to determine which control variables were of relevance in this study, both the literature and correlation matrix, given in table 1a and 1b, were studied. Literature examining organizational commitment indicated that tenure and gender can have an effect on both organizational commitment (Wright & Bonett, 2002) and justice perceptions (Lee, Pillutla, & Law, 2000; Meyer et al., 2002; Sweeney & McFarlin, 1997). Also, the relevance of taking into account demographics when studying organizational commitment and justice perceptions was stressed in several studies (Loi, Hang-Yue, & Foley, 2006; Mathieu & Zajac, 1990; Welbourne, Balkin, & Gomez-Mejia, 1995). The importance of demographics, including age, gender and tenure, was also indicated in studies examining turnover intentions (Russ & McNeilly, 1995) and self-rated employability (De Cuyper & De Witte, 2011). Based on these findings, age, gender and tenure were included in the survey. Part time or full time work also had an effect on both organizational commitment (Lee & Johnson, 1991) and turnover intentions (Peters,
Jackofsky, & Salter, 1981; Martin & Hafer, 1995). Based on this, ‘hours spent as gig worker’ was asked for. The other control variables included in the survey were specified for the context of the gig economy, including; payment form, contract form, short/long-term intention, platform working most for, location work executed and performance management system. It was expected that these characteristics of work-on-demand platforms might differ significantly and were, therefore, of importance in the current model.
In order to determine the control variables relevant for the tested hypothesis, the correlation matrix was studied. Control variables were chosen based on their significant correlation with variables as well as the above mentioned relevance. This resulted in including the control variables ‘Hours spent working for a platform’, ‘Short/long-term intention’, ‘working mostly for Uber’, ‘Payment form’, ‘Location working; USA’ and ‘Performance management system’.
3.5 Data analysis
All means, standard deviations, correlations and Cronbach’s alphas are given in table 1a and 1b. There was a significant negative correlation between Perceived justice of PMS and
turnover intention (r = -.44, p <.01). This negative correlation with turnover intention was found
for both procedural (r = -.35, p <.01) and distributive justice (r = -.42, p <.01). Important to note is, however, that these correlations were moderate to weak. There was a positive correlation between Perceived justice of PMS and Organizational commitment, (r = .63, p <.01). Positive correlations with Organizational commitment were found for both procedural (r = .72, p <.01) and distributive justice (r = .71, p <.01) as well. All correlations of the three scales for justice with Organizational commitment were found to be strong. It was also indicated that external SRE was moderately and positively related to both Perceived justice of PMS (r = .57, p <.01) and Organizational commitment (r = .51, p <.01). Finally, there was a significant
negative, moderate, correlation between Organizational commitment and turnover intention (r = -.57, p <.01).
Table 1a
Means, standard deviations, correlations of all scales and control variables
Variables M SD 1 2 3 4 5 6 7 8 9 10 1. Turnover intention 2.77 0.91 1 (.75) 2. Justice 3.80 0.85 -.442** 1 (.91) 3. Procedural justice 3.73 0.90 -.353** .810** 1 (.85) 4. Distributive justice 3.77 0.83 -.416** .948** .954** 1 (.82) 5. Organizational Commitment 3.87 0.92 -.573** .626** .719** .708** 1 (.91)
6. Self-rated External Employability 3.76 0.84 -0.09 .567** .551** .587** .507** 1 (.84) 7. Age 33.82 10.20 -.355** .193* 0.095 0.15 .288** 0.144 1
8. Gender 1.41 0.49 -.189* 0.044 0.089 0.071 .217** -0.016 0.047 1
9. Tenure 3.58 4.92 -0.009 -0.032 -0.077 -0.058 0.011 0.15 .170* -0.026 1
10. Hours working for platform 25.45 12.85 -.264** .279** .258** .282** .332** .242** 0.136 -0.033 0.117 1 11. Source of income 1.28 0.45 0.151 -0.008 -0.088 -0.052 -0.088 0.006 0.034 0.049 -.185* -.318** 12. Short/long-term intention 3.16 1.18 -0.041 .271** .345** .325** .260** .307** 0.145 -0.062 0.018 .225** 13. Uber 0.55 0.50 -0.159 .400** .371** .405** .422** .407** .177* -0.031 0.002 .202* 14. Deliveroo 0.12 0.33 0.059 -.203* -0.146 -.183* -0.124 -0.121 -0.111 0 0.066 -.188* 15. Foodora 0.16 0.36 0.16 -.375** -.376** -.395** -.350** -.323** -0.164 -0.081 0.01 -0.029 16. Temper 0.03 0.17 .241** 0.102 0.116 0.115 -0.061 0.121 -0.14 0.031 -0.09 -0.139 17. UberEats 0.09 0.29 -0.11 0.045 -0.003 0.021 -0.127 -.194* -0.151 0.132 -0.151 0.065 18. Helpling 0.01 0.12 -0.063 0.098 0.103 0.106 0.061 0.092 0.15 0.022 .316** -0.028 19. Payment form - Task 0.6 0.49 -0.165 .191* 0.151 .179* 0.138 .179* 0.114 0.042 0.033 -0.027 20. Payment form - Hours 0.35 0.48 .183* -.241** -.175* -.218** -.211* -.188* -0.14 -0.138 -0.076 -0.018 21. Payment form - Set amount 0.04 0.20 -0.048 0.142 0.09 0.121 0.159 0.061 0.097 .181* 0.122 0.097 22. Contract - Zero Hour 0.26 0.44 0.122 -.272** -0.147 -.219** -.196* -.214* -0.158 -0.027 -0.051 -0.022 23. Contract - Set Hours 0.36 0.48 -0.137 0.081 0.076 0.083 .171* 0.028 0.015 0.031 0.158 .167* 24. Contract - Self-employed 0.38 0.49 0.027 0.165 0.056 0.114 0.006 0.165 0.127 -0.006 -0.11 -0.145 25. Location working - USA 0.57 0.50 -.333** .433** .376** .424** .449** .308** .410** 0.107 -0.01 .230** 26. Location working - UK 0.21 0.41 -0.047 -0.157 -0.073 -0.119 0.078 -0.024 -0.128 0.038 0.009 -0.061 27. Location working - NL 0.22 0.42 .444** -.364** -.378** -.390** -.612** -.344** -.365** -0.165 0.003 -.215* 28. Performance management system - App 0.79 0.41 -.260** .474** .385** .450** .456** .401** .304** 0.069 -0.079 .226** 29. Performance management system -
Evaluation
Table 1b
Means, standard deviations, correlations of all scales and control variables
Variables M SD 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 11. Source of income 1.28 0.45 1 12. Short/long-term intention 3.16 1.18 -0.03 1 13. Uber 0.55 0.50 0.154 .192* 1 14. Deliveroo 0.12 0.33 -0.088 0.025 -.412** 1 15. Foodora 0.16 0.36 -.184* -.273** -.478** -0.159 1 16. Temper 0.03 0.17 -0.013 -0.059 -.190* -0.063 -0.073 1 17. UberEats 0.09 0.29 0.071 0.02 -.355** -0.118 -0.137 -0.054 1 18. Helpling 0.01 0.12 0.058 0.086 -0.133 -0.044 -0.052 -0.02 -0.038 1 19. Payment form - Task 0.6 0.49 0.134 0.085 .422** -0.094 -.522** -0.12 0.063 -0.023 1 20. Payment form - Hours 0.35 0.48 -0.138 -0.086 -.348** 0.09 .539** 0.141 -0.134 -0.089 -.900** 1 21. Payment form - Set amount 0.04 0.20 0.023 0.032 -0.164 0.03 -0.091 -0.036 .176* .272** -.256** -0.156 1 22. Contract - Zero Hour 0.26 0.44 -0.008 -0.036 -0.095 0.133 0.107 -0.1 -0.018 -0.07 -0.048 0.076 -0.043 1 23. Contract - Set Hours 0.36 0.48 -.245** 0.001 -0.036 -0.052 .246** -0.129 -0.087 0.159 -.222** .244** -0.086 -.441** 1 24. Contract - Self-employed 0.38 0.49 .249** 0.032 0.121 -0.068 -.339** .217** 0.102 -0.095 .263** -.310** 0.123 -.461** -.593** 1 25. Location working - USA 0.57 0.50 .223** 0.09 .525** -.430** -.421** -.199* 0.076 0.103 .343** -.382** 0.11 -.187* -0.128 .294** 1 26. Location working - UK 0.21 0.41 -0.165 0.127 -0.072 .512** -.219** -0.087 -0.041 -0.061 0.062 -0.047 -0.02 0.104 0.128 -.220** -.591** 1 27. Location working - NL 0.22 0.42 -0.106 -.231** -.556** 0.014 .716** .322** -0.051 -0.064 -.470** .501** -0.112 0.121 0.028 -0.136 -.617** -.270** 1 28. Performance management system - App 0.79 0.41 0.087 .232** .566** -.297** -.603** 0.087 0.041 -0.087 .367** -.393** 0.107 -0.145 -0.128 .257** .485** 0.042 -.620** 1 29. Performance management 0.21 0.41 -0.087 -.232** -.566** .297** .603** -0.087 -0.041 0.087 -.367** .393** -0.107 0.145 0.128 -.257** -.485** -0.042 .620** -1.00** 1
4. Results
4.1 Preliminary analysis
Respondents were asked, before filling in the questions regarding perceived justice, if they receive performance ratings by clients in their work-on-demand app or on the website, or performance evaluations by the company working for (e.g. by management or HR). It was, therefore, of interest to analyse whether perceived justice of PMS differs between these two methods. In order to analyse this, a one-way ANOVA was done. This analysis revealed that there was a significant difference between gig workers receiving client ratings in the app/website and gig workers receiving performance evaluations by their company in their perceived justice (F(1,42) = 37.198, p <.05). Showing that gig workers receiving performance ratings in the app/website indicated to perceive a higher level of justice than gig workers receiving other performance evaluations. This is plotted in figure 2. Both distributive (F(1,42) = 25.822, p <.05) and procedural justice (F(1,42) = 41.871, p <.05) were also significantly different between gig workers receiving client ratings in the app/platform and gig workers receiving performance evaluations by their company, in that perceived justice of ratings in the app/website was significantly higher than perceived justice of other performance evaluations. Based on these findings, and the before mentioned correlations, the type of performance management system was included as a control variable.
Figure 2. Comparing justice for different Performance management systems. 4.2 Hypothesis testing
First of all, it was tested whether perceived justice of PMS and turnover intentions were associated, independent of organizational commitment (Baron & Kenny, 1986). This association was supported by the significant total effect of perceived justice of PMS on turnover
intentions (c1 = -.394, SE = .098, p <.05). This implied that, without taking into account the
organizational commitment of gig workers, an increase of perceived justice of performance management by one unit leads to a decrease of turnover intentions with c1 = .394 units. This
was also indicated by the bias-corrected bootstrap confidence interval given for the total effect, differing significantly from zero (CI [-0.587 to -0.200]). These results supported the interest of examining the relationship between perceived justice of PMS and turnover intentions.
Hypothesis 1 predicted that perceived justice of PMS would have a negative relationship with turnover intention, when organizational commitment is taken into account. This hypothesis was rejected, indicated by a non-significant direct effect of perceived justice of PMS on turnover intention (c1′ = -.057, SE = .106, p =.591). This implied that gig workers differing
