The mediating effect of employee engagement on the relationship between performance management and individual work performance

(1)

1

University of Amsterdam

The mediating effect of employee engagement on the relationship

between performance management and individual work

performance

Master Thesis

Executive Programme in Management studies

Track: Leadership & Management

Author: Rogier van Loenen

Student number: 11141409

Supervisor: dr. W. van Eerde

Date: December 28, 2017

Version: final

(2)

2

Statement of originality

This document is written by student Rogier van Loenen 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

(3)

3

Abstract

This study examined the relationships between performance management, employee

engagement and three dimensions of individual work performance. Employee engagement is believed to play an important role between performance management and the desired

outcome: increased performance. This study is conducted at the Dutch Tax Authorities (D.T.A.), which faces major changes in its workforce because of automation, digitalization, and the increasing role of data and analytics. To manage this change it is essential for the D.T.A. to understand what practices and policies contribute to an engaged workforce. Using self-ratings of engagement, task performance, contextual performance and

counterproductive work behavior. I found that employees found themselves to perform better when they were highly engaged in their work. Further, this study confirms that engagement is a more proximal outcome of performance management than individual work performance itself. Significant evidence was found that all performance management variables are

positively related to employee engagement. Probably the most convincing contribution of this study is that employee engagement is found to be an important mechanism that explains how performance management influences individual work performance. Thus instead of influencing individual performance directly this study proved that performance management influences engagement and in turn engagement positively influences individual performance.

(4)

4

1. Introduction

The Dutch Tax Administration (D.T.A.) faces hard times. The service and its employees have been in turbulent waters for years through outdated computer systems and complex

legislation that are difficult to implement. Over the years, new tasks were added such as the payment of benefits, while budget cuts and organizational changes increased. Within the Tax Administration, human capital has been the most important production factor which is

increasingly combined or replaced by technology. The remaining workforce must be

functioning optimally to be able to cope with these disruptive changes. Traditionally, a lot of attention within the organization is given to the knowledge and skills of employees, but little attention to motivation end engagement is present, and as a consequence little is known about how to create a motivated and engaged workforce.

When I started working for the D.T.A. 2,5 years ago, I slowly recognized how little attention is given to performance management in this organization. First, I wondered whether it would be more present at higher levels of the organization and not visible for employees, but that doesn’t seem to make sense. Although automation is progressing in recent years, the employees are still the core of the organization. Still more than 25.000 employees work at the D.T.A. These employees all contribute to the performance of the organization. In order to obtain the desired results of the organization there has to be a management process that drives the direction, effort and persistence of the employees. The D.T.A. has a standard of one performance appraisal per year. The performance appraisals that I experienced were not satisfying to me. It consisted of a quick review of goals that were agreed on one year ago and a general review of the development of my knowledge, skills, and abilities. Further attention was given to my future development, opportunities for training and educational progress. During the year very little feedback was given on the progress of my goals, or behavior that was desired or not. The invisibility of performance management in the D.T.A. and the dissatisfaction of my performance appraisal resulted in a lack of engagement for me. In previous jobs at other companies (all in the private sector) I never experienced

(6)

6

organization. The need for a motivated and engaged workforce and the apparent

shortcoming of performance management within the D.T.A leads to the central question of this study: To what extent is there a relationship between performance management and employee engagement, and is employee engagement in turn related to employee job performance?

The Dutch Tax Administration (D.T.A.) is a public service organization with more than 25.000 employees. Every working day approximately one billion euros are collected. Every month the D.T.A. pay out around eight million benefits and the customs department annually checks about 12 million containers. The administration also handles a large amount of relatively small workflows and exceptions. The content as well as the process of the tax legislation is complex. The legislation changes along with the development of society, which leads to constant changes and often an increase in complexity.

In May 2015, a large scale investment program which can best be translated as the

Investment Agenda was launched to increase the performance of the D.T.A. By applying new technology like big data analysis and information-driven services, the D.T.A. expects to generate more tax revenues at lower costs and aims to improve interaction with companies and citizens. This has consequences for the human capital of the organization. The forecast was that on the one hand 5.000 mostly administrative jobs would become redundant and on the other hand 1.500 new jobs would be created, for example in the area of data analysis. The D.T.A. is not the only organization that is confronted with technological changes that leads to a rapidly changing workforce. A lot of organizations in the private as well as the public sector are confronted with these changes and probably all these organizations struggle with the question how to obtain a workforce of employees that feel energetic and dedicated. Scholars have recognized this transition and the past decade a large number of scientific studies have been conducted in the field of engagement. So what do we already know about engagement?

Schaufeli et al. (2002, p74) define work engagement as “a positive, affective motivational state of fulfillment that is characterized by vigor, dedication, and absorption”. Vigor has to do

(7)

7

with the level of energy while working; dedication refers to being strongly involved in one’s work and experiencing a sense of significance, enthusiasm, and challenge; and absorption refers to being fully concentrated and have all one’s attention and interest at work. According to Schaufeli et al. (2002) engagement is not a momentary and specific state, it is “a more persistent and pervasive affective-cognitive state that is not focused on any particular object, event, individual, or behavior” (p. 74). Thus, engaged employees have high levels of energy and are enthusiastic about their work and are often fully immersed in their work so that time flies (Bakker & Demerouti, 2007).

Saks and Gruman (2014) did a comprehensive review on the research that has been done on engagement. In the first part of their article they address the issues of the meaning of engagement as well as the theories and the measurement of engagement. I will discuss these subjects later in this paper. First, I want to review what relations with engagement have been researched in order to explain what this study can contribute to the discussion of

engagement.

According to Saks and Gruman (2014) many job resources have been found to antecedents of employee engagement but autonomy, performance feedback and a supportive

environment have been reported frequently and consistently to be positively related to engagement. Leadership and individual differences are also believed to be important for engagement but research on these subjects received much less attention than research on job resources. Further, challenging job demands have been found to be positively related to engagement, while hindrance job demands are negatively related. Consequences of

employee engagement have also been researched extensively. Employee engagement has been found to be positively related to many work-related outcomes such as job satisfaction, job performance, organizational commitment, organizational citizenship behavior and health and wellness of employees. However, Saks and Gruman (2014) emphasize that almost all of the research that has been done so far is correlational and not experimental and most studies have used self-report measures for all of the variables. This leads to caution on causal conclusions and raises concerns about same-source bias such as central tendency,

(8)

8

social desirability, and halo effects (e.g. Bass & Avolio, 1989). Christian et al. (2011) also recognized the issue of causality by noting that it is possible that the relationship between engagement and other variables represent reverse or reciprocal causality. Thus, instead of concluding that more performance feedback leads to more engaged employees it might as well be that engaged employee receive performance feedback more often. And, instead of claiming that engaged employees perform better it might as well be true that higher

performing employees might become more engaged.

This study is interesting for a number of reasons. First, it contributes to the existing research by examining a unique combination of variables that has never been researched before. Elements of performance management and employee performance have been researched in relation to engagement but research on a comprehensive set of performance management and employee performance variables has not been done before. Second, performance management has been researched intensively for several decades but only in the past decade scholars and practitioners have been increasingly interested in a more strength-based approach instead of a deficit strength-based-approach of managing performance. Employee engagement can be seen as an important element in a strength-based approach of

performance management by helping and facilitating employees’ change and then leading to improvement in their performance (Kim et al., 2013). Third, the D.T.A. has to manage major changes in its workforce because of the aging workforce, and 5.000 employees that leave the organization in the coming three years. The nature of the remaining work will change significantly during the coming decade because of automation, digitalization and the

increasing role of data and analytics. To manage this change it is essential for the D.T.A. to understand what practices and policies contribute to an energized, dedicated and

concentrated workforce.

Research question

To what extent does employee engagement mediate the relationship between performance management and employee job performance?

(9)

9

2. Literature review and hypotheses

First, I will explain the dependent variable employee job performance by explaining what it is and what dimensions of job performance can be distinguished. Second, I will explain what employee engagement is, how it is linked to employee job performance and how it facilitates the process of performance management. Third, I will review the independent variable of performance management and will identify which elements of performance management are possible antecedents of employee engagement. Finally, I will discuss the mediating role of employee engagement.

Individual work performance

What is performance? It is difficult to find an overall definition of performance that is applicable across different jobs and situations. Before coming to a clear definition of performance I first need to discuss three aspects of performance that matter for this study. The first is the level of performance. Within an organization, individual performance, team performance, unit performance, and organizational performance can be distinguished. Individual performance is what comes first because it is the basic building block on which the entire economy is based (Kim & Ployhart, 2014). Individual performance is a prerequisite for the outcome of all other levels of performance because without individual work performance an organization will not perform at all. Therefore, the focus of this study is on individual performance.

The second aspect of performance is the distinction between the determinants of performance and performance itself. The determinants of performance such as the

knowledge, skills, and choice behavior affect performance but should not be confused with performance itself. Performance itself is what directly facilitates the organizational goals (Campbell & Wiernik, 2015).

The final distinction is the difference between the outcome of performance and performance itself. The outcome of performance can be defined as the products or services that are

(10)

10

produced and contribute to the overall strategic goals of an organization. The outcome of performance is not included in this study because it is influenced by many other factors than the individual level of performance and can be seen as a more distal outcome of

performance itself.

The focus of this study is the individual level of performance and consists of the actions or behaviors that employees undertake in their daily work situation. Individual work performance can therefore be defined as “things that people actually do, actions they take, that contribute to the organization’s goals” (Campbell & Wiernik, 2015, p48).

Koopmans et. al (2011) distinguished four dimensions of individual work performance in a systematic review of performance management literature. These elements are (1) task performance, (2) contextual performance, (3) adaptive performance, and (4)

counterproductive work behavior (CWB). (1) Task performance can best be described as the competency with which one performs central job tasks. Indicators of this performance

dimension are completing job tasks, work quantity, work quality, job skills and job knowledge. Task performance can also be defined as in-role performance. And can be described as those officially required outcomes and behaviors that directly serve the goals of the

organization (Motowidlo & Van Scotter, 1994); (2) “Contextual performance can be defined as individual behaviors that support the organizational, social, and psychological

environment in which the technical core must function.” (Koopmans et al.,2011, p861). Indicators of this dimension of performance are extra tasks, effort, initiative, enthusiasm, persistence, pro-activity and creativity; (3) Adaptive performance is defined as the extent to which an employee adapts to changes in a work-system or work-role. Indicators of this dimension are generating new ideas, learning new tasks, being flexible and open minded and understanding other groups or cultures; (2) contextual and (3) adaptive performance are also labeled as extra-role behavior or organizational citizenship behavior (OCB) and can be defined as positive, discretionary behaviors on the part of an employee that are believed to directly promote the effective functioning of an organization without necessarily directly influencing an employee’s productivity (MacKenzie et al., 1991); (4) counterproductive work

(11)

11

behavior (CWB) is behavior that can harm the organization. Indicators of this dimension are off-task behavior, too many or longer breaks, absenteeism, presenteeism, complaining, gossiping, and theft.

In a later study Koopmans et al. (2012) found that adaptive performance is not a separate dimension, but rather an aspect of contextual performance. Both performance aspects can be considered supporting the organizational, social and psychological environment and both are extra-role behaviors that do not directly contribute to the central job tasks. Therefore, in this study adaptive performance will not be considered as a separate dimension.

For this study, the remaining three dimensions of employee performance are relevant, because each dimension represents a different facet of individual work performance. Task and contextual performance have been researched extensively before in relation to

engagement, because it is believed that engaged employees will be more focused on their work tasks, which increases task performance, and are more likely to step out of their job to facilitate the organization, which increases contextual performance (Christian et al., 2011). CWB is interesting, because employees who are low in employee engagement might engage in CWB as means of retaliating against their employees for proving an unpleasant work environment (Ariani, 2013).

Engagement

Employee engagement has been a very popular subject for the past 15 years among researchers. But also consultants and organizations have a great interest in this subject because researchers have shown that employee engagement is an important factor for an organization’s success (Rich et al., 2010). Researchers have also found that employee engagement in general has declined over the past generation (Bates, 2004; Richman, 2006). The disengagement of employees is also referred to as the “engagement gap” which could be costing billions of dollars because of lost productivity (Bates, 2004; Johnson, 2004; Kowalski, 2003).

(12)

12

The subject of engagement is somewhat controversial among researchers. There is no consensus about the definition, the meaning and the name of the construct. The names employee engagement (Saks & Gruman, 2014), job engagement (Rich et al., 2010) and work engagement (Schaufeli & Salanova, 2011) have been used as a name for the same

construct. There is also no consensus about how employee engagement should be measured.

Kahn (1990) was one of the first researchers to study engagement. He defined engagement as “the harnessing of organization members’ selves to their work roles; in engagement, people employ and express themselves physically, cognitively, and emotionally during role performances” (p. 694). On the contrary he defines disengagement as “the uncoupling of selves from work roles; in disengagement, people withdraw and defend themselves

physically, cognitively, or emotionally during role performances” (p. 694). Kahn (1990, 1992) also argued that engagement also means that individuals are psychologically present when occupying and performing an organizational role. When people are psychologically present they are attentive, connected, integrated, and focused in their role performances (Kahn, 1992).

Before further explaining what engagement is, it is worth discussing similar employee well-being constructs. Christian et al. (2011) explain the difference between engagement and job attitudes such as job satisfaction (JS), affective organizational commitment (AC) and job involvement (JI). JS is more similar to satiation and engagement is related to activation. Further, JS is an evaluative description of job conditions, whereas engagement is a

description of experience resulting from the work itself. AC is an emotional attachment to the organization that results from shared interests and values, whereas engagement represents perceptions that that are based on the work itself. Macey and Schneider (2008) suggests that AC might be a facet of engagement but not sufficient for engagement. Engagement differs from JI because JI is a more cognitive belief that a job satisfies one’s needs. As a results JI may be considered as a facet of engagement (Christian et al., 2011). Rich et al. (2010) argue

(13)

13

that JI focuses on the cognitive energy individuals invest to maintain identities related to work. JS focuses on affective reactions and the need to maintain happiness.

Engagement has also been described by several researchers as the opposite of burnout (Maslach et al., 2001). Research on burnout and engagement has found that the core dimensions of burnout (exhaustion and cynicism) and engagement (vigor and dedication) are, indeed, opposites of each other (Gonzalez-Roma, et al., 2006).

Schaufeli et al. (2002) define engagement as “a positive, fulfilling, work-related state of mind that is characterized by vigor, dedication, and absorption” (p. 74). They note that vigor involves high levels of energy and mental resilience while working; dedication refers to being strongly involved in one’s work and experiencing a sense of significance, enthusiasm, and challenge; and absorption refers to being fully concentrated and engrossed in one’s work. Three theoretical models that explain the construct of engagement can be found in the literature. The first model was that of Kahn (1990). He found that a person’s degree of engagement was a function of psychological meaningfulness, psychological safety, and psychological availability. Psychological meaningfulness is about the extent to which people experience meaning from their work and feel that they are worthwhile, useful and valuable. Psychological safety has to do with being able to be yourself without the fear of negative consequences (Kahn, 1990). Psychological availability refers to the belief of a person to have the physical, emotional, and psychological resources available to perform during work (Kahn, 1990). May et al. (2004) found that meaningfulness, safety, and availability were significantly related to engagement. They also found that job enrichment and role fit were positively related to meaningfulness; rewarding coworker and supportive supervisor relations were positively related to safety while adherence to coworker norms and self-consciousness were negatively related; and resources available were positively related to psychological

availability while participation in outside activities was negatively related. The link between Kahn’s theory and performance management as antecedents of engagement, which will be discussed later, is in the motivational characteristics of performance management. These

(14)

14

characteristics motivate workers cause experiences of meaningfulness, responsibility and knowledge of results (Hackman & Oldham, 1976).

The second theory of engagement is based on the literature on job burnout. According to Maslach et al. (2001) job burnout has six major antecedents: workload, control, rewards and recognition, community and social support, perceived fairness, and values. According to this theory, engagement is associated with the opposites of these antecedents: sustainable workload, feelings of choice and control, appropriate recognition and reward, a supportive work community, fairness and justice, and meaningful and valued work. Schaufeli and Salanova (2011) criticize this theory. They argue that “from a psychological perspective, the assumption of a perfectly inverse relationship of burnout and work engagement is not feasible because not feeling burned-out doesn’t necessarily mean that one feels engaged, and not feeling engaged doesn’t necessarily mean that one is burned-out. In fact,

engagement and burnout may co-occur, at least to some extent” (p41). A meta-analysis of Halbesleben (2010) showed that correlations between work engagement and burnout range from –.24 to –.65, depending on the dimensions involved (Halbesleben, 2010). This is much less than –1.0, which would result when both were perfect counterparts. It can be concluded that burnout and engagement should be measured independently.

A third theory of employee engagement is the Job Demands–Resources (JD-R) model (Bakker & Demerouti, 2007). A unique feature of the JD-R model, in which it differs from the other two theories, is that it poses two parallel processes. The first is a health impairment process and explains what workers can do. The second is a motivational process and

explains what workers will do (Demerouti & Cropanzano, 2010). Key antecedents in these

twin processes are job demands and job resources. Job demands are those aspects of the work that require physical, emotional or psychological effort from an employee. Because of this effort they are associated to psychological costs such as exhaustion. They have the

potential to reduce health and thereby reduce performance (Demerouti & Cropanzano,

2010). Examples of these features are: work overload, job insecurity, role ambiguity and time pressure (Saks & Gruman, 2014). Job resources are those aspects of a job that are

(15)

15

functional to achieving goals, reduce job demands and stimulate personal growth, learning, and development. Examples of job resources are: pay, career opportunities, social relations, role clarity, participation in decision making skill variety, task identity, task significance, autonomy, and performance feedback (Saks & Gruman, 2014). The idea of the JD-R model is that job resources activate a motivational process that can lead to higher levels of

engagement, positive attitudes, and well-being and lower the potential for burnout (Bakker & Demerouti, 2007; Crawford et al., 2010). On the other hand, high job demands lead to a depletion of energy and increased stress that can cause disengagement, burnout, and health problems (Bakker & Demerouti, 2007). “The basic premise of the model is simply that the more resources an employee has, the more engaged he/she will be” (Saks & Gruman, 2014, p163). The link between this theory and performance management as antecedents of

engagement, which will be discussed later, is because performance management elements like feedback, goal setting, and communication can be seen as job resources which in turn activates a motivational process that can lead to higher levels of engagement.

Most research has been done on the JD-R model, but Crawford et al. (2010) criticize it and state that “a limitation of the job demands–resources model is that it does not include all relevant predictors of employee engagement” and its “greatest use is to broadly categorize working conditions as either resources or demands in predicting engagement” (p. 844). The relationship between employee engagement and performance has also been

researched extensively. Engagement is thought to be related to the persistence and intensity with which one pursues task performance (Burke, 2008; Rich et al., 2010). Christian et al. (2011) argue that engagement is closely aligned with task-specific motivation. “Because engaged employees experience a high level of connectivity with their work tasks, they strive toward task-related goals that are intertwined with their in-role definitions and scripts, leading to high levels of task performance” (Christian et al., 2011, p120). I expect engaged

employees will be more focused on their work tasks and thus engagement will be positively related to task performance.

(16)

16

Rich et al. (2010) argue that engaged employees invest themselves more fully while at work than those who are less engaged. They should be more willing to step outside the

boundaries of their formally defined jobs and engage in behavior that could ultimately benefit the organization. Kahn (1990, 1992) argues that the physical, cognitive, and emotional energies of engagement foster active, complete role performances through behavior that is extra conscientious, interpersonally collaborative, innovative, and involved. This positive work behavior contributes more indirectly to the goals of the organization and can be labeled as contextual performance. I expect employee engagement is positively related to contextual performance.

Hypothesis 2: Employee engagement is positively related to contextual performance

So far, little research has been done on the relationship between engagement and

counterproductive work behavior (CWB). There is some evidence that positive emotions at work are negatively related to employee engagement. Sulea et al. (2012) found that higher levels of engagement are beneficial for organizations because they reduce CWB. Ariani (2013) also found a negative relationship between employee engagement and CWB. She argues that employees who are engaged in their job maintain a positive perception of their work, and individuals who are not engaged may have negative perceptions of their situation at work. When employees have unfavorable perceptions of their situations at work, they are more likely to engage in CWB (Colbert et al., 2004)

Hypothesis 3: Employee engagement is negatively related to CWB

Thus, employee engagement has been linked to employee performance and Kahn’s theory as well as the JD-R model point out that performance management variables are indicated as antecedents of engagement. The role of performance management as antecedent of engagement will be discussed in the next section.

(17)

17

Performance management

What is performance management? The name implies that it is a practice to control

performance. Castka et al. (2003) state that the overall goal of performance management is to ensure that all parts of the organization are working together in an optimal way to achieve the results desired by the organization. This implies that performance management aims at all levels of performance. Gruman and Saks (2011, p123) define performance management as “all organizational policies, practices and design features that interact to produce

employee performance.” From this point of view performance management aims at the individual level of performance which is consistent with the focus of this study.

Roberts (2001) made an extensive list of examples of policies and practices that are part of performance management: the setting of corporate, departmental, team, and individual objectives, the use of appraisal systems, reward strategies, training and development schemes, individual career planning, feedback, communication, coaching, interventions and even culture management. In this study I am looking for the elements of performance management that influence individual work performance through employee engagement. From the extensive list of Roberts (2001) the performance management practices were selected for this study on the basis of two criteria. First, evidence must exist in the literature that this practice influences employee engagement. Second, the practice must be suitable for interventions in order to be of practical value for the organization.

Feedback

Several theories emphasize the importance and positive effects of feedback for employees’ task performance (e.g. Locke & Latham, 2002). Task performance is related to cognitive variables such as learning and knowledge. According to Belschak and Den Hartog (2009) feedback does not only effect cognitive variables but it also elicits emotional reactions. They further note that the literature on emotions suggests positive feedback will generally lead to positive emotions, such as pride and happiness, and negative feedback will generally result in negative emotions, such as disappointment or guilt. Feedback can be related to

(18)

18

engagement in two ways. First, when feedback is given in a constructive way it can create a feeling of safety. Psychological safety is one of the antecedents of engagement (Kahn, 1990) as discussed in the section of engagement. Second, feedback can be seen as a job resource because it activates a motivational process that leads to higher levels of engagement as discussed earlier in the engagement section (Bakker & Demerouti, 2007; Crawford et al., 2010).

Hypothesis 4: Feedback is positively related to employee engagement

Goal setting

Gruman and Saks (2011) note the importance of goals for initiating employee engagement, because goals stimulate the intensity or the feeling of engagement. Macey et al. (2011) go even further and argue that engagement can only occur when an employee has a specific goal, and this individual goal needs to be aligned with the organizational goal. This alignment creates a strategic focus to engagement, because it ensures that the individual goals

contribute to the goals of the organization.

Hypothesis 5: Goal setting is positively related to employee engagement

Communication

D.T.A. can be seen as an organization that is increasingly being dominated by knowledge intensive jobs. Pulakos (2008) note that it is difficult to manage and set objectives for knowledge intensive jobs because such work is more varied and subtle. Therefore, in knowledge intensive organizations performance management must also focus on managing the context in which performance management takes place. Therefore, also communication as a more general variable of performance management will be included in this study. Internal communication is defined by Kalla (2005, p303) as the “social interaction through messages” According to Welch and Jackson (2007) internal communication reflects the management’s ability to build relationships between internal stakeholders at all levels within an organization. Mishra et al. (2014) argue that if an organization shares information

(19)

19

frequently and honestly, employees feel that they belong to that organization and that they share a mission with their employer. This leads to employee engagement, because the organization and the employee develop a bond of trust. Macey and Schneider (2008) go even further by stating that trust is central to the network of antecedents of employee

engagement, because engaged employees “invest their energy, time, or personal resources trusting that the investment will be rewarded (intrinsically or extrinsically) in some meaningful way” (p22). They continue by reasoning that trust in the organization and the leader is essential, because only under that condition employees will display engagement behavior. They also link trust to psychological safety (Kahn 1990), which has been discussed earlier in the section of engagement. They state that if there is psychological safety, employees have the belief that they will not be punished for their personal engagement. An example of punishment is performing above the norms of a group and then being socially punished by that group.

Hypothesis 6: Communication is positively related to employee engagement

Fairness of the performance appraisal

A key component of performance management is the appraisal and evaluation of employee performance (Gruman & Saks, 2011). Dobbins et al. (1993) note that performance appraisals sometimes evaluate employees on criteria that are irrelevant, or over which they have no control. Such appraisals are not useful and are perceived by employees as unfair. “In order for employees to feel comfortable employing and expressing themselves fully during role performances (i.e. displaying engagement) they must trust that their organization and

managers will treat them fairly and justly” (Gruman & Saks, 2011, p131). Leiter and Maslach (2008) found evidence that once people experience problems with fairness in the workplace, they are more likely to develop burnout over time. In contrast, people who were not

experiencing problem with fairness in the workplace are more likely to display engagement behavior. They argue that fairness might play a unique role in defining the workplace as either a good place or a bad place to be, because once people begin to feel angry about

(20)

20

inequities and have no faith that organizational processes will correct these inequities, this may lead to an increase of negative reactions on the job.

Hypothesis 7: Performance appraisal fairness is positively related to employee engagement

Training

Theory on training and performance addresses a direct effect as well as an indirect effect. The effect is direct when performance problems have less to do with motivation but are

designed to increase the ability of employees to perform better (Murphy & DeNisi, 2008).An

indirect effect through engagement is also proposed. Gruman and Saks (2011) give three examples of how training can influence job performance through engagement. First, because training can be an important resource for preparing employees to cope with job demands. Second, because “training can also make employees feel more secure about their ability to perform their job thereby lowering their anxiety and increasing feelings of availability” (Gruman & Saks, 2011, P131). Third, because training is especially relevant for providing employees with resources that will make them feel available to fully engage in their roles. The indirect effect of training on performance through engagement can also be linked to psychological capital (PsyCap). PsyCap is a higher order construct that consists of the dimensions self-efficacy, hope, optimism and resilience. As described by Kahn (1990), individuals are more available when they feel secure, and self-efficacy, hope, optimism and resilience are important dimensions of security. All of the PsyCap variables can be

developed through training (Gruman & Saks, 2011). Schaufeli and Salanova (2008) suggest that enhancing engagement can be promoted by offering employees training that provides experiences of vocational success, encouragement, and reducing the fear of failure.

(21)

21 Mediating role of employee engagement

To this point, I have argued that performance management leads to higher levels of

employee engagement and, in turn, employee engagement leads to higher levels of task and contextual performance and lower levels of CWB. In other words, I have described a model in which employee engagement mediates relationships between performance management and individual work performance. Although it is the ultimate objective of performance management, increased task and contextual performance can be considered a distal outcome of the performance management process. More proximal outcomes include the cognitive, affective outcomes that precede changes in performance (Gruman & Saks, 2011). Performance management practices are expected to lead to Kahn's (1990) three

psychological conditions that produce engagement because performance management processes provide the necessary resources to employees that lead to employee

engagement. Engaged employees experience a high level of connectivity with their work tasks. They strive toward task-related goals, leading to higher levels of task performance (Christian et al., 2011). Therefore, I expect that employee engagement mediates the relationship between all elements of performance management and task and contextual performance. On the other hand, I expect that employees that are connected with their work and have a positive state of mind regarding their work, have less reason to involve

themselves in counterproductive work behavior. Therefore, I expect that performance management leads to higher levels of employee engagement as a proximal outcome and in turn decrease the levels of CWB.

Hypothesis 9: Feedback is positively related to task performance through increased employee engagement.

Hypothesis 10: Feedback is positively related to contextual performance through increased employee engagement.

Hypothesis 11: Feedback is negatively related to CWB through increased employee engagement.

(22)

22 Hypothesis 12: Goal setting is positively related to task performance through increased

employee engagement.

Hypothesis 13: Goal setting is positively related to contextual performance through increased employee engagement.

Hypothesis 14: Goal setting is negatively related to CWB through increased employee engagement.

Hypothesis 15: Communication is positively related to task performance through increased employee engagement.

Hypothesis 16: Communication is positively related to contextual performance through increased employee engagement.

Hypothesis 17: Communication is negatively related to CWB through increased employee engagement

Hypothesis 18: Performance appraisal fairness is positively related to task performance through increased employee engagement.

Hypothesis 19: Performance appraisal fairness is positively related to contextual performance through increased employee engagement.

Hypothesis 20: Performance appraisal fairness is negatively related to CWB through increased employee engagement.

Hypothesis 21: Training is positively related to task performance through increased employee engagement.

Hypothesis 22: Training is positively related to contextual performance through increased employee engagement.

Hypothesis 23: Training is negatively related to CWB through increased employee engagement.

(23)

23

The conceptual model with all hypotheses is depicted in figure 1.

(24)

24

3. Method

Research design

My research philosophy is positivism. Performance management, employee engagement, and individual performance are observable realities to me. I’m looking for relationships between these variables. With a deductive approach I use existing theory on employee engagement to develop my hypotheses and test them, leading to the further development of theory on employee engagement. Since I use data to test my theory, my main research design is quantitative. A surveys was used to the measure the constructs. Besides a contribution to the literature, the research should also provide practical points to improve employee engagement. Therefore, the challenge in the research design is to find a balance between the number of variables and the size and complexity of the research. More

independent variables might explain more variance in the dependent variable. This also comes with disadvantages. A variable cannot be measured with a single question. In that way scale reliability cannot be determined. In other words: with one single question

understanding may be ambiguous due to the subtle differences that language may induce. A higher number of questions for each variable gives more possibilities to increase scale reliability of each variable. Of course there are limitations to the number of questions in a survey. A survey that contains too many questions and requires more than eight minutes will result in a lower response rate and will affect the reliability of the whole study because participants tend to rush to the end of the survey. The variables were operationalized with likert scales. An open question was added to the survey to gather some qualitative

information on employee engagement. The survey was taken at one moment in time, which makes a cross-sectional study. This is a major limitation of this study because it limits the conclusions on causality. The survey is shown in appendix A.

(25)

25 Data collection procedure

Three departments within the D.T.A. agreed to participate in this study. A total of 900 employees were invited to take part in the survey. All employees received an e-mail containing a brief description of the study and an anonymous link to the online survey. A reminder was sent after one and a half week. The total data collection period lasted three weeks.

Sampling technique

The population of this study consisted of all the employees of the D.T.A. Non-probability sampling was used in this study. The reason for this is because the D.T.A. performed an employee satisfaction research in the same period for the entire organization and top management was not willing to cooperate in this study. Therefore, I had to find departments that are willing to participate in this study. I found three different departments that wanted to participate. This can be described as convenience sampling because the participants were selected because they are most conveniently available. This comes with limitations that will be discussed later.

Measures

Employee engagement was measured using the validated nine-item Dutch version (UBES-9)

of the Utrechtse Work Engagement Scale (UWES; (Schaufeli, et al, 2006). The original version contains 17 items. The short version was chosen because this study consists of nine variables and the number of questions of this survey had to be limited in order to prevent unfinished surveys. Employee engagement consists of three dimensions (vigor, dedication, and absorption) and each dimension was measured with three items. An example item is: “At my job I feel strong and vigorous”. All items were scored on a seven-point rating scale

ranging from “never” to “always”. Although confirmatory factor analyses have supported the three-dimensional structure of the UWES, the dimensions are very closely related (e.g.,

(26)

26

Demerouti et al., 2001; Schaufeli et al., 2006). Therefore, one overall score was be used as the only indicator of employee engagement.

Individual work performance was measured using the validated Dutch version (IWPV;

Koopmans et al., 2014a) of the Individual Work Performance Questionnaire (IWPQ;

Koopmans et al., 2014b). It was developed as a self-report questionnaire. Several downsides accompany self-reporting of performance, instead of managerial ratings. Jamarillo et al. (2005) showed that self-ratings have a lower correlation with objective performance than managerial ratings. Further, self-ratings are known to show leniency effects (Van der Heijden & Nijhof, 2004). That means that employees tend to present themselves in a more favorable way. Self-report scales were chosen for several reasons. First, in many jobs, objective measures of performance are not easily obtainable (Jaramillo et al., 2005). Especially for knowledge work like jobs at the D.T.A., direct measures of behaviors or outcomes such as production quantity are very difficult to measure (Koopmans et al., 2012). Second,

employees have much more opportunity to observe their own behaviors than peers or managers do (Van der Heijden & Nijhof, 2004). This may be especially true for

counterproductive behaviors, because most of these behaviors are intended to be

unobservable. Supervisors have little or no observations at all for judging counterproductive behaviors (Dalal, 2005). Third, compared to managerial ratings, self-reports have practical advantages such as ease of collection, less issues with confidentiality, and less missing data (Schoorman & Mayer, 2008). The IWPQ consists of three dimensions (task performance,

contextual performance, and counterproductive work behavior) with a total of 18 items. All

items had a recall period of three months and a five-point scale (“seldom” to “always” for task and contextual performance, “never” to “often” for counterproductive work behavior). An example item is: “I managed to plan my work so that is was done on time” (task

performance).

Goal setting was measured with a scale that was developed by Wright (2004) based on

Locke and Latham’s (1990) Goal Setting Questionnaire and an instrument developed by Steers (1976). The scale consists of the dimensions job goal specificity (four items) and job

(27)

27

goal difficulty (five items). An example item is: “My responsibilities at work are very clear and specific”. The original scale was measured on a six-point scale. In order to reduce the number of different scales and the complexity of the questionnaire, all items were measured on a seven-point scale (“strongly disagree” to “strongly agree”). Since the questionnaire was in Dutch, all items were translated to Dutch and then translated back into English by an independent person (native English speaking colleague) to check the if the questions were still valid.

Feedback was measured using parts of the Feedback Environment Scale (FES) which was

developed and validated by Steelman et al. (2004). The scale consists of seven dimensions (source credibility, feedback quality, feedback delivery, favorable feedback, unfavorable feedback, source availability, and promotes feedback seeking) and 32 items. Since the questionnaire had to be limited to 70 items, three dimensions were selected that are

expected to be most relevant for this study: feedback quality (five items), favorable feedback (four items), and unfavorable feedback (4 items). An example items is: “My supervisor gives me useful feedback about my job performance”. All items were translated to Dutch and then translated back into English to check for consistent validity. One feedback quality item was left out because two translated items in Dutch resembled too much. All items were measured on the original seven-point scale ranging from “strongly disagree” to “strongly agree”.

Communication was measured using parts of the Communication Satisfaction Questionnaire

(Downs & Hazen, 1977). The CSQ consists of ten dimensions: horizontal and informal communication, communication climate, communication with supervisor, media quality, top management communication, organizational integration, organizational perspective,

interdepartmental communication, personal feedback, and relationship with subordinates. For this study only the dimensions top management communication (4 items) and

communication with supervisor (5 items) were used because these items are believed to be core parts of the performance management system and distinctive from other variables in this study. The items in the CSQ were measured on a seven-point scale (“strongly

(28)

28

scale that measures agreement instead of satisfaction in order to reduce the number of different scales and reduce the complexity of the questionnaire. An example item is: “My supervisor trusts me” (communication with supervisor).

Performance appraisal fairness was measured with a six item scale that was validated by

Tang and Sarsfield-Baldwin (1996). This measure used a five-point scale from “very unfairly” to “very fairly”. All items were rewritten to fit them in the seven-point scale that measures agreement (“strongly disagree” to “strongly agree”) instead of fairness in order to reduce the number of different scales and reduce the complexity of the questionnaire. An original example item is: “How fair do you feel your last performance appraisal was?”. A rewritten example item is: “My last performance appraisal was fair”. All items were translated in Dutch and back translated to ensure the validity of the original items.

Training utility was measured with a six-item scale used by Van Eerde (2008), a combination

of Ford and Noe’s (1987) scale and items from Morgan and Casper’s (2000) scale. This measure was intended to measure training utility from the perspective of the employer. Therefore, all items were rewritten to fit to the perspective of the employee. A rewritten example item is: “I have been able to apply what I have learned in training to the job”. The original five-point scale was changed in a seven-point scale ranging from “strongly disagree” to “strongly agree”. All items were translated in Dutch and back translated to ensure the validity of the original items.

(29)

29

4. Results

In this chapter the results of the quantitative analyses are reported, which consists of three parts. First, the necessary preliminary steps of data preparation will be described. Second, the correlations between all dimensions in the model are reported. In the last part all hypotheses of the conceptual model are examined. All calculations were done with IBM SPSS Statistics version 23. The PROCESS macro was installed in order to test the indirect effects.

Data preparation

The first step in the data preparation was the recoding of counter indicative items. Four items were phrased so that a high level of the item represented a low level of the construct being measured. An example of a counter indicative item is: “I seldom receive praise from my supervisor”. All four items were recoded. A respondent that answered seven was recoded to one, six was recoded to two, five was recoded to three, etc.

The second step was computing the scale reliability (Cronbach’s Alpha) of the continuous variables. Cronbach’s alpha (CA) reliability coefficient ranges between zero and one. The closer the CA coefficient is to one the greater the internal consistency of the items in the scale. CA’s higher than .7 are considered to be acceptable (Gliem & Gliem, 2003). All initial CA’s are higher than .7 except for goal setting. Initial CA for goal setting with all nine items was .678. The scale reliability was increased to .705 by deleting the item: “It is difficult to evaluate success or failure on my job”. By deleting this item, all CA’s are higher than .7 and therefore acceptable. Final CA’s are reported in table 2, which will be outlined in the

correlation analysis.

The third step was computing the scale means of all variables. Scale items were computed into scale means by calculating the average score of all items for every variable. With

descriptive statistics the mean, standard deviation, skewness and kurtosis of all scale means were calculated. Skewness and kurtosis were calculated in order to determine the normal distribution of the data for each variable. Normal distribution of the data is required to perform

(30)

30

the preferred parametric statistical methods, such as linear regression. The results show some negative skewness for goal setting (-1.02) and performance appraisal fairness (-1.20). Values for skewness between -2 and +2 are considered acceptable in order to prove normal distribution (George & Mallery, 2010), therefore nu further changes were made to the data.

Control variables

Data for the following control variables was collected during the data collection: manager role, age, gender, education, tenure of the employment and tenure of the current job.

Education is a multi-categorical variable and was dummy coded to be able to test the effects in a regression analysis. The model is statistically significant and explains 8% of the variance

(Adj. R2 = .080, p<.05). The effect of manager role (ß = .214, p<.05) and Education

University/Master are significant in this model ((ß = .245, p<.05). Therefore, in the further analysis the effects will be controlled for manager role and high education level. The results are shown in table 1.

Table 1

Results of linear regression analyses of all Control Variables as predictors for Engagement

Engagement

Variable R2

Adj. R2

R2

change B Std. Error Beta t Sig.

Model 1 .116 .080 .116 .001

Manager .892 .260 .214 3.437 .001

Age .008 .013 .072 .610 .543

Gender .053 .161 .022 .330 .742

Tenure employment .002 .010 .026 .208 .836

Tenure current job .000 .009 -.002 -.029 .977

Education HAVO/VWO .188 .274 .062 .685 .494

Education MBO .289 .278 .099 1.038 .301

Education HBO/Bachelor .002 .258 .001 .008 .994

Education University/Bachelor -.684 .564 -.083 -1.212 .227

Education University/Master .757 .296 .245 2.561 .011

Note: N=250; all tests were two-tailed. p<.05 Correlation analysis

In table 2 an overview is given of the means, standard deviations, correlations and scale reliabilities of all the variables in the model. The first observation that stands out is that engagement shows significant moderate correlations with all other variables. This might indicate that engagement indeed plays an important role as mediating variable between performance management and individual work performance. The highest correlation is found

(31)

31

between engagement and contextual performance (r=.44, P<.01). The second noticeable observation is that no significant correlations is found for goal setting with task performance. This is unexpected because goal setting is generally thought to be positively related to task performance (Locke & Latham, 2002). Further, CWB shows negative correlations with all other variables, except task and contextual performance. This shows that someone who often engages in behavior that harms the organization, shows lower levels of feedback, goal setting, communication, performance appraisal fairness, training, engagement, and vice versa. The last observation is that the relationships of feedback with communication (r = .61, p < .01) and feedback with performance appraisal fairness (r = .57, p < .01) both show significant strong positive correlations.

Table 2

Scale, Mean, Standard Deviation, Pearson Correlation and Scale Reliability

Variables Scale M SD 1 2 3 4 5 6 7 8 9 10 11

1. Manager role (1=manager role) 0-1 .09 .28 -

2. Highly educated (1=university) 0-1 .16 .37 .05 -

3. Engagement 1-7 4.11 1.16 .22** .20** (.94) 4. Task Performance 1-5 3.37 .74 .06 .12 .24** (.79) 5. Contextual Performance 1-5 3.04 .74 .19** .17** .44** .30** (.82) 6. CWB 1-5 2.23 .67 .01 -.02 -.28** -.11 -.058 (.77) 7. Feedback 1-7 4.48 1.12 .12* .09 .29** .16** .16** -.25** (.91) 8. Goal Setting 1-7 5.01 .84 .11 .04 .40** .00 .15** -.11* .37** (.71) 9. Communication 1-7 4.76 1.00 .11 .18** .38** .19** .23** -.33** .61** .27** (.88) 10. Performance Appraisal Fairness 1-7 5.43 1.02 -.26 .03 .26** .20** .12* -.23** .57** .37** .48** (.93) 11. Training 1-7 4.76 1.27 .31 .24** .42** .09 .24** -.16** .33** .42** .40** .28** (.94)

N=276. Reliabilities (Cronbach Alpha's) are reported along the diagonal M: Mean; SD: Standard Deviation

* correlation is significant at the .05 level (two-tailed) ** correlation is significant at the .01 level (two-tailed)

Hypothesis testing

To test the mediating effect of engagement between performance management and individual work performance the PROCESS macro was installed in SPSS. For each of the five performance management variables the effects were analyzed for all three dimensions of individual work performance separately. Model four was chosen in the PROCESS macro which supports the analysis of a simple mediation. From the analyses conclusions about the total effect, the direct effect and the indirect effect can be drawn. The analyses will be

(32)

32

discussed for each performance management variable separately. Therefore the order of analysis for the hypotheses in this section is not the same as in the literature section. All analyses were controlled for manager role and highly educated.

Feedback

The results of the analyses for feedback are shown in table 3. The effect of engagement on task performance is statistically different from zero (p<.01) and therefore support is found for H1. The effect of engagement on contextual performance is statistically different from zero (p<.001) and therefore support is found for H2. The effect of engagement on CWB is statistically different from zero (p<.001) and therefore support is found for H3.

The effect of feedback on engagement is statistically different from zero (p<.001) and therefore support is found for H4.

The indirect effect of feedback on task performance is statistically different from zero, as revealed by a 95% BC bootstrap confidence interval that is entirely above zero (.007 to .064). The direct effect of feedback on task performance is not significant (p>.05). The conclusion can be drawn that engagement fully mediates the relationship between feedback and task performance and therefore support is found for H9.

The indirect effect of feedback on contextual performance is statistically different from zero, as revealed by a 95% BC bootstrap confidence interval that is entirely above zero (.027 to .094). The direct effect of feedback on contextual performance is not significant (p>.05). Since the indirect effect is statistically different from zero and the direct effect is not, the conclusion can be drawn that engagement fully mediates the relationship between feedback and contextual performance and therefore support is found for H10.

The indirect effect of feedback on CWB is statistically different from zero, as revealed by a 95% BC bootstrap confidence interval that is entirely below zero (-.067 to -.011). The direct effect of feedback on CWB is statistically different from zero (p<.001). Since the indirect effect, as well as the direct effect are statistically different from zero, the conclusion can be

(33)

33

drawn that engagement partially mediates the relationship between feedback and CWB and therefore no support is found for H11.

Table 3

Results of analyses for Engagement as mediator between Feedback and Task Performance, Contextual Performance and CWB

Predictor Consequent R2

Coeff p LLCI ULCI

Model 1 .139 .000

Feedback Engagement .223 .000 .104 .342

Manager Engagement .837 .001 .358 1.316

Highly Educated Engagement .525 .004 .165 .886

Model 2 .070 .001

Engagement Task Performance .123 .004 .050 .206

(direct effect) Feedback Task Performance .068 .099 -.012 .146

Manager Task Performance -.006 .971 -.331 .319

Highly Educated Task Performance .140 .257 -.103 .384

Total effect .095 .020 .015 .175

Indirect effect .027 .007 .064

Model 2 .205 .000

Engagement Contextual Performance .252 .000 .174 .329

(direct effect) Feedback Contextual Performance .013 .744 -.063 .089

Manager Contextual Performance .267 .086 -.038 .572

Highly Educated Contextual Performance .164 .159 -.065 ,392

Total effect .069 .091 -.011 .148

Model 2 .104 .000

Engagement CWB -.147 .000 -.221 -.073

(direct effect) Feedback CWB -.101 .000 -.174 -.128

Manager CWB .180 .228 -.113 .472

Highly Educated CWB .068 .543 -.151 .286

Total effect -.134 .000 -.206 -.061

Indirect effect -.033 -.067 -.011

Note: N=256; all tests were two-tailed. p<.05; 5000 bootstrap samples were used to test the indirect effect Goal Setting

The results of the analyses for goal setting are shown in table 4. The effect of goal setting on engagement is statistically different from zero (p<.001) and therefore support is found for H5. The indirect effect of goal setting on task performance is statistically different from zero, as revealed by a 95% BC bootstrap confidence interval that is entirely above zero (.028 to .145). The direct effect of goal setting on task performance is not significant (p>.05). The conclusion can be drawn that engagement fully mediates the relationship between goal setting and task performance and therefore support is found for H12.

The indirect effect of goal setting on contextual performance is statistically different from zero, as revealed by a 95% BC bootstrap confidence interval that is entirely above zero (.076 to .195). The direct effect of goal setting on contextual performance is not significant (p>.05). Since the indirect effect is statistically different from zero and the direct effect is not, the

(34)

34

conclusion can be drawn that engagement fully mediates the relationship between goal setting and contextual performance and therefore support is found for H13.

The indirect effect of goal setting on CWB is statistically different from zero, as revealed by a 95% BC bootstrap confidence interval that is entirely below zero (-.145 to -.040). The direct effect of goal setting on CWB is not significant (p>.05). Since the indirect effect is statistically different from zero and the direct effect is not, the conclusion can be drawn that engagement fully mediates the relationship between goal setting and CWB and therefore support is found for H14.

Table 4

Results of analyses for Engagement as mediator between Goal Setting and Task Performance, Contextual Performance and

CWB

Coeff p LLCI ULCI

Model 1 .218 .000

Goal Setting Engagement .488 .000 .337 .639

Model 2 .071 .001

(direct effect) Goal Setting Task Performance -.099 .085 -.213 .014

Manager Task Performance .021 .897 -.303 .346

Total effect -.018 .741 -.126 .090

Model 2 .207 .000

(direct effect) Goal Setting Contextual Performance -.046 .396 -.152 .060

Highly Educated Contextual Performance .161 .166 -.067 .389

Total effect .085 .118 -.022 .191

Model 2 .078 .000

Engagement CWB -.172 .000 -.251 -.093

(direct effect) Goal Setting CWB .008 .884 -.096 .111

Manager CWB .149 .322 -.147 .445

Total effect -.076 .131 -.176 .023

Note: N=256; all tests were two-tailed. p<.05; 5000 bootstrap samples were used to test the indirect effect

Communication

The results of the analyses for communication are shown in table 5. The effect of communication on engagement is statistically different from zero (p<.001) and therefore support is found for H6.

The indirect effect of communication on task performance is statistically different from zero, as revealed by a 95% BC bootstrap confidence interval that is entirely above zero (.007 to .097). The direct effect of communication on task performance is not significant (p>.05). The

(35)

35

conclusion can be drawn that engagement fully mediates the relationship between communication and task performance and therefore support is found for H15.

The indirect effect of communication on contextual performance is statistically different from zero, as revealed by a 95% BC bootstrap confidence interval that is entirely above zero (.057 to .146). The direct effect of communication on contextual performance is not significant (p>.05). Since the indirect effect is statistically different from zero and the direct effect is not, the conclusion can be drawn that engagement fully mediates the relationship between communication and contextual performance and therefore support is found for H16.

The indirect effect of communication on CWB is statistically different from zero, as revealed by a 95% BC bootstrap confidence interval that is entirely below zero (-.083 to -.011). The direct effect of communication on CWB is statistically different from zero (p<.001). Since the indirect effect, as well as the direct effect are statistically different from zero, the conclusion can be drawn that engagement partially mediates the relationship between communication and CWB and therefore no support is found for H17.

Table 5

Results of analyses for Engagement as mediator between Communication and Task Performance, Contextual Performance and

CWB

Coeff p LLCI ULCI

Model 1 .202 .000

Communication Engagement .388 .000 .258 .519

Model 2 .070 .001

(direct effect) Communication Task Performance .081 .097 -.015 .177

Manager Task Performance .012 .941 -.312 .337

Total effect .125 .007 .034 .216

Model 2 .206 .000

(direct effect) Communication Contextual Performance .029 .529 -.061 .119

Highly Educated Contextual Performance .157 .181 -.073 .386

Total effect .124 .007 .034 .214

Model 2 .146 .000

Engagement CWB -.111 .004 -.186 -.035

(direct effect) Communication CWB -.192 .000 -.277 -.108

Manager CWB .154 .287 -.130 .439

Total effect -.235 .000 -.315 -.155

The mediating effect of employee engagement on the relationship between performance management and individual work performance

University of Amsterdam