Management control in the public sector: How does job-related tension impact employee’s motivation?

Management control in the public sector: How

does job-related tension impact employee’s

motivation?

Wilco Boerman

S3805654

w.boerman@student.rug.nl

University of Groningen

Faculty of Economics and Business

Master Thesis

Supervisor: Dr. Lucia Bellora-Bienengräber

Date: 22-06-2020

Word count: 13390

This study analyses the association between different uses of management control and work motivation, considering the moderating effect of job-related tension. The study draws Simons’ (1995) Levers of Control framework and the self-determination theory to examine the influence of the belief system and interactive system on autonomous motivation. Enhancing autonomous motivation should be of great importance for organizations, since motivation is positively associated with organizational outcomes. This study is placed in a public sector context in which the level of work stress has increased in the last decade due to undergoing reforms. In addition, this study uses role theory by Kahn et al. (1964) as supporting theory to examine if job-related tension might weaken the association between the belief system and interactive system with autonomous motivation. Provided with a secondary database of the University of Groningen containing a total sample of 259, multiple regression models are used to test the hypothesis. Finding indicate that the belief system is positively associated with one type of autonomous motivation. No evidence is found for a positive association between the interactive system and types of autonomous. Job-related tension does not moderate the association between the belief system and autonomous motivation but does moderate the association between the interactive system and autonomous motivation.

1. INTRODUCTION

“You are just always struggling to keep up”, is one of the sincere comments of a respondent from a survey in The Guardian (Dudman et al. 2015), regarding the work stress in the public sector. The article shows that at least 40 percent of the respondents feel stress in relation to their jobs and another 44 percent feels job-related stress a lot of the time, due to budget cuts and higher caseloads. The increasingly use of budgets may go hand in hand with the undergoing reforms of the public sector in the last decade. Probably the most important reform in the public sector is known as New Public Management (NPM). The cornerstone of the NPM-movement is that the use of business-like management models leads to higher organizational performance (Hood, 1995). NPM is derived and inspired by the way of working of the private sector and emphasizes the value of business-like management by e.g. setting standards and measuring performance to increase economic efficiency and transparency in decision-making (Van der Kolk et al., 2019). However, the NPM-movement is increasingly criticized. Van der Kolk et al. (2019) argue that the public sector differs too much from the private sector. For example, the performance of goals of organizations in the public sector are often harder to accurately specify than organizations in the private sector (Speklé & Verbeeten, 2014), since the goals are more focused on social and organizational aspects economic targets. Ter Bogt & Scapens (2012), show in their study that the pressure to meet the performance standards and the level of anxiety increased among employees, due to new implemented performance systems. The article in the Guardian underlines these findings by sketching the increasing level of stress experienced in the public sector, taking its toll on employees’ motivation. Well-motivated employees should of great importance for organizations, since motivation is positively associated with organizational outcomes, like job satisfaction (Deci et al., 1989) and higher employee performance (Baard et al., 2004). Beforementioned studies and the article in The Guardian show that implementing business-like controls is not always associated with better performance. The movements in the public sector that are inspired by the management philosophy of the private sector are still subject of an ongoing debate (Van der Kolk et al., 2019). This study examines the association between management control and motivation considering the work stress perceived, focused on employees of higher professional educational organizations.

because employees in this sector seem to be more attracted by autonomous motivation, rather than controlled motivations, such like extrinsic rewards (Van der Kolk et al., 2019). Besley & Ghatak (2005), argue that this could be because of the social objectives public organizations have.

The question that raises is how public organizations can maintain or increase the autonomous motivation. Organizations ensure that the interests and behavior of employees are aligned with the goals and interests of the organizations by using management control (MC) (Merchant & Van der Stede, 2007). Examples of management control mechanisms are planning systems, monitoring procedures and performance measurement systems (Henri, 2006). The information that is gathered by management controls is used for decision-making or to make adjustments where necessary. The set of management controls used by an organization is defined as the Management Control System (MCS) of the organization. A comprehensive and dominating framework to classify different forms of control is Simon’s levers of control framework (Tessier & Otley, 2012). The idea behind this framework is that different tensions in an organization exist, which should be balanced by the use of MC (Simons, 1995). For example, the amount of freedom an employee experiences are challenged by constraints and direction (e.g. standard operating procedures or rules). Simons developed four levers of control to better manage this type of tensions in an organization: belief, interactive, boundary and diagnostic. The four levers together frame the different control mechanisms that form the MCS an organization uses. This study focuses on the belief system and interactive system, because they seem to be especially relevant for maintaining autonomous motivation.

The article in The Guardian indicates the growing work stress in this sector. This is in line with the findings of Pop-Vasileva et al. (2011), which show that Australian academics experience high work stress due to excessive workloads. Work stress is often conceptualized in management control literature as job-related tension. Job-related tension expresses the individual feelings of the employee about being bothered by work-related factors (Ivancevich & Donnelly, 1974). As supporting theory this study uses the role theory provided by Kahn et al. (1964). Job conflict and job ambiguity are often conceptualized as job-related tension and are associated with a changing dynamic and nature of organizations (Wooten et al., 2010). Therefore, the theory seems especially relevant for the public sector, considering the implemented NPM and the high experienced work stress. When job-related tension is defined as the individual feelings of the employee about being bothered by work-related factors, one might expect that the relationship between MC and motivation might be weakened.

I will investigate the relationship between the employee’s perception of MCS and the employee’s autonomous motivation, considered with the moderating effect of job-related tension. The SDT and role-theory are used to ground the hypotheses. Based on the literature gap identified and explained in the introduction, the research question of this study is as follows:

RQ: What is the association between the employee’s perception of the MCS and

This study makes two theoretical contributions in the management control field. First, Wright (2001) indicates in his literature review that motivation is an understudied topic has failed to achieve interest among public sector scholars and suggests that empirical evidence of characteristics of public organizations must be combined with psychological theories of work motivation. This study complements this gap by raising understanding of different forms of MC on the autonomous motivation of employees in higher professional educational organizations. This is done by using the SDT as an empirically based approach to human work motivation and Simons’ levers of control to frame MCS. Prior research that used the levers of control have mainly focused on managerial and organizational outcomes. This study relates two levers (belief and interactive) to individual employee’s work motivation in the public sector, which to the best of my knowledge, has never been done before. Second, it is needed to gain better understanding of the task environment, MC and motivation (Van der Kolk et al., 2019). This study uses role theory as supporting theory to asses job-related tension. Interacting the levers with job-related tension is completely new but should be extremely helpful in gaining understanding in which conditions the levers are able to maintain or maybe even increase autonomous motivation. Next to theoretical contributions, the study has a practical contribution as well. It can be imagined that when employees are confronted with a strongly present MCS to enhance motivation, but at the same time are subjected to high levels of job-related tension, this enhancement can lose its ‘power’. The findings should make management aware about the impact of job-related tension and in addition take appropriate action. This could be developing management policies or creating tasks packages with a balanced workload, which would motivate employees and ensure a effectively use the MCS.

2. HYPOTHESIS DEVELOPMENT

This study investigates how different types of management controls associate with the autonomous motivation of an individual employee in the public sector. Additionally, it looks how the individual employee role in the organization moderates the association between the perception of management control and autonomous motivation. Figure 1 summarizes the study’s theoretical model. This section introduces the model by using Simons’ framework of the levers of control and explains how it draws on self-determination theory and role theory.

2.1 Management Control System

Management control should ensure that the employee’s interests and behavior are aligned with the interest of the organization (Merchant & Van der Stede, 2007). Anthony (1965) defines management control as processes utilized by management to achieve organization’s strategy by influencing the behavior of employees to organization’s interest. Although the definition of management control differences across scientific literature, they centralize a specific and important factor: the employee’s behavior. Management control offers different mechanisms in order to effect employee’s behavior, so that the application of those mechanisms ensures objectives of the organization are realized in continuity. The set of management control mechanisms may differ across organizations, because every organization has different needs and objectives.

The set of the management control mechanisms used by an organization is defined as Management Control Systems (MCS) (Merchant & Van der Stede, 2007). Simons (1995) defines a MCS as the formalized procedures and systems to collect information and use specific information to maintain or alter the organization’s activities. Examples of management control mechanisms are planning systems, monitoring procedures and performance measurement systems (Henri, 2006). The information that is gathered from the systems are used in organizational decision making. There are several ways to taxonomy MCSs, based on different criteria.

communicate the core values and direction that senior management wants their employees to embrace” (Simons, 1995, p. 6). The interactive control system “are the formal information systems that managers use to involve themselves regularly and personally in the decisions of subordinates” (Simons, 1995, p. 13). The interactive control system should minimize strategic uncertainties of an organization (Tessier & Otley, 2012). The boundary system “delineates the acceptable domain of activity for organizational participants” (Simons, 1995, p. 9), it should identify the boundaries of the territory of the organization’s participants. At last, the diagnostic system is the set of informal information systems that “help managers track the progress of individuals, departments or production facilities towards strategically important goals” (Simons, 1995, p. 4).

To define the main advantage of using Simons’ framework for this study, I will compare it to another dominating framework in management control literature, the framework of Merchant & Van der Stede (2007). Merchant & Van der Stede (2007) distinguish four types of control based on the influence management control has over personnel, culture, actions or results to analyze the behavioral aspects of organizations’ management control systems. The strength of this model is the relatively clearly distinction and recognition of organizational aspects (personnel, culture, actions and results), which are easily linkable to different management control elements (Van der Kolk et al, 2019). However, a main difference between the frameworks is that Merchant & Van der Stede analyze the behavioral aspects of an organization to provide a better understanding on MSC operations, whereas Simons’ construct incorporates organizational social aspects next to its economic approach (Hared et al., 2013). Simons’ levers of control address the fundamental problem managers face to exercise control in organizations, without losing flexibility, innovation and creativity of their employees (Simons, 1995). To conclude, Simons’ levers of control focuses on the usage of management control (e.g. belief system, rather than the types of management control which are presented in Merchant & Van der Stede’s framework (e.g., cultural controls) (Van der Kolk et al., 2019).

Van der Kolk et al. (2019) mention that the main drawback of Simons’ framework is the way the different forces created by the MCS, are represented in Simons’ levers of control. Simons distinguishes positive forces (e.g. freedom) and negative forces (e.g. constraint) in his framework. Van der Kolk et al. (2019), argue that presenting management controls in such a way could influence survey respondents, instead of presenting them in a more ‘neutral’ way. However, the word negative in the context of Simons’ framework should not be explained as controls with a “bad influence” on employees, but as important controls that guide employees’ behavior (Tessier & Otley, 2012). Simons (1995) calls this ‘the power of negative thinking’. Telling people what not to do allows for more innovation and creativity, than telling people what to do (e.g., standard operating procedures or strict rules).

2.2 Management control in the public sector

The public sector is an attractive sector to investigate because the sector has undergone MC-related developments in the last decade (Van der Kolk et al., 2019). For example, higher education institutions are undergoing reforms of marketization (Mohammadi & Mirzamohammadi, 2019), wherein the most important one is called New Public Management (NPM). Hood (1995) defines the basis of NPM as: “removing the differences between the public and private sector and shifting emphasis from process accountability towards accountability of results”. The idea of NPM is that public organizations should be more output-orientated, to increase economic efficiency and transparency in decision-making (Van der Kolk et al., 2019). NPM introduces business-like management controls, like performance measurements, which should lead to higher organizational performance (Van Helden & Jansen, 2003). However, concerning MC in higher education, various studies have investigated the implementation of NPM and its impact on the higher education field, on working environment and conditions of academics, of which several studies were highly critical of the effects of NPM in higher education institutions (Heinicke & Guenther, 2019). For example, Pop-Vasileva et al. (2011) show that Australian academics are experiencing a higher level of work stress and lower level of job satisfaction due to excessive workloads. Ter Bogt & Scapens (2012), conclude that new implemented performance measurement systems increased the pressure to meet performance standards, but also increased the level of anxiety. Literature shows that anxiety is negatively associated with job performance and productivity (Sirivastava & Krishna, 1980) and work stress negatively affects job performance (Imtiaz & Ahmad, 2009). This seems contradictory with the goal of NPM to increase organizational efficiency and performance.

organizational objectives. Therefore, more emphasis on a result-orientated view and pre-defined targets can be especially harmful for public organizations. For this reason alone, one would expect that work motivation in the public sector would be a great deal. However, Wright (2001) notices that work motivation in the public sector has failed to achieve similar interest among scholars when compared to the private sector.

Concluding, there is an ongoing debate in the public sector regarding how to balance different management controls (Van der Kolk et al., 2019) and differences between motivation of individual employees in the private sector and public sector are still unexplored (Georgellis et al., 2011). I will investigate the association between management control in the public sector (higher education) and employee’s motivation.

2.3 Motivation and Self-Determination Theory

As explained, organizations use management control to alter or maintain employee’s behavior to organization’s interests. If you look at the beforementioned definition for management control, one can see that it is important to understand what drives behavior. The driving force behind an individual’s behavior is motivation (Rabideau, 2005). Rabideau argues that motivation captures the dynamics of our behavior, which could be needs, ambitions or desire. Pinder (1998), defines motivation as an energy that has intensity, persistence and direction, which drives people to action. Ryan & Deci (2000), argue that motivation refers to the reason for doing and performing something. Motivation should be of great importance for organizations, because motivation is positively associated with important organizational outcomes. Motivation is associated with job satisfaction (Deci et al., 1989) and higher employee performance (Baard et al., 2004). Different theories emerged in literature to better understand motivation in work environments. This study will draw upon the Self-Determination Theory (SDT).

the freedom to make an independent choice and to self-determine to perform tasks (Niemiec & Ryan, 2009). The need for competence refers to the experience of individual’s behavior as effectively enacted. (Niemiec & Ryan, 2009), it expresses the possibilities of self-development and the feeling of being able to meet challenges (Van der Kolk et al., 2019). Relatedness indicates the tendency to feel connected to the social environment or people and develop a sense of belonging (Ryan & Deci, 2000). SDT centralizes the extent to which individuals are able to satisfy the three needs (Gagne & Deci, 2005). Gagne & Deci (2005), argue that promoting satisfaction of autonomy, relatedness and competence in work climates will enhance employee’s intrinsic motivation and internalization of extrinsic motivation, which results in a higher autonomous motivation.

For organizations it is particularly important to understand the type of motivation. For example, organizations use extrinsic incentives to motivate employees, which could have an undesired effect on he autonomous motivation (Gneezy et al., 2011). When an employee is autonomously motivated to perform an activity and a bonus is attached later to that activity, the employee’s motivation for the activity may shift from autonomous motivation to controlled motivation, because the employee may focus more on gaining the bonus than enjoying the task. This crowding out of autonomous motivation is known as the ‘crowding out’ effect (Gneezy et al., 2011). Crowding out of autonomous motivation is dangerous for organizations, because autonomous motivation is assumed to be more powerful than controlled motivation and when motivation shifts from autonomous to controlled, this shift can hardly be undone (Groen et al., 2017).The SDT delivers a continuum that ranges from controlled motivation to autonomous motivation, based on the level of internalization (Figure 1). Figure 1 shows different motivations that can be characterized to the degree in which they are controlled or autonomous motivated. The higher the degree of internalization, the more autonomous motivated behavior will be. Each form of motivation differs in their underlying processes and experiences (Gange & Deci, 2005).

At the far left, we find amotivation. Amotivation lacks any form of motivation. The individual has no intention or interest to complete his or her task.

Extrinsic motivation vary form as the level of internalization raises. Extrinsic motivation is needed for tasks that employees find not interesting. The completion of these tasks is dependent upon desired consequences or outcomes such as rewards or punishments. Extrinsic motivation undermines the individual’s own interest (Ryan & Deci, 2000). The first form of extrinsic motivation is external regulation, which is considered as the ‘purest’ form of controlled motivation. Gagne & Deci (2005), argue that people’s behavior is completely initiated by external demands, which actions only are intended to those ends. There is no degree of internalization at all (e.g., I am doing my homework because my parents are watching me). In absence of the external contingency, the employee will not remain engaged in the activity or eventually quit. Introjected regulation is a form of moderately controlled extrinsic motivation, where a regulation is taken in by an individual, but still does not accept it as his or her own idea. In this form, people behave in order to maintain self-worth, out of guilt, to feel worthy or to satisfy their ego’s (e.g. I do my homework because then I feel worthy towards my parents) (Gagne & Deci, 2005). The individual will still be engaged in their tasks without the presence of the external contingency, because the external contingency is taken in the person.

Further to the right, we find extrinsic motivation underlying a degree of autonomous motivation, which means that people behave in a way for their own self-selected goals. Identified regulation is the form of extrinsic motivation where people feel freedom and willingness because their actions are more aligned with their own identity and goals (Gagne & Deci, 2005). This occurs when an individual has personally identified with the importance of the behavior and accepted it as a regulation of his own, because it benefits him in achieving a self-determined goal (e.g. I do my homework to become better in math). The task is seen as personally important or useful but does not have to accompanied with enjoyment. The task is used to achieve an end to positively affect an individual’s well-being, but it is still extrinsic motivation because the activity is instrumental to achieve the goal. Integrated regulation is considered as the “purest” form of extrinsic autonomous motivation, because it has the fullest degree of internalization (Gagne & Deci, 2005). Here, the individual integrated the motivation fully in their own values and needs. Ryan & Deci (2000, p. 71), define integrated regulation as “the process through which individuals fully transform their identified values and behaviors into the self”. The drive for behavior does not come from others, but from within the person itself. The behavior is driven by an individual’s values towards extrinsic environment (e.g. I tutor math because I value good education for people who need tutoring), and these values are self-determined (Gagne & Deci, 2005). Here too, the motivation is still extrinsic, because the individual is not interested in the activity itself but sees the activity as important to serve own values.

(2000), describe intrinsic motivation as the activities that are the most important source of enjoyment and vitality in an individual’s life. Intrinsic motivation shares qualities with integrated regulation (extrinsic), because they are both autonomous. However, intrinsic motivation is driven by the emotions that emerge while performing the activity (e.g. joy), whereas integrated regulation is driven by personal values and goals and the activity is seen as an instrument to reach these values and goals (Gagne et al., 2010). Gagne & Deci (2005) state that intrinsically motivated behavior is prototypically autonomous.

Maintaining autonomous types of motivation may be particularly relevant in the public sector (Van der Kolk et al., 2019). Wright (2001), argues that employees in the public place a high value on helping others and public service, hence the core values and mission of a public organizations. This study draws upon the belief system and interactive system and relates them with autonomous motivation. 2.4 The belief system, interactive system and autonomous motivation

I argued earlier that the difference of being autonomous or controlled motivated depends on the degree of internalization and the satisfaction of the three psychological needs. Like previously explained, the SDT views external rewards or interventions as undermining the autonomous motivation due to the crowding-out effect. However, Frey (2012) argues that the opposite of this effect could also appear, the so-called ‘crowding-in effect’. This occurs when external control is perceived as supportive and the SDT claims that a need-supportive environment can foster the autonomous motivation by satisfying the three psychological basic needs. Van der Kolk et al. (2019), argue that various types of management control may sustain or hinder the fulfillment of the need of autonomy, competence and relatedness and may either contribute or obstruct a need-supportive environment. Van der Kolk et al. (2019) point out that this could be relevant especially for the public sector, because employees are more attracted by autonomous motivation than by extrinsic rewards. The belief and interactive system are defined by Simons (1995) as controls that motivate and reward employee’s behavior and that promote learning. According to the SDT, these are the levers that would sustain the fulfillment of the need of autonomy, competence and relatedness and would therefore foster the need-supportive work environment perceived by the employee, creating the crowding-in effect. I expect that the belief system and interactive system play an important role in fostering autonomous motivation and therefore will only focus on the belief and interactive systems in this study, which will be further elaborated.

they feel more self-regulated and autonomous. Organizations could use management and employee meetings to communicate their mission and vision to level with their employees and increase feasibility. Feasibility leads to higher internalization of employees, which leads to higher autonomous motivation (Gagné, 2010). The belief control system will foster more autonomous motivation than controlled motivation, because the belief system controls foster the need of autonomy, relatedness and competence of an employee. Thus, I hypothesize:

H1: The employee’s perception of the emphasis given to the belief control system is positively associated with autonomous motivation.

Simons (1995) describes the interactive control system as are the formal information systems that managers use to involve themselves regularly and personally in the decisions of subordinates. Through the interactive control system, senior managers participate in the decision-making process of subordinates to focus attention on key strategic issues (Tekavčič et al., 2008). The system should facilitate cross-hierarchical communication in all levels of the firm. This ensures that there is frequent dialogue between employees and managers to align the organization and to look in the same direction (Widener, 2007). This increases the participation of employees in an organization. Groen et al. (2017), show that employee participation in the budgeting process can increase the autonomous motivation of employees, because the feeling of relatedness is satisfied through the social interaction. Also, the interactive control system should be organized in such a way that is not obstructive for subordinates. It should create a ‘positive informational environment that encourages information sharing between organizational actors and does not usurp the decision rights of the subordinates’ (Simons, 1995; Simons, 1987). In view of SDT, this would increase the feeling of relatedness towards the organization and would not violate the feeling of autonomy of the subordinates, because the rights of these subordinates should not be affected when the interactive controls are well used. The interactive control system will foster more autonomous motivation than controlled motivation, because the belief system controls foster the need of autonomy, relatedness and competence of an employee. Thus, I hypothesize:

H2: The employee’s perception of the emphasis given to the interactive control system is positively associated with autonomous motivation.

2.5 The moderating effect of job-related tension

political and legal constraints than the private sector, because the public sector is predominately concerned with regulation and has the goal of maintaining or improving social and economic conditions (Macklin et al., 2006). After the implementation of NPM in the public sector, one would expect that the two sectors would be more comparable, because the public sector should be more output orientated. However, Steen & Schott (2019) argue that the implementation of NPM added an extra challenge in the working environments, because now employees in the public sector need to balance the ‘traditional’ public values (e.g. integrity, neutrality and legality) and business values (e.g. output and efficiency). This is supported by Pop-Vasileva et al. (2011), which show that Australian academics are experiencing a higher level of work stress and lower level of job satisfaction due to excessive workloads after the implementation of NPM.

In most organizational research, stress in working environments is been conceptualized as job-related tension (Wooten et al, 2010). Job-job-related tension expresses the individual feelings of the employee about being bothered by work-related factors (Ivancevich & Donnelly, 1974), by tensions arising from circumstances in the job environment (Kenis, 1979). An often-used theory that is concerned with organizational stress is the ‘role theory’ by Kahn et al. (1964), a dominating model in psychological organizational literature. The role theory describes the roles of individuals in the organization, which could give insights in the job environment of an organization. Unclear roles can lead to role conflict and role ambiguity. Role conflicts arise when the role has responsibilities or requirements that are incongruous or contradictory with the expectations from within or outside the individual’s role (House & Rizzo, 1972; Kahn et al, 1964). When the role demands are vague or/and the expectations of the role are unpredictable, role ambiguity arises (Schuler, 1981). Examples of causes of role ambiguity are changing environments of organizations, technological change or rapid growth (Lyons, 1971). These events cause that the expectations or demands of the role of the employee changes. In research, role conflict and role ambiguity are often captured in the variable job-related tension (Wooten et al, 2010), which will be the variable in this study.

job-related tension does perceive his or her environment as need-supportive, so that crowding-effect can occur due to the belief and interactive system.

To conclude, according to role theory, it is expected that the degree of job-related tension affects the association between the variables in hypotheses 1 and 2. The belief system and interactive system may be associated with responses differently under different degrees of job-related tension. For instance, the influence of a belief system or interactive system may be more positively/negatively associated with autonomous motivation when there is less/ more job-related tension, by whether the crowding-in effect occurs or not. Accordingly, two hypotheses are developed to examine the moderating effect of job-related tension. I hypothesize:

H3: Job-related tension negatively moderates the association between the employee’s perception of the belief control system and autonomous motivation.

H4: Job-related tension negatively moderates the association between the employee’s perception of the interactive control system and autonomous motivation.

3. METHODOLOGY

The aim of this study is to examine the effect of the belief system and interactive system perceived by the employees in the public sector on the autonomous motivation, and how this association is moderated by the effect of job-related tension. The method follows a quantitative approach since this study aims to test the hypotheses based on numerical survey data.

3. 1 Data collection and sample characteristics

For testing the proposed hypotheses, a secondary dataset is used. The data is collected from an existing database that is provided by researchers of the University of Groningen. Data is gathered about

management control- and working environment aspects, by more than 500 managers and employees of four higher professional educational public organizations in The Netherlands. The organizations are selected from the researchers’ personal network which provided the opportunity to distribute the surveys among employees of the public organizations. The survey is constructed by adopting constructs and measurements of existing literature. The items in the database represent different management controls (e.g., the levers of control by Simons (1995)), contingency variables (e.g., strategy and culture) and employee motivation, expressed on the continuum presented by the SDT (e.g., amotivation, controlled motivation and autonomous motivation).

The program Qualtrics is used to conduct the survey. A preliminary questionnaire is pre-tested by colleagues and professors of the University of Groningen in the field of management control. This questionnaire is revised based on their input, resulting in a survey consisting of 28 items, consisting of multiple sub-questions. Before sending the surveys, the permission of the boards of directors are gained. Due to Dutch privacy regulations, it is not possible to contact or send the surveys without the permission of the board. The respondents’ anonymity of the respondents is assured due to privacy regulations as well. This is communicated to the respondents in the email with the link to the survey. E-mailing the surveys resulted in more than 500 completed surveys (response rate: organization 1 75%; organization 2 unknown; organization 3: 20% and organization 4: 20%). The average completion time of one survey is between 20 and 30 minutes. Due to confidentiality reasons, this research uses 259 respondents of one higher professional educational public organization from one higher professional educational public organization as sample.

3.2 Procedure

The statistical procedures and tests of this study are performed in IBM SPSS version 24. Before testing the hypothesis, key assumptions in the dataset are checked to examine whether the dataset is suitable for further analysis.

Missing values. The items of the survey are measured with the use of scales. The Likert-scale is a commonly used measurement instrument to measure attitude and uses a range of answer possibilities to a given question, ranging in a particular order (Jamieson, 2005). For example, a Likert-scale used in this study for the items measuring the employee’s perception of the belief system varies from “to an extremely low extent” (1) to “an extremely high extent” (7). However, some items in the dataset have the answering possibility “does not apply (DNA)”, where the respondent has the opportunity to indicate that the item is not applicable to him or her. Hair et al. (2018), describe that questions that are not applicable can be seen as missing values because respondents do not answer the question and thus create missing data. Hair et al. (2018), offer in their book different remedies to resolve missing values. At first, the extent of the missing values is checked, because variables with as little as 15% missing data or less could be candidates for deletion (Hair et al., 2018). Higher levels of missing

TABLE 1

SAMPLE CHARACTERICS

Variable Frequency Percentage Variable Frequency Percentage Gender

Respondents

259 100% Educational background

259 100%

Male 134 51.74% Secondary Vocational 25 9.65%

Females 125 48.26% Secondary education 3 1.16%

Bachelor’s degree 81 31.27%

Language 259 100% Master’s degree 150 57.92%

NL 220 84.94%

EN 39 15.06% Job Activities 259 100%

Educational staff 141 54.44%

Age of respondents 259 100% Support staff 118 45.56%

25 - 35 42 16.22%

36 - 45 62 23.94% Type of contract 259 100%

46 - 55 78 30.12% Fixed contract 237 91.50%

56- 65 76 29.34% Temporary contract 22 8.50%

data should be remedied. In this study, 37 cases answered one time or more DNA on the 15 JRT-items, which is about 14% of the total sample. However, I find deletion not desired, because 14% is a significant part of the sample and all of the remaining data of the 37 cases will be lost. Second, it is examined whether the missing values can be classified as “missing at random (MAR)” or “missing completely at random (MCAR)”. The difference between MCAR and MAR is that MAR has no relationship between the missing values and the data observed or missing, whereas MCAR has a relationship between the missing values and the data observed. An independent sample t-test is performed by forming two groups: cases that answered DNA one time or more on the 15 JRT-items (group 1) and cases that answered valid values (group 2). No significant differences between the means of the two groups were found. Afterwards, the Little’s MCAR test is performed as a second approach to test whether the pattern of the missing values off all items is comparable with the expected pattern for a random missing data process (Hair et al., 2018). The Little’s MCAR showed to be significant (p < 0.05) in untabled related results, and thus the missing values cannot be classified as MCAR. Because the t-test found no significant differences between the two groups and the Little’s MCAR was significant, the missing data can be classified as MAR (Hair et al., 2018). Third, based on the classification, the most appropriate remedy offered by Hair et al. is selected to resolve the missing values in the data. Because a MAR missing data pattern is found, the Expectation Maximization (EM) estimation is applied in this study. Any other remediation method would result into bias (Hair et al., 2018). The EM method makes the best possible estimates for the missing values and then estimates the parameters again. This is an ongoing process until the best estimated value for the missing value is negligible and replace the missing value with another value used in the Likert-scale.

Non-response bias. This study controls for response bias. Lahaut et al. (2013), define non-response bias as the bias that occurs when the respondents and the non-respondents of the survey differ in the outcome variable. The consequence of this bias is that the parameters of the population can be over- or underestimated. However, this study has no available data on non-respondents. A widely used alternative is using late respondents as a proxy for non-respondents and compare them with the early respondents (Lahaut, 2013). At first, the average starting date of the respondents is calculated. Respondents who finished the survey before the average starting date are marked as ‘early respondent’ (group 1) and respondents after the average starting date are marked as ‘late respondent’ (group 2). This resulted in 178 early respondents and 81 late respondents. Second, an independent sample t-test is conducted for all items in the survey that are used to measure the constructs. This resulted in no significant differences (p < 0.05) in the means of the items between the early respondents and the late respondents (Appendix A). No evidence for non-response bias is found.

particular moment (Jakobsen & Jensen, 2015). Common method bias is often conceptualized as common method variance, which is the variance that is attributable to the used measurement method, rather than the construct that is represented by the measurement (Podsakoff et al., 2003). Due to Dutch privacy regulation the survey data is anonymous. An anonymous survey should mitigate common method bias in responses when employees are asked sensitive questions about the organization or their own position in the organization. When respondents are anonymous, it is less likely that responses are edited in a way with how respondents think that the researcher wants them to respond or in a socially desirable way (Podsakoff et al., 2003). Another way to detect common method bias is using a statistical approach to test for common method variance, which is tested by using the Harman’s single factor test (Podsakoff et al., 2003, p. 899): “an unrotated factor analysis is performed and based on the results there are two basic assumptions: (1) either a single factor emerges from the analysis or (2) one single factor will account for the majority of covariance”. An unrotated factor analysis showed that there are seven factors emerging. When pre-setting that the factor analysis should extract one general factor, in untabled related results about 24% of the total variance is explained by that factor. This result is far below the agreed threshold of 50% (Eichhorn, 2014). Based on the two unfilled assumptions no evidence for common method bias is found.

Normality. The normality of the constructs used in the theoretical model is tested. The reason to do this is that statistical procedures have been derived based on the underlying distribution of the data, especially procedures based on normal distributions (Shapiro & Wilk, 1965). The distribution of the data indicates which tests should be used. First, a Shapiro-Wilk tests is performed to test if the distribution of the constructs is significantly different to a normal distribution. The test shows significant outcomes (untabled) for all four constructs, which means that the distribution of the constructs is significantly different to a normal distribution. However, Field (2005), argues that the Shapiro-Wilk test has a big limitation, because with large samples it is very easy to get significant outcomes, which is due to small deviation from normality. Because of this, Field (2005) strongly recommends plotting the data to check for normality by using Q-Q-plots. The Q-Q-plot shows the values you would expect (normal) against the values that are actually seen. Second, Q-Q plots are created to check for normality (Appendix B). Mostly all values of the constructs follow the normal distribution line in the Q-Q-plots. At last, the univariate kurtosis and skewness of the constructs are calculated to indicate whether the complete set of indicators departs from a normal distribution. The skewness indicates the lack of symmetry of the distribution, whereas the kurtosis indicates if the distribution is left- or right-tailed. The agreed threshold for univariate skewness is between -2 and 2 and the univariate kurtosis -7 and 7 (Kline, 2011). Based on the large sample size (n = 259), the limitation of the Shapiro-Wilk test with large samples and the presented Q-Q-plots, it is assumed that the data of the constructs are normally distributed.

significance of p < 0.05. Second, the underlying constructs of survey items is uncovered to examine the construct validity. An exploratory factor analysis is performed to explore whether the survey items actually measure the construct for which they are intended (Appendix C). The Direct Oblimin-rotation is used because the items of the independent variables are expected to correlate, supplemented with the Kaiser normalization, which means that components with Eigenvalues below 1 are eliminated. The model fits when the factor loadings with their component are high enough. Five items of job-related tension had to be deleted due to their low factor loadings with their component or high factor loadings with other components (see Appendix C). Third, the internal consistency of the questions per construct are assessed. This is done by calculating the Cronbach’s alphas, i.e., assessing how close the items are related within the group that measures the variable. The Cronbach’s alpha should be higher than 0.70 for the construct measurement to be internally reliable, otherwise it cannot be used for statistical analysis (Nunnally, 1978). The Cronbach’s alphas of the constructs exceed Nunnally’s threshold of 0.70 and are reported in the measurement section, which is the following section. At last, the data is checked for multicollinearity. Multicollinearity occurs when what to be appear separate variables are actually measuring the same thing (Kline, 2011). The data is checked for multicollinearity by using the regression analysis approach, which calculates the Variance Inflation Factors (VIF-values) (Kline, 2011). Kline argues that these values should be lower than 10, otherwise the variable is redundant. All VIF-values of the tested constructs of the theoretical model were lower than the threshold of 10, with figures between 1.13 and 1.2. No evidence for multicollinearity has been found.

3.3 Measurement

correlated with each other, they are all integral parts of JRT. Jarvis et al. (2003), argue that such constructs can be treated as unidimensional, which is the case in this study. Therefore, JRT will be treated as reflective as well. The variables used in the theoretical model are created by using factor scores, which follows a reflective logic. The specification of the variables is following. A robustness check is also conducted using average scores.

Independent variables

Belief system. One of the independent variables, the employee’s perception of the belief system, is measured in the way Kruis, Spekle and Widener (2016) describe in their research. This measurement is based on four questions as developed by Widener (2007), regarding the mission statement and core values of the organizations. The respondents are asked four questions to measure the perception of an employee of the belief system with statements like: “The mission statement clearly communicates the organization’s core values to me”. The questions are measured on a fully anchored seven-point semantic scale, varying from 1 (to an extremely low extent), till 7 (to an extremely high extent). The second item (BELIEF 2) had to be deleted due to its high factor loading with the interactive system in the factor analysis. The Cronbach’s alpha of the three remaining questions related to the belief system is 0.81, so the questions are considered as internal consistent to measure the perception of the belief system.

Interactive system. The employee’s perception of the interactive system is captured by the way Bedford & Malmi (2015) report in their research, which is based upon the measurement model of Bisbe et al. (2007). The measurement is based on five different statements in how performance measures are used, regarding: intensive use of top management, intensive use by operating management, face to face challenge and debate, focus on strategic uncertainty and facilitating involvement. The statements about intensive use of top management and intensive use by operating management is captured as one question in this study. Based on these five properties, respondents are asked to give answers varying from 1 (to an extremely low extent), till 7 (to an extremely high extent) to four statements like: “performance measurements are used to provide a frequent agenda to management activities”. The Cronbach’s alpha of the four questions related to the employee’s perception of the interactive system is 0.90.

Dependent variable

regulation. Identified regulation, integrated regulation and intrinsic motivation represent autonomous motivation (Gagne et al., 2010). Because integrated regulation is not captured in this study, making one construct value for autonomous motivation is not ideal, since it would not represent all types of autonomous motivation and would not follow literature. In addition, each discrete type of motivation reflects different reasons for behaving (Gagne & Deci, 2005). Combining different types of motivation would lose valuable information and does not make fully use the SDT-continuum. Ryan & Deci (2000) argue that intrinsic motivation is highly autonomous and can be seen as the prototypic instance of self-determination. Gagne & Deci (2005) state that intrinsically motivated behavior is prototypically autonomous. To summarize, intrinsic motivation can be seen as the “purest” form of autonomous motivation. Because this study investigates the association between the employee’s perception of the belief/ interactive system and autonomous motivation, the survey questions of intrinsic motivation are taken into consideration to measure autonomous motivation. The intrinsic motivation is captured by three statements such as: “the work I do is interesting”. Respondents are asked to rate these statements on a seven-point Likert scale, varying from 1 (not at all) till 7 (completely/ entirely). The Cronbach’s alpha of the three questions related to intrinsic motivation is 0.91. To capture the other type of autonomous motivation, the entire main analysis is replicated with identified regulation as dependent variable instead of intrinsic motivation as robustness check. The Cronbach’s alpha of the three questions related to identified regulation is 0.84. Found differences between intrinsic motivation and identified regulation will be discussed in the discussion section.

Moderating variable

Job-related tension. The moderator variable, job-related tension, is measured according to the measurement provided by Wooten et al. (2010). The authors provide a job-related tension index (JRTI), based on the 15-item JRTI provided by Rogers et al. (1994). The JRTI is based on the role theory provided by Kahn et al. (1964), which is explained in the literature review. The factors are based on the conceptual framework of role theory. Based on the 15-item JRTI, 15 survey questions are developed. Each question is measured using a six-point Likert scale, varying from 1 (“never”) to 5 (“always”) or 6 (“does not apply”), that capture statements like: “I know what is expected from me”. Three items had to be removed, due to their low level of correlation with this construct in the factor analysis (JRT 4, JRT 6, JRT 9, JRT 13). The Cronbach’s alpha of the remaining 11 questions related to job-related tension is 0.87.

Control variables

Boumans et al. (2011), argues that people until the age of 44 are striving for advancement and try to find a niche where they can accelerate. However, after the age of 44, employees are more interested in maintaining their position because of changing environment or competition from younger employees. In this study, it is assumed that age will influence the employee’s perception of the belief and interactive system and therefore will control on age. Age will be measured in number of years.

Type of contract. The motivation of the employee is impacted by the employee’s relation to the organization (Grund & Thommes, 2017). Their study found that contract type moderates the relationship between public service motivation (intention of “doing good for society”) and being employed in the public sector. For example, part-time contracts negatively moderate the relationship between employment in the public sector and motivation, because social involvement of employees with a part-time contract in the organization may be limited. In this study, it is assumed that the type of contract will influence the autonomous motivation of employees and therefore will control on the type of contract. A statement is developed in the survey about the type of contract the employee has, where 1 is ‘temporary contract’ and 2 is ‘fixed contract’.The above-mentioned control variables are considered in the hypothesis testing, only if the control variables have significant correlation with the dependent variable and/ or the moderating variable. Otherwise the control variables are not taken into account and therefore not used further in the study.

TABLE 2 Description of variables

Label Name Measurement

BELIEF Employee’s

perception of the use of the belief system

7-point Likert Scale, where 1 is “to an extremely low extent” and 7 is “to an extremely high extent”

INTER Employee’s

perception of the use of the interactive system

7-point Likert Scale, where 1 is “to an extremely low extent” and 7 is “to an extremely high extent”

AUTMOT Level of employee’s autonomous

motivation

7-point Likert Scale, where 1 is “not at all” and 7 is “completely/ entirely”.

JRT Level of employee’s job-related tension

6-point Likert Scale, where 1 is “never”, 5 is “always” or 6 “does not apply”.

AGE Age of respondent Age in years.

3.4 Data analysis

The hypothesis testing will be performed in IBM SPSS 24. Before testing the hypothesis, an examination of the descriptive statistics will be performed per variable to get more statistical knowledge about the variables. Thereafter, the data will be further analyzed by performing the Pearson-correlation test and investigating the correlations between the variables to explore the direction of the association between the variables. According, the hypotheses are tested by using multiple regression. At first, a regression analysis is performed to investigate the direct interaction of the control variables, age and type of contract, with employee’s autonomous motivation. Second, multiple regression is performed to test the direct effect of the employee’s perception of the belief system and interactive system with employee’s autonomous motivation, which are hypothesis 1 and 2. Third, a multiple regression analysis is performed to investigate the interaction of the moderating effect of job-related tension with respect to the association between employee’s perception of the belief system/ interactive system and autonomous motivation. At last, robustness checks are performed to test how conclusions change when assumptions or statistical procedures change.

4. RESULTS

This section reports the findings of the data analysis described in the method section. At first, the descriptive statistics are reported. After, a correlation table of the constructs of the theoretical model will be presented and discussed. According, the results of the multiple regression analysis will be presented and explained.

4.2 Descriptive statistics

(Van der Kolk et al., 2019). The scores on autonomous motivation are in line with the beforementioned expectations. What is remarkable, is that the mean of job-related tension variable is relatively low, with a mean of 2.32 and a mode of 2.00. The previously stated held notion was that stress levels in the working environment of the public sector would be higher than the working environments in the private sector (Macklin et al., 2006). Although private organizations are not included in this study, a higher job-related tension score would be expected.

4.3 Correlation analysis

Table 4 presents the correlations between the constructs, which aims to show the existence between different concepts, respectively to examine whether change in one concept is met with similar change in another concept. A correlation analysis is an important step to make before conducting the regression analysis. This is done with the use of the factor scores mentioned in the method section. Again, the control variable ‘type of contract is left out’, due to its nominal character of data, calculating a correlation with this variable would not be appropriate.

At first, the individual constructs of belief system (r=.19, p< 0.01) and interactive system (r=.11 p<0.10) are significantly positively correlated with autonomous motivation. These results are in accordance with the expectations described in the theory section, where the expectation is created that the belief system and interactive system are positively associated with autonomous motivation. However, the correlations are not considered as strong, on the contrary, they will be probably sooner identified as weak correlations. Furthermore, the belief system variable and interactive system variable are significant positively correlated (r=.33, p<0.01), which is in accordance with literature as well. Kruis

TABLE 3

DESCRIPTIVE STATISTICS

et al. (2015) argue that the levers are expected to complement and balance each other.Job-related tension negatively correlates with the constructs belief system and interactive system (r=-.19, p<0.01; r=-.21, p<0.01), as well as with the variable autonomous motivation (r=.30, p<0.01). Age only correlates significantly with the interactive system (r=.13, p<0.05), but this correlation cannot be expressed as strong. Overall, it can be concluded that the correlations follow the expectations sketched in the theory development section.

4.4 Regression analysis

The correlation analysis, presented in table 4, presents significant correlations between the constructs of the theoretical model. To further examine the hypothesis, multiple regression analyses are performed to tests the relationship between the constructs, as defined in the hypotheses development section. Different models are defined to analyze the hypotheses. Model 1 (control variables), only tests the association between the control variables and autonomous motivation to examine whether the control variables (age and type of contract) are alternative explanations before testing the hypotheses. Model 2 (main effects), tests the first two hypothesis, which examines the expected positive association between the belief system/ interactive system and autonomous motivation. Model 3 (moderating effects), tests the third and fourth hypotheses, examining the expected negative moderating effect of job-related tension on the association between the belief system/ interactive system and autonomous motivation. Job-related tension will be included in model 3 as main effect. F-tests tested the fit of the models with respect to the data. Model 1 turned out to be not significant, where model 2 and 3 were significant (p < .01; p < .10). Therefore, evidence is found that model 2 and 3 fit to the data, which is not the case for model 1. The multiple regression results are presented in table 5.

Model 1 - control variables. Model 1 only included the control variables ‘age’ and ‘type of contract’ to test their association on autonomous motivation. It was expected that age and type of contract could be alternative explanations while testing the hypotheses. The correlation analysis showed

TABLE 4 CORRELATION STATISTICS Variable 1 2 3 4 5 1. BELIEF x 2. INTER .33*** _x 3. AUTMOT .19*** _.11* _x 4. JRT -.19*** _-.21*** _-.30*** _x 5. AGE -.02 -.13** _{.02 -.08 x}

that age and type of contract do not significantly correlates with autonomous motivation. Model 1 shows that there is no significant relationship between age and type of contract with autonomous motivation. This study theorized that age and type of contract would be associated with the employee’s autonomous motivation, but no evidence has been found whatsoever. Because of the non-significant correlations with autonomous motivation, the non-significant associations between the control-variables and autonomous motivation and the non-significant F-test, both control variables will not be included in further analysis for simplicity reasons.

Model 2 - main effects. Model 2 tests the first and second hypothesis. Hypothesis 1 hypothesized that the employee’s perception of the belief system is positively associated with autonomous motivation, whereas hypothesis 2 hypothesized a positive association between the interactive system and autonomous motivation. For belief system a significant positive association with autonomous motivation is found (β = .17, p < .01). Evidence is found that the belief system is positively associated with autonomous motivation, hence hypothesis 1 is supported. No evidence is found for the hypothesized positive association between the interactive system and autonomous motivation. Because no evidence is found, hypothesis 2 should be rejected. However, the percentage of variance of the independent variable explained by model 2 is little (adjusted R2 _{= .03).}

job-related tension and autonomous motivation (β = -.28, p < .01). This is in line with the results found in the correlation analysis, where job-related tension was significant negatively correlated with autonomous. Model 3 explained more percentage of the variance than model 2 (adjusted R2 _{= .15). This} is possibly due to the added job-related tension construct as main effect in model 3.

TABLE 5

REGRESSION ANALYSIS

Autonomous Motivation Variable Model 1 Model 2 Model 3 Control Age .00 (.01) Type of contract .12 (.23) Main effects Belief system .17*** .09* (.07) (.06) Interactive system .05 .04 (.07) (.06) Job-related tension -.28*** (.07) Two-way interaction

belief system x job-related tension .16

(.06) Interactive system x job-related tension -.16***

(.06)

F-value .20 5.08*** 8.60***

R-square .00 .04 .15

Adjusted R-square .00 .03 .13

4.5 Robustness check

Identified regulation as dependent variable. As explained in the method section, autonomous motivation is captured by the MAWS with two types of motivation: intrinsic and identified regulation. In this additional analysis the entire analysis is replicated with identified regulation, instead of intrinsic motivation. The replication follows the same approach as the main analysis, e.g. using factor scores and one-tailed testing. The correlation results are shown in table 6. As in the main analysis, the belief system and interactive system are correlated (r = 0.33, p < .01). The belief system, interactive system and job-related tension show insignificant correlations with identified regulation. The regression analysis is shown in table 7. Following main model 3, no significant association is found. This differs from the main analysis, where a significant association for intrinsic motivation was found with job-related tension, the belief system and an interaction variable (interactive system x job-related tension). When identified regulation is used as dependent variable, all four hypotheses of this study are rejected.

TABLE 6

CORRELATION STATISTICS – ROBUSTNESS CHECK

Variable 1 2 3 4 5 1. BELIEF x 2. INTER .33*** _x 3. IDENTIFIED .09 -.03 x 4. JRT -.23*** _-.20*** _{-.07 x} 5. AGE -.03 -.12** _{.00 -.08 x}

Variable measurement. Jarvis et al. (2003), argue that measurement model misspecification is fairly pervasive across published literature, which could lead to different conclusions. This research used factor scores as construct variables to tests the hypotheses, which follows a reflective logic as explained in the method section. To test the robustness of the models, the regressions are rerun. The construct values are created by averaging the item scores belonging to that specific construct. This would follow a formative logic. The results are presented in appendix D. Based on the final model, no new significant or insignificant outcomes are found which means the results are qualitatively unchanged. Based on the results of this robustness test, there is no difference in the acceptance or rejection of the hypotheses tested between the main analysis and this robustness test.

TABLE 7

ADDITIONAL ANALYSIS – IDENTIFIED REGULATION Autonomous Motivation

(identified regulation) Variable Model 1 Model 2 Model 3 Control Age .00 (.01) Type of contract .47*** (.23) Main effects Belief system .12** .05 (.07) (.07) Interactive system -.07 -.06 (.07) (.06) Job-related tension -.05 (.07) Two-way interaction

belief system x job-related tension .19

(.07) Interactive system x job-related tension -.01

(.06)

F-value 2.15 1.54 2.80**

R-square .01 .01 .05

Adjusted R-square .00 .00 .03

5. DISCUSSION AND CONCLUSION 5.1 Summary of research

The goal of this research is to investigate the associations between employee’s perception of MCS and motivation in the public sector and to investigate whether job-related tension negatively moderates this association. This study used responses from 259 respondents of one higher professional educational organization. Based on their perceptions and responses, this research tries to answer the research question: What is the association between the employee’s perception of the MCS of the organization and employee’s autonomous motivation in the public sector, and how does job-related tension moderate this association? To answer this research question, the question is decomposed to four different hypotheses.

At first, it was hypothesized that the belief system and interactive system positively associate with employee’s autonomous motivation. Findings show indeed that the belief system is positively associated with autonomous motivation. This implies that the stronger the belief system is perceived by the employee, the more autonomously motivated the employee will be. Furthermore, it was expected that the interactive system and autonomous motivation should associate as well. No evidence indicated that this expectation would exist, since no significant positive association is found. Second, it was hypothesized that job-related tension would negatively moderate the associations between the belief system/ interactive system and autonomous motivation. Findings show that job-related tension does not negatively moderate the association between the belief system and autonomous motivation. However, the interaction term between job-related tension and the interactive system was significant, implying that job-related tension negatively moderates the association between the interactive system and autonomous motivation. The above applies when intrinsic motivation is used as autonomous motivation.

The robustness checks showed that when identified regulation is used as autonomous motivation, all four hypotheses would be rejected. The belief system and interactive system are not positively associated with identified regulation. Job-related tension does not moderate this association either. Concluding, the effect of the two levers on autonomous motivation depends on which type of autonomous motivation is applied for an individual employee.

5.2 Discussion

system on the intrinsic autonomous motivation of the employee will be enhanced, no matter how much job-related tension is felt by the employee.

The same approach was theorized for the interactive system, which is the collection of controls that managers use to involve themselves regularly and personally in the decision of subordinates. Involving employees in decision-making processes was expected to satisfy the need for relatedness A moderating effect is found of job-related tension on the association between the interactive system and intrinsic motivation. This implies that when the level of job-related tension rises, the association of the interactive system on intrinsic motivation is influenced negatively. However, no association is found between the interactive system and types of autonomous motivation. A possible explanation is that the association could be significant for different subgroups in the sample, but that the association is not significant for the whole population. For example, the interactive system may be used differently across departments in the organization, since different subordinates are involved. It is possible that job-related tension is perceived as high in one department but as low in another department, due to different roles.

A possible explanation for finding differences in the significance of the main effects of the belief system and interactive system on intrinsic motivation is, is how and why the systems are used, and how this usage is perceived by the employee. The interactive system should focus on identifying opportunities for strategy to emerge and top managers are focused on emerging new strategy and ideas (Kruis et al., 2016). However, the belief system is concerned to share the strategy as a perspective, it is already “aimed” to engender the feeling of relatedness by employees. Top management uses the belief system to ingrain the firms’ vision deeply throughout the organization, with the use of controls like the mission statement and core values (Kruis et al., 2016) Wright (2001), argues that employees in the public sector place a high value on helping others and public service, hence the core values and mission of a public organizations. This could be a possible reason that the belief system fulfills the need of relatedness better, whereas for the interactive system involving employees in the decision-making process is not enough to fulfill the need of relatedness.

However, although it was not hypothesized, a direct significant negative association is found between job-related tension and intrinsic autonomous motivation. Wani (2013), argues that not many studies have been conducted to examine the association between job tension and employee’s motivation. Wani (2013), finds that an increased level of job stress leads to a decrease in motivation level of employees. Potential causes for this relationship are lesser scope for personal growth, underutilization of abilities and ambiguous organizational policies. The findings of Wani (2013), substantiate the found significant negative association between job-related tension and autonomous motivation in this study. 5.3 Contributions