Management controls in a
knowledge-intensive context: the effect on autonomous
motivation
“An assessment of higher professional educational organizations”
By:
Daan Jan Jacob Winter
S3840557
D.j.j.winter@student.rug.nl
Master Thesis
MSc. BA - Management Accounting and Control
Supervisor: prof. dr. ir. P.M.G. (Paula) van Veen-Dirks
University of Groningen
Faculty of Economics and Business
Date: 22-06-2020
Keywords: Management Control, Self-determination Theory, Job Characteristics Model,
Autonomous Motivation, Knowledge Characteristics of Tasks, Levers of Control
2
Management controls in a knowledge-intensive
context: the effect on autonomous motivation
ABSTRACT
3
1. Introduction
Optimal resource utilization is the utopia of every manager, in both the private and public sectors. In knowledge-intensive1 organizations, such as higher professional educational organizations (HPEO), employees are the most valuable resources. Since autonomous motivation of employees is positively related to job performance (Gagné and Deci, 2005; Chen, Lill, and Vance, 2019), optimization of autonomous motivation will be desired to obtain this optimal resource utilization. Ryan and Deci (2000, p.69) stated: “Motivation is highly valued because of its consequences: Motivation produces”. Literature suggests that the way management controls are perceived affects employee motivation and creativity because MCs can be perceived as controlling or enabling (Mundy, 2010), whereas the first constricts and limits employees, which harms the sense of autonomy, and the latter stimulates and encourages employees, which makes them believe they have control over their actions and enhances the sense of autonomy. Chenhall (2003) requested scholars to focus attention on contemporary dimensions of MC, context and organizational and social outcomes. Van der Kolk, Van Veen-Dirks and Ter Bogt (2019) did such an in-depth study and showed that the four different types of MCs (framework of Merchant and Van der Stede (2007)) have different effects on the motivation of employees in the public sector. They subsequently pointed out that examining the effect of the task complexity on the relation between MC and motivation could enhance the understanding and fill a gap in literature. Therefore, I aim to fill the gap and enhance the understanding of the effects of MCs by examining the effect of the four levers of control (Simons, 1995) on autonomous employee motivation in HPEOs, and how the knowledge characteristics of their tasks affect these relations. Studies argue that when tasks become more complex, other levels of autonomy are demanded (Ghosh, 2014; Sia and Appu, 2015). Since MC have the potential to facilitate autonomy, the context with knowledge-intensive jobs is interesting.
The Levers of Control (LoC) framework of Simons is assumed to be most appropriate because of the dichotomy of positive and negative controls, and their different effects on autonomy. Because autonomy is essential for autonomous motivation, this framework offers the opportunity to argue, conclude, and advise based on this dichotomy. Although there has been some criticism in management control research (Ahrens and Chapman, 2004; Bisbe et al., 2007; Ferreira and Otley, 2009), the levers of control framework of Simons (1995) has gained a prominent position, with 4564 citations in Google Scholar at June 20th, 2020. He distinguishes the controls based on their “purpose for managers attempting to harness the creativity of employees’’ (Simons, 1995) because he thinks that businesses, who are in a competitive environment, need their employee's initiatives and opportunity-seeking behavior to survive.
1 Knowledge-intensive applies to organizations in which knowledge has more importance than other inputs and
4
In motivation and management related literature, the Self-Determination Theory (SDT) is often used to assess the source and type of motivation. The SDT approaches motivation based on the fulfillment of three basic needs of humans to feel self-determinant, which are competence, relatedness, and autonomy, and feeling self-determinant is associated with a higher degree of intrinsic motivation (Simons, 1995). As a supporting theory, the Job Characteristics Theory of Hackman and Oldham (1976) is used to assess the influence of the knowledge characteristics of tasks (KCT). This theory separates job characteristics into three categories; social, motivational and contextual. The first is about the social and interpersonal work environment, the second is about the characteristics that enrich the work to make it more motivating and satisfying, which subsequently can be sub-divided into work characteristics that reflect the task and knowledge requirements of work, the latter category is about characteristics of the physical and environmental context within which work is executed. Because this study focusses on autonomous motivation, the domain of motivational characteristics is being used. More specifically, the 'knowledge characteristics of work' part from the dichotomy of motivational characteristics of work is used because the study is focused on employees with knowledge-intensive jobs.In this study, I focus on employees of higher professional educational organizations, which belong to the public sector. Supporters of New Public Management (NPM) suggest that MCs in the public and profit sector should be used similarly, but there is some criticism. For example, Hammerschmid, Van de Walle, Andrews and Mostafa (2019) found that the strategic choice for downsizing public organizations (i.e. reformations to cut costs through redundancies or the closing of facilities) to achieve efficiency, which is typical for NPM, resulted in a decrease of quality delivered. Since the public sector is not about making a profit, but about supporting the community, it is a questionable issue that quality is negatively affected to achieve efficiency.
With this study, I aim to fill the gap and contribute to the understanding of the effects of different types of MCs on the autonomous motivation of employees by examining how these effects are influenced by the knowledge-intensity of tasks. To the best of my knowledge, this is the first study that examines the influence of task characteristics on the effects of MCs on autonomous motivation in a knowledge-intensive context, which therefore fills the gap in the literature as pointed out by Van der Kolk et al. (2019). Based on the arguments above, the following research question is formulated:
RQ: “How do management controls influence the autonomous motivation of employees in higher professional educational organizations, and how do knowledge characteristics of tasks affect this relationship?’’
5
Next to filling the gap as mentioned by Van der Kolk, Van Veen-Dirks, and Ter Bogt (2019), this study has some additional academic contributions. It provides initial results regarding the influence of job characteristics on the relation between MCs and employee motivation. Also, it contributes to the SDT since the fulfillment of certain psychological needs in the work domain seem to depend on the knowledge characteristics of tasks as well (Van den Broeck, et al., 2010).Since this study aims to contribute to the existing MC and SDT literature, I will conduct a theory-testing approach. This approach consists of four stages, to begin with describing the issue and identifying the literature gap. Second, the formulation of the hypotheses and conceptual model will be done. Third, the data analysis will take place, to end with the interpretations, conclusions, discussion and implications of the findings. The paper is presented in this chronologic order.
2. Theoretical background and hypotheses development
This study is focused on the functioning of MCs in HPEOs because it offers an interesting field of research due to the NPM-stream and findings of differences between the functioning of MCs in the public sector and private sector. For example, Boyne (2002) and Wright (2001) found that public sector employees are less oriented to their rewards and more oriented towards providing help to others when compared to employees in the private sector, while similar levels of work motivation have been found (Frank and Lewis, 2004). This suggests that MCs designed for rewarding employees might have less impact on employees in the public sector. However, in the ‘90s and ‘00s, NPM-reforms have been on the agenda of many public organizations. In general, NPM aims to improve transparency, economic efficiency and is output-oriented, by implementing more commercial management styles (van Helden and Jansen, 2003) which also includes rewarding for performance. In educational settings, NPM is becoming undeniable too. Concepts like rankings, competition, quality control, efficiency, and monitoring have taken place in educational organizations (Mohammadi and Mirzamohammadi, 2020). Since it can be argued that the employees in the public sector are less reward oriented, it is interesting to look at how MCs affect employee's autonomous motivation, especially because of the rise of NPM.
2.1 Self-Determination Theory
6
stimulated (Ryan and Deci, 2000). During the development of the theory, three basic psychological needs were discovered (competence, autonomy, and relatedness) which are assumed to be essential for achieving self-motivation and personality integration. The need for competence refers to the degree employees are confident about their ability to perform the assigned tasks and the degree to which they perceive that they are in charge of the outcomes of these tasks (Ryan and Deci, 2000). Autonomy is assumed to be necessary to achieve autonomous employee motivation and refers to the perception of the level of self-possessiveness. In other words, the degree to which people think they execute tasks because of intrinsic stimuli and convincement (Deci and Ryan, 1987). The need for relatedness refers to the need for a social environment that supports individuals and makes them feel safe and accepted, in order to make people feel confident and perform in the most optimal way (Ryan and Deci, 2000).2.2 Motivation
SDT argues that there are roughly two types of motivation, autonomous and controlled (Gagne, Forest, Aube, Morin, and Malorni, 2010), which both play a part in meeting the three psychological basic needs. Autonomous motivation exists where a person finds something interesting or enjoyable to do, so it comes from within the individual with limited external influences. For example, Prendergast (2008) expects that employees who need to perform a task they like, are autonomously motivated. To achieve autonomous motivation, the fulfillment of the psychological need for autonomy is essential. The purest form of autonomous motivation is called intrinsic motivation. Ryan and Deci (2000, p. 70) describe it as “the inherent tendency to seek out novelty and challenges, to extend and exercise one's capacities, to explore, and to learn”
On the other hand, controlled motivation exists when there are external influences that make people do things. Activities that are less or not interesting need a different type of motivation also called extrinsic motivation. Extrinsic motivation is often stimulated with instrumental reasons, such as implicit approval or tangible- and/or verbal rewards. This means that the satisfaction of activities does not come from the activity itself, but rather the extrinsic reward attached to it (Gagne and Deci, 2005).
7
control but includes the feeling of employees themselves (e.g. fear or pressure) to make them perform the tasks (Gagné and Deci, 2005; Ryan and Deci, 2000). Identified regulation motivates employees because they identify themselves with the activity according to their values or goals. Identified employees experience more sense of freedom and volition since they perceive the activities as in line with their personal values and goals (Gagné and Deci, 2005). Integrated regulation is closely linked to intrinsic motivation and motivates employees because they have internalized the values and needs of the people around them and act according to them. Although it is their own choice to act along these values and needs, it is still assumed as an external motivator.Figure 1 - The SDT continuum based on Ryan and Deci (2000)
The example as presented by Ryan and Deci (2000) about the motivation of a student who does his homework for personal development as a preparation for the future compared to a student who does it purely because his parents force him to, visualizes the differences perfectly. In this example, the first student has much and the latter fewer intrinsic motivation. Gagne et al. (2010) split the continuum in half by categorizing the external and introjected regulation as controlled motivation and identified and integrated regulation, and intrinsic motivation as autonomous motivation.
Especially for designers of management control systems (MCS), the type of motivation (see SDT continuum) is relevant because of its potential to influence the autonomous employee motivation, which is important for improving performance (Chen, Lill, and Vance, 2019). When motivation is related to employees in the public sector, several studies argue that it is likely that intrinsic motivation is relatively more present in this context because it is plausible that employees chose to work in this sector because of intrinsic rather than extrinsic reasons (Boyne, 2002; Wright, 2001; Wright, 2007; Georgellis et al., 2011; Van der Kolk et al., 2019). Besides, studies have found that intrinsic rewards are more important for research scientists than instrumental rewards, meaning that intrinsic motivation prevails (Sauermann and Cohen, 2008; Ryan, 2014). Because the work of employees in HPEOs can be considered as knowledge-intensive, it can be argued that in this context intrinsic motivation prevails too.
2.3 Management controls
8
designed to direct their behavior. The structure of the SDT offers a good foundation to hypothesize about MCs and the source of motivation because it discusses the extent to which employees are self-determined and autonomous. MCs are often a combined and designed set of rules and statements, which together influence the employee behavior to make them achieve the general business goals (Merchant and Van der Stede, 2007). According to Simons (1995), four types of controls can be distinguished, the belief system, boundary system, diagnostic control system (DCS), and interactive control system (ICS). Belief controls are designed to inspire employees to be innovative due to the formulation of core values and an organization its mission. Interactive controls stimulate the interaction among employees and between them and their supervisors within organizations to give feedback and share information, to help each other, and to create organization-wide (strategic) improvements. Boundary controls form the figurative boundaries where employees must act within. It prevents companies from being exposed to unnecessary risks and is often communicated via bylaws or policy statements. Diagnostic controls are in place to assure employees do what they need to do, monitor the progress of these activities, and to base rewards or punishments upon.Extensive literature argues that the MCS of an organization is a balanced, interrelated mix of different types of control (i.e. Malmi and Brown, 2008; Mundy, 2010; Kruis, Speklé and Widener, 2016). Simons (1995) stated that the power of these four levers of control lies in how they are balanced to achieve complementarity and cooperation in between. Several studies argue that all four levers need to be present in the MCS to create a dynamic tension that is beneficial for the organization (i.e. Tuomela, 2005; Widener, 2007). However, this balance is different for every organization and context-dependent, which makes it a challenge to balance them in an optimal way (Kruis, Speklé and Widener, 2016).
The LoC framework (Simons, 1995) dichotomies the four levers of control into positive and negative controls, according to their ‘forces’. The belief system and ICS are categorized as 'positive controls' since they promote positive, inspirational forces and support autonomy by giving employees the freedom of choice in their work. The boundary and diagnostic control systems are categorized as 'negative controls' since they constrain employees with the aim to achieve structure and desired behavior by formulating clear targets and expectations, and monitoring performance, which on its turn ensures predictability. Both are needed to ensure long-term success since one creates innovation and strategic improvements and the other ensures predictability and goal achievement (Kruis, Speklé and Widener, 2016). However, the word 'negative' may cause a bad association although negative controls are not bad controls. For example, the boundary system is like breaks on a car, without them, they crash (Simons, 1995, p. 41).
9
be present to realize dynamic tensions that create effective control. However, they criticize Simons' definition of the dichotomy to be broad and general. They argue that the concept of positive and negative controls can be linked to a more general concept about the dual role of controls, which is gaining attention in MCS literature (Mundy, 2010; Wouters and Wilderom, 2008; Groen, Wouters and Wilderom, 2017; Speklé, van Elten and Widener, 2017). For example, Mundy (2010) refers to the creation of the dynamic tension in an organization by the dual role of controls with the aim to gain unique organizational capabilities and competitive advantages. She describes a comparable dichotomy based on the use of MCs and formulates the two groups as 'enabling' (positive) and 'controlling/coercive' (negative), which is in line with previous studies in this field (Ahrens and Chapman, 2004; Adler and Borys, 1996) and used by more recent studies (Tessier and Otley, 2012). Empirical studies showed that the LoC framework is an interrelated system. Tuomela (2005) concluded that the functions of the levers affect each other, a change in one lever has consequences for the other(s) in the corresponding MCS. Also, Widener (2007) found evidence for the interrelated system. She found that the levers together have more impact on performance than the sum of the individual impact of levers. Therefore, it can be argued that the four levers are complementary and create a dynamic tension that produces (Mundy 2010; Henri 2006). Since the positive controls are assumed to enhance creativity and the negative controls to protect the organization from the environment, Speklé, van Elten and Widener (2017) found that the two can coexist and do not necessarily counteract each other. This implies that managers can create a dynamic tension that produces, while no trade-offs in the use of positive or negative controls are needed. Additionally, the LoC framework will be used in this study because of the dichotomy he made between positive and negative, which is not provided by other frameworks such as Merchant and Van der Stede (2007), to contribute to the understanding of the dynamic tension between the two classifications, and translate potential findings to useful practical implications.Based on the arguments above, in this study, the concepts of positive and negative control systems will be used to develop the hypotheses about the relation between the individual LoCs and autonomous employee motivation in the public sector.
2.4 Positive management control systems and autonomous motivation
10
and represents a significant part of the organization's culture. Moreover, Wright (2007) stated that employees in the public sector are more motivated for intrinsic rewards (e.g. task, mission or public service) rather than performance related extrinsic rewards. Since the belief system represents parts of this intrinsic rewards, it can be assumed that it is important for motivation in the public sector. From an SDT point of view, this system relates to the need for ‘relatedness’ of people and enhances intrinsic motivation, because culture makes people feel connected to each other and the organization (Merchant and Van der Stede, 2007). Also, Ryan (1995) mentioned that when employees internalize the formulated goals, values and strategic view of an organization, the perception of being controlled weakens and they feel more autonomous because when employees have internalized these culture-related aspects, they are intrinsically driven to achieve these goals and strategies.The second control system which is part of the positive control system is the ICS, possibly the most ambiguous concept of the LoC framework because of the wide variety of applications. Bisbe et al. (2007) conducted an in-depth analysis to extrapolate an appropriate definition and concluded that the ICS should be a multi-dimensional construct. They found the following five dimensions that are different in nature but together are representing characteristics of ICS: "(1) an intensive use by top management; (2) an intensive use by operating managers; (3) a pervasiveness of face-to-face challenges and debates; (4) a focus on strategic uncertainties; and (5) a non-invasive, facilitating and inspirational involvement." In other words, the ICS is used to provide constructive feedback for improvement, and to establish communication between managers and subordinates to make the strategic view and decisions debatable and gain input from the operational level of organizations (Abernethy and Lillis, 1995; Henri, 2006; Speklé, 2001; Chen, Lill and Vance, 2019). The cross-sectional interactions enable managers to determine priorities and steer upon the strategic goals because they gain useful knowledge from the operational level (Ahrens and Chapman, 2004; Wouters and Wilderom, 2008). It also offers the opportunity for feedback among employees or between managers and subordinates, which stimulates organizational learning and improves task performance (Henri, 2006). These feedback opportunities create a work environment that enables employees to encourage, stimulate, and complement each other when needed and stimulates opportunity- seeking behavior. Although interactive controls tend to be time-consuming and costly because of the attention needed of senior managers to realize the cross-sectional interaction (Widener, 2007), this intensive use will enhance the involvement of top management at the work floor. Hence, the intensive use can be associated with an increase of the relatedness among employees and management.
11
among employees could enhance the social cohesion and sense of being related. The constructive feedback will help employees to learn and improve their capabilities, which enhances the feeling of competence. Combined, the growing competence and sense of relatedness might cause manager to give employees more responsibilities and become more independent, which in turn enhances the autonomous employee motivation.Since both of the positive management control systems are assumed to be autonomy supporting, hence, are likely to enhance autonomous motivation, I hypothesize the following:
H1: The stronger the belief system of an organization is perceived by employees, the more autonomously motivated employees are.
H2: The stronger the interactive control system of an organization is perceived by employees, the more autonomously motivated employees are.
2.5 Negative management control systems and autonomous motivation
The negative management control system, as mentioned by Simons (1995), consist of boundary system and diagnostic control system (DCS). Boundary systems are in place to ensure that employees do not take excessive risks and are aware of the rules they need to obey. Often, these controls specify punishments when employees cross the lines, or rewards when obeying. Practical examples of boundary systems are codes of conduct, policies, and bylaws. As mentioned before, the SDT distinguishes controlled and autonomous motivation based on the fulfillment of the three psychological basic needs. Controlled motivation is realized by external forces who make people do things, such as rules and regulations and the attached rewards or punishments. When looking at the SDT continuum, introjected motivation can be linked to the boundary system perfectly (Ryan and Deci, 2000). Introjected motivation refers to obeying rules to avoid guilt or anxiety or to gain pride, but not accepting them as your own (Ryan and Deci, 2000). Concluding, it can be expected that the use of boundary systems harms the autonomy of employees, hence, it is expected to negatively influence the autonomous motivation.
12
controls’ in the public sector: performance reports about the satisfaction of customers or the number of completed activities within a certain time lapse. Examples of results controls in HPEO’s could be rankings, competitions or quality control. The results of this monitoring are often compared to pre-set targets to identify variances from these targets. Rewards and punishments can be attached to these targets. Henri (2006) argues that looking at performance standards, and especially the negative variances, enhances the motivation of employees. However, according to SDT (Deci and Ryan, 2000), self-determination is negatively influenced by the use of external controls and rewards or punishments, such as monetary rewards, because the focus of employees may shift from their internal drive for performing their tasks to achieving targets for achieving the external rewards or preventing punishments. It reduces the sense of being autonomous among employees because they are being controlled and constrained since they are not free in their choice of task performance (Ryan and Deci, 2017; Frey, 2012). This effect is also called the ‘crowding-out effect’. Frey (2012) argued that external interventions could affect the intrinsic motivation of employees negatively because they perceive it as being controlled. According to Frey, Homberg, and Osterloh (2013) the following three conditions are needed for the crowding-out effect to become apparent; (1) intrinsic motivation was present before the appearance of the effect, (2) the employees must perceive the external rewards as controlling and (3) the rewards must not compensate the reduction of intrinsic motivation.Although achieving individual targets could make individuals feel competent, and related to a group when achieving group targets, the need for autonomy is crucial when autonomous motivation is desired (Van den Broeck et al., 2010; Van der Kolk et al., 2019). As argued before, intrinsic motivation will be relatively more present in the public sector, which makes the ‘crowding-out effect’ even more relevant for this study. Concluding, the presence of negative controls and its consequences may cause a decrease in intrinsic motivation, which will reduce the autonomous motivation of employees.
Based on the arguments above, it can be expected that the negative controls harm the autonomous motivation of employees. Hence, the following hypotheses are formulated:
H3: The stronger the boundary system of an organization is perceived by employees, the less autonomously motivated employees are.
H4: The stronger the diagnostic control system of an organization is perceived by employees, the more autonomously motivated employees are.
2.6 Knowledge characteristics of tasks and management control systems
13
complex, autonomous, and uncommon (Ghosh, 2014; Sia and Appu, 2015) because employees will need to deal with these uncertainties. Hence, it can be argued that the need for autonomy is increasing whenever jobs or tasks become more complex.The job design survey (JDS) of Hackman and Oldham (1975) formed the base of the Job Characteristics Theory (JCT) (Hackman and Oldham, 1976), and is designed to assess job characteristics and to determine if, and how, jobs can be crafted to achieve desired employee behavior, and to evaluate the influence of these changes on employees. The JDS mentions three essential psychological needs to achieve positive personal and work outcomes (e.g. autonomous motivation and productivity), namely perceiving the work as being meaningful, feeling responsible for the work outcomes, and knowing about the work results. The fulfillment of these needs depends on the presence of different job characteristics as mentioned in the model. When linked to the SDT, the first need can be linked to the need for relatedness because work is meaningful when it makes an impact or adds value to the people whom you perform the work for. Second, the need for responsibility is in line with the need for autonomy. The third need can be linked to competence because receiving feedback contributes to improving work performance, which is likely to enhance the feeling of competence of employees since they will learn new skills. In other words, it can be argued that jobs can be crafted to meet the needs of employees, leading to positive work outcomes. Moreover, several studies already argued that job characteristics have motivational potential because they influence the fulfillment of these psychological needs (Hackman and Oldham, 1976; Shalley and Gilson, 2004) and creativity (Chia et al., 2019). Although all three psychological needs are important, I will focus primarily on the need for autonomy as the main line of reasoning when developing the hypotheses because, as argued above, especially this need is affected by the level of KCT.
Morgeson and Humprey (2006) have redesigned the JDS into the Work Design Questionnaire (WDQ) because there was a lot of criticism on the JDS due to the neglection of other work characteristics (Parker, Wall, and Cordery, 2001), internal consistency issues (Taber and Taylor, 1990) and problems with the factor structure (i.e. Idaszak and Drasgow, 1987; Kulik, Oldham, and Langer, 1988). Morgeson and Humphrey (2006) have designed the WDQ by categorizing work characteristics into three main concepts: motivational, social and contextual characteristics. First, the motivational characteristics are assumed to represent the overall complexity of work in the literature, and is subcategorized into task- and knowledge characteristics based on Campion and McClelland (1993) because it enlightens the fact that jobs can be created or adjusted to increase the demand of the individual characteristics. Second, the social characteristics are about the broader social context in which work is executed. Third, the contextual characteristics are about the physical and environmental context in which work is performed.
14
together cover ''the kinds of knowledge, skill, and ability demands that are placed on an individual as a function of what is done on the job '' (Morgeson and Humprey, 2006., p. 1323), and are called: job complexity, information processing, problem-solving, skill variety, and specialization. Job complexity is about the complexity of the tasks in the sense of difficulty in performing it. Information processing is about the amount of data that has to be attended and processed. Whenever this amount of data is high, it makes tasks more challenging. Problem-solving is about the extent to which jobs require creativity and capacity to solve complex issues, which also reflects the need for intensive cognitive processing. Skill variety refers to the extend jobs require the ability to apply various skills to perform a task. It is distinct of task variety because using skills for different tasks is not as challenging as having to combine multiple skills to perform one single task. Specialization reflects the extent to which a job requires specific skills or in-depth-knowledge about a topic.Since the knowledge characteristics refer to the extent to which jobs and tasks are challenging, the SDT-concept ‘competence’ can be related to it. Individuals have their own capacity to which it can perform complex tasks, hence, individuals have their own desires and limits regarding the ability to perform tasks. Whenever tasks or jobs meet the capacity, it can be argued that the need for competence will be met in general. In addition, having to perform tasks that are challenging and need specific skills and knowledge might give employees the feeling of being in charge of the task and its result, which enhances the sense of autonomy. However, if tasks or jobs exceed the capacity of employees, it can make them feel incompetent and, therefore, harm the autonomous motivation.
With regard to motivation, knowledge-intensive jobs (e.g. scientific research) are intensively associated with intrinsic motivators such as task significance, challenge, autonomy, and skill variety (e.g., Lawler and Hall, 1970; Lounsbury et al, 2012). Also, Jindal-Snape and Snape (2006) found that scientists are motivated intrinsically to deliver high quality and discover new knowledge, are demotivated by the lack of feedback from management, and that instrumental rewards are predominantly not perceived as motivational or demotivating. Gagné and Deci (2005) stated that it is more intrinsically motivating when complex tasks are completed, compared to easy or routine tasks. Simons (1995) mentioned that MCs are used by managers to stimulate the creativity and gain commitment of employees. The positive levers of control are in place to enable employees to be creative, innovative and self-directing, and to create a common culture, group feeling and social cohesion, which matches the intrinsic motivators that are associated with knowledge-intensive jobs. Therefore, when high levels of knowledge characteristics of tasks are present, it can be argued that the relation between the positive levers of controls and autonomous motivation will be enhanced.
15
by the internalization of this values and goals and will enhance the sense of autonomy even more. Since knowledge-intensive jobs are associated with a higher need for autonomy and the belief system enables employees to be self-determined, it can be argued that the positive effect of the belief system stronger in a knowledge-intensive context.Second, as mentioned in the development of hypothesis 2, the ICS is used to provide constructive feedback for improvement, and to establish interaction between managers and subordinates to discuss the strategy and gain input from the operational level of organizations. Since complex tasks could demand interaction and cooperation among employees and/or supervisors due to different specializations, it can be assumed there is more interaction present in this context which enhances the feeling of relatedness. Additionally, more feedback will be provided in a knowledge-intensive context to improve the competency and motivation of employees, since easy or routine tasks do not need continual feedback. Since ICSs tend to cause an increase of competence and sense of relatedness, possibly resulting in more responsibilities and independence for employees, the autonomous employee motivation stimulating effect will be stronger in this context. When combined with the sense of autonomy employees will obtain from having an impact on the strategy formulation, it is assumed that when the level of KCT increases, the positive relation between ICS and autonomous employee motivation will be strengthened.
Concluding, the condition of having high levels of knowledge characteristics is creating a context where the effects of the positive control system on the autonomous motivation of employees in HPEOs is enhanced. Therefore, the following hypotheses are formulated:
H1a: The higher the level of knowledge characteristics of tasks, the more positive the relationship between the belief system and employees' autonomous motivation will be.
H2a: The higher the level of knowledge characteristics of tasks, the more positive the relationship between the interactive control system and employees' autonomous motivation will be.
When assessing the negative relation between the negative control systems (boundary system and DCS) and autonomous motivation, the main expected cause of this negative relation is the controlling characteristics of the system which could harm the autonomous motivation of employees.
First, the boundary system is in place to indicate the boundaries within which employees must carry out their work. In other words, it limits employees to some degree in their freedom to perform their tasks. This limit of freedom has a negative effect on the feeling of autonomy of employees. Also, this limitation could make employees feel incompetent regarding tasks that are beyond the boundaries set. Since work with high knowledge characteristics requires a certain degree of autonomy to perform the tasks, this negative effect can be more damaging compared to jobs involving easy or routine tasks.
16
knowledge-intensive jobs and tasks are characterized by the fact that it is not always clear upfront what the outcome will be (van Elst, Aschoff, Bernardi, Maus and Schwarz, 2003). This could imply that employees need to make concessions with regard to the outcome of the tasks to perform them in a way that contributes to achieving the formulated objectives, at the cost of the way that is best in their view, when these are not in line. This consideration will harm the sense of autonomy and stimulate the sense of being controlled, as employees cannot blindly choose what they believe to be the best option. As mentioned before, the DCS is used to enable managers to move their focus to other issues after the formulation of targets for employees (Simons, 1995), which causes a decrease in interaction and feeling of relatedness. However, whenever tasks are complex, they are possibly multi-disciplinary and only executable by cooperation, which could enhance the sense of relatedness. Often monetary rewards are linked to the targets set for the employees, which may cause the crowding-out effect. Since knowledge-intensive jobs are largely associated with high levels of intrinsic motivators (e.g., Lawler and Hall, 1970; Lounsbury et al, 2012; Jindal-Snape and Snape, 2006), it can be argued that this crowding-out effect could be more effective within this context.Based on the arguments above, it can be argued that having high levels of knowledge characteristics of tasks is likely to strengthen the harming effect of negative controls on autonomous motivation. Therefore, the following hypotheses are formulated:
H3a: The higher the level of knowledge characteristics of tasks, the more negative the relationship between the boundary system and employees' autonomous motivation will be.
H4a: The higher the level of knowledge characteristics of tasks, the more negative the relationship between the diagnostic control system and employees' autonomous motivation will be. 2.7 Conceptual model
Figure 2 - Conceptual model
17
3. Methodology
3.1 Data collection and sample
The aim of this research was to examine the effect of positive and negative MCs on the autonomous motivation of employees in the public sector, and how these effects are influenced by the level of knowledge characteristics of their work. To do so, the scope of this research was set on employees of higher professional educational organizations because jobs in these organizations are often characterized as knowledge-intensive.
The dataset for this study is drawn from a database that was created by two faculty members of the University of Groningen and is provided by my thesis supervisor, also of the University of Groningen. This database contains information about management control- and job-related aspects of more than 500 managers and employees of four higher professional educational organizations, which are collected via surveys of 28 items of multiple questions with the survey instrument Qualtrics. The survey was established by using existing frameworks and measures to ensure the validity of the variables, which will be explained later on in the variable-section, and the quality is guaranteed by the execution of pre-test by colleagues from the field and colleague researchers of the faculty members.
The surveys were spread among organizations where the faculty members had personal access to the board of directors because, due to privacy regulations (GDPR), it is not possible to send emails to employees without their own permission. After obtaining this permission, the surveys were sent to the employees of the organizations. This personal approach by conversations and presentations created commitment of the boards and access to email addresses, which caused the high response rates. This approach contributed to the reduction of the risk for a non-response bias in the dataset because the involvement of the board of directors creates shared interests. Also, tests were performed after the data collection to reduce this risk for a non-response bias. Although a personal approach was used, the anonymity of the respondents was assured to them to avoid the results being influenced by the common method bias.
18
3.2 Data analysis
The data analysis started with a check for missing and extreme data, but none were found. To check if the sample is appropriate to use the KMO and Bartlett’s Test of Sphericity are conducted. The KMO measure of sample adequacy score of 0.882 suggested the sample was adequate. Bartlett’s Test of Sphericity was significant, which indicated an exploratory factor analysis (EFA) was useful to do. Next, the construct validity check consisting of a factor- and reliability analysis was executed. The results of the principal component EFA are reported in Appendix A according to their standardized correlation values. The Direct Oblimin-rotation was used because the independent variables were expected to correlate. First, separate factor analyses were performed for the group of independent, moderator and dependent variables with an extraction according to the Kaiser criterion (eliminating components with Eigenvalues below 1). Second, a factor analysis is performed containing all the constructs. However, when extracted according to the Kaiser criterion, two levers ended up in the same
Table 1 - Descriptive statistics sample
Variable Frequency Percentage
Age (N=259) 20 to 30 14 5% 31 to 40 50 19% 41 to 50 74 29% Older than 50 121 47% Organizational tenure (N=259) 0 to 10 years 134 52% 11 to 20 years 88 34% 21 to 30 years 29 11% 31 years or more 8 3% Educational background (N=259) Primary education 0 0% Secondary education 3 1%
Secondary vocational education 25 10%
Bachelor's degree 81 31%
Master's degree or higher 150 58%
Main activities (N=259)
Educational staff 141 54%
Educational support staff 118 46%
19
construct. Therefore, a second factor analysis for all the constructs is performed with fixed number of 10 extractions. The reliability check consisted of calculating the Cronbach Alpha of each variable, with a critical value of >0.7 to be usable (Nunnally, 1978), after the removal of several items based on the factor analysis. This tested the reliability of the measurement regarding the internal consistency, that is, the level of relatedness between items in a group. Next, a test for normality was performed concerning the dependent variable with a Shapiro-Wilk test and Q-Q plot (Appendix B). Also, a test for collinearity between the variables was conducted, based on the Variance Inflation Factor (VIF). To reduce the risk of a non-response bias, an independent sample t-test is used to check whether there is a significant difference in the results of employees who completed the survey in the first quartile of the collection period, compared to the last quartile relatively, each consisting of 65 respondents (Appendix C).The formulated hypotheses were tested according to multiple linear regression analyses because the independent variables contain ratio scale data and the dependent ratio scale data. Before conducting the regression analyses, the independent-, dependent-, moderating- and control variables were tested on interrelations by correlating them by a PEARSON-test. Although the underlying items of the variables were answered according to a 5- or 7-point Likert scale, the PEARSON-test was used rather than the SPEARMAN-test. The categorical variables in this study were treated as interval data because of the large sample size and use of 5- and 7-points scales (i.e. Dolan, 1994; Johnson and Creech, 1983). This result showed whether it was possible to use all these variables and gave insights into the correlations among the variables used. Then, the hypotheses were tested in chronological order. Lastly, the interaction variables of the levers and KCT (‘lever’ multiplied with KCT) to test for moderation were added. To prevent the results of multicollinearity, the variables were standardized to z-scores before creating the IVs for the moderation testing. The control variables that correlated significantly with the dependent and moderating variables were used in every test.
3.3 Measures
20
Independent variables
As stated in the theoretical background, Simons (1995) describes the belief system and ICS as positive MCs and the boundary system and DCS as negative MCs. Therefore, the first positive MC variable Belief system (BELSYS) is measured similarly to Kruis, Speklé, and Widener (2016). They used four questions regarding the organization’s core values and mission statement, as developed by Widener (2007) and validated the measure by correlating the results of two related additional questions to the scores of the existing measure. The questionnaire could be answered by a 7-point scale from ‘’not descriptive’’ to ‘’very descriptive’’. In this sample, one question is removed due to cross-loadings, after removing this question the variable had a Cronbach alpha of .812.
Second, the Interactive control system (INTER) is measured according to the measurement used by Bedford and Malmi (2015). The design of this instrument is based on the model of Bisbe et al. (2007) and distinguishes five properties regarding the use of interactive controls, which are all measured by a single question. The formulation of these questions is in line with the formulations in the studies of Widener (2007), Henri (2006) and Bisbe and Otley (2004), although the first two questions of Bedford and Malmi (2015) were combined, ending up with four questions. After removing one item due to a factor loading below .4, which enhanced the validity of the measurement, the variable had a Cronbach alpha of .916.
Although the study is about the influence of the positive management controls, the two variables are not combined in the tests with the aim to obtain as much information as possible. In addition, the Cronbach alpha of the combined variable (.837) is affecting the score of INTER negatively and the score of BELSYS just slightly positively.
Next, the first variable of the negative MCs is the Boundary system (BOUSYS). This variable is captured by using the instrument designed by Kruis, Speklé, and Widener (2016), which is based on four questions regarding the organization’s codes of conduct, as formulated by Widener (2007). The questions are designed to ask about the employee perception of behavior directing effects of the codes of conduct and other regulations. Kruis, Speklé, and Widener (2016) again correlated the metrics with two questions to validate the instrument. In this sample, this variable has a Cronbach alpha of .861.
Second, the Diagnostic control system (DIAGN) is captured by using the tool of Bedford and Malmi (2015). This measurement consists of five questions that are related to the use of accounting as control mechanisms, which are based on the theories of Henri (2006), Widener (2007) and Simons (1995). In this sample, this variable has a Cronbach’s alpha of .964.
21
Moderator variable
The knowledge characteristics of tasks (KCT) is measured by the instrument developed by Morgeson and Humphrey (2006), which takes the following five aspects of knowledge-related task characteristics into account: (1)Job complexity, (2)Information processing, (3)Problem solving, (4)Skill variety, and (5)Specialization. These characteristics are measured according to a 5-point scale from ‘totally disagree’ to ‘totally agree’, with four items per aspect. A separate factor analysis for this variable, with an extraction based on the Kaiser criterion, led to four constructs (Appendix A), which is not in line with the construct as designed by Morgeson and Humphrey (2006). KCT is treated as one variable because the aim of this study is to assess the general impact of KCT. Also, the constructs found by the factor analysis are strongly intercorrelated and a high Cronbach alpha for KCT as one variable is found. In this sample, the Cronbach’s alpha of this variable is .812, which is good, although three items had to be excluded, one due to a factor loading below the .4 and two because of cross-loadings.
Dependent variable
22
Control variables
To prevent the results of being affected by other factors that have an influence on the autonomous motivation of employees, the variables main activities and educational background are used. These control variables are also used in the study of Cai et al. (2019), who also studied the moderating role of job characteristics in work motivation related context.
Educational background was used as control variable because I assume that when employees are higher educated, in general, their tasks will contain higher levels of KCT. Employees were asked to fill in the highest degree of education obtained. The variable is measured and coded in the dataset according to the following scale: 1 = Primary school, 2 = Bachelor’s degree, 3 = Master’s degree or higher, 4 = Secondary education, 8 = Secondary vocational education. A dummy variable was created to use the level of education as a control variable, whereas 0 is lower educated (Primary, secondary and secondary vocational education) and 1 is higher educated (Bachelor’s degree or higher).
The variable ‘main activities’ was used in the hypothesis testing because it is plausible that employees who perform educational tasks are predominantly associated with knowledge-intensive tasks, compared to educational support staff. The variable was coded as 1 = educational staff, 2 = educational support staff.
The variables age, organizational tenure and type of contract were tried during the hypotheses testing because they are often used in literature. However, they were not included since they did not correlate significantly to the moderating or dependent variable and influenced the (adjusted) R² negatively.
4. Results
Since there were no respondents excluded due to missing data of extreme values, the sample consisted of 259 respondents. To perform the statistical tests, IBM SPSS 24 was used.
The test for normality of the dependent variable (AUTMOT) showed a significant result (p = .00) on the Shapiro-Wilk test, indicating that the variable is not normally distributed. However, this could be caused by the large sample size (Field, 2009). Therefore, a Q-Q plot was created for checking normality of the dependent variable. This plot showed that the variable was spread relatively equally along the line, indicating that the dependent variable was normally distributed. The results for multicollinearity among the independent variables did not exceed the critical value of 10 (O’Brien, 2007) since the VIF scores ranged between 1.01 and 2.16, so multicollinearity was not an issue. The test for a non-response bias on early and late responses showed no significant difference between the groups (Appendix C).
4.1 Correlation results
23
level, which suggests the theoretical arguments of the levers as an interrelated system could be true. Also, the belief system is correlating at a 1% significance level with intrinsic and autonomous motivation. Also, KCT correlates significantly with the DCS, identified regulation, intrinsic- and autonomous motivation, level of education, and main activities. Intrinsic motivation correlates significantly with the four levers, KCT, and identified regulation and autonomous motivation. Last, the variable ‘main activities’ is correlating significantly to the belief system, boundary system, KCT, and the level of education, and the boundary system is correlating significantly to the level of education. When looking at the mean scores, it can be concluded that the sample is relatively intrinsically motivated (5.62 out of 7), highly educated (0.89 out of 1), and performs knowledge-intensive tasks (3.63 out of 5).4.2 Regression results
In table 3, the five models show the results of the regression analyses to test the hypotheses. The column ‘Main effects’ presents the results for hypotheses 1 to 4. Hypothesis 1 predicted that the belief system is positively related to autonomous employee motivation. The coefficient of this variable is significant and positive (B = .23, p = .00), hence, H1 is supported. Hypothesis 2 predicted that ICS is positively associated with autonomous employee motivation. The coefficient of this variable is positive but not significant (B = -.04, p = .52), leaving H2 unsupported. Hypothesis 3 predicted that the boundary system is negatively related to autonomous employee motivation. The coefficient of this variable is close to zero and not significant (B = .00, p = .95). Therefore, H3 is not supported. Last, hypothesis 4 predicted that DCS is negatively related to autonomous employee motivation. The coefficient of this variable is not significant (B = .01, p = .86), leaving H4 unsupported.
Table 2 - Correlations and descriptives statistics
CONSTRUCT (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
(1) Belief Sys. 1.000
(2) Boundary Sys. .41** 1.000
(3) Inter. Con. Sys. .39** .44** 1.000
(4) Diagn. Con. Sys. .35** .34** .69** 1.000
(5) Knowledge Ch. .06 -.05 .03 .13* 1.000 (6) Identified Reg. .09 -.01 -.04 -.01 .18** 1.000 (7) Intrinsic Mot. .33** .17** .16** .18** .31** .48** 1.000 (8) Autonomous Mot. .24** .09 .06 .10 .28** .87** .85* 1.000 (9) Level of Education .01 -.21** -.09 -.01 .27** -.01 .01 -.03 1.000 (10) Main Activities .14* .15* .11 .11 -.19** -.06 -.08 -.01 -.38** 1.000 Observations 259 259 259 259 259 259 259 259 259 259 Mean 4.47 4.00 3.72 4.14 3.63 5.62 5.56 5.59 0.89 1.46 Std. Dev. 1.02 1.44 1.27 1.19 0.40 1.15 1.06 .95 0.31 0.49 Skewness -.48 -.22 -.35 -.64 -.31 -1.04 -.87 -.75 -2.54 .18 Std. Error of Skewness .15 .15 .15 .15 .15 .15 .15 .15 .15 .15 Kurtosis .48 -.81 -.17 .77 .43 1.42 1.19 .59 4.48 -1.98 Std. Error of Kurtosis .30 .30 .30 .30 .30 .30 .30 .30 .30 .30 Min 1 1 1 1 2.35 1 1.33 2 0 1 Max 7 7 7 7 4.53 7 7 7 1 2
25
Variable Main effects Interaction 1 Interaction 2 Interaction 3 Interaction 4
Constant 2.79*** 2.59*** 2.71*** 2.68*** 2.76*** (.61) (.62) (.62) (.62) (.61) Controls Main activities -.16 -.17 -.17 -.15 -.18 (.12) (.12) (.12) (.12) (.12) Level of education -.38* -.37* -.37* -.35* -.39* (.20) (.20) (.20) (.20) (.20) Independent variables Belief system .23*** .25*** .24*** .24*** .25*** (.06) (.06) (.06) (.06) (.06) Interactive control system -.04 -.04 -.02 -.04 -.05 (.07) (.06) (.07) (.07) (.07) Boundary system .00 .01 .00 .01 .00 (.05) (.05) (.05) (.05) (.05) Diagnostic control system .01 .02 .00 .01 .02 (.07) (.07) (.07) (.07) (.07) KCT .67*** .69*** .68*** .67*** .67*** (.15) (.15) (.15) (.15) (.15) Interaction variables Belief system * KCT -.12** (.05) Interactive control system * KCT -.08 (.05) Boundary system * KCT -.09 (.06) Diagnostic control system * KCT -.09* (.05) R² .14 .16 .15 .15 .16 ! R² .02 .01 .01 .01 Adjusted R² .12 .13 .12 .13 .13
Table 3 - Regression results with autonomous motivation as dependent variable
26
4.3 Additional analyses
Since the factor analysis showed two distinct constructs of motivation, two additional analyses were performed to investigate if the results are different when these two types of motivation are assessed individually.
First, the results of the same tests as in the previous section but with identified regulation as the dependent variable (Table 5) show that the main effects are the same as with autonomous motivation. Also, the interaction variable ‘Belief system*KCT’ shows a significant negative moderation (B = -.13, p = .06). However, the results of the interaction variable ‘Boundary system*KCT’ turns out significant and negative (B = -.14, p = .05), indicating that whenever the level of KCT gets higher, the effect of the boundary system on identified regulation becomes more negative. The significant score of the interaction variable ‘Diagnostic control system*KCT’ in table 3 is not present when using identified regulation as a dependent variable. When comparing the adjusted R² of the models in table 5 to the models in table 3, it can be concluded that the explained variance has decreased in every model.
Second, the results with intrinsic motivation as a dependent variable (Table 6) show that the main effects are the same as with autonomous motivation and identified regulation. Again, the interaction variable ‘Belief system*KCT’ shows a significant negative moderation (B = -.11, p = .06). Contradicting to the test with identified regulation, the interaction variable ‘Boundary system*KCT’ is not significant in this test (B = -.04, p = .54). However, negative significant moderations are found for the interaction variables ‘Diagnostic control system*KCT’ (B = -.14, p = .01) and ‘interactive control system*KCT’ (B = -.11, p = .05). This indicates that, for the three significant interaction variables in this test, whenever the level of KCT increases, the relation between the levers and intrinsic motivation is influenced negatively. Looking at the adjusted R² of the models in table 6 ranging between .19 and .20, it can be concluded that these scores are higher than the scores of the models in table 5, ranging between .03 and .04, and the models in table 3, ranging between .12 and .13.
Table 4 - Overview of outcomes
Dependent variable
Hypothesis Autonomous motivation Identified regulation Intrinsic motivation
H1 Supported Supported Supported
H2 X X X H3 X X X H4 X X X H1a X* X* X* H2a X X X* H3a X Supported X
H4a Supported X Supported
27
Variable Main effects Interaction 1 Interaction 2 Interaction 3 Interaction 4
Constant 4.01*** 3.79*** 3.97*** 3.83*** 3.99*** (.78) (.78) (.78) (.78) (.78) Controls Main activities -.20 -.21 -.21 -.18 -.21 (.16) (.16) (.16) (.16) (.16) Level of education -.31 -.30 -.31 -.26 -.32 (.26) (.26) (.26) (.26) (.26) Independent variables Belief system .16** .18** .16** .17** .17** (.08) (.08) (.08) (.08) (.08) Interactive control system -.08 -.08 -.07 -.08 -.08 (.08) (.08) (.08) (.08) (.08) Boundary system -.02 -.02 -.02 -.01 -.02 (.06) (.06) (.06) (.06) (.06) Diagnostic control system .00 .00 -.01 -.01 .00 (.08) (.08) (.09) (.08) (.09) KCT .51*** .54*** .52*** .52*** .52*** (.19) (.18) (.19) (.18) (.19) Interaction variables Belief system * KCT -.13* (.07) Interactive control system * KCT -.04 (.07) Boundary system * KCT -.14* (.07) Diagnostic control system * KCT -.05 (.06) R² .06 .07 .06 .07 .06 ! R² .01 .00 .01 .00 Adjusted R² .03 .04 .03 .04 .03
Table 5 - Regression results with identified regulation as dependent variable
28
Variable Main effects Interaction 1 Interaction 2 Interaction 3 Interaction 4
Constant 1.57** 1.39** 1.45** 1.53** 1.53** (.66) (.66) (.66) (.67) (.65) Controls Main activities -.12 -.13 -.14 -.11 -.15 (.13) (.13) (.13) (.13) (.13) Level of education -.45** -.44** -.44** -.44** -.46** (.22) (.22) (.22) (.22) (.22) Independent variables Belief system .31*** .32*** .32*** .31*** .33*** (.07) (.07) (.07) (.07) (.07) Interactive control system .00 .00 .03 -.01 -.02 (.07) (.07) (.07) (.07) (.07) Boundary system .03 .03 .02 .03 .03 (.05) (.05) (.05) (.05) (.05) Diagnostic control system .03 .03 .00 .02 .04 (.07) (.07) (.07) (.07) (.07) KCT .82*** .84*** .84*** .82*** .82*** (.16) (.16) (.16) (.16) (.16) Interaction variables Belief system * KCT -.11* (.06) Interactive control system * KCT -.11** (.06) Boundary system * KCT -.04 (.06) Diagnostic control system * KCT -.14** (.05) R² .21 .22 .22 .21 .23 ! R² .01 .01 .00 .02 Adjusted R² .19 .20 .20 .19 .20
Table 6 - Regression results with intrinsic motivation as dependent variable
29
5. Discussion and Conclusion
5.1 Summary of results and discussion
The results regarding the hypotheses for positive MCs only support H1, which implies that the stronger the belief system is perceived, the more autonomously motivated employees are. Since the sample consists of employees of a public organization, it can be assumed that these employees are especially driven by core values and goals of the organization (Wright, 2007), which supports the findings. Although H1a expected a strengthening moderating relation, the contradicting results showed a significant negative moderating effect of KCT. This implies that whenever tasks become more complex, the relation of H1 is mitigated. As stated before, it is likely that when the level of KCT rises, the need for autonomy increases. When employees get more autonomy, they can perform their tasks in their way. From an SDT point of view, a possible explanation for this mitigating moderation can be that employees may become more self-determined and less affected by the belief system when the level of KCT rises.
The additional analyses showed more information about the nature of autonomous motivation. The direct relations for both positive controls and the moderating effect for the belief system were the same when testing for the two underlying types of motivation, as when testing for autonomous motivation. However, the result regarding the moderating effect for the ICS was different when testing for intrinsic motivation. A negative significant moderating effect was found, indicating that when the level of KCT rises, the effect of ICS on intrinsic motivation is influenced negatively. However, no significant direct relation is found regarding the ICS and any type of motivation. This suggests that the moderation of KCT exists but is probably not applicable to this sample as a whole but rather on a subgroup within the sample. A possible explanation could be that the results are significant for certain departments within the organization. Since the ICS is designed by the top-management and operating managers (Bisbé et al., 2007), differences in the uses of the ICS may exist within the organization.
The results for negative MCs in the main analysis with autonomous motivation indicated no direct relations. A negative significant moderating relation was found when testing for H4a, indicating that when the level of KCT rises, the relation between the diagnostic control system and autonomous motivation will become influenced negatively, which is in line with the hypothesis.
30
system diminishes the sense of freedom and volition to perform tasks. Identified regulation is motivating because of the matching of tasks and personal values and goals, which increases the sense of autonomy regarding the performance of work because people feel identified with the task. Because the boundary system limits peoples' freedom and makes people feel controlled, the identification of employees with their tasks may be lost, explaining the negative effect on identified regulation.A reason no significant moderating effect for intrinsic motivation is found might be that it is because intrinsic motivation is the purest motivation originating from within the individual by its interests, and not caused by external forces (e.g. Ryan and Deci, 2000). Mostly the boundary system is not specified on the individual level but rather on the company level, which reduces the impact on specific job characteristics. Therefore, it can be argued that the belief system has less impact on intrinsic motivation when compared to identified regulation.
Opposite to the boundary system, a negative significant moderating effect was found when testing the relation between diagnostic control systems and intrinsic motivation. Again, this could be linked to the impact on the job characteristics and the need for autonomy. Since the diagnostic control system is about monitoring and communicating performance metrics (e.g. targets or budgets), it is likely to have an impact on the characteristics of jobs and tasks (e.g. Simons, 1995). As mentioned in the theory section, Frey (2012) argued that external interventions could harm the intrinsic motivation of employees because they perceive it as being controlled. Since the findings are in line with Frey's argument, it can be argued that the intrinsic motivation of employees is harmed when employees are limited at the task level by the DCS. A potential reason no significant moderating effect for identified regulation was found could be that because identified regulation is caused by external forces, lower levels of autonomy are needed by employees. Therefore, the controlling character of DCS maybe not as effective on motivation when compared to intrinsic motivation.
Although arguments have been made regarding the positive and negative controls, no clear pattern can be found in the results. Therefore, the arguments cannot be supported by results about the cohesion between the levers that belong to their category and their influence on any type of autonomous motivation.
Since the results of this study show large differences in (adjusted) R2 between identified regulation (table 5) and intrinsic motivation (Table 6), it can be argued that any form of controls has more impact on the intrinsic employee motivation when compared to identified regulation.
Another result that stands out is that all significant moderating effects found were negative, which could indicate that for all the types of control the relation towards motivation is influenced negatively whenever the level of KCT rises. In other words, the results suggest that when jobs or tasks become more complex, employee motivation can be stimulated by reducing the presence of controls.
31
5.2 Contributions
As stated in the introduction, this study aims to fill the literature gap as pointed out by Van der Kolk, Van Veen-Dirks, and Ter Bogt (2019) by examining the effect of the KCT on how MCs influence the autonomous motivation of employees in the public sector. Since the results suggest the KCT have an influence on some of the relations between MCs and the autonomous motivation of employees in an HPEO, the findings contribute to the knowledge about MC- and work motivation literature. Also, the findings also contribute to the SDT-stream in the work domain since it combines the theory with the JCM and therefore, confirms and shows a relation between job characteristics and work motivation determinants (Van den Broeck, et al., 2010). Moreover, additional analyses regarding the two underlying constructs of motivation offer new insights regarding the effect of MC on different types of motivation. Lastly, the four LoC are assessed separately which enables future researchers to make distinctions between the types of controls and their effects on employee motivation.
The managerial contribution of this paper could be that when MCs are designed, the level of knowledge-intensity can be taken into account to foster autonomous employee motivation. For example, in a knowledge-intensive organization were creativity is needed, managers need to take the tasks of employees into account when designing MCs. When employees perform knowledge-intensive tasks, they may consider keeping the number and intensity of MCs relatively low compared to employees with more routine tasks. This can result in more autonomous employee motivation, leading to higher performance (Chen, Lill, and Vance, 2019) and better resource utilization. Also, when employees are more intrinsically motivated, they are more self-determined and need less management of the manager. This enables managers to spend their time and effort on other tasks.
5.3 Limitations
32
5.4 Future Research
Future studies can use these results for meta-analytical studies to obtain a more generalizable view about the influence of KCT on the motivating character of MCs. Also, when taking into account the before mentioned limitations regarding the context, replication of this study in other contexts such as organizations with a knowledge-extensive character and organizations with both knowledge-intensive and -extensive jobs could be interesting to do. Especially, since I found no significant direct relations for the boundary system DCS and ICS but significant moderating effects, it could be interesting to look into this. Also, results could be different when using organizations from the private sector or other types of public organizations, such as the government or public hospitals. Also, these organizations could be used together to make appropriate comparisons between the different contexts. When looking at the limitation regarding the method used, more specific relations could be discovered when using the underlying concepts of ‘KCT’. For example, it could be interesting to look at the influence of an individual concept of KCT on motivation itself, or a specific type of motivation. Lastly, studying the influence of KCT on the design of MC based on the LoC, because of their interrelations and complementarities (Widener, 2007), could be interesting, since it will enhance the understanding of the influence of job characteristics on MC.
5.5 Conclusion
