Performance management systems within the public sector:
Effects on employee job performance
“An empirical analysis on the effect of the type of PMS on employee job performance within the
public sector”
Author: Robert Bleeker
Student number: 2210800
Track: Organizational and Management Control (O&MC)
Abstract
This is the first study on the micro level within the public sector that examines the use of performance management systems (PMSs) and their effect on employee job performance. I hypothesize that two different types of PMSs have contrasting effects on employee job performance. The hypotheses were tested using survey data from 119 employees of the municipality of Groningen. My findings indicate that
employee job performance is not affected by the use of any type of PMS. However, I did find that the developmental type of PMS is perceived to be more present than the judgmental type. These findings suggest that in practice the judgmental type is not seen as an improvement over the developmental type. This contradicts New Public Management who are an important source of inspiration for management and policy makers within the public sector. Furthermore, the two types of PMS are not mutually exclusive and
are in fact positively correlated. The idea that management controls operate in isolation is therefore contradicted. These results can have a couple of interesting implications, both for practice and for policy
within the public sector.
Supervisor: Wilmar de Munnik
Co-assessors: Jim Emanuels and Kristina Linke Date: 20-01-2018
2
1. Introduction
The New Public Management (NPM) movement has gained a foothold within the public sector over recent decades. In general terms, it can be stated that NPM refers to the introduction and use of private sector management styles and instruments within the public sector. NPM focuses on improving performance and, more generally, the efficiency and effectiveness of the public sector (Ter Bogt & Scapens, 2012).
Changes in the spirit of the NPM movement have led to adjustments in the management of public sector organizations, based on the adoption of private sector management techniques and the assumption of competitive markets (Groot & Budding, 2008; Hood, 1995; Pollitt, 2002; Walker et al., 2011). In order to develop an efficient, effective, and accountable public sector, NPM supports a rather mechanistic concept of performance management in which measurable and specific pre-set performance targets should lead the efforts of civil servants towards the achievement of their organizations’ objectives (Bevan & Hood, 2006; Newberry & Pallot, 2004. Judgmental performance management systems (PMSs) are assumed to be objective and to support greater clarity, and are considered by NPM as an improvement over the former (developmental) PMSs that were more uncertain and subjective (Ter Bogt & Scapens, 2012). Despite these assumed benefits of greater clarity and objectivism, some studies suggest that the increasing use of these judgmental PMSs could have opposite effects and seem to be related to negative consequences. For example, these PMSs can put serious pressure on employees and they can create anxiety and stress (cf. Thorsen, 1996; Winefield et al., 2003; Tytherleigh et al., 2005; Woods, 2010). It can also encourage individuals to play safe and get the output they need. This means that employees focus merely on the quantity of their work and not on the quality, because they fear a negative appraisal for not meeting their performance targets. In the longer term this could be damaging, because individuals are pursuing their own goals, or the goals of their managers, but not the goals of the organization. This misalignment between the individual and organizational goals could be harmful for the overall organizational performance.
Regardless of the rise in the use of judgmental PMSs within the public sector, and their assumed consequences, existing research has been primarily of explorative nature (cf. Ashton et al., 2009; Bevan & Hood, 2006; De Bruijn, 2002; Dill & Soo, 2005; Marginson & van der Wende, 2009; Merisotis & Sadlak, 2005; Newberry & Pallot, 2004; Ter Bogt & Scapens, 2012;). Speklé and Verbeeten (2014) did provide empirical evidence on the effect on performance within the public sector. However, they focused on performance measurement systems and business unit performance, so the macro level. But larger scale empirical evidence on the effect of the use of PMSs on employee job performance (micro level) in the public sector is, to the best of my knowledge, still missing. The question as to the effect of the use of a PMS on employee job performance cannot be answered conclusively.
3
employee job performance within the public sector and to provide larger sample quantitative evidence of this effect. To do this, this study gathered perceptions regarding the use of judgmental- and
developmental PMSs and analyzed their effects on employee job performance within the setting of the Dutch public sector. Thus, employee perceptions, and not objective data, form the input for this study. When management and policy makers understand the perceived impact of these types of PMSs, they can change the daily operations. These practical insights could lead to a more effective and better performing public sector. Attempting to improve the performance of the public sector is undoubtedly very important.
This study has three major contributions to existing scientific and practitioner literature. It is the first to examine the micro level effect of PMSs on an individual’s behavior within the public sector. Second, it provides evidence that the ideas and views of the NPM movement are not as apparent in practice. Third, it provides evidence that the different types of PMSs do not operate in isolation and are positively correlated to each other.
The remainder of this paper is structured in the following way. In the subsequent section the research question is presented. Then the literature section starts with an overview of the most important literature which results in hypotheses and a conceptual model. The methodology section describes the sample and measurement. Subsequently the results of the analysis can be found. In the final section this study provides a discussion with conclusions, limitations, and suggestions for future research.
1.1. Research question
Based on the afore mentioned empirical gap and the purpose of my study, I have formulated the following research question:
RQ: How does the type of PMS affect employee job performance within the public sector?
2. Literature review
In the sections below performance management and performance management systems (PMSs) will be defined. In addition, two types of PMS and task autonomy will be introduced and their effect on employee job performance will be hypothesized.
2.1. Performance management
Performance management is “a continuous process of identifying, measuring, and developing the
performance of individuals and teams and aligning performance with the strategic goals of the organization” (Aguinis, 2013, p.2). Performance management demands that managers assure that the
4
relation between employee performance and organizational goals and makes the employees’ contribution to the organization clear (Aguinis, 2013). Performance management can be used to measure the
performance of different business units, teams, employees, or even processes. It is a systematic process to make sure that the organizational mission and goals are constantly met in an efficient and effective way. Performance management systems (PMSs) are the evolving formal and informal processes, mechanisms, systems, and networks used by organizations for transferring the key objectives and strategic goals of the organization, for assisting the strategic process and ongoing management. The definition performance management system is used to embrace these more general processes and include both the formal, but also the more subtle, yet important, informal controls that are used (Chenhall, 2003; Ferreira & Otley, 2009; Malmi & Brown, 2008).
Performance measurement helps managers to understand where they currently are, to identify what they want or need to improve, or to influence their subordinate’s behavior (Neely, 1998). Performance measurement refers to the “process of assigning numbers to the represent aspects of
organizational behavior and performance (Flamholtz, 1983, p. 156). A performance measurement system
is the set of metrics used to quantify both the efficiency and effectiveness of actions (Neely et al., 1995). This study is interested in the usage of PMSs. Therefore, in the remainder of this study,
judgmental and developmental types of PMS refer to different ways of using a performance management system.
2.1.1. PMS as a management control
The PMS is traditionally viewed as a necessary component of planning and control (Katic et al., 2011). Management control (MC) refers to a set of processes and mechanisms used by managers to influence the behavior of individuals and groups towards more or less predetermined objectives (Flamholtz et al., 1985; Chenhall & Langfield-Smith, 2007; Speklé, 2001). The MC literature identified two different roles for MC elements: first, they are aimed at constraining the subordinates by monitoring and incentivizing them, so that they will act in line with the organization’s goals (Jensen & Meckling, 1976). The second role is a facilitating management control and is focused on fostering autonomy and facilitating the subordinates in better serving their organization (Davis et al., 1997).
2.1.2. Types of PMS
2.1.2.1. Judgmental PMS
5
judgmental types of PMS focus on achieved or historical performance (cf. Anderson, 1994; Mc Gregor, 1972; Randell, 1989,1994; Redman, 2001; Townley, 1997). Different authors describe and use various definitions and types of PMSs that are corresponding with the judgmental type. It can be argued that a judgmental PMS is conceptually similar to the incentive-oriented PMS described by Speklé and Verbeeten (2014), Simons’ (1995) diagnostic use and the cybernetic management controls of Green and Welsh (1988). In order to illustrate this similarity, the definitions of these various types of PMS, that are used by these authors, can be found in figure 1.
Thus, using a judgmental PMS actually constrains employees in the freedom to perform their work. But how may this affect their behavior? Think for example about university professors. Professors need to write a certain number of articles in a certain amount of time, as a result of the judgmental PMS. If they fail to write enough articles, they get ‘punished’ and receive a bad evaluation. To make sure that they won’t receive this negative appraisal, the amount of articles to publish becomes more important than the quality of those same articles. They change their behavior, due to the pressure and quantitative way of performance management of the judgmental PMS, in such a way that they will play safe just to publish enough articles. Meeting their performance targets becomes more important than the quality of the work. The stimulus and opportunity to perform a groundbreaking research will be strongly reduced because the qualitative outcomes might not be satisficing enough for the quantitative oriented judgmental PMS.
Type of PMS Definition
Judgmental A judgmental system places the organization’s concerns with control and a centrally coordinated information system to the center. Used as a tool for resource allocation, appraisal is part of the immediate manager’s responsibility, documentation is accessible to central administration and is the basis of compensation, promotion and disciplinary decisions. Links to an organizational system of punishment and reward, however, tend to compromise the accuracy or veracity of the information individuals are willing to elicit (Townley, 1997).
Incentive-oriented Speklé and Verbeeten (2014) use the term incentive-oriented when the role of the performance measurement system emphasizes target setting, incentive provision, and rewards.
Diagnostic Diagnostic control systems are the formal information systems that managers use to monitor organizational outcomes and correct deviations from preset standards of performance (Simons, 1995, p.59)
Cybernetic By cybernetic, we mean a process in which a feedback loop is represented by using standards of performance, measuring system performance, comparing that
6
the system, and modifying the system's comportment (Green and Welsh, 1988, p. 289)
Figure 1: conceptually similar types of PMS (judgmental)
2.1.2.2. Developmental PMS
The developmental type seeks to identify strengths and weaknesses of individuals and to develop their skills. It recognizes performance management as a way of acquiring the commitment and trust of individuals and it is largely focused on the future (cf. McGregor, 1972; Randall, 1989; Townley, 1997). Conceptually associated to the developmental type, as can be seen in figure 2, are the exploratory use of Speklé and Verbeeten (2014), and Simon’s (1995) interactive use of PMS. Furthermore, these types of PMSs could assist in identifying policy areas that are in specific need of political or managerial attention, facilitate selective intervention and priority setting, and facilitate a valuable search for new policy approaches. This gently leads to a common frame of reference as to what is regarded as satisficing performance and it provides input on how this can be achieved (cf. Burchell et al., 1980; Mundy, 2010; Simons, 1995; Speklé, 2001). Instead of focusing on keeping things on track, this PMS is about the discovery of sensible tracks, fostering autonomy and facilitating the subordinates in better serving their organization. For instance, university professors are not afraid or feel pressured to perform. They are given the responsibility to determine which research topic will have the most promising results within their literature field. There is no additional pressure or fear of negative consequences that affects their behavior. Thus, in contrast to a judgmental PMS, a developmental PMS actually gives them the freedom and confidence in executing their work.
Type of PMS Definition
Developmental A developmental system sees organizational benefits accruing from individual commitment to, and trust in, the scheme. These schemes are designed to identify individual strengths and weaknesses and develop skills and abilities. Intended to facilitate the flow of information and reduce mistrust, they deny the link between appraisal and reward/punishment of traditional or judgmental appraisal schemes (Townley, 1997).
Exploratory The exploratory use corresponds to strategy formation and communication of goals, strategy management and learning, and the strategic decision making role (Speklé and Verbeeten, 2014).
7
organization. They provide frameworks, or agendas, for debate and motivate information gathering outside of routine channels (Simons, 1995, p. 96). Figure 2: conceptually similar types of PMS (developmental)
In sum, this study mainly focuses on the use of developmental- and judgmental PMSs. Both types are used differently and can have different purposes and effects on the behavior of employees. PMSs are part of the management controls used by an organization and it depends upon the context and goals of the organization which type should be used. In subsequent sections the context and its effect for using a type of PMS will be explained.
2.2. Employee job performance
When we look at the definition of performance management, employee job performance can be defined as: the degree of alignment of an individual’s performance with the strategic goals of the organization. More alignment leads to better employee job performance, where less alignment leads to worse
performance. To create this alignment, management controls can be used (a set of processes and
mechanisms) to influence the behavior of individuals towards the strategic goals. A PMS accommodates this purpose and is a management control that focuses on controlling employee behavior in order to create goal alignment. Since goal alignment leads to better employee performance, it can be stated that there is a direct relationship between PMSs and employee job performance.
2.3. Context and the effect of PMS on employee job performance
This study focuses on two types of PMSs that are used within the public sector: judgmental and
8
illustrate a situation of high contractibility think about individuals working on a garbage truck. They have a designated path from which they need to collect the garbage, a strict time frame to perform this task, and their performance is being measured in a very quantitative way (is all the garbage collected that we are responsible for within the designated time frame?). These employees cannot determine which garbage to pick up, how much garbage, and the amount of time needed. In such a situation using a judgmental PMS is suited, because goals regarding the amount of garbage to pick up and accurate measures of performance are easy to determine. Finally, they know and control the process that transforms efforts into results (picking up garbage), and they are capable in predicting the expected outcomes of alternate courses of action (not picking up garbage).
In contrast, when facing low contractibility, the type of PMS should be different from situations of high contractibility. In situations of low contractibility performance measures will only produce a partial representation of the companies’ ultimate goals. In addition, during these circumstances it is recognized that the selection of appropriate performance measures is problematic (Hyndman & Eden, 2000). Strongly focusing on result targets will probably produce dysfunctional consequences because, as a result of the accompanied incentives, organizational actors will aim attention at the realization of their own targets instead of focusing on the organizational goals (Speklé and Verbeeten, 2014). Take for example a police officer. Defining goals regarding the number of arrests to make, or the maximum amount of time available to solve a crime, is problematic. Moreover, accurate measures of performance are hard to determine due to the varied nature of the job content. Finally, it is not always clear how and if a crime can be solved and if there are other options or potential solutions. In such a situation of low contractibility, the focus should not be on meeting predetermined targets (judgmental PMS), but on double loop learning, mutual consultation, experimentation, and consecutive adaptation to emerging insights, gently leading to a joint set of goals and an understanding of how these goals can be achieved (Speklé and Verbeeten, 2014). A developmental PMS provides this and is therefore appropriate in situations of low contractibility.
Interestingly, Speklé and Verbeeten (2014) found a positive performance effect of the exploratory use (conceptually similar to developmental type) which appears to exist independent of the level of contractibility. Moreover, they found that an incentive oriented performance measurement system (conceptually similar to judgmental PMS) negatively influences performance. Therefore, I hypothesize: H1: The developmental PMS positively affects employee job performance within public organizations in situations of low contractibility.
9 2.4. Task autonomy
In the subsequent sections task autonomy will be introduced. The degree of task autonomy can be seen as a consequence of the type of contractibility (high or low) that an individual is facing. In addition, the degree of task autonomy is also affected by the type of PMS that is being used. The effect of high- and low contractibility and the two types of PMS on task autonomy will be made clear. In addition, the effect of task autonomy on employee job performance will be hypothesized.
2.4.1. The type of PMS and its effect on task autonomy
Task autonomy is defined as the degree to which an individual is given substantial freedom,
independence, and discretion in carrying out a task, such as scheduling work and determining procedures to follow (Hackman, 1980). According to Humphrey and Morgeson (2006), task autonomy can be defined by three indicators: (1) work scheduling autonomy; (2) decision-making autonomy; and (3) work methods autonomy. Low levels of task autonomy can be found in situations of high contractibility. Recall the example of individuals working on a garbage truck. They are assigned to do a specific job and do not schedule their own work and time, they cannot determine themselves which garbage to pick up, and the way to fulfill their work is pre-determined (use the garbage truck). The degree of task autonomy can therefore be seen as an expression of freedom. By using a judgmental PMS organizational actors will be constrained in their freedom, independence, and discretion in carrying out a task. Thus, it can be argued that using a judgmental PMS will diminish an employees’ degree of task autonomy. In contrast, by emphasizing the use of developmental PMSs the degree of task autonomy will be increased. This type of a PMS is about the discovery of sensible tracks, fostering autonomy and facilitating the subordinates in better serving their organization. Accordingly, it can be concluded that the degree of task autonomy is affected by the type of PMS that is being used.
2.4.2. Task autonomy and employee job performance
Based on the aforementioned literature it can be argued that employees should be given higher degrees of task autonomy in situations of low contractibility. In such a situation, the selection of appropriate
10
constitutes performance and is given discretion to carry out his task. Even in general it is likely that giving task autonomy to employees will result in higher satisfaction, motivation, and performance (Argote & McGrath, 1993; Dwyer, Schwartz, & Fox, 1992; Loher, Noe, Moeller, & Fitzgerald, 1985; Spector, 1986; Langfred & Moye, 2004). In addition, the commonly acknowledged causal mechanism that links task autonomy to performance is motivation. According to Hackman and Oldham (1975), the motivating potential of a job is determined by five job characteristics. Task autonomy is one of these five
characteristics. The standard principle is that jobs will be more motivating and satisfying if high levels of these characteristics are present (Humphrey & Morgeson, 2006). This is of course valuable for
organizations because motivated employees are likely to show better performance (Drake, Wong, and Salter, 2007).
Since high levels of task autonomy are more appropriate in situations of low contractibility, than in situations of high contractibility, and due to its positive effect on performance, I hypothesize:
H3: The degree of task autonomy positively affects employee job performance within the public sector in situations of low contractibility.
Furthermore, let us remember once again the example of the individuals working on the garbage truck. These civil servants cannot determine which garbage to pick up, how much garbage, and the amount of time needed to do this, as a result of the judgmental PMS. However, as stated previously, it can be argued that in such a situation of high contractibility a judgmental PMS, and therefore lower levels of task autonomy, are suitable. In contrast, a developmental PMS increases an individuals’ task autonomy which is more appropriate in situations of low contractibility. In the above section the positive effect of task autonomy on employee job performance in situations of low contractibility has been hypothesized. Thus, employee job performance is affected indirectly by the type of PMS used, due to the direct effect of the type of PMS on task autonomy. This tells us that task autonomy is a mediator. “A simple mediation model
is any causal system in which at least one causal antecedent X variable is proposed as influencing an outcome Y through a single intervening variable M” (Hayes, 2013, p. 86). In plain words: X influences Y,
through M. In our model X is the type of PMS, Y is employee job performance and M is task autonomy. Therefore, I hypothesize:
H4: A judgmental PMS is expected to negatively affect employee job performance in situations of low contractibility within the public sector, through its negative effect on task autonomy.
11 2.5. Control variables
According to Chenhall (2003) the appropriate design(s) of MCs will be influenced by the context within which they operate. Two of these variables are organizational culture and organizational structure. These variables together with age, gender, and educational background are the control variables within this study. Control variables are included to prevent alternative explanations for our findings (Schmitt & Klimoski, 1991), to reduce error terms, and to increase statistical power (cf. Schwab, 1999).
2.5.1. Organizational structure
Organizational structure is about the formal specification of different roles for organizational members, or tasks for groups, to ensure that the activities of the organization are carried out. Burns and Stalker (1961) discuss structure, generically, in terms of mechanistic and organic approaches. The means to achieve these forms of structure involve mechanisms such as rules, procedures and openness of communications and decision processes. Structural arrangements influence the efficiency of work, the motivation of
individuals, information flows and control systems and can help shape the future of the organization. Structure also includes the motivation with which work is performed.
2.5.2. Organizational culture
Because mine is a single country study, national culture does not play a role in our control configurations. Instead, I examine organizational culture, which may actually have a stronger effect on the design of MC than national culture (Chenhall, 2003). Hofstede (1991) defines organizational culture as the manifestation of practices/behaviors evolving from the shared values in the organization.
2.5.3. Age
According to Ng and Feldman (2008) age is strongly related to several dimensions of job performance. Their pattern of findings suggest that older employees are more likely to control their emotions at work, are good citizens, and are less likely to engage in counterproductive behaviors.
2.5.4. Gender
12
2.5.5. Education level
“Education level refers to the academic credentials or degrees an individual has obtained” (Ng and Feldman, 2009, p. 91). These authors found that educational level is positively related to task performance. In addition, they argue that education level is negatively related to two specific counterproductive work behaviors. Finally, highly educated workers are likely to contribute more effectively to noncore activities at work as well (Pennings, Lee, & van Witteloostuijn, 1998).
2.6. Conceptual model
Figure 3: Conceptual model
H1 Developmental PMS + Employee job performance
H2 Judgmental PMS - Employee job performance
H3 Task autonomy + Employee job performance
H4 Judgmental PMS - Task autonomy +
*
Employee jobperformance
H5 Developmental PMS + Task autonomy +
**
Employee jobperformance Figure 4: Hypotheses overview
* -
*+
= -
**+
*+ = +
3.
Methodology
The following chapter provides an overview of the approach used within this study, in which way the data is collected, and what the sample of this study is. In addition, an overview is provided of the different measurements.
3.1. Approach
13
However, this phenomenon has not been completely discussed by the academic literature. Literature streams related to this business phenomenon are quite elaborated and not scattered. Nevertheless, the evidence on the theoretical explanations of the business phenomenon is still inconclusive, so there is a gap in the literature field that must be (partly) closed (van Aken, et al., 2012). I conduct a theory testing approach since this is the case for existing management accounting research which focuses on performance management within the public sector.
3.2. Data collection and sample
The municipality of Groningen is an example of local government within the Dutch public sector. Thus, to capture the effect of the use of the two types of PMS and task autonomy within the Dutch public sector, this analysis is based on survey data from 119 employees of the Dutch municipality of Groningen. Within the municipality a subsequent distinction is made towards what is called the social domain (sociaal domein). This municipality and the subsequent distinction is chosen due to the access and availability of data. The use of a survey allows me to gather the perceptions of employees on the micro level and is most commonly employed for theory testing in management accounting research (Van der Stede et al., 2005). The questionnaire is spread out by sending an anonymous link towards a wide variety of different departments within the municipality. By using an anonymous link I tried to obtain as much data as possible. Before sending this link, the approval of the department’s manager was obtained. Specific instructions should ensure that only relevant employees did fill in the questionnaire. Finally, after distributing the questionnaire, two reminders were send during the data collection period in order to increase the response rate.
This study focusses on the use of two types of PMSs in situations of low contractibility.
14
Moreover, the legal standard suggests that survey questions should always be pre-tested to assess whether they can ‘‘... be correctly understood by respondents and easily answered by them’’ (Morgan, 1990, p. 64). The survey instrument is pre-tested, as is suggested by Dillman (1978; 1999), by sending them to the scrutiny of two groups of people: colleagues and prospective respondents. Instead of
colleagues, fellow students were used. They are addressing the same topic in their study. From each group two persons were selected. After receiving feedback from both groups of people, adjustments were made. According to Dillman (1978; 1999) the questionnaire should also be pre-tested by the users of the data. However this group is not relevant for this study and can therefore not be used. It is important to note that, before the pre-testing occurred, a translation was needed (from English to Dutch) since our respondents are Dutch. To increase the reliability of the translation, this process was also done together with a fellow student.
3.3. Measurements
I rely as far as possible on instruments validated in previous studies.
3.3.1. Employee job performance (EmPerf)
I measure employee job performance with a well-established instrument developed by Van de Ven and Ferry (1980). This instrument is specially created to capture performance in public sector organizations. These seven performance dimensions include: (1) productivity, (2) quality or accuracy of work produced, (3) the number of innovations(4) reputation for work excellence, (5) attainment of production or service level goals, (6) efficiency of operations, and (7) morale of employee. Respondents are asked to indicate their score on each dimension, using a five point semantic scale (1 = far below average; 5 = far above average).
3.3.2. Type of PMS (Judg; Develop)
To capture the effect of the type of PMS on the performance of employees, I ask respondents to reflect on the degree to which the judgmental type and the developmental type are being used (1 = very low extent; 7 = very high extent). The questions and categories are based on the survey instrument used by Bedford, Malmi and Sandelin (2016). They gather perceptions regarding the use of diagnostic and interactive performance measurement systems. Since these are conceptually similar to judgmental- and
developmental PMSs, they provide an appropriate instrument to use within this study.
3.3.3. Task autonomy (TaskAut)
15
potential manifestations of this construct. Humphrey and Morgeson (2006) define task autonomy by three indicators: (1) work scheduling autonomy; (2) decision-making autonomy; and (3) work methods
autonomy. A five point semantic scale (1 = very low extent; 5 = very high extent) is used to indicate the score of employees on each of the dimensions.
3.3.4. Organizational culture (Flexibility; Control)
For organizational culture I follow the approach of Kruis et al. (2016). They follow Henri (2006) who distinguishes flexibility- and control values. Flexibility values reflect adaptability, openness to change and responsiveness, whereas control values reflect stability, formality, and predictability (Henri, 2006). Furthermore, I used the questionnaire provided by Kruis et al. (2016) who form four variables: group culture, developmental culture, hierarchical culture, and rational culture. The flexibility value is created by combining the items of group culture and the items of developmental culture. In addition, the items for rational culture and hierarchical culture together form the control value. A seven point scale is used (1 = very low extent; 7 = very high extent).
3.3.5. Structure (Structure)
We again use the questionnaire provided by Kruis et al. (2016). Five questions measure the construct of structure. Lower scores of this construct refer to the use of mechanistic structural controls while higher scores refer to the use of organic structural controls. A seven point scale is used (1 = very low extent; 7 = very high extent).
4. Analysis and results
Chapter four provides an overview of the data that was obtained within this study and how this data was transformed to make it suitable for data analysis. Subsequently, the descriptive results, regression results and impact on my hypotheses are provided.
4.1. Raw Data
After receiving the input from my questionnaires, I tested the raw dataset for missing values as is
16
excluding the control values, range from 0,735 to 0,905, indicating adequate reliability (Kline, 1999). For each of the multi-item variables, I have used the respondents’ ratings of the relevant items of each construct and divided them by the number of items to obtain the composite scale for each variable.
Variable
Items
Perform (employee job performance) Cronbach’s a – 0,735
Judg (use of judgmental PMS) Cronbach’s a - 0,856
Develop (use of developmental PMS) Cronbach’s a – 0,850
Taskaut (the degree of task autonomy) Cronbach’s a - 0,905
FlexC (Flexibility values) Cronbach’s a - 0,806
Productivity
Quality or accuracy of work produced
The number of innovations Reputation for work excellence Attainment of production or service
level goals
Efficiency of operations Morale of employee
Identify critical performance variables
Set targets for critical performance variables.
Monitor progress
Provide information to correct deviations
Review key areas of performance Provide overview of top
management activities
Provide overview of subordinate activities.
Challenge and debate
Focus on strategic uncertainties Dialogue and information sharing Freedom in scheduling work Freedom in determining work order Freedom in planning
Personal initiative or judgment Freedom in decision making Autonomy in making decisions Freedom in determining work
methods
Independence and freedom in performing the work.
17
ContC (Control values) Cronbach’s a - 0,374*
Struct (Structure) Cronbach’s a - 0,762
Meeting new challenges Bureaucratic environment Formal policies Goal accomplishment Stability Communication of control information Accessibility of operational information Content of work-related communication
The operating management philosophy (decision making) The operating management
philosophy
(coordination/adaptation)
*The control values are dropped from the analysis because reliability is too low. If the stability question is deleted the Cronbach’s alpha would rise to 0,495 which is still insufficient.
Figure 5: Overview of variables, their components and their reliability
In addition, I tested my dependent and independent variables for outliers, however the results of my boxplot did not show any problems. Furthermore, as recommended by Field (2009), the dependent variable was tested for normality and homoscedasticity. First of all, to test for normality, I used the Shapiro-Wilk test. The Shapiro-Wilk test compares the scores in the sample to a normally distributed set of scores with the same mean and standard deviation. If the test is non-significant (p > .05) it tells me that the distribution of the sample is not significantly different from a normal distribution. If, however, the test is significant (p < .05), then the distribution in question is significantly different from a normal
distribution (Field, 2009). Results showed that my dependent variable is non-normally distributed (p = 0,006). However, this test has its limitations because with large sample sizes it is very easy to get
significant results from small deviations from normality. Even though a large sample remains undefined, the significant test does not necessarily tell me whether the deviation from normality is enough to bias any statistical procedures that I apply to the data. As recommended by Field (2009), I also plotted the data and tried to make an informed decision about the extent of non-normality. The normality plot actually shows that the cloud of dots is more or less evenly spaced around the line, which does indicate a normal
18
and based on the output coefficients of the obtained values of respectively Judg (p = 0,987), Develop (p = 0,171), and Taskaut (p = 0,350) it can be concluded that these values are greater than the significance level of 0,05. This indicates homoscedasticity and that there is not a heteroscedasticity problem (Glejser, 1969). The outcomes and plots for these different tests can be found in appendix B. As recommended by Belsley, Kuh, and Welsch’s (1980) my independent variables are tested for multicollinearity, but no problems were found; the Variance Inflation Factors (VIF) ranged from 1,044 – 1,420. Thus, mean-centering all
variables, as recommended by Aiken and West (1991), was not necessary.
Variable VIF
Judgmental (Judg) 1,373 Developmental (Develop) 1,420 Task Autonomy (Taskaut) 1,044 Figure 6: Variance Inflation Factors
After the multicollinearity test, I have standardized relevant variables to equate their impact on employee job performance. Variables with larger ranges and variances could have a bigger impact on the dependent variable (Henry, Tolan, & Gorman-Smith, 2005). It is necessary to standardize, when the theoretical ‘weight’ of each of the variables is considered similar (Milligan, 1996). Since the variables are considered of similar importance, they are standardized by transforming them into z-scores. After standardizing the variables, the dataset was completed for analysis.
4.2. Results
Figures 7, 8, and 9 provide an overview of the descriptive statistics for the variables used within this study. Our sample consists of 34 men (28,6%) and 85 women (71,4%), so a total of 119 respondents with an average age of 44 years (Mage = 44,40, SD = 10,937). Furthermore, three employees have secondary education (2,5%), twelve finished intermediate vocational education (10,1%), seventy four finished higher vocational education (62,2%) and 30 have a university degree (25,2%). Thus, the major part of my sample (87,4%) has higher education or a university degree. This is not a surprising result since my study focusses on situations of low contractibility which in most situations require at least higher vocational education. Furthermore, the descriptive statistics show that the employees perceive a higher presence of
19
The performance of both men and women can be considered equal with a slightly higher standard deviation for women. At last, both men and women perceive a higher presence of organic structural controls than mechanistic structural controls. Because both groups also perceive a higher presence of developmental PMSs, these results are not surprising.
Variable Mean Std. dev Theoretical range Minimum Maximum Education level 5,100 0,669 1-6 3,00 6,00 Age 44,400 10,937 - 22,00 63,00 Judg 3,653 1,166 1-7 1,00 6,00 Develop 4,329 1,082 1-7 1,20 6,40 Perf 3,516 0,401 1-5 2,71 4,71 Taskaut 3,907 0,545 1-5 2,44 5,00 FlexC 4,661 0,803 1-7 2,00 6,67 Struct 4,668 0,994 1-7 1,40 7,00
Figure 7: Descriptive statistics of total sample
Gender: Men Mean Std. dev Theoretical range Minimum Maximum Education level 5,150 0,744 1-6 3,00 6,00 Age 47,47 11,076 - 25,00 63,00 Judg 3,882 1,226 1-7 1,80 6,00 Develop 4,470 1,237 1-7 1,20 6,40 Perf 3,500 0,396 1-5 2,71 4,29 Taskaut 3,875 0,534 1-5 2,78 5,00 FlexC 4,607 0,818 1-7 3,33 6,67 Struct 4,505 1,148 1-7 1,40 6,40
Figure 8: Descriptive statistics of gender men
Gender: Women
Mean Std. dev Theoretical range Minimum Maximum Education level 5,080 0,640 1-6 3,00 6,00 Age 43,180 10,753 - 22,00 63,00 Judg 3,562 1,135 1-7 1,00 6,00 Develop 4,272 1,016 1-7 1,60 6,00 Perf 3,522 0,404 1-5 2,71 4,71 Taskaut 3,920 0,552 1-5 2,44 5,00 FlexC 4,682 0,801 1-7 2,00 6,50 Struct 4,734 0,924 1-7 1,60 7,00
Figure 9: Descriptive statistics of gender women
20
positive correlation can also be found between task autonomy (Taskaut) and the use of the developmental PMS. This is not unexpected, since theoretically the degree of task autonomy is also positively associated with the developmental PMS. In addition, the only significant correlation with employee job performance (Perf) is structure (Struct). Moreover, gender and age are not significantly correlated to either the
dependent variable or to any of the independent variables. As a result, these two variables will be removed as control variables during further analysis.
*p < 0,05; **p < 0,01 (1-tailed)
Figure 10: Correlation matrix 4.3. Regression
To test the first three hypotheses, I performed a multiple regression using forced entry. A multiple regression analysis is a way of predicting an outcome variable from several predictor variables (Field, 2009). Forced entry is a method in which all independent variables are forced into the model
simultaneously. This method relies on good theoretical reasons for including the chosen independent variables and the researcher makes no specific decision about the order in which variables are entered (Field, 2009). However, one of the assumptions for performing a regression is that the type of data should be continuous or should be measured at least at the interval level (Field, 2009). This is not the case for our data which is measured at the ordinal level. Despite this difference, researchers often use continuous methods in spite of the variables’ ordinal nature (Rhemtulla et al., 2012). When the number of categories reaches four or five, researchers generally work under the assumption that ordinal variables are very much alike to continuous variables, and will produce good results (cf. Beauducel & Herzberg, 2006; DiStefano, 2002; Dolan, 1994; Johnson & Creech, 1983; Muthén & Kaplan, 1985). Since my variables range from five to seven categories, I chose to treat my ordinal variables as continuous data.
21
Subsequently, I regressed my independent variables together with the control variables structure,
flexibility culture, and education level, which will be reported as model 2. Having two models shows how much of the variation in our dependent variable is determined by the independent variables and by the control variables (Field, 2009). The results of the regression analysis are presented in figure 11.
B SE B β Model 1 Constant 3,440 ,293 Judg -,064 ,037 -,186 Develop ,054 ,041 ,145 Taskaut ,020 ,069 ,027 Model 2 Constant 3,401 ,432 Judg -,056 ,038 -,162 Develop ,026 ,049 ,071 Taskaut -,014 ,081 -,020 Education level -,007 ,060 -,011 FlexC ,002 ,061 ,004 Struct ,061 ,045 ,151 Note: Model 1 R2 = .030, ∆ R2 = 0.005; for Model 2 R2 = ,046, ∆ R2 = -,005 (p < .001). * p < .001.
Figure 11: Regression results for employee job performance
My first model has a R2 of 0,030. This means that model 1 only explains 3% of the variance in
performance. The second model (including the control variables) explains slightly more of the variance, a total of 4,6%. Furthermore, my adjusted R2 (∆ R2) gives us an idea of how well my model generalizes and
ideally the value should be the same, or very close to, the value of R2 (Field, 2009). The difference
between the two values for model 2 is 0,051 (5,1%). This reduction means that if the model were derived from the population, instead of a sample, it would account for about 5,1% less variance in the outcome (Field, 2009).
Hypothesis1 predicted that the use of a judgmental PMS would have a negative impact on employee job performance. The regression did not show a significant correlation between the use of a judgmental PMS and employee job performance (p = 0,143). There is no evidence found for this
22
on employee job performance. The results did not find a significant relationship between task autonomy and employee job performance (p = 0,859). We reject H3.
To test my two mediation hypotheses (H4 and H5), I used Hayes’s PROCESS tool. According to Field (2013) this is by far the best way to tackle mediation. Andrew Hayes (2012) developed the tool and it computes all of the effect size measures that I have hypothesized in my two mediation hypotheses. Thus, this test provides a direct test of the indirect effect of my independent variables on my dependent variable through the mediator. This indirect effect represents how employee job performance is influenced by the type of PMS through a causal sequence. In this causal sequence the type of PMS influences task
autonomy, which in turn influences employee job performance (Hayes, 2013). To perform the test, I employed the bootstrapping approach, thereby making no assumption about the distribution of indirect effect and it provides confidence intervals for the estimate (Preacher & Hayes, 2004; Shrout & Bolger, 2002). Results indicated that the indirect effect of the use of a judgmental PMS on employee job
performance was not significant: (b = 0,0006; 95% CI [-0,0055, 0,0137]. The fact that the range includes 0, means that it is likely that there is not a real indirect effect (Field, 2013). In addition, the use of a judgmental PMS is not a significant predictor of task autonomy (b = -0,018; p = 0,681). Furthermore, task autonomy is also not a mediator between the use of a developmental PMS and employee job performance (b = 0,0006; 95% CI [-0,0082, 0,0178]. In addition, the use of a developmental PMS does not significantly predict task autonomy (b = -0,042; p = 0,453). The results again did not support hypotheses 4 and 5. Figure 12 provides an overview of the main analytical results.
23
5.
Discussion and conclusion
First of all, in the discussion and conclusion section the results are summarized. In this summary, theoretical explanations are given for the observed results. Afterwards the limitations, implications, and directions for future research are provided.
5.1 Summary of Results
This paper is one of the first that researched empirically the effect of the perceived use of a PMS on an individual’s performance within the public sector. The findings did not provide the expected results and donot match to the body of research underneath this study.
First of all, the results did not show any significant impact of the use of a judgmental PMS on employee job performance. These findings were unexpected because the economics and behavioral literature suggests that these PMSs are only appropriate in situations of high contractibility. In addition, Speklé and Verbeeten (2014) did find the evidence for a negative impact of a judgmental PMS on performance. Furthermore, explorative research revealed that the use of judgmental PMSs can have several negative consequences like stress, anxiety, and playing safe which were expected to negatively influence performance. However, it is possible that certain people are in general just more susceptive to these feelings or negative consequences, while the majority is not influenced. The locus of control
concept, developed by Rotter (1966), is a framework for understanding people’s perceptions regarding the controlling factors in their lives. In this framework a distinction is made between an internal locus of control and an external locus of control. People that have an internal locus of control have the tendency to believe that they control their own reactions and emotional states. In contrast, people with an external locus of control have the tendency to believe that their emotions, perceptions, and lives are circumstance dependent. Thus, people that are leaning more towards an internal locus of control could actually not perceive these negative consequences, or their behavior is just not influenced by it.
Second, the idea of a developmental PMS is that it gently leads to a common frame of reference as to what is regarded as satisficing performance, and it provides input on how this can be achieved.
However, no evidence was found for a positive effect of the use of a developmental PMS on employee job performance. Again, the locus of control could provide a potential explanation. Even though employees perceive the presence of this type of PMS, their behavior is not influenced by it due to a more internal locus of control. Another possible explanation could lie within the work of Schneider, Beatty, and Baird (1987). They classified a PMS into three phases, namely a development and planning phase, a managing and reviewing phase, and a rewarding phase. It could be that, despite the perceived presence of this PMS, the effect is not being managed or reviewed by management. Thus, employees do understand what
24
expected performance. Management should intervene in such a situation, however, due to a lack of reviewing and managing performance, they are not able to intervene and get the performance back on track. Perhaps management is overly relying on the use of the developmental PMS and also a judgmental type of PMS is needed to manage and review the performance.
Third, despite that the degree of task autonomy can be considered high (mean of 3,90), no positive effect of task autonomy on employee job performance was found. Several explanations can be found within the literature. First of all, even though in general empirical support can be found for a positive relationship between task autonomy and performance (Spector, 1986), according to Langfred and Moye (2004) the effect size remains modest. Therefore, the positive effect of task autonomy hypothesized within this study could in practice be not as prevalent as was expected. This line of reasoning is corresponding to the work of Godard (2001); Wall, Kemp, Jackson, & Clegg (1986); and Langfred & Moye (2004). They argue that the positive effect of task autonomy on performance is much more subtle than the theoretical models have suggested. In addition, Langfred and Moye (2004) conclude that autonomy should definitely not be given to an employee who is not motivated by autonomy. Thus, another explanation could also lie within the personality and motivation of the employees.
Fourth, due to NPM an increase in the use of judgmental PMSs was noticed in recent decades. However, my descriptive results showed that employees still perceive a higher presence of developmental PMSs than judgmental PMSs. These results can also be observed in the study of Speklé and Verbeeten (2014). The perceived higher presence of developmental PMSs contradicts the idea of NPM that judgmental PMSs are seen as an improvement over the developmental PMS, because in practice the developmental type is still more prevalent. In addition, a positive correlation between the two can be found; the presence of one type of PMS also affects the presence of the other type of PMS. Thus, the usage of the different types of PMSs are not mutually exclusive. This is in line with the reasoning of Widener (2007) who argues that the two different types of PMSs actually complement each other and are interrelated. Furthermore, due to this positive correlation, one can question if the theoretical framework underneath this study is actually appropriate. Chenhall (2003) argues that much of Management Control Systems (MCS) research examines single themes or practices. However, even though these themes or practices and their contexts appear to be not related to each other, they are in fact part of a broader control system. This is supported by Malmi and Brown (2008) who also argue that MCS do not operate in
isolation. Accordingly, it is questionable that a clear distinction between the existence of two different types of PMS is in fact correct. This study, and the arguments provided by Chenhall (2003), and Malmi and Brown (2008), could indicate that there is in fact just one type of PMS, which consists of a
25 reason for not finding any significant relationships.
Thus, the obtained data from our sample was not able to provide significant evidence of the hypothesized relationships. The research question: How does the type of PMS affect employee job performance within the public sector, cannot be answered conclusively. Our findings suggest that the use of a type of PMS does not affect employee job performance within the public sector.
5.2. Limitations
This study has multiple limitations. The data was obtained solely from one organization within the Dutch public sector. No other data was gathered to study the extent to which this particular context varies across other Dutch public sector organizations or across different countries. Moreover, management was
excluded from the sample. Excluding management diminishes generalizability across the entire Dutch public sector. In addition, this study controlled only for one part of culture, namely the flexibility values. . Eliminating the control values, due to inadequate reliability, means that no overall conclusion can be drawn regarding the controlling power of culture.
Another limitation is that this study made use of an anonymous link in order to obtain as much data as possible. Even though we tried to make as clear as possible that there are certain requirements (paygrade; and non-management) to qualify for this questionnaire, the possibility is that the results are biased. Potentially people could have filled in the questionnaire who do not work in situations of low contractibility or are part of management.
Furthermore, the assumption made that pay-grade is a representative border for situations of low contractibility could also be questionable. It could be possible that situations for people with lower paygrades are actually characterized by low contractibility and vice versa.
Subsequently, our respondents were asked to rate their own performance with respect to their surrounding colleagues who are probably subjected to the same type of PMS. Comparing performance within departments, and not between departments like Speklé and Verbeeten (2014) did, could be a limitation. When we compare the performance variable’s given scores between the two studies, our range is indeed smaller. Besides that, their analysis is based on data from 97 different business units. Chances are within our study that relatively a large amount of the respondents work within the same department (that is the department I am working in) and smaller amounts vary across a large(r) degree of different departments. Perhaps this resulted in an insufficiently diversified sample due to too many respondents working within the same environment. Furthermore, my study also does also not meet the threshold of 200–300 respondents which appear to be able to achieve a valid degree of face validity in court (Morgan, 1990). These factors potentially could explain why our model only predicts 4,6% of employee job
26
difference between R2 and adjusted R2 which could indicate that the cross-validity of our model is not very
good (Field, 2009).
Another limitation can be found regarding the measurement of the variables. The findings are based on perceptions and perceptions can be biased. For example, this study relies exclusively on self-reported performance, and does not research if, for instance, management or supervisors are satisfied with the achieved results of personnel.
Despite my confidence in validated instruments of previous research and pre-testing the
questionnaire, the proxies could potentially misrepresent some important concepts in the analyses due to inappropriate measures or insufficient interpretation of the questionnaire (cf. Ittner and Larcker, 2001). A final limitation is that additional factors such as behavioral controls, the organizational- or departmental size, type of employment, or mutual trust among employees and managers could impact the use and effect of PMSs. These variables are not taken into account within this study. In addition, the effect of the use of the different types of PMSs are tested individually on employee job performance. As argued before, it is plausible that these management controls do not operate in isolation and are part of a broader management control package. Chenhall (2003) claims that studying MCS elements in isolation could lead to the possibility of model under specification.
In despite of these limitations, this study appears to have a couple of interesting implications, both for practice and for policy within the public sector. Trying to improve the overall performance of
employees, and thus for the entire public sector, is unquestionably very important for policy makers and management. However, emphasizing the use of a judgmental PMS, as is recommended by NPM, is probably not the entire solution. Actually, our results indicate that the two different types of PMS are not mutually exclusive, probably need each other to function well, and could operate as one package. For example, focusing more on judgmental components of a PMS could be necessary to make sure that employees are really going to use the performance information and guidance provided by the
developmental components of a PMS. This indicates the need for a more balanced use of both types of PMSs. Even more, the results show that a developmental PMS is used more than the judgmental PMS. Thus, to enhance the efficiency of the public sector policy makers should not merely focus on the ideas of NPM and emphasize judgmental PMSs, but they should expand their frame of reference.
5.4. Future research
27
objective and better information to make such a comparison. Our respondents were asked to rate their own performance with respect to their surrounding colleagues who are probably subjected to the same type of PMS. It could be fruitful to combine this study with the one of Speklé and Verbeeten (2014). For example, by letting employees rate the degree of the different types of PMS being used within their subunit, and subsequently let management rate the performance of the unit.
Previous research indicated that people who have a more internal locus of control are likely to be more self-reliant and benefit from having a greater opportunity for self-control. In contrast, people with an higher degree of the external locus of control appear to listen to others’ judgments and benefit from the structuring of tasks by others (cf. Cromwell et al., 1961; Julian and Katz, 1968; Crowne and Liverant, 1963). Thus, employees’ locus of control could be an interesting mediator for future research, in examining the relationship between PMSs and employee job performance.
Finally, the results indicated that the sample was probably not strong enough to explain the predicted relationships. Future research could conduct a similar study, with a bigger and more diversified sample in order to obtain more valid results. Accordingly, this future research could examine to which extent our findings can be generalized to other institutional settings. However, in doing any of this future research, it is wise to consider if such a clear distinction can and should be made between different types of PMSs. The PMSs are not mutually exclusive and potentially do not operate in isolation. Thus,
understanding how these two types of PMSs, or their components, actually obstruct and need each other, and perhaps constitute as just one PMS, is necessary and could provide very interesting practical and theoretical insights.
References
Aguinis, H. (2013). Performance Management (3rd ed.). Pearson.
Aiken, L. S., West, S. G. (1991). Multiple Regression: Testing and Interpreting Interactions. Newbury Park: Sage Publications.
Anderson, G. (1994). Performance appraisal. In: B. Towers (Eds.) The Handbook of Human Resource
Management, 186–207. Oxford and Cambridge: Blackwell.
Argote, L., & McGrath, J. E. (1993). Group process in organizations. In C. L. Cooper & I. T. Robertson (Eds.), International review of organizational psychology, 333–389. New York: Wiley.
28
Baker, G. (2002). Distortion and risk in optimal incentive contracts. Journal of Human Resources, 37, 728-751.
Barber, B.M., & Odean, T. (2001). Boys Will Be Boys: Gender, Overconfidence, and Common Stock Investment. Quarterly Journal of Economics, 116, 261-292.
Beauducel, A., & Herzberg, P. Y. (2006). On the performance of maximum likelihood versus means and variance adjusted weighted least squares estimation in CFA. Structural Equation Modeling, 13, 186-203. Bedford, D. S., Malmi, T., & Sandelin, M. (2016). Management control effectiveness and strategy: An empirical analysis of packages and systems, Accounting, Organizations and Society, 51 (2016), 12-28. Belsley, D. A., Kuh, E., & Welsch, R. E. (1980). Regression diagnostics: identifying influential data and
sources of collinearity. New York: Wiley.
Bevan, G. & Hood, C. (2006). What’s measured is what matters: targets and gaming in the English public health care system, Public Administration, 84(3), 517-538.
Bogt ter, H. J., & Scapens, R. W. (2012). Performance management in universities: Effects of the Transition to More Quantitative Measurement Systems. European Accounting Review, 21(3), 451-479. Borman, W.C., & Motowidlo, S. (1993). “Expanding the criterion domain to include elements of contextual performance”, in Schmitt, N. and Borman, W.C.A. (Eds.), Personnel Selection in
Organizations, 71-98 Jossey-Bass, San Francisco, CA.
Burchell, S., Clubb, C., Hopwood, A., & Hughes, J. (1980). The roles of accounting in organizations and society. Accounting, Organizations and Society, 5, 5-27.
Burgess, S., & Ratto, M. (2003). The role of incentives in the public sector: issues and evidence. Oxford
Review of Economic Policy, 19, 285-299.
Burns, T., & Stalker, G. M. (1961). The management of innovation. London: Tavistock.
Chenhall, R. H. (2003). Management control systems design within its organizational context: findings from contingency-based research and directions for the future. Accounting, Organizations and Society, 28, 127-168.
Chenhall, R. H., & Langfield-Smith, K. (2007). Multiple Perspectives of Performance Measures.
European Management Journal, 25(4), 266-282.
Cromwell, R., Rosenthal, D., Shakow, D., & Kahn, T. 1961. Reaction time, locus of control, choice behavior and descriptions of parental behavior in schizophrenic and normal subjects. Journal of
29
Crowne, D. P., & Liverant, S. 1963. Conformity under varying conditions of personal commitment.
Abnormal Social Psychology, 66, 547-555.
Davis, J. H., Schoorman, F. D., & Donaldson, L. (1997). Toward a stewardship theory of management. The Academy of Management Review, 22(1), 20-47.
De Bruijn, H. (2002). Performance measurement in the public sector: strategies to cope with the risks of performance measurement. International Journal of Public Sector Management, 15, 578-594.
Dill, D. D., & Soo, M. (2005). Academic quality, league tables, and public policy: a cross-national analysis of university ranking systems, Higher Education, 49(4), 495-533.
Dillman, D. (1978). Mail and telephone surveys: The total design method. New York: Wiley. Dillman, D. (1999). Mail and internet surveys: The tailored design method. New York: Wiley.
DiStefano, C. (2002). The impact of categorization with confirmatory factor analysis. Structural Equation
Modeling, 9, 327-346.
Dolan, C. V. (1994). Factor analysis of variables with 2, 3, 5, and 7 response categories: A comparison of categorical variable estimators using simulated data. British Journal of Mathematical and Statistical
Psychology, 47, 309-326.
Drake, A. R., Wong, J., & Salter, S. B. (2007). Empowerment, Motivation, and Performance: Examining the Impact of Feedback and Incentives on Non-management Employees. Behavioral Research in
Accounting, 19(1), 71-89.
Dwyer, D. J., Schwartz, R. H., & Fox, M. L. (1992). Decision-making autonomy in nursing. Journal of
Nursing Administration, 22, 17-23.
Feltham, G.A., & Xie, J. (1994). Performance measure congruity and diversity in multi-task principal/agent relations. The Accounting Review, 69, 429-453.
Ferreira, A., & Otley, D. (2009). The design and use of performance management systems: An extended framework for analysis. Management Accounting Research, 20, 263-282.
Field, A. (2009). Discovering statistics using SPSS (3rd. ed.). London: Sage Publications.
30
Gibbons, R. (1998). Incentives in organizations. Journal of Economic Perspectives, 12(4), 115-132. Glejser, H. (1969). A New Test for Heteroskedasticity. Journal of the American Statistical
Association, 64 (235), 315–323.
Gneezy, U., Niederle, M., & Rustichini, A. (2003). Performance in Competitive Environments: Gender Differences. Quarterly Journal of Economics, Aug,1049-1074.
Godard, J. (2001). High performance and the transformation of work? The implications of alternative work practices for the experience and outcomes of work. Industrial and Labor Relations Review, 54, 776-805.
Green, S., Welsh, M. (1988). Cybernetics and dependence: reframing the control concept. Academy of Management Review, 13(2), 287–301.
Groot, T., & Budding, T. (2008). New public management’s current issues and future prospects. Financial
Accountability & Management, 24, 1-13.
Hackman, J. R., & Oldham, G. R. (1975). Development of the Job Diagnostic Survey. Journal of Applied
Psychology, 60, 159-170.
Hayes, A. F. (2012). PROCESS: A versatile computational tool for observed variable mediation,
moderation, and conditional process modeling [White paper]. Retrieved from http://www.afhayes.com/
public/process2012.pdf.
Hayes, A. F. (2013). Introduction to Mediation, Moderation, and Conditional Process Analysis: a
Regression-Based Approach. New York: Guilford Publications.
Henri, J. F. (2006). Organizational culture and performance measurement systems. Accounting,
Organizations and Society, 31, 77-103.
Henry, D. B., Tolan, P. H., & Gorman-Smith, D. (2005). Cluster Analysis in Family Psychology Research.
Journal of Family Psychology, 19, 121–132.
Hofstede, G. (1981). Management control of public and not-for-profit activities. Accounting,
Organizations and Society, 6, 193-211.
Hofstede, G. H. (1991). Cultures and Organizations: Software of the mind. London: McGraw-Hill. Hood, C. (1995). The ‘new public management’ in the 1980: variations on a theme. Accounting,
31
Humphrey, S. E., & Morgeson, F. P. (2006). The Work Design Questionnaire (WDQ): Developing and Validating a Comprehensive Measure for Assessing Job Design and the Nature of Work. Journal of
Applied Psychology, 91(6), 1321-1339.
Hyndman, N., & Eden, R. (2000). A study of the coordination of mission, objectives and targets in U.K. executive agencies. Management Accounting Research, 11, 175-191.
Ittner, C.D., Larcker, D.F., (2001). Assessing empirical research in management accounting: a value-based perspective. Journal of Accounting and Economics, 32, 349-410.
Jensen, M. C., & Meckling, W. H. (1976). Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure. Journal of Financial Economics, 3(4), 305-360.
Julian, J. W., & Katz, S. B. 1968. Internal versus external control and the value of reinforcement. Journal of Personality and Social Psychology, 8, 89-94.
Johnson, D. R., & Creech, J. C. (1983). Ordinal measures in multiple indicator models: A simulation study of categorization error. American Sociological Review, 48, 398-407.
Katic, D., Majstorovic, V., & Colak, I. (2011). Performance Measurement Review, Annals of DAAAM &
Proceedings: 22(1), 515-516.
Kline, P. (1999). The handbook of psychological testing (2nd ed.). London: Routledge.
Kruis, A. M., Speklé, R. F., & Widener, S. K. (2016). The levers of control framework: an exploratory analysis of balance. Management Accounting Research, 32 (2016), 27-44.
Langfred, C. W., & Moye, N. A. (2004). Effects of Task Autonomy on Performance: An Extended Model Considering Motivational, Informational, and Structural Mechanisms. Journal of Applied Psychology, 89 (6), 934 – 945.
Loher, B. T., Noe, R. A., Moeller, N. L., & Fitzgerald, M. P. (1985). A meta-analysis of the relation of job characteristics to job satisfaction. Journal of Applied Psychology, 70, 280-289.
Malmi, T., Brown, D.A., (2008). Management control systems as a package—opportunities, challenges and research directions. Management Accounting Research, 19, 287–300.
Marginson, S. and van der Wende, M., (2009). Europeanisation, international rankings and faculty mobility: three cases in higher education globalisation, in: Centre for Educational Research and
Innovation, Higher Education to 2030, 2, 109–144.
