Amsterdam Business School
The effect of leader-member exchange quality on the relationship
between Performance Indicators (PI) participation and PI
contractibility:
A survey study among operational employees
Name: Anne Op ‘t Landt Student number: 10259295
Thesis supervisor: dr. ir. B.A.C. Groen Date: 19-06-2016
Word count: 16,907
MSc Accountancy & Control, specialization Control
Statement of Originality
This document is written by student Anne Op ‘t Landt who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
The aim of the current study is to study the effect of the quality of the relationship between the employee and the supervisor on the relationship between participation of employees in the development of the performance indicators and the contractibility of the performance indicators. The research is based on a survey among 70 pairs of operational employees and their direct supervisors in various jobs and industries. The following is found: firstly, according to the first hypothesis, if operational employees participate in the development of the performance indicators, supervisors perceive the performance indicators as more sensitive, precise and verifiable. Secondly, contrary to the second hypothesis, theinteraction analysis revealed that the relationship between Performance Indicators (PI) participation and PI contractibility is not affected by the quality of the relationship between the employee and the supervisor. The current study contributes to the existing knowledge about participation in the development of the performance indicators and recommends supervisors to allow the participation of the employee in the development of the performance indicators, no matter what the quality of the relationship between the supervisor and the employee is. If the supervisor allows this, the contractibility of the performance indicators will be higher.
Keywords: Operational performance indicators; Employee participation; Leader-member exchange; Contractibility
Contents
1 Introduction ... 5
2 Theory ... 9
2.1 Literature review ... 9
2.1.1 Performance indicators ... 9
2.1.2 The contractibility of performance indicators ... 10
2.1.3 Agency theory ... 11
2.1.4 Organizational justice ... 12
2.1.5 Leader-member exchange ... 13
2.1.6 PI Participation ... 15
2.2 Hypotheses development ... 15
2.2.1 PI Participation and the contractibility of PIs ... 15
2.2.2 The quality of leader-member exchange on the relationship between PI participation and PI contractibility ... 18
3 Research Method ... 22 3.1 Respondents ... 22 3.2 Variable measurement ... 23 3.3 Statistical analysis ... 28 4 Results ... 31 5 Discussion... 35 References ... 41
Appendix A. Items used in the scales ... 49
1 Introduction
Many organizations use performance indicators to translate the strategy of the organization into the actions of the employees. Supervisors can measure both the efficiency and work
effectiveness of employees via performance indicators (Neely et al., 1995). The success of an organization depends on individual employee actions (Burney et al., 2009) and performance indicators can measure these actions, so it is necessary that the performance indicators which are used in an organization are contractible, this means they should be sensitive, precise, and
verifiable (Moers, 2006). However, developing contractible performance indicators is a cost intensive process. Especially for operational employees, who are studied in this research, it is challenging. It is challenging because the performance indicators must be based on detailed aspects of how their work must be performed (Groen et al., 2016a).
Previous studies have investigated how organizations can improve operational
performance indicators. Englund and Gerdin (2015) have reported that performance indicators will be more valid and transparent to supervisors if supervisors based the performance indicators more on operational knowledge. The operational employees have this operational knowledge, so if the operational employees participate in the process of developing performance indicators, the performance indicators will be more valid and transparent. Groen et al. (2016a) have examined this. They performed a research on the effectiveness of performance indicators for supervisors in case employees were involved in the development of them. Their research showed that
supervisors perceive the performance indicators of higher contractibility and more useful in case employees participated in the development of the performance indicators (Groen et al., 2016a). The current study also finds that supervisors perceive the performance indicators of higher contractibility if the operational employees participate in the development of the performance indicators.
If the employee participates in the development of the performance indicators, the supervisor and the employee have to cooperate and communicate to develop the indicators. But to what extent does the relationship between the employee and the supervisor influences the impact of the participation process on the contractibility of the performance indicators? Will employee participation lead to worse contractible performance indicators if there is a bad relationship or will employee participation lead to better contractible performance indicators if there is a good relationship? The aim of the current study is to determine the effect of the quality of the relationship between the employee and the supervisor on the relationship between
participation in the development of the performance indicators and the contractibility of the performance indicators. To define the relationship between the supervisor and the employee a
behavioral theory is used: the leader-member exchange theory. The research question which will be answered is: Does the quality of the leader-member exchange affect the relationship between participation in the development of the performance indicators and better contractible
performance indicators? The answer which the current study finds is that the quality of the leader-member exchange does not affect the relationship between participation in the
development of the performance indicators and better contractible performance indicators.
Fig. 1. Interaction model of the research question.
There are several reasons why the current study is relevant, interesting and why it contributes to the existing literature. Firstly, performance indicators can be very helpful for the organization, because they can enable employees to perform better which can lead to a higher employee performance (Neely et al., 1995). Therefore it is important that the performance indicators are correct and accurate. The more accurate the performance indicators are, the more they are valued (Hölmstrom, 1979; in Indjejikian and Matĕjka, 2011, pp. 261-262). An
organization wants accurate and correct performance indicators which are specific enough to be meaningful to the supervisors and employees who are using them. This is a difficult and cost intensive process. That is why it is crucial to develop the right indicators in the easiest way possible. The result of the current study will help organizations to develop accurate indicators in an efficient way. The current study will show organizations when PI (performance indicators) participation can lead to better contractible performance indicators.
Secondly, this research is interesting because the relationship between a supervisor and Contractibility performance indicators Participation in performance indicators development Leader-member exchange
an employee can be very different between and in organizations. The relationship can be good, so friendly to each other, or bad, so hostile to each other. I expect that the quality of this relationship will affect if the participation in the development of indicators will lead to better contractible performance indicators. If there is no good communication and cooperation
possible between the supervisor and the employee the participation can fail and can lead to lower contractibility of the performance indicators. To avoid that participation lead to lower
contractible performance indicators it is important to know how the quality of the leader-member exchange affects the relationship between participation and the contractibility of the performance indicators and the current study will provide this knowledge.
Thirdly, in the current study I will include a behavioral theory, the leader-member exchange theory, and a psychological concept, organizational justice. According to Merchant et al. (2003) there is a need to include more behavioral theories in the accounting research related to the design and effects of organizational incentive systems. Performance indicators are a major element of most incentive systems. That is why including a behavioral theory and especially the leader-member exchange theory is a contribution to the existing literature. To my best
knowledge this is the first paper which connects leader-member exchange theory to performance indicators and especially to PI participation and the contractibility of the performance indicators. In this manner, management accounting literature is connected to psychology literature.
Fourthly, most previous studies investigated the relationship between PI participation and how employees perceive the co-developed performance indicators (Abernethy and Bouwens, 2005; Groen et al., 2012, 2016b; Kleingeld et al., 2004), while the current study investigates the relationship between PI participation and how the supervisors perceive the co-developed
performance indicators. So a contribution is that the current study measured participation, LMX leader-member exchange (LMX) and contractibility at different levels, while previous studies measured them at the same level. It is important to know how the supervisors perceive the co-developed performance indicators as the supervisors decide if the employee is allowed to participate and if they are going to use the performance indicators.
The remainder of this research is structured as follows: section 2.1 discusses existing literature about performance indicators, the contractibility of performance indicators, the agency theory, the psychological concept organizational justice, leader-member exchange and
participation in the development of performance indicators. Based on this literature the
hypotheses will be developed in section 2.2. Section 3 describes how this research is conducted. In section 4 the results are presented. Finally, in section 5the implications of the results,
suggestions for further research and the strengths and limitations of the current study will be discussed.
2 Theory
In this section (2.1) the existing literature is discussed in the literature review. In the hypothesis development section (2.2) the hypotheses are developed based on the existing literature.
2.1
Literature review
In this section the existing literature will be discussed to understand the concepts and definitions of the topic performance measurement and to be able to develop the hypotheses. This section will start with the explanation of the definition of performance indicators and contractibility. After this, the agency theory will be discussed, because this is the basis for the use of
performance indicators. However, the agency theory does not include psychological concepts, which are necessary to be aware of, so next the psychological concept organizational justice will be discussed. Then there will follow an explanation of the variables leader-member exchange and PI participation.
2.1.1 Performance indicators
In this section there will be explained what performance indicators are and why organizations use performance indicators.
Performance indicators can also be called key performance indicators, performance measures or performance metrics. Examples of performance indicators are return on capital employed, sales to asset ratio, mystery shopper rating, inventory levels and employee culture surveys (Atkinson et al., 2011). To use the performance indicators to translate the strategy of the organization into the actions of the employees, the performance indicators should be aligned to the strategic objectives of the organization (Atkinson et al., 2011).
For lower level employees/supervisors performance indicators can be used to teach them what is important for the organization and to see/show how well they are performing. If the employees see how well they are performing, this can increase their understanding of what is happening in the organization and can enable them to perform better (Perrin, 1998, p. 377). It is a way to monitor the employees in their productivity and performance and also a way to
motivate the employees to improve this (Elrod et al., 2013). Previous studies indicated that firms who are using performance indicators can generate commitment, enhance learning and motivate their employees (Chenhall et al., 2014, p. 2). It is essential that employees know what important is for the organization and what the strategic objectives of the company are. Performance indicators can help employees to act in the interest of the strategic objectives, because the
compensations and rewards based on the employee’s decisions are aligned to the strategic objectives(Abernethy et al., 2010).
A performance measurement system is the combination of all the performance indicators which are used in an organization. Organizations use performance measurement systems to perform multiple roles: communicating the company’s strategic objectives, motivating employees to help the company achieving its strategic objectives, evaluating the performance of supervisors, employees and operating units, helping supervisors allocating resources to the most productive and profitable opportunities and providing feedback on whether the company is making progress in improving processes and meeting the expectations of customers and shareholders (Atkinson et al., 2011). Performance indicators can only perform these roles if they are sensitive, precise and verifiable; this is also called contractibility. The contractibility of the performance indicator should be as high as possible, in the next section a further explanation of contractibility will be given.
2.1.2 The contractibility of performance indicators
The contractibility of a performance indicator must be high to be able to perform the multiple roles which are described in the last section. In this section the contractibility of a performance indicator will be defined.
The contractibility of performance indicators depends on the performance measure properties sensitivity, precision and verifiability (Moers, 2006). Sensitivity is about how well the performance indicators measure the relevant aspects of the performance of the employee. Precision is about the influence the employee has on the performance indicators and the influence external factors have on the performance indicators. Verifiable is about how objective the performance indicators are and if the data which is used can be checked (Moers, 2006). These performance measure properties should be high, to have a high contractibility.
In the current study the survey is held with operational employees. Operational employees are work-floor employees in line positions, such as operators and professionals (Groen et al., 2016a). They are directly involved in the creation of the actual products or services of the organization. It can be an accountant in an accounting firm, a consultant in an advisory firm or a teacher in high school. The work of operational employees is often context specific; each work floor has its own peculiarities. Developing sensitive, precise and verifiable
performance indicators is already a cost intensive process, but this makes it more challenging. Especially for operational performance indicators it is important that the performance indicators are based on detailed aspects of how work must be performed (Groen et al., 2016a). Detailed
aspects are for example how the activities are exactly carried out and which resources are used for these activities.
2.1.3 Agency theory
Most of the reasons why organizations use performance indicators are described in section 2.1.1. These reasons are based on the agency theory. The agency theory is a traditional theory which is used for many years in the management accounting literature. The principle and the agent mentioned in the agency theory are called the supervisor and the employee in the current study. The assumption of the agency theory is that the interests between the supervisor and the employee are different (Kunz and Pfaff, 2002). The supervisor wants to maximize the profit, while the employee wants the best compromise between work and leisure. The supervisor does not have all the information about the characteristics and the capabilities of the employee and the employee has more knowledge about his job than the supervisor. This assumption is called the information asymmetry (Abernethy et al., 2004). Because of the information asymmetry and the different interests between the supervisor and the employee, the supervisor wants to control the employee. Controlling and monitoring can be performed by measuring the performance outcome of the employee (Jensen and Meckling, 1976). An example of a performance indicator which the supervisor can use to control the employee and align the different interests is
economic value added; if the employee retrieves this he will receive a bonus. In this case, leisure will be more costly for the employee and the employee will do his best to reach this certain profit, because it personally benefits him (Bonner and Sprinkle, 2002).
There are a few assumptions on which the agency theory is based. The most important assumption is based on the case that the employee is opportunistic and rationalistic (Douma and Schreuder, 1998). Opportunistic does mean that the employee makes a decision while he does not consider other principles and other people and that he adapts every situation for his own benefit. Rationalistic does mean that the employee based his decision on facts of and with knowledge from antecedents and consequences. The employee will take care that his expected proceeds are higher than the expected costs. In these decisions he will act from his self-interest (Douma and Schreuder, 1998). These assumptions are based on an economic perspective and do not include the psychological perspective, while this is also important according to Merchant et al. (2003), so in the next section there will be considered a psychological concept.
2.1.4 Organizational justice
The agency theory is the basic theory in the management accounting literature; however, the agency theory does not include psychological concepts. The expected outcomes based on the agency theory can deviate when psychological concepts are taken into consideration. An example of such a psychological concept is organizational justice. During the last decade, organizational justice has been among the most frequently researched topics (Johnson et al., 2006).
Organizational justice is the employees’ sense of moral propriety of how they are treated; it is defined as the very essence of individuals’ relationship to supervisors (Cropanzano et al., 2007). In the case of performance indicators it is the extent to which employees feel that all aspects around the performance indicators are fair.
Organizational justice has many positive outcomes for employees and for the
organization. These positive outcomes are greater trust and commitment, improved employee performance, more helpful citizenship behaviors, improved employee satisfaction (Colquitt et al., 2001), diminished conflict (Cropanzano et al., 2007), increased motivation (Zapata et al. 2009; Taylor et al. 1995) and organizational commitment (Colquitt et al., 2013). This ultimately affects organizational success (Johnson et al., 2006).
Most of the studies on organizational justice have focused on two dimensions of organization justice: the distributive and procedural dimension (Burney et al., 2009). However, there is also another dimension of organizational justice, this is interactional justice (Colquitt, 2001). Distributive justice is about the amount of fairness of the decision outcomes (Colquitt, 2001), it is fostered when employees feel that the allocation of the outcome is consistent with the goal of the situation and match with their inputs or contributions (e.g. time, effort and skill put into jobs) (Adams, 1965; Leventhal, 1976). In the current study it is the extent to which
employees feel the performance indicators are fair.
Whiledistributive justice is focused on the fairness of the outcomes, procedural justice is focused on the fairness of the procedures which lead to the outcomes. Procedures are fair when employees have voice or influence in the processes (Thibaut and Walker, 1975) or when the process is characterized as consistent (across persons and time), bias-free (neutral decision makers), accurate, ethical, correctable (opportunities to correct bad outcomes) and representative (all interested parties are considered) (Leventhal, 1980). In the current study it is the extent to which employees feel the process of developing the performance indicators is fair.
While procedural justice concerns the formal aspects of the procedures, interactional justice concerns the informal aspects of the procedures (Tyler and Bies, 1990). Interactional justice consists of interpersonal justice and informational justice. Interpersonal justice is the
extent to which employees feel their supervisor has treated them fairly during the procedures (Zapata et al., 2013). In the current study interpersonal justice is the extent to which employees feel their supervisor has treated them fairly during the process of developing the performance indicators. Fairly treated does mean: treated in a polite manner, with respect and dignity
(Colquitt, 2001). Informational justice is the extent to which employees feel their supervisor has explained and communicated the procedures well (Zapata et al., 2013). The supervisor has to justify, so explain the basis for decisions, and be truthful (Colquitt, 2001). In the current study it is the extent to which employees feel their supervisor has explained and communicated the process of the development of the performance indicators well.
In this section the psychological concept organizational justice is discussed.
Organizational justice is the employees’ sense of moral propriety of how they are treated; it is defined as the very essence of individuals’ relationship to supervisors (Cropanzano et al., 2007). To measure the quality of the relationship between the supervisor and the individual, in this case the employee, the behavioral theory leader-member exchange is used and will be discussed in the next section.
2.1.5 Leader-member exchange
Besides the psychological concept organizational justice, the current study also considers a psychological theory; the leader-member exchange theory. The leader-member exchange theory (LMX) was for the first time studied in 1975. Dansereau et al. (1975) define LMX as the quality of the social exchange between leader and subordinate. The LMX theory describes how effective leadership relationships develop between dyadic partners in organizations. In the current study the supervisor and the employee are the dyadic partners (Graen and Uhl-Bien, 1995). Supervisor-employee dyadic exchanges necessarily reflect supervisor’s decisions concerning execution of responsibilities, resource allocation and evaluation of information about employee behavior. All of these affect the employees because of their lower hierarchical status in the dyad (Scandura et al., 1986). The LMX theory presumes that supervisors often differentiate among employees and that they develop different exchange relationships with these employees (Liden et al., 1997). These different exchange relationships have different qualities. In the current study high and low quality are presumed.
A low quality LMX is characterized by the employee having low levels of support from the supervisor, low levels of responsibility, low decision influence, and the employee does not have the feeling that the supervisor is satisfied with her/him (Maslyn and Uhl-Bien, 2001). These employees are in the out-group. Maslyn and Uhl-Bien (2001) stated that a low quality LMX is
more based on the transactional part of the employment relationship, such as the economic exchange that focuses on pay for performance, while a high quality LMX is more based on the social nature of the employment relationship, such as the social exchange that focuses on loyalty, commitment, support, and trust (Dulebohn et al., 2012).
A high quality LMX is characterized by the employees having a lot of support from the supervisor, high levels of responsibility, decision influence, access to resources and the employee has the feeling that the supervisor is satisfied with her/him (Maslyn and Uhl-Bien, 2001). These employees are in the in-group. A high quality LMX is based on mutual obligation and reciprocity (Liden et al., 1997). Reciprocity is a social rule that state that individuals repay kind (harmful acts) with kindness (retribution) (Falk and Fischbacher, 2006).
The quality of the relationship between the supervisor and employee is dependent on the personality of the supervisor and the employee. However, it also depends on how much effort the supervisor and employee put in the relationship (Maslyn and Uhl-Bien, 2001) and how effective the supervisor is (Graen and Uhl-Bien, 1995). If the supervisor and employee put more effort in the relationship and the supervisor is more effective, the quality of the LMX will increase.
Duarte et al. (1994) stated that there is little to no relationship between the quality of leader-member exchange and performance. However, there are more studies in the management literature which show that there is a relationship between leader-member exchange and
performance. Three of these studies are very reliable because they discuss all studies about leader-member exchange together in one study, which make their conclusions stronger. These meta-analyses were examined by Dulebohn et al. (2012), Ilies et al. (2007) and Gerstner and Day (1997). They discovered that LMX quality has links to various employee outcomes, such as employee satisfaction and employee performance. If the quality of the leader-member exchange is higher, the employee satisfaction, employee performance, organizational citizenship behavior and role clarity (Gerstner and Day, 1997) are higher, while the role of conflict (Gerstner and Day, 1997) and the stress level are lower (Nahrgang et al., 2009).
Concluded, the LMX theory is about the quality of exchange between the supervisors and the employees (Dansereau et al. 1975). This quality can be high or low. A low quality leader-member exchange is more based on economic exchange, while a high quality LMX is more based on social exchange (Maslyn and Uhl-Bien, 2001). A high LMX quality has a positive influence on employee performance (Dulebohn et al., 2012). To be able to answer the research question about the effect of the leader-member exchange quality on the relationship between PI participation
and the contractibility of the performance indicators, it also important to define PI participation, this will be accomplished in the next section.
2.1.6 PI Participation
Performance Indicators (PI) participation is the co-development of performance indicators between the supervisor and the employee (Groen, 2012). It is about how supervisors and
employees work together to design and implement a new or modified performance measurement system. The supervisor and the employee have to communicate and cooperate to develop the performance indicators. For the employee PI participation is the amount of influence he feels he had on the design, implementation, and maintenance of the performance indicators he is
measured by (Abernethy and Bouwens, 2005; Groen et al., 2016b).
It is important to know that PI participation is much broader than only participation in goals setting or participative budgeting. PI Participation also includes participating in the conceptualization of the indicators, identifying the required data, designing graphs and tables, adapting IT-systems and even producing the periodic performance reports (Groen et al., 2012). So PI participation is about having substantial influence in the development of all aspects of the performance indicators during all its phases, so during the design phase, when the definitions and appearances are developed and the data is provided, during the implementation phase, when the performance indicators are incorporated in the daily routine, and during the maintenance of the performance indicators, when the revisions are made (Neely et al., 1997, 2002).
2.2
Hypotheses development
Based on the theory and prior research discussed in section 2.1, the hypotheses will be presented below. The first hypothesis is PI participation will lead to better contractible PIs (from the opinion of the supervisor) and the second hypothesis is the impact of PI participation on the contractibility of PIs is more positive (negative) if the leader-member exchange quality is high (low).
2.2.1 PI Participation and the contractibility of PIs
Section 2.1 gave the definitions of the contractibility of performance indicators and PI participation. To be able to answer the research question about the effect of leader-member exchange quality on the relationship between PI participation and the contractibility of
performance indicators, it is important to know what the relationship is between PI participation and PI contractibility.
The concept information asymmetry is defined in the agency theory; the employee has more specific information about his/her job than the supervisor has (Abernethy et al., 2004). The supervisor is afraid that the employee will hold this information private and uses this information only for his own advantage in the participation process. However, there are two reasons why this is not the case. First, the PI participation process is the possibility for employees to communicate specific job knowledge and to include this job knowledge in the performance indicators (Shields and Shields, 1998). Earlier research has shown that including job specific knowledge in the performance indicators increases the contractibility (Wouters and Wilderom, 2008). This is especially the case with operational employees, because for operational performance indicators it is important that the performance indicators include detailed aspects of the work and employees know these detailed aspects. Secondly, like employees, supervisors also have private information and employees do not know what supervisors know and what they do not know. This makes it more difficult for the employees to keep information private and to use this information for their own advantage (Groen et al., 2016a). These two reasons show that PI participation can increase the contractibility of the performance indicators because there is more information sharing and job-specific knowledge is included in the performance indicators.
Supervisors are afraid that if employees participate in the development of performance indicators they only want to develop the performance indicators for their own benefit. This view corresponds to the agency theory that assumes the employee is opportunistic and rationalistic (Douma and Schreuder, 1998). However, the agency theory does not include psychological concepts, like organizational justice. Brown et al. (2009) have shown that employees are more honest in participation situations than would be expected based on the assumptions made in the agency theory. When employees have voice or influence in the process, like it is the case with PI participation, the procedural justice will be higher. This is also proved by the study of Derfuss (2009). If there is more organizational justice, employees will respond in a reciprocal manner because they feel they are fairly treated, so they also want to treat the supervisor fairly (Fehr and Gächter, 2000). This will lead to mutual benefit instead of only the benefit of the employee. Based on the case that employees are more honest and organizational justice increases in
participation situations, it is expected that the employee will develop the performance indicators for the mutual benefit and this will increase the contractibility of the performance indicators.
PI participation has similarities with participative budgeting. Research has shown that participative budgeting has a positive impact on for example budget relevance, attitude towards budget and budget usefulness (Derfuss, 2009). If participative budgeting has a positive impact, it is expected that PI participation also has a positive impact on the contractibility of the
performance indicators.
The contractibility of a performance indicator depends on the measurement properties sensitivity, precision and verifiability (Moers, 2006). The supervisor wants sensitive performance indicators because he wants to see the difference between good and bad performance. The supervisor also wants precise indicators because he wants the performance indicators to be influenced by the employees who are measured and not by other employees or other external factors. If the performance indicators are imprecise they do not give the right information to the supervisor (Abernethy et al., 2004) and the supervisor is not able to give a fair bonus to the employee. The supervisor also wants to control the performance indicators, so he/she wants them to be verifiable (Groen et al., 2016a).
Like the supervisors, the employees also wants sensitive, precise and verifiable performance indicators. When performance indicators are not representing the impact of the work of the employees, so when they are not sensitive, employees are very frustrated (Chenhall et al., 2014). This means that employees want sensitive performance indicators, because they want to be rewarded on the effort they make. They do not want to put effort in their job if this effort will not be measured. Employees are assumed to be risk-averse, so they do not want external factors influencing the outcome which is measured (Keeping and Levy, 2000). They want precise performance indicators instead of general indicators. Employees want their performance indicators to be verifiable as well, because they want to be able to check the data which is used in the performance indicators (Jordan and Messner, 2012). If employees do not participate in the development of performance indicators, employees cannot give input and information. In this case there is more chance that supervisors develop performance indicators which are insensitive, imprecise and not verifiable, while both supervisors and employees want sensitive, precise and verifiable performance indicators. If they work together with the same goal, it is more likely that the contractibility of the performance indicators will increase.
Concluded, PI participation can increase the contractibility of the performance indicators because the employee shares job-specific knowledge and the organizational justice increases. In addition, participative budgeting has a positive impact on the contractibility of performance indicators and both the supervisor and the employee want performance indicators which are sensitive, precise and verifiable.
So the first hypothesis is:
2.2.2 The quality of leader-member exchange on the relationship between PI participation and
PI contractibility
The previous section shows that the expectation is that PI participation will lead to better
contractible performance indicators. In this section a hypothesis about the effect of the quality of leader-member exchange on the relationship between PI participation and better contractible performance indicators is developed. A low quality leader-member exchange is more based on economic exchange, while a high quality LMX is more based on social exchange (Maslyn and Uhl-Bien, 2001). There are several theoretical underpinnings for why a positive or negative interaction effect should be expected.
Firstly, a positive interaction effect is expected because a high quality leader-member exchange leads to less conflict. Sometimes supervisors are concerned about PI participation because they think it will lead to disagreement and conflict (Groen et al., 2016a). However, when the quality of the leader-member exchange is high, the role of conflict will decrease (Gerstner and Day, 1997). So when there is a high quality LMX the PI participation will not result in conflict, because the chance of conflict is already decreased due to the high quality LMX. When PI participation does not lead to conflict, the more likely it is that PI participation leads to better contractible performance indicators. Contrary to a high quality leader-member exchange, an employee and a supervisor can also have a low quality LMX. The role of conflict will increase when there is a low quality LMX (Gerstner and Day, 1997). If there is a conflict, the expectation is that this will only increase when the supervisor and employee have to work together, so this will lead to lower contractible performance indicators in participation situations.
Secondly, a positive interaction effect is expected when there is a high quality leader-member exchange because the employee knows the supervisor is satisfied about his performance and the employee wants to be rewarded for that performance. When there is a high quality leader-member exchange, employees have support from the supervisor and they know that the supervisor is satisfied about them and their work (Maslyn and Uhl-Bien, 2001). Employees want to be rewarded for the effort they put in their work, so they want performance indicators to be representing the impact of their work, which does mean the indicators have to be sensitive (Chenhall et al., 2014). For the employee who knows his job performance is good, it is very important that the performance indicators are precise and verifiable. Section 2.2.1 showedthat the supervisor also want sensitive, precise and verifiable performance indicators. The more alignment there is between the goals of the employee and the supervisor, the more likely it is that these goals will be achieved when the performance indicators are developed together. In this case the performance indicators will have a higher contractibility when employees have influence on
their design. Contrary, when there is a low quality LMX, a negative interaction effect will be expected because the employee does not have the feeling that the supervisor is satisfied with him (Maslyn and Uhl-Bien, 2001). The employee will think that the supervisor is not satisfied with his effort and his work outcomes. Therefore the employee does not want performance indicators which are dependent on his work outcomes, so he wants performance indicators which are imprecise, insensitive and not verifiable. The supervisor wants precise, sensitive and verifiable performance indicators. When the supervisor and employee are participating in the development of the performance indicators but do not have the same goals, the contractibility of the
performance indicators will decrease.
Thirdly, a positive interaction effect is expected because in a situation of a high quality leader-member exchange employees want very sensitive performance indicators and share all information. When there is a high quality LMX, the employee has higher levels of responsibility, more decision influence and more access to resources (Maslyn and Uhl-Bien, 2001). This does also mean that the employee has more job-specific knowledge in comparison to the supervisor. This is called information asymmetry (Abernethy et al., 2004). This can have disadvantages in the participation process and on the contractibility of the performance indicators, because the employee can choose which information he shares with the supervisor and so he can influence the performance indicators only for his own benefit. However, when there is a high leader-member exchange quality the employees want their performance indicators to be very sensitive, so they will share all specific-job knowledge in the participation process to get the most sensitive performance indicators as possible. So job specific knowledge in relation to a high quality LMX doesn’t affect the relationship between PI participation and the contractibility of the
performance indicators negatively. It is even a positive effect on the relationship between PI participation and the contractibility of the performance indicators if the employee shares all the job-specific knowledge. However, a negative interaction effect is expected when there is a low quality LMX and when there is information asymmetry. The employee does not want sensitive performance indicators so the employee will not give the right information when developing the performance indicators and this will have a negative impact on the contractibility of the
performance indicators.
Fourthly, a positive interaction effect is expected when the organizational justice increases because there is a high quality leader-member exchange. Employees who have a high quality LMX with their supervisor are in the in-group (Maslyn and Uhl-Bien, 2001). Employees in the in-group feel they belong to a group and it is expected that they feel they are fairly treated during the participation process of developing performance indicators in comparison to the
employees in the out-group; this will increase the interpersonal justice (Zapata et al., 2013). The expectation is that the informational justice of employees who have a high quality LMX will increase because the communication between the employee and the supervisor will be more effective (Zapata et al., 2013). The higher organizational justice, the more likely it is that
employees will respond in a reciprocal manner to the supervisor which leads to employees acting for the mutual benefit instead of only for their own benefit. If this is the case in the PI
participation process, the contractibility of the performance indicators will increase. In comparison to a positive interaction effect, a negative interaction effect is expected when the organizational justice decreases because there is a low quality leader-member exchange.
Fifthly, a positive interaction effect is expected when there is better communication and cooperation because the leader-member exchange quality is high. A high quality LMX is based on mutual obligation and reciprocity (Liden et al., 1997). Reciprocity results in affective
attachment between supervisors and employees (Ferris et al., 2009). High quality leader-member exchanges are characterized by loyalty, commitment, support, and trust (Dulebohn et al., 2012). To develop performance indicators together, the supervisor and the employee has to
communicate and cooperate. The more loyalty, commitment, support and trust there is between the supervisor and the employee, the better the communication and cooperation will be. This is the result of the supervisor and the employee working together and being open for each other’s ideas. The better the communication and cooperation in the PI participation process, the higher the contractibility of the performance indicators. However, a negative interaction effect is expected when there is worse communication and cooperation because the leader-member exchange quality is low.
Sixthly, a positive interaction effect is expected because the employee and the supervisor put more effort in the relationship when the quality of the leader-member exchange is high. This is because the quality of the LMX is dependent on the amount of effort the supervisor and the employee put in their relationship (Maslyn and Uhl-Bien, 2001). If the quality of the leader-member exchange is high, the supervisor and employee had put more effort in this relationship. It is expected that the supervisor and employee will in this case also put more effort in the communication and cooperation through the PI participation process which will increase the contractibility of the performance indicators. This differs from a low quality LMX in which less effort was inputted. It is expected that the supervisor and employee in this case will also put less effort in the communication and cooperation through the PI participation process which will decrease the contractibility of the performance indicators.
effect should be expected are outlined in this section. A high quality LMX will decrease conflicts, while a low quality LMX will increase conflicts. When there is a high quality LMX the employee wants more sensitive performance indicators, will share more job-specific information, the organizational justice will increase and the cooperation and communication in the PI
participation process will be better; this has a positive impact on the relationship between PI participation and the contractibility of performance indicators. In contrast, when there is a low quality LMX the employee wants less sensitive performance indicators and will hide job-specific information, the organizational justice will decrease and the cooperation and communication in the PI participation process will be worse; this has a negative impact on the relationship between PI participation and the contractibility of performance indicators. This leads to the second hypothesis:
H2: The impact of PI participation on PI contractibility is more positive (negative) if the leader-member exchange quality is high (low).
Fig. 2. Expected interaction effect of the quality of the leader-member exchange on the relationship between PI participation and PI contractibility.
3 Research Method
To test the hypotheses a survey method was used because publicly available archival data for testing the hypotheses was not available. This section describes how the data was collected and who the respondents were, how the variables were measured and which statistical analyses were used.
3.1
Respondents
The hypotheses were tested with a survey among pairs of operational employees and their direct supervisors. The respondent pairs of supervisors and employees had to meet three criteria: (1) the employees had to carry out operational activities of their organization; (2) the employees must have worked in their current functions for at least one year; (3) the organization must have performance indicators about employees’ performance.
Together with eight other students the respondent pairs were found. Each student had to find at least six pairs of respondents to be allowed to use the collected data.Students found their respondents pairs in their network of friends, family and (ex) employers. I found ten pairs of respondents by asking friends, family and recruiters of (ex) employers if they know supervisors and employees who want to participate and who meet the criteria or if they want to participate and meet the criteria themselves. Instead of random sampling, I used respondent-driven (or ‘snowball’) sampling. If I found a potential respondent I asked him/her if he/she knows more potential respondents. According to Salganik and Heckatorn (2004), respondent-driven sampling can lead to samples as good or even better compare to random sampling. It was difficult to find respondents, because many potential respondents could not be respondents because their organizations had not yet developed a comprehensive performance measurement system.
The coordinator of the master thesis survey project sent the respondents of all students e-mail messages with the request to participate and a link to the web page containing the survey. The respondents participated on voluntarily basis and were informed that their answers would be kept strictly confidential and anonymous. Besides the nine students who collected data, there were also five Chinese students who collected respondents from China. However, these respondents are excluded because the organizational culture between Europe and China is too different (Verburg et al., 1999). Including the Chinese respondents would impair the internal and external validity of the current study.
The survey was completed by 70 pairs of respondents because some students found more than six pairs of respondents. Eight days after the invitation a reminder was send to the employees and the supervisors who had not yet completed the survey. There is a non-response
bias, because there are significant differences in the scores of PI contractibility between early and late respondents (t (86) = -2.16, p < 0.05). The test for non-response bias is based on the ‘early/late hypothesis’, this means that early and late respondents are compared because late respondents are sometimes similar to non-respondents (Pace, 1939; Ferber, 1948). Table 1 gives an overview of the respondents’ characteristics.
Table 1
Respondent characteristics.
3.2
Variable measurement
In this section the variables will be operationalized, the constructs which were used can be find in appendix A. Respondents had to rate all items on a seven-point fully anchored Likert scale: (1) totally disagree, (2) disagree, (3) moderately disagree, (4) neutral, (5) moderately agree, (6) agree, 7) totally agree.
The independent variable in this research is PI participation. To measure this construct, sources developed by Abernethy and Bouwens (2005) are used. Cronbach’s alpha of this scale is 0.965 and the questions in the scale are completed by the employee.
The dependent variable in this research is PI contractibility. The constructs to measure PI contractibility are the measurement properties of the performance indicators (Moers, 2006).
Characteristic Employees Supervisors
Sex 55.1% male
44.9% female 80.3% male 19.7% female
Education 4.3% lower (vocational)
education 20% intermediate general/vocational education 44.3% higher general/vocational education 31.4% scientific education
0% lower (vocational) education
14.3% intermediate general/vocational education 35.7% higher general/vocational education 50% scientific education
Age Mean = 33 (SD = 10) Mean = 39.1 (SD = 9.9)
Nationality Dutch 53 Turkish 9 British 1 Belgian 2 South African 1 Russian 2 French 1 Bosnian 1 Dutch 61 Turkish 7 British 1 Spanish 1
Departmental tenure Mean = 4.6 (SD = 4) Mean = 6.1 (SD = 5.2)
Span of control N/A Mean = 16.4 (SD = 15.7)
Number of employees in
organization Mean = 2789 (SD = 10197) min 2.8, max 70000 Same as for employees, as supervisor-employee pairs belong to the same organization Job type of employee 17.4% operator
27.5% service employee 55.1% professional
The measurement properties are sensitivity, precision and verifiability. To assess the convergent validity of the original scale and to make the scale relevant for operational employees, an exploratory factor analysis was performed. The analysis was performed before testing the hypotheses. Table 2 shows that the varimax rotated principal component analysis of the final data yielded three factors with an eigenvalue higher than 1. These results are the same as the results of Groen et al. (2016a), so the same decisions were made and four items of the original scale were used in the relevant scale. These items best reflects the contractibility of the performance indicators (from the supervisors’ opinion) and are a combination of precise, sensitive and verifiable items. Only strong factors loading 0.5 or higher were retained on the relevant construct. Four negatively (recorded) precision items were loaded on another factor, so these items were deleted. They loaded on another factor than the other precision items because they were negatively formulated and people tend to react different on negative questions than positive questions. Three sensitivity items were deleted because these items were about employee effort, which is the same as employee input, but if the supervisor considers the contractibility he will be interested in employee output instead of employee input. The variable PI contractibility is not normally distributed, because the skewness indicates a significant deviation from normality
(𝑍𝑆𝑘𝑒𝑤𝑛𝑒𝑠𝑠 = -3.84, p < 0.001). A solution for skewness or kurtosis can be deleting outliers.
Outliers were defined by inspecting the Boxplot whether SPSS identifies outliers. After deleting the outlier, the skewness still indicates a significant deviation from normality (𝑍𝑆𝑘𝑒𝑤𝑛𝑒𝑠𝑠 = -2.81,
p < 0.01). Because the scores tend to accumulate on the right of the distribution, a second power transformation was performed. After this transformation the variable PI contractibility is normally distributed (𝑍𝑆𝑘𝑒𝑤𝑛𝑒𝑠𝑠 = -1.07, ns). Cronbach’s alpha of this scale is 0.814 and the
questions in the scale are completed by the supervisor.
As interacting variable in this research leader-member exchange is used. To measure leader-member exchange the sources developed by Graen and Uhl-Bien (1995) are used. The variable is not normally distributed, because the skewness and kurtosis indicates a significant deviation from normality (𝑍𝑆𝑘𝑒𝑤𝑛𝑒𝑠𝑠 = -3.66, p < 0.001) and (𝑍𝐾𝑢𝑟𝑡𝑜𝑠𝑖𝑠 = 2.07, p < 0.05). After deleting the three lowest scores which are outliers the leader-member exchange variable is normally distributed (𝑍𝑆𝑘𝑒𝑤𝑛𝑒𝑠𝑠 = 1.40, ns) and (𝑍𝐾𝑢𝑟𝑡𝑜𝑠𝑖𝑠 = -0.98, ns). Cronbach’s alpha of this
scale is 0.906. The questions are completed by the employee, because the questions of the PI participation variable are also completed by the employee, different types of exchange relationships also depend on the attitude and behaviors of employees (Van Breukelen et al., 2006) and a positive perception of the relationship is how the employee perceives the supervisor’s course of action and motives (Hollander, 1992).
Components Table 2
Factor leadings of the variable PI contractibility. Items
Precision: 1 2 3
My employee’s performance expressed in the performance indicators is strongly affected by …
1… changes in economic conditions 0.78
2… decisions made in other parts of the organization 0.78 3… changes in the behavior of parties outside the organization, such as
customers, suppliers or competitors
0.79
4… factors beyond his/her responsibility 0.82
5. The performance indicators measure only what my employee can
actually influence 0.81
Sensitivity:
6. The performance indicators express accurately whether my employee
functions well or not 0.76
7. If my employee performs well, it is directly reflected in the performance indicators
0.70 8. Working hard leads to better performance on the performance
indicators 0.80
9. Devotion and effort in the job leads to better performance on the performance indicators
0.75 10. Providing effort in his/her job leads to better performance on the
performance indicators 0.88
Verifiability:
11. The performance indicators are objective and verifiable 0.71 Table 3 shows the descriptive statistics of the survey items; PI participation, PI contractibility and leader-member exchange. Table 4 presents the factor leadings of the
measurement model. As can be seen, only strong factor loadings of 0.5 or higher were included. The control variables used in the current study are sex, educational level, departmental tenure, length of time the performance indicators have been implemented, information asymmetry and the number of subordinates the supervisor has (ln span of control). By testing the first hypothesis leader-member exchange was also included as a control variable.
Demographic variables like sex, educational level and departmental tenure are included because the study of Ali and Davies (2003) has shown that sex and departmental tenure may explain differences in employee performance. Based on this finding the expectation is that sex,
educational level and departmental tenure can also explain differences in PI participation and PI contractibility. Based on these reasons age could also be included as a demographic control variable but as table 5 shows age correlates with departmental tenure, for the employee (r = 0.72, p < 0.01) and for the supervisor (r = 0.40, p < 0.01). To avoid multicollinearity and because departmental tenure has more influence on PI participation and PI contractibility, I decided to
eliminate age as control variable. Table 3
Descriptive statistics.
Table 4
Factor loadings of the measurement model.
Variables Components PI participation 1 2 3 Item 1 0.94 Item 2 0.92 Item 3 0.92 Item 4 0.93 Item 5 0.85 PI contractibility Item 6 0.68 Item 7 0.86 Item 8 0.86 Item 9 0.78 Leader-member exchange Item 10 0.72 Item 11 0.81 Item 12 0.75 Item 13 0.87 Item 14 0.78 Item 15 0.77 Item 16 0.80
Variables Α N M SD Min Max
PI participation 0.965 Item 1 70 2.86 1.788 1 6 Item 2 70 2.96 1.813 1 6 Item 3 70 3.03 1.744 1 6 Item 4 70 2.87 1.785 1 6 Item 5 70 3.19 1.796 1 7 PI contractibility 0.814 Item 6 70 4.63 1.505 1 7 Item 7 70 4.70 1.536 1 7 Item 8 70 5.17 1.444 1 7 Item 9 70 5.20 1.246 1 7 Leader-member exchange 0.906 Item 10 70 5.80 1.150 2 7 Item 11 70 5.66 1.273 2 7 Item 12 70 5.59 1.198 2 7 Item 13 70 5.67 1.224 1 7 Item 14 70 4.73 1.444 1 7 Item 15 70 5.50 1.294 1 7 Item 16 70 6.14 1.011 1 7
Table 5
Correlations including potential control variables
a 1 = male; 2 = female
b 1 = primary education; 2 = lower vocational education; 3 = intermediate general education; 4 =
intermediate vocational education; 5 = higher general education; 6 = higher vocational education; 7 = scientific education
c 1 = operator; 2 = service employee; 3 = professional
* p < .10 (two-tailed) ** p < .05 (two-tailed) *** p < .01 (two-tailed)
Length of time the performance indicators have been implemented is included as a control variable because it may be related to PI participation (if performance indicators are longer
implemented employee influence is less likely) and to PI contractibility (if performance indicators are longer implemented the expectation is that the contractibility of the performance indicators is higher).
Information asymmetry is included because PI participation will only increase the contractibility of the performance indicators if there is information asymmetry. When the supervisor does not have all the information about the characteristics and the capabilities of the employee and the employee has more knowledge about his job than the supervisor, there is information asymmetry (Abernethy et al., 2004). This variable is included because PI participation will only lead to better contractible performance indicators in the eyes of the supervisor, if the supervisor knows less than what the employees knows and the employee can share this knowledge in the participation process. To measure information asymmetry the
construct sources developed by Dunk (1993) are used. Supervisors could also choose N/A in the survey which is defined as a missing value in SPSS. Cronbach’s alpha of this scale is 0.934 and the questions in the scale are completed by the supervisor.
The number of subordinates the supervisor has (ln span of control) is included as a control variable because it may be related to leader-member exchange. This can be the case because the quality of LMX depends on how much effort the supervisor and employee put in
Item 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 PI participation 2 PI contractibility 0.28** 3 Leader-member exchange 0.43*** 0.06 4 Sex supervisor a -0.20 -0.07 0.13 5 Education supervisor b 0.22* -0.01 -0.09 -0.16
6 Tenure department supervisor 0.16 0.08 0.19 0.15 -0.20
7 Length PI implementation -0.01 0.29** -0.16 -0.16 -0.03 0.23*
8 Information asymmetry -0.09 0.25** 0.01 0.01 0.11 -0.01 -0.08
9 Ln Span of control -0.00 -0.01 -0.04 0.15 -0.02 -0.04 0.25** 0.11
10 Sex employee a -0.01 -0.09 0.06 0.08 0.14 -0.17 -0.27** 0.02 -0.02
11 Education employee b 0.16 -0.09 0.08 -0.24* 0.21* 0.17 0.05 0.01 -0.12 -0.06
12 Tenure department employee 0.10 -0.09 -0.12 -0.01 -0.08 0.16 -0.01 0.01 0.08 -0.14 -0.16
13 Job type c 0.02 0.08 0.06 0.02 0.03 -0.01 -0.01 0.20* 0.03 0.19 0.18 -0.39**
14 Age supervisor 0.06 0.01 0.08 0.04 -0.06 0.40*** 0.21* 0.08 0.17 0.15 -0.11 0.11 0.13
15 Age employee 0.02 -0.05 -0.05 0.05 -0.09 0.05 -0.14 0.06 0.20 -0.01 -0.16 0.72*** -0.21* 0.22*
the relationship (Maslyn and Uhl-Bien, 2001) and how effective the supervisor is (Graen and Uhl-Bien, 1995), if there are many subordinates it will be more difficult for the supervisor to put numerous effort in every relationship and to be effective. The number of subordinates the supervisor supervises can influence the PI participation. If there are many subordinates it will be more difficult to give every subordinate the opportunity to participate. Besides the number of subordinates the supervisor supervises (ln span of control), the number of employees in an organization could also be included as a control variable but as table 5 shows ln number of employees correlates with ln span of control (r = 0.43, p < 0.01), with departmental tenure (r = 0.26, p < 0.05) and with length of time the performance indicators have been implemented (r = 0.24, p < 0.05). To avoid multicollinearity between the control variables, ln number of employees was eliminated as control variable.
Type of job is included as control variable because a different type of job can have a different influence on PI participation and PI contractibility (if a professional participates the supervisor may take his ideas more seriously than if an operator participates, so the supervisor has a different view on the contractibility of the performance indicator after participation) and on leader-member exchange (an operator may be in a lower hierarchical status in the dyad than a professional (Scandura et al., 1986), this affects the relationship between the supervisor and employee).
3.3
Statistical analysis
The collected data from the surveys was analyzed using SPSS.
The first hypothesis, PI participation will lead to better contractible PIs (from the opinion of the supervisor); consist of a dependent and an independent variable. This was tested using a linear regression analysis, because the independent variable is measured on at least an ordinal scale (Pallant, 2013). The estimated regression model (including control variables) for H1 is:
𝐶𝑂𝑁𝑇𝑅𝐴𝐶𝑇𝐼𝐵𝐼𝐿𝐼𝑇𝑌𝑖 = 𝛽0+ 𝛽1𝑃𝐴𝑅𝑇𝐼𝐶𝐼𝑃𝐴𝑇𝐼𝑂𝑁𝑖+ 𝛽2𝐿𝑀𝑋𝑖+ 𝛽3𝑆𝐸𝑋(𝑠𝑢𝑝𝑒𝑟𝑣𝑖𝑠𝑜𝑟)𝑖+ 𝛽4𝐸𝐷𝑈𝐶(𝑠𝑢𝑝𝑒𝑟𝑣𝑖𝑠𝑜𝑟)𝑖+ 𝛽5𝑇𝐸𝑁𝑈𝑅𝐸𝐷𝐸𝑃(𝑠𝑢𝑝𝑒𝑟𝑣𝑖𝑠𝑜𝑟)𝑖+ 𝛽6𝐿𝐸𝑁𝐺𝑇𝐻𝑃𝐼𝑖+
𝛽7𝐼𝑁𝐹𝑂𝐴𝑆𝑆𝑌𝑖+ 𝛽8𝐿𝑁𝑆𝑃𝐴𝑁𝑖 + 𝛽9𝑆𝐸𝑋(𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑒)𝑖 + 𝛽10𝐸𝐷𝑈𝐶(𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑒)𝑖 +
𝛽11𝑇𝐸𝑁𝑈𝑅𝐸𝐷𝐸𝑃(𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑒)𝑖 + 𝛽12𝐽𝑂𝐵𝑇𝑌𝑃𝐸(𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑒)𝑖 + 𝜀𝑖𝐶𝑂𝑁𝑇𝑅𝐴𝐶𝑇𝐼𝐵𝐼𝐿𝐼𝑇𝑌
The second hypothesis, the impact of PI participation on PI contractibility is more positive (negative) if the leader-member exchange quality is high (low); consist of a dependent variable, an independent variable and an interacting variable. This was tested using an interaction
(Pallant, 2013). The estimated regression model (including control variables) for H2 is: 𝐶𝑂𝑁𝑇𝑅𝐴𝐶𝑇𝐼𝐵𝐼𝐿𝐼𝑇𝑌𝑖 = 𝛽0+ 𝛽1𝑃𝐴𝑅𝑇𝐼𝐶𝐼𝑃𝐴𝑇𝐼𝑂𝑁𝑖+ 𝛽2𝐿𝑀𝑋𝑖+ 𝛽3(𝑃𝐴𝑅𝑇𝐼𝐶𝐼𝑃𝐴𝑇𝐼𝑂𝑁𝑖 ∗ 𝐿𝑀𝑋𝑖) + 𝛽4𝑆𝐸𝑋(𝑠𝑢𝑝𝑒𝑟𝑣𝑖𝑠𝑜𝑟)𝑖 + 𝛽5𝐸𝐷𝑈𝐶(𝑠𝑢𝑝𝑒𝑟𝑣𝑖𝑠𝑜𝑟)𝑖 + 𝛽6𝑇𝐸𝑁𝑈𝑅𝐸𝐷𝐸𝑃(𝑠𝑢𝑝𝑒𝑟𝑣𝑖𝑠𝑜𝑟)𝑖 + 𝛽7𝐿𝐸𝑁𝐺𝑇𝐻𝑃𝐼𝑖 + 𝛽8𝐼𝑁𝐹𝑂𝐴𝑆𝑆𝑌𝑖+ 𝛽9𝐿𝑁𝑆𝑃𝐴𝑁𝑖+ 𝛽10𝑆𝐸𝑋(𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑒)𝑖+ 𝛽11𝐸𝐷𝑈𝐶(𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑒)𝑖+ 𝛽12𝑇𝐸𝑁𝑈𝑅𝐸𝐷𝐸𝑃(𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑒)𝑖+ 𝛽13𝐽𝑂𝐵𝑇𝑌𝑃𝐸(𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑒)𝑖 + 𝜀𝑖𝐶𝑂𝑁𝑇𝑅𝐴𝐶𝑇𝐼𝐵𝐼𝐿𝐼𝑇𝑌
To test this model a regression was performed that examined how the effect of the independent variable on the dependent variable changes for different levels of the interacting variable. This was performed by constructing a new variable (𝛽3) (Pallant, 2013) consisting of
the product of the mean centered independent variable and the mean centered interacting variable, in this case (mean centered) participation * (mean centered) LMX. The interacting variable, LMX, was also added to the model to control for the effect of the interacting variable on the dependent variable.
Before the hypotheses with a linear regression analysis or an interaction regression analysis were tested, there were assumptions which had to be met or tested. The most important issues for linear regression are assessing equality of variance, assessing autocorrelation, assessing multicollinearity, assessing normality, identifying and dealing with outliers and homoscedasticity (Pallant, 2013). The normality of the variables is already described in section 3.2.
To check for robustness of the results, the same regression analyses were run but with the (standardized) residuals of the participation variable and the leader-member exchange variable, when performing bootstrapping and with the questions of the leader-member exchange variable completed by the supervisor instead of the employee.
Firstly, the same analyses were run but with the residual of participation instead of the ‘normal’ participation variable and with the residual of the leader-member exchange instead of the ‘normal’ leader-member exchange variable. The residuals of the variables are used because there is a significant chance that participation and leader-member exchange are related. The expectation is that the higher (lower) the leader-member exchange quality the more (less) participation takes place. The residual of participation consists only of the variation of
participation which is not explained by LMX. The residual of participation was found by running a linear regression analysis with participation as the dependent variable and LMX as the
independent variable. So the residual of participation is what is left over after explaining the variation in participation using LMX. The residual of LMX consists only of the variation of LMX which is not explained by participation and was found by running a linear regression with LMX as the dependent variable and participation as the independent variable.
Secondly, bootstrapping is performed on the regression analyses, because the current study is conducted on a small sample (70 pairs), the scores of PI contractibility were not
normally distributed before the power transformation and the demographic control variables are not normally distributed (Moony and Duval, 1993). Bootstrapping involves resampling the data with replacement to generate an empirical estimate of the entire sampling distribution of a statistic (Moony and Duval, 1993). In the current study the bootstrapping generates 1000 random samples of 70 observations.
Thirdly, the same regression analyses were run but with the questions of the leader-member exchange variable completed by the supervisor instead of the employee. The reason for this is that the employee and the supervisor of the same pair both can give different answers belonging to the leader-member exchange questions because supervisors’ and employees’ views of LMX are not isomorphic (Gerstner and Day, 1997; Sin et al., 2009). Other reasons are that the quality of the leader-member exchange depends on how effective the supervisor is (Graen and Uhl-Bien, 1995) and if the supervisor has positive expectations of the employee, it will stimulate the relation in ways of support and guidance (Liden et al., 1993).
