Pay: a Dangerous Change Agent
Incentive pay and the effectiveness of organizational change
Master Thesis, MscBA, Specialization Change Management University of Groningen, Faculty of Business and Economics
March 2009 Linda Visser Student number 1385658 Paramaribostraat 95 3531 KR Utrecht Tel: 06 30872508 e-mail: visser-linda@hotmail.com
Supervision and co-assessment University: Dr. B.J.M. Emans
&
Dr. P.H. van der Meer
Supervision at Hay Group: L. Bosman, managing consultant
PAY: A DANGEROUS CHANGE AGENT Incentive pay and the effectiveness of organizational change
ABSTRACT
TABLE OF CONTENTS
1. INTRODUCTION... 5
2. THEORY AND HYPOTHESES... 8
2.1 Pay as a motivator of desired behaviour ...8
2.2 Predictors of pay-for-performance in changing behaviour ...8
Perceived transparency...8 Perceived fairness...9 Perceived controllability...10 Pay-out amount...10 Pay-out frequency...11 Participation...11 Conclusion...12
2.3 Predictors of Pay-for-performance in Enhancing Change Effectiveness ...13
Alignment of change objectives and pay indicators...14
2.4 Conclusion...15
3. METHODOLOGY...17
3.1 Research Design ...17
Research focus and unit of analysis...17
Criteria for the selection of organizations...17
3.2 Quantitative Research ...18
Data collection: questionnaire...18
Data analyses...20
3.3 Qualitative research...21
Data collection: interviews & documents...22
Data analyses: multiple-embedded case study design...22
4. RESULTS...24 4.1 Respondents ...24 Company A...24 Company B...24 Company C...25 Company D...25 Company E...26 4.2 Descriptions of Results...26
Mean ratings of variables...26
Comparing the different organizations...27
Summary of descriptive results...33
4.3 Exploratory Results ...33
4.4 Pay-for-performance and the Effectiveness of Organizational Change...43
Desired and realised change in behaviour...43
Effectiveness of the change effort...45
Critical success factors...46
5. DISCUSSION...48
5.1 Interpretation of the Results ...48
5.3 Practical Implications ...52
5.4 Limitations ...52
Response and reliability...52
Validity...52
5.5 Suggestions for Further Research ...53
6. REFERENCES ...55
1. INTRODUCTION
Research has shown that most change efforts in established organizations fail to meet expectations because of internal barriers to change. Research states that approximately seventy percent of the change processes do not achieve their goals (Boonstra, 2000, Pettigrew, 1987). Many researchers are provoked to search for an explanation why so many change processes fail. According to Worley and Lawler (2006) the truth is that the effectiveness of change efforts is largely determined by organizational design, or how a company’s structure, process, reward systems and other features are orchestrated over time to support one another as well as the company’s strategy. Therefore, the only way to ensure that organizations will be able to change is to design them to change. Although there are a variety of features that contribute to an organization that is “built to change” and effective, the role of rewards and motivation in promoting change is one of the more important ones (Lawler & Worley, 2006, Worley & Lawler, 2006, Wruck, 2000). That is important since approximately seventy per cent of unsuccessful change projects can be explained by human behaviour (Cozijnsen & Vrakking, 2003).
According to Demers, Forrer, Leibowitz and Cahill (1996) ‘change on paper matters very little if human behaviour does not change.’ Cozijnsen and Vrakking (2003) take as a basic assumption that ‘successful change will continue to be work of man.’ Management must not only realize that change has an impact on the organization’s employees, but as much attention should be paid to the human side of change as to the planning of the technical and structural aspects of change. However, organizations tend to be rather good at the latter, while being poor at the former (Demers et al., 1996). That is striking, especially when one considers that the success of a change operation is assumed to be determined by the level of cooperation of the employees in the lower levels of the organization (Wissema, Messers & Wijers, 1991). The human side of change might therefore be an interesting subject to both practitioners and researchers. Since rewards affect both behaviour and motivation (Wruck, 2000), it is worth researching the role that rewards play in promoting change in organizations. The relationship between reward, motivation and change effectiveness is visualized in figure 1.1.
FIGURE 1.1:
Role of Reward in Promoting Change
Strangely, the literature concerning the practice of organizational change is silent with respect to pay and its role in organizational change efforts. Books and articles on organizational change often have little or nothing to say about what role, if any, pay can play in an effort to change the organization. At first glance, it seems difficult to explain that this component of organizations does not receive more attention. After all, pay has the potential to influence organizational effectiveness and it can play an important role in determining behaviour. As such, it would seem to be a strong candidate for attention in most organizational change efforts.
There is no single explanation for why pay has been so frequently ignored in the field of organizational change. It is the result of a number of circumstances. According to Lawler (1981), the major reason seems to be that the pay administration is a highly technical issue that requires considerable expertise, just like engineering or accounting. Change agents rarely have this expertise and, as a result, have difficulties in discussing pay issues with technical experts. In addition the value orientation of many change practioners account for the low priority given to pay. The organization development movement has always had a humanistic orientation (Burnes, 2004, Cummings & Worley, 2001), whereas pay and pay decisions emphasize the non-humanistic aspects of work organizations. The result is that many people, who have gone into organizational change because of its attention to humanistic values, find discussions on work and pay to be counter to their basic interests and values.
In most cases, however, it is a mistake for change agents not to be concerned with pay. There are a number of reasons why pay should play a vital role in most organization change efforts. They stem form the nature of pay and its central role. It is worth reviewing these reasons briefly, since they serve as a justification for the argument that pay should have an important role in organizational change efforts. Among the factors that argue for the inclusion of pay in organizational change efforts are: its importance as both a cost factor and a potential motivator of behaviour, its systemic ties to important variables in the organization and its potential for impacting the total organization (Lawler, 2000a). This Master Thesis is a piece of research that fits in within the field of linking pay to the effectiveness of change efforts. This research will focus on pay-for-performance as a tool for organizational change. Pay-for-performance pre-eminently can be used to reward employees for showing new behaviour. The following research question will be answered in this thesis:
In which way can a pay-for-performance system improve the effectiveness of organizational change?
package plays the most important role in improving motivation and in promoting desired behaviour (Armstrong, 1993, Baker, Jensen & Murphy, 1988, Kreitner, Kinicki & Buelens, 2002, Lazaer, 1998). In this research the definition of a pay-for-performance system encompasses only the monetary part. In order to focus and simplify the discussion of change efforts, this research is concentrated on a particular type of organizational change: efforts in which performance change is the objective. An effective change effort is a change effort that reaches its objective in the specified time frame with the specified resources (Golembiewski, 1990).
2. THEORY AND HYPOTHESES
Organizations continually distribute rewards to their members since rewards have a significant effect on the attitudes and behaviours of employees. An understanding of the relationship between rewards and individual behaviour is a crucial ingredient in understanding the role of pay in organizational change efforts. In this chapter, the research and theory on how reward influences motivation, attitudes and behaviours of individuals will be summarized. The predictors of a pay-for-performance system that supports behavioural changes will be described as well as the predictors of pay-for-performance that link these behavioural changes to the effectiveness of change efforts.
2.1 Pay as a motivator of desired behaviour
It is beyond the scope of this research to go into great detail about the many impacts that reward systems have on individual and organizational behaviour, but it is important that they have multiple impacts, several of which are important to change efforts. The most important impacts involve their effect on motivation, and the attraction, retention and development of employees. In essence rewards affect motivation when they are effectively tied to performance and significant amounts of money are given. In this regard, pay-for-performance systems can play an important role. A pay-for-performance plan refers to a situation were pay is not guaranteed (Miceli & Heneman, 2000). It has to be re-earned and is depended on specific results like individual performance, team performance or company performance. The performance measures can range from very concrete behavioural ones; accidents and absenteeism, to more aggregated measures of performance such as productivity, costs, quality, customer satisfaction, delivery and cycle time (Emans, 2007). In the sections 2.2 and 2.3 the predictors of pay-for-performance that support behavioural changes which are linked to the effectiveness of a change effort will be described.
2.2 Predictors of pay-for-performance in changing behaviour
In the literature many factors can be found that influence the effectiveness of pay-for-performance. According to Thierry (1987) the effectiveness of a pay-for-performance system depends on three characteristics, namely (1) transparency, (2) fairness and (3) controllability. Fairness and controllability are also mentioned by Moers (2001). Thierry (2002) adds the (4) pay-out amount and (5) frequency of pay-outs as important characteristics. Based on research of Vroom (1964) it can be argued that (6) participation of employees in the development of a pay-for-performance system will make it more effective. These concepts will be explained in more detail in this section, resulting in seven hypotheses.
compensation system, but also of the objectives of the firm (Armstrong, 1993). Clear communication of these rules towards the personnel will enhance the understanding of the methodologies, measures and targets used and thereby create a better basis of support and commitment to pay-for-performance. Without appropriate explanations of the system, openness about how the compensation system operates and frequent feedback about performance, the necessary climate of trust may not evolve (Milkovich & Newman, 1987, Wiscombe, 2001). According to Globerson (1985) the system must be simple to understand and ideally an employee must be able to calculate its own incentive. In sum, the perception of transparency and openness is expected to have a positive relationship with motivation to change behaviour. Diminishing the risk of exerting effort without being rewarded accordingly is expected to have a positive effect on the willingness to exert effort in line with the desired new behaviour. This leads to the following hypothesis:
Hypothesis 1. Perceived transparency and openness of the different elements of the pay-for-performance system leads to a change in organizational behaviour.
Hence, distributive justice (i.e. fairness of outcomes) and procedural justice (fairness of processes) are expected to be determinants of behaviour. This leads to the following hypotheses:
Hypothesis 2. Perceived fairness of the outcomes of pay-for-performance leads to a change in organizational behaviour.
Hypothesis 3. Perceived fairness of the process of pay-for-performance system leads to a change in organizational behaviour.
Perceived controllability The third characteristic for evaluating pay-for-performance’s effectiveness is the perceived relationship between effort and incentive. Expectancy theory, as originally formulated by Vroom (1964) states that motivation will be strong when individuals can reasonably expect that their efforts and contributions will produce worth-while rewards. This implies that motivation is only likely when a clearly perceived relationship exist between performance and outcome and the outcome is seen as attractive and as satisfying the needs of the employee. Thus, an incentive pay only works in situations where the link between effort and pay is clear and the value of the reward is worth the effort. Each behaviour is also associated with a certain expectancy of probability of success (Roy, 1952). This expectancy represents the individual’s perception of how hard it will be to achieve such behaviour and the probability of successful achievement of that behaviour.
All too often, the criteria used in the pay-for-performance plan are unrelated to the individual worker’s effort (Lawler, 1981). This strive for ‘noise reduction’ is one of the two main criteria that determine the choice of the optimal performance measure: the effect of effort on the performance measure should vary as little as possible in order to have control of one’s incentive compensation (van Herpen, van Praag & Cools, 2003). Since a change effort might require totally new behaviour from employees, a training that develops the needed skills and/or knowledge might be essential to give the employees a feeling of controllability (Wiscombe, 2001). Lastly, employees must feel that the goals are reasonable and attainable (Moers, 2001). In sum, the perception of controllability is expected to have a positive relationship with behavioural change. This leads to the following hypothesis:
Hypothesis 4. Perceived controllability over one’s incentive compensation leads to a change in organizational behaviour.
percentage of incentive pay is more attractive, regardless of the current wage level. However with a relatively high bonus percentage the increase in attractiveness becomes curvilinear. Additionally, the research showed that the smallest bonus amount that was found attractive is about 5%. Thierry (2002) concludes that the difference must be perceived as large and attractive. Expectancy theory argues the same by stating that motivation is only likely when the outcome is seen as attractive and satisfying the needs of the employee. Hence, this leads to the following hypothesis:
Hypothesis 5. A pay-out amount that is perceived as large and attractive leads to a change in organizational behaviour.
Pay-out frequency With regard to the frequency of the pay-outs, Wiscombe (2001) notes that pay outs should be made quarterly or at least more often than annually. According to reflection theory, rewards have more effect on organizational behaviour when individuals validate more meaning in their rewards (Thierry, 2002). Frequent pay-outs of the bonuses give employees a better knowledge of their functioning and the need for a new working method.
Hypothesis 6. Pay-outs that are made frequently leads to a change in organizational behaviour.
Participation Another intervention that is appropriate for enhancing the effectiveness of pay-for-performance is the involvement of employees in the development of the pay-for-pay-for-performance system. Change management literature is unanimously in declaring that employee involvement during organizational change is critical for its success (Chawla & Kelloway, 2004). Coch and French (1947) were the first authors to stress the importance of ‘participation’ in their search for ways to overcome resistance to change. By means of active participation employees can gather information and can observe actions, events and people they would have otherwise not encountered (Fishbein & Ajzen, 1975). These effects of participation may influence the acceptance of the information (Fishbein & Ajzen, 1975) as people tend to believe own findings rather than what is said by others, but not yet experienced. Gathering new information and enduring new experiences may cause people to change their beliefs (Fishbein & Ajzen, 1975). This in turn can perhaps influence one’s willingness to change. If employees are encouraged to participate in the change process and their input is enlisted, this participation is expected to increase commitment and performance, to reduce resistance and to enhance the acceptance of (even unfavourable) decisions according to Chawla and Kelloway (2004).
might be a good possibility precisely because it is so important and difficult to deal with. Because of the importance of the pay area, success here is likely to produce a general organizational climate that is characterized by mutual influence and trust. This is congruent with the literature which suggest that participation is likely to be successful when it involves issues that are important for employees (Vroom, 1964). There is no better indicator of the seriousness of an organizational change effort and of the trustworthiness of management than to turn over the pay administration to employees.
Summarizing, participation in pay decisions can decrease resistance to change behaviour since it produces an organizational climate of trust and recognition. Hence, the following hypothesis is developed: Hypothesis7: Involvement of employees in the development of a pay-for-performance system leads to a change in organizational behaviour.
FIGURE 2.1:
Predictors of Pay-for-performance Effectiveness
2.3 Predictors of Pay-for-performance in Enhancing Change Effectiveness
If pay-for-performance can have a powerful, positive effect on employee behaviour, can it also help in making the change effort successful? Just as pay systems can create resistance to change, it potentially can create a willingness to change. The expectancy theory (Vroom, 1964) can be helpful in understanding when pay systems will facilitate and when they will hinder change. According to Vroom (1964), when pay systems are perceived as rewarding change, they will encourage it, and when they are seen as punishing change, they will discourage it. For example, when individuals are offered bonuses for successfully and rapidly implementing organizational change, it can speed change, and when individuals
perceive that a change will lead to a better pay system, change is encouraged. Thus, compensation must be aligned with the change objective in order to make it effective.
Alignment of change objectives and pay indicators A review of compensation design literature shows that the alignment perspective has become almost universal among authors in the academic media. This perspective argues that no particular pay-for-performance plan is effective or ineffective in the abstract. Different pay-for-performance systems may be effective in different contexts. Rather, a given pay-for-performance plan is effective to the extent that it is aligned with business needs or objectives or, in this case, with the proposed organizational changes (Crosby, 1989).
Crosby (1989) correctly observes that poorly designed and implemented pay-for-performance systems destroy firm value by rewarding people for doing the wrong things. The critical issue is whether pay-for-performance and the resulting employee behaviour are associated with a contribution to the effectiveness of the change effort. It is, for example, easy to see how individually based piece rates can damage efforts to foster quality and teamwork. If teamwork contributes significantly to productivity, incentives based on individual performance destroy value by discouraging teamwork and cooperation. Also, piece rates encourage employees to trade off quality for increased quantity in output.
Bonus blindness and gaming. Bonus blindness (Cools, 2005) is a phenomena that people with a pay-for-performance plan only perform tasks that are related to the pay-for-performance plan, even though this can be very harmful for the company or its clients. This phenomena is enchantingly mentioned in the title of a famous article by Kerr (1975) as ‘ The folly of paying for A while hoping for B, or getting what you pay for’. The number of examples of bonus blindness is infinite. For example, piece rates for employees in a production hall led to an increase in productivity with 20%, while the number of returns grew with 30%. Employees at a warehouse received a bonus per repair resulting in advising twice as many repairs than necessary to their clients. Because of the bonus blindness, pay-for-performance systems often lead to undesired and value destroying behaviour. In the examples above the use of piece-rates and bonuses resulted in high turnovers on the short term. On the longer term, however, it has led to high costs (high bonuses), dissatisfied clients and a decrease in sales.
Another damaging effect of pay-for-performance systems is ‘gaming’ or fraud (Cools, 2005). This contains the manipulation of the measured performance, without really improving the performance. Recent studies (Burns & Kedia, 2004, Denis, Hanouna & Sarin, 2005, Erickson, Hanlon & Maydew, 2004) showed a positive relationship between incentive pay and fraud by top management.
The behaviour of the employees is good for the company value. The pay-for-performance system cannot guide its own direction (Crosby, 1989). It may hit the target, but it also may damage to the place from which it was launched. For a pay-for-performance plan to impact organizational change effectiveness, it must be given direction. The change strategy helps in giving direction to the compensation system by indicating the types of employee behaviour and abilities that the company needs to be successful in the market.
Keeping in mind the assumption that a pay for performance plan should support the change mission, the key issue here is that there should be a clear link between change goal attainment and bonus attainment resulting from a pay for performance plan. In terms of the expectancy theory (Vroom, 1964), this would mean that attainment of the change goals should be instrumental for the attainment of a (valued) bonus. The change goal indicators, the pay for performance indicators as well as the goals set on those indicators should be related (van Vijfeijken, 2004).
Pay-for-performance systems can also facilitate the communication process during change efforts since they establish the measures that define the new visions and goals. This can help employees in understanding what actions they should take to behave according to the new standard. Compensation systems serve as a directing or informational role (Fisher, Sprinkle & Walker, 2008). If goals are revised, but the reward system is not, employees get conflicting signals. Most subordinates solve this conflict by following the reward system (Wruck & Jensen, 1994). Reward systems that pay for performing in line with the new visions and policies can help making the change effort a success.
In sum, change objectives and pay for performance indicators have to fit together in terms of content. First, the desired change in behaviour should be determined; next a pay-for-performance plan can be installed that encourages this behaviour. Hence, the following hypothesis is developed:
Hypothesis 8: A fit between the change objectives and the pay-for-performance indicators is positively related to the effectiveness of the change effort.
2.4 Conclusion
FIGURE 2.2:
Conceptual Model Pay-for-performance & Change Effectiveness
Pay Perceived Transparency Participation Pay-out Frequency Perceived Fairness Perceived Controllability Pay-out amount Effective Pay-for-Performance system Change in organizational behaviour Effective change effort
Fit change objective &
3. METHODOLOGY
In the previous chapter the research and theory on how reward influences behavioural change was addressed. In addition, eight hypotheses with respect to pay-for-performance’s effectiveness during change efforts were developed. This chapter will deal with the methodology underlying this research. The design of the research and the means to obtain and to process data will be addressed in this chapter.
3.1 Research Design
This section will elaborate on a number of issues related to design choices I made in this research. Five organizations were selected in order to test the eight hypotheses that were developed in the previous chapter. The leading principle in the data collection phase was triangulation. This is the process of collecting multiple sources of evidence on a particular issue. A questionnaire, interviews and documents were used to establish data triangulation. The research contains a quantitative and a qualitative part, this will be elaborated further in respectively sections 3.2 and 3.3. First, the research focus and the selection of participating organizations will be discussed.
Research focus and unit of analysis To avoid confusion on what has actually been studied, the unit of analysis in this research is defined as an individual in an organization. So the employee has been the starting point, in the light of which the effectiveness of a pay-for-performance plan in enhancing organizational change effectiveness has been evaluated.
3.2 Quantitative Research
For the quantitative part of the research a questionnaire was used. The questionnaire consisted primarily of closed questions. The questionnaires were used to collect data from a large number of employees within the participating organizations, and to gather information in a consistent and comparable way. In this section the data collection and data analysis of the questionnaire are elaborated further.
Data collection: questionnaire The questionnaire was sent to 265 employees of the five participating organization by mail1 and were returned anonymously by 42% of the employees. The questionnaire renders information about employee perceptions of the hypothesized predictors of pay-for-performance, as well as an individual assessment of the level of change in their behaviour. Since a large number of employees can be addressed, the use of a questionnaire as a means to collect data is considered to be appropriate. In addition, in the assessment of attitudes it is common to make use of a questionnaire (Keller & Warrack, 2003). It has been stressed that no data would be published that would be reducible to a single person. This guarantee of anonymity intended to remove any hesitations with regard to the responses to the questionnaire.
The questionnaire can be found in appendix A and consisted of two parts, starting with general questions considering the respondents’ gender, age, education, discipline and tenure in the organization. The existing questionnaire of Bennebroek-Gravenhorst (2004) acted as a model for this part of the questionnaire. The second part of the questionnaire addressed the respondents’ perception of the pay-for-performance system and the level of motivation to change their behaviour. The hypotheses introduced in the preceding chapter have been assessed through various items. By making use of known multi-item scales that have proven to be valid and reliable, a solid base for the current research has been created. The items have been derived from previous studies (Bouma & Emans 2005, Hay Group Total Remuneration Survey, van Herpen et al., 2002, Kersten, 2004, Patterson, West, Shackleton, Dawson & Lawton, 2005) and have been adapted to the context under study. This includes the specification of the organization (i.e. Company A, Company B, Company C, Company D and Company E) the specification of a pay-for-performance plan and the specification of a change effort (i.e. the use of pay-for-pay-for-performance plan for influencing the behaviour during a change effort). Additionally, all questions were posed in Dutch, which meant that some of the items needed translation. Employees were asked to react on statements by providing answers on a five point Likert-scale ranging from ‘completely disagree’ to ‘completely agree’. To ensure the reliability of the measurement, the Cronbach’s Alpha (α) has been calculated for the measured concepts (Allen & Yen, 2002). This measure was chosen because it is frequently and widely
1
used as a reliability measure in several scientific areas. The higher the resulting Cronbach’s alpha the better the reliability of the scale. A high reliability indicates that all items included in the factor measure the same and are, for that reason, affected by error to a low degree. As a rule a minimum Cronbach’s Alpha of 0.70 has been applied to consider the constructed item as reliable (Allen & Yen, 2002). The items that decreased the reliability of the scale considerably when included have been removed. High reliabilities (0.95 or higher) are not necessarily desirable, as this indicates that the items may be entirely redundant. Consecutively, the factors and the items that are used to measure them are elaborated further.
Transparency. The perceived transparency and openness of pay-for-performance have been assessed by asking questions concerning the complexity and the clarity of the pay-for-performance plan. In total 5 statements were formulated in order to capture transparency. The items are based on the studies of Herpen et al. (2002), Kersten (2004) and Patterson (2005). Most of the items contained the word clear, referring to the complexity and communication dimensions of this item. For this item the cronbach’s alpha was equal to 0.898.
Fairness. The perceives fairness was measured by 11 statements. 6 statements were focussed on the perceived fairness of the outcomes. This was measured by asking employees whether they felt that the match between pay and performance was fair. The statements that measure the perceived fairness of the outcomes of the pay-for-performance plan were derived from the study of Herpen et al (2002). Cronbach’s alpha for this item equals 0.893. 5 statements were focussed on the perceived fairness of the process. This was measured by asking whether the employees felt that the pay-for-performance system treats them fairly. The statements were derived from the study of Kersten (2004). For this item the cronbach’s alpha was equal to 0.819.
Controllability. Controllability was measured by 6 statements concerning the employees’ influence on the pay-for-performance system. The statements were derived from the study of Herpen et al. (2002) and based on the literature concerning controllability. Initially, Cronbach’s alpha for this item equals 0.669. One statement was worded negatively and this resulted in a need to exclude this item. The exclusion of this item increased the reliability from α = 0.669 to α = 0.733.
Pay-out amount. This item was measured by a question concerning the size of the pay-out amount and a statement concerning the motivating potential of the size of this pay-out amount. The ranges of the pay-out amounts were based on the average size of variable pay as stated in the Total Remuneration Survey 2007 of the Hay Group.
question concerning the pay-out frequency and a statement concerning the motivating potential of the frequency of pay-out.
Participation. Participation was measured by 5 statements concerning the employees’ involvement in the development of the pay-for-performance system. The statements were derived from the studies of Patterson et al. (2005) and Bouma & Emans (2005). Cronbach’s alpha for this item equals 0.839.
Motivation to change behaviour. In order to grasp the level of change in behaviour as a result of pay-for-performance, the respondents were asked to react on 5 statements. All statements measured the link between pay-for-performance and the motivation to show new behaviour. Three of those statements were company specific and concerned specific behaviours that were formulated by management as the desired new behaviours. Since no empirical research has been done in linking pay to change, the statements are developed by the author herself. Cronbach’s alpha for this item equals 0.874.
Data analyses In order to generate results from the collected data various kinds of analyses have been conducted, making use of SPSS. The statistical choices will be addressed in this section.
Descriptive analysis. Firstly, the means on the different predictors and on the outcome variable were analysed. Since there are differences between the designs of the pay-for-performance system of the different participating organizations, it is reasonable to expect that the members of the different organizations have distinct perceptions of the pay-for-performance system and the resulting change in their behaviour. Looking at the overall scores will lead to biased conclusions and therefore the means of the variables were measured more in depth at a company level.
Exploratory Research. Firstly, a Pearson correlation analysis has been conducted. The resulting correlations made it possible to indicate whether there were relationships between the distinguished factors. Most interesting was the question whether the presumed predictors of an effective pay-for-performance system correlated significantly with the dependent variable, i.e. a change in behaviour. According to Miles and Shevlin (2001) an absolute correlation between .10 and .29 could be regarded ‘small’, an absolute correlation between .30 and .49 could be regarded ‘medium’, and an absolute correlation of over .50 could be regarded ‘large’.
behaviour and change effectiveness. By means of series of regression analyses it has been tested what variables of a pay-for-performance plan are related to motivation to change behaviour and change effectiveness.
Collinearity. Important to note is the issue of collinearity, which occurs when independent variables correlate to a large extent. Collinearity may cause problems in executing regression analyses, because it is difficult to determine which factor is important (Field, 2005). If multiple independent factors are important predictors, there will be increased uncertainty (measured by standard errors) and inaccuracy (measured by the regression coefficients) (Miles & Shevlin, 2001). A collinearity problem might be indicated by a high correlation when two factors are considered, although a lower correlation does not indicate the absence of collinearity problems. If there are multiple predictors, it is the multiple correlations that matter (Miles & Shevlin, 2001). With the help of SPSS collinearity was diagnosed by determining the tolerance (Field, 2005). A tolerance value of 0 indicates complete collinearity, while a tolerance value of 1 indicates that the considered predictor is uncorrelated with the other predictors. Closely related to tolerance is the variance inflation factor (VIF), which can be determined by dividing 1 by the tolerance level. The greater the VIF, the greater the possibility of collinearity. If collinearity causes serious problems, i.e. VIF > 2, it would be best to collect new data (O’ Briens, 2007). Collinearity was also checked by analysing the eigenvalues and the condition index. Since it would not have been an option in this research to collect new data, it seemed best to remove or combine items of the distinct predictors, as is suggested as well (Field, 2005). High correlation between predictors implied that rather similar constructs were measured. A principal component analysis (similar to a factor analysis) was used to regroup items in to new uncorrelated predictors (components). These uncorrelated predictors were used as independent predictors in regression analyses.
3.3 Qualitative research
With the help of the statistical test as described in the previous section the relationship between pay-for-performance plans and behavioural change can be described. However, this does not tell us anything about the relation of this new behaviour and the success of the change effort. In order to get a grasp on this relation, interviews have been conducted with the management of the organizations under study.
Data collection: interviews & documents Semi-structured interviews were conducted with Human Resource Managers and team leaders to gather information concerning the design and implementation of the pay-for-performance system and to measure the influence of pay-for-performance on the effectiveness of the change effort. The focus of the interviews was on questions concerning the desired and realised changes in behaviour and the role of pay-for-performance in obtaining the organizational change objectives. In addition, interviewees were asked to give their opinion on why pay-for-performance worked or did not work. All the interviews were held in Dutch. Confidentiality on the outcomes of the interviews has been guaranteed in order to obtain reliable results.
The existing data used for this research consisted of many different things, such as descriptions of the organization, the pay-for-performance plan, job descriptions and socio-demographic data for the employees under study. This information was mainly used as background information about the participating organizations and their pay-for-performance plan. In addition, this information was used to compare the different cases and to establish literal and theoretical replication. A consultant of Hay Group compared the pay-for-performance systems with the results of the research in order to provide some help with the explanations for differences and similarities between the different participating organizations.
Data analyses: multiple-embedded case study design The interviews have been tape recorded and transcribed on paper in Dutch. The responses to the open-ended interview questions have been coded. These coded responses have been grouped into themes. Within a case the frequencies of the themes indicate the importance of the suggestions and comments. Per theme the responses have been compared within the different participating organizations. The transcriptions of the interviews have been archived in the case study database. This was done in order to make the research verifiable.
TABLE 3.1:
Four Basic Case Study Designs (Source: Yin, 1994)
Single-case designs Multiple- case design Holistic (single unit
Of analysis)
Embedded (multiple Units of analysis)
In this study, a multiple-embedded case study design has been followed. It is multiple because several case studies were conducted, and it is embedded because several employees (units of analysis) within an organization (case) were studied. When applying a multiple cases design, Yin (1994) stresses the importance of following a ‘replication logic’ rather than a ‘sampling logic’. Whereas in a sampling logic approach multiple cases are considered similar to multiple respondents, multiple cases are considered as multiple experiments under a replication logic approach. Following a replication logic in a multiple-cases design means that the focus lies on replication of findings across different cases, thus the emphasis when interpreting the findings should be on cross-case conclusions, rather than on an isolated, within-case analysis approach. Analogous to experimental replication, Yin (1994) distinguishes two types of replication: literal and theoretical replication. Literal replication takes place when results found in a first case, are predicted and found in a second case. Theoretical replication takes place when a second case produces results that contrast with the findings of a first case, but for a predictable reason (Yin, 1994).
Design 1 Design 2
4. RESULTS
In this chapter the result of the empirical research will be presented. First the participating organizations and their pay-for-performance system will be introduced. Next the mean ratings of the measured factors will be described. Thereafter, the responses of the employees of the different organizations will be compared and the proposed hypotheses will be tested. Finally the results of the qualitative research will be described. The role of pay-for-performance in stimulating behavioural change that enhances the effectiveness of the change effort in the different participating organizations will be discussed here.
4.1 Respondents
Five companies participated in this research. In this section the different companies will be introduced and a short description of the pay-for-performance system and its goals will be given.
Company A The Netherlands is a distribution country par excellence. Each day, millions of tonnes of goods enter and leave our country: by land, sea and air. Obviously, these goods movements have to be monitored. This is the responsibility of Company A. In performing these activities, customs deals with political, policy and legal rules, both at European and at national level. By performing its monitoring duty as effectively and efficiently as possible, Company A ensures that obstacles to trade are restricted to a minimum. Company A has three core duties: 'stopping' goods at the border, 'monitoring' the correct application of laws and regulations, 'levying and collecting' taxes. About 10 years ago pay-for-performance has been implemented at all offices of company A in the Netherlands. The implementation has been forced by the ministry of internal affairs of the Netherlands in order to change passive civil servants to professional workers, thus stimulating employees to put forward more effort and initiative. Incentive pay is paid out once a year and depends on the performance appraisal. There are no guidelines or frameworks for determining which employees get incentive pay, the only specifications given are that 30% of all employees should get a bonus and a specific budget is given to be shared amongst these 30% of employees. This means that employees could earn about a half to one month salary extra. Interviews were held with the HR manager and two team leaders. The questionnaire was send to all employees of the unit Amersfoort, 41 of the 75 employees filled in the questionnaire.
organization managed and financed by its members. Through its activities Company B establishes a bridgehead between the insurance industry and other sectors of society. Company B employs about 120 employees, mostly policy officers. In order to enhance the quality of work a pay-for-performance system has been implemented. The objectives of the pay-for-performance were more focus on results, output en development of employees. Incentive pay is paid out once a year and depends on personal and SMART formulated appointments with the team leader. These appointments are focused on personal results and developmental issues and these appointments are the input for the performance appraisal. The different appraisals (A,B,C,D & E) are coupled to different percentages of incentive pay. An interview was held with the HR Manager and the questionnaire was send to all employees. 31 of the 116 employees filled in the questionnaire.
Company C Company C is a company that delivers energy. Internationally, Company C occupies top rankings in its core businesses; electricity, gas, water & wastewater, waste disposal & recycling. In the Netherlands, Company C is a relatively small party. It delivers gas and electricity to 350.000 households and 30.000 companies in the Netherlands. About 350 employees take care of everything that is needed to supply this energy. In 2007 the energy market in the Netherlands was liberalized. This meant a transformation from a public services company to a more commercial and market orientated firm in order to keep up with the competition in the liberalized energy market. In order to manage this change, the management of Company C implemented pay-for-performance. The objective of pay-for-performance was to institutionalize a cultural change. People had to think in customers instead of in connections. People were given a stimulus to put more effort in gaining new clients and in addition they were stimulated to take more initiative, to look for opportunities and to cash the opportunities found. Incentive pay is partly based on individual performance (realised volume) and partly on the profitability of the company as a whole. A maximum of 2,5 month salary can be earned extra and incentive pay is paid out once a year. The research took place in the sales force (B2B and B2C) of Company C. Interviews were held with the managers of B2B, B2C and HR. The questionnaire was send to all account managers, 8 of the 20 account managers returned the questionnaire.
Since the objective of the pay-for-performance system is to stimulate growth, customers which resign are subtracted from the incentive pay amount. The research took place in the account management department of Company D. An interview was held with the director of Company D and the questionnaire was send to all account managers. 3 of the 6 account managers returned the questionnaire.
Company E Company E is an organization that provides project management, program management, IT management and financial management to clients in the telecom, financial, energy and government sector. Employees of Company E are sent to one of the clients on a temporary duty for a limited period of time. Pay-for-performance was implemented to facilitate the development towards a more professional organization, i.e. an organization with satisfied clients, growth through references and long term relations with both clients and employees. Incentive pay is focused on performance and depends on the following formula: extra billable hours (counting starts above 1400 hours) x tariff x percentage depending on performance appraisal. The performance appraisal depends on personal development, client evaluation, attitude and special efforts for Company E. The incentive pay is paid out once a year. Interviews were held with managers of Finance, Information and HR. The questionnaire was send to all professionals. 29 of the 48 professionals filled in the questionnaire.
4.2 Descriptions of Results
In this section some information is provided on the mean ratings of the factors. In total 112 of the 265 employees filled in the questionnaire, this corresponds with a response rate of 42 %. 88 men and 24 women filled in the questionnaire and this corresponds with the gender distribution in the participating organizations. With respect to age, most respondents felt in the category above 50 (namely 35,7%). First a general overview of the results is given. Next the different participating organizations are compared on all predictors at item level. Finally, the results of the descriptive analysis are summarized.
other on each variable. The items of each variable can be found in appendix A. Table 4.1 also shows the standard deviation for the different organizations. One can discover that the standard-deviation within the different organizations is low, as such the mean is an accurate representation of the variables.
TABLE 4.1:
Mean Ratings and Standard Deviations of the Predictors and Dependent Variable
Mean Std. Deviation Std. Deviation Company A Std. Deviation Company B Std. Deviation Company C Std. Deviation Company D Std. Deviation Company E Behavioural change 1,98 ,860 ,56 ,92 ,35 ,50 ,72 Transparancy 3,32 1,02 ,95 ,97 ,63 ,58 ,70 Outcome fairness 2,78 ,87 ,86 ,70 ,63 ,58 ,64 Process fairness 2,79 ,78 ,74 ,75 ,40 ,31 ,43 Controllability 2,98 ,70 ,74 ,67 ,42 ,42 ,43 Participation 2,75 ,85 ,70 ,94 ,61 ,14 ,46 Motivating potential of
size of pay out amount 2,62 1,17 ,83 1,26 ,87 ,00 ,77 Motivating potential of
frequency of pay out 2,39 ,99 ,88 1,06 ,92 ,56 ,72 N
112 41 31 8 3 29
Comparing the different organizations In this section the participating companies will be compared with each other on each variable.
FIGURE 4.1: Level of Behavioural Change
Transparency. Table 4.1 showed that transparency was the only predictor that scored above 3 (µ = 3.32). However, this is still not a very high rating, meaning that employees did not find the pay-for-performance system transparent, clear and understandable. However, when the different organizations are compared (figure 4.2), one can discover that this low rating is caused by Company A. At Company D, on the other hand, employees rated the predictor transparency very high, meaning that pay-for-performance is communicated well and that it is clear for the employees how pay-for-performance works. The employees of the other three organizations rated the variable transparency between average and good.
FIGURE 4.2: Level of Transparency
questions about pay-for performance are answered individual results are
communicated well procedures are well
communicated understands pay-for-performance system clarity of pay-for- performance 5 4,5 4 3,5 3 2,5 2 Mean Company E Company D Company C Company B Company A company
pay caused change in behaviour focus at performance indicator 2
focus at performance indicator 1 degree of change in behaviour
Outcome Fairness. The predictor ‘outcome fairness’ scored a mean rating of 2.78 (table 4.1), this means that, overall, employees perceived that they did not receive a fair amount of incentive compensation relative to their input. When the different organizations are compared (figure 4.3), one can notice that at Company D the employees perceived the outcome distribution of the pay-for-performance system as fair. Again, the employees of Company A rated the variable low, counting for the low overall mean rating. The employees of Company B, Company C and Company E rated the fairness of the distribution of the incentive pay as average.
FIGURE 4.3: Level of outcome fairness
Process Fairness. Table 4.1 showed that the overall mean rating for process fairness was low (µ = 2.79). When zoomed in (figure 4.4), it becomes clear that the employees of Company A rated the accuracy of the performance assessment and the adherence to the defined pay-for-performance policy lowest, immediately followed by the perceptions of the employees of Company B and Company C. Again the pay-for-performance system of Company D was rated to be the most adequate. A fact that is worth mentioning with regard to process fairness is that all employees agree that their performance assessment is partly based on a subjective assessment of their manager.2
2 In the bar chart the item is not recoded for visual reasons. For all analysis this item was recoded (see:
methodologies)
colleagues who get incentive pay, earn this good performers earn
more feels recognized by
incentive pay pay in line with
meaning for org pay in line with
devotion pay in line with
FIGURE 4.4: Level of Process Fairness
Controllability. The predictor ‘controllability’ scored a mean rating of 2.98 (table 4.1), this means that overall employees did not expect that their efforts and contributions would produce worth-while rewards. However, when the different organizations are compared (figure 4.5), one can discover that the Company A employees rated controllability considerably lower. Interestingly, the item ‘performance goals are reasonable’ is rated much higher than the other items. The reason for this could be the fact that performance goals are not linked to variable pay at Company A. Only at Company D employees believed that the pay-for-performance system uses correct criteria and that they could influence the size of their incentive pay. The employees of Company B, Company C and Company E perceived the link between effort and rewards as rather weak (score between 2.5 and 3.5).
FIGURE 4.5: Level of Controllability
management follows incentive pay procedures strictly partly subjective assessment of manager objective assessments accurate assessments Pay-for-performance is regulated well 5 4 3 2 1 Mean Company E Company D Company C Company B Company A Company
my manager made sure I had the capabilities to get
a bonus performance goals are
reasonable getting a bonus depends
on myself by working hard I can
influence my incentive pay system uses correct
Size and Frequency. Table 4.1 shows that, overall, the motivational potential of the size of the pay-out amount was rated slightly negative (µ = 2.62). However, by looking at table 4.2 one can discover that only Company A and Company B rated the size of their bonus as too low to stimulate them to change their behaviour. Company D, Company C and, though in a lesser degree, Company E perceived the size of the incentive pay amount to be high enough to stimulate them to change their behaviour. It is worth mentioning that this information correlates perfectly with the information from the desk research on the pay-out amounts. This correlation shows that the higher the pay-out amount the higher the perception of employees that the amount stimulated them to change their behaviour. With regard to frequency, only Company D rated the frequency of pay-out as slightly stimulating them to change their behaviour (figure 4.6). The desk research showed that pay-out frequency is the same for all participating organizations (i.e. once a year) and therefore it can be concluded that the motivating potential of frequency of pay-out varies per employee and per organization.
TABLE 4.2:
Motivating Potential of Pay-out Amount
Size of the maximum pay-out amount Mean rating Motivating potential Company D 20 000 euro 4,00 Company C 20 % of annual salary 3,75 Company E extra billable hours
(counting starts above 1400 hours) x tariff x percentage depending on performance appraisal.
3,32
Company B 8% of annual salary (depends on performance and Compa Ratio3)
2,61 Company A 8% of annual salary
1,83
FIGURE 4.6:
Level of Motivation to Change Behaviour by Frequency of Pay-out
Participation. Table 4.1 shows that the overall mean rating for participation is 2.75. When zoomed in at item level (figure 4.7), one can discover that employees have a reasonable understanding of the changes in the pay system and the goals of the pay-for-performance system. However, when looking at the item ‘degree of participation’4 it becomes clear that most employees were only informed about the changes in the pay system and its goals. Only the employees of Company D indicated that they were consulted for their opinion before the implementation of the pay-for-performance system. Figure 4.7 shows that the employees of Company A noted that they were totally kept out.
FIGURE 4.7:
Level of Participation in Incentive Pay Decisions
4 The scale of this item is different: 1.kept out, 2.informed, 3.consulted for opinion, 4.directly involved, 5. consulted
Company E Company D Company C Company B Company A 4 3 2 1 0 Mean degree of participation opportunities to give opinion participation in payment decisions understands goal variable
pay understands reason for
Summary of descriptive results The descriptive analysis showed that all employees indicated that they did not change their behaviour since the introduction of the pay-for-performance system. Employees of Company A rated all moderating predictors low. This means that the employees of Company A perceive pay-for-performance system as badly designed. The employees of Company B, Company C and Company E rated all predictors neutral. This means that the pay-for-performance system is not perceived to be badly designed. However, it does also not indicate that the pay-for-performance system is well designed. The employees of Company D rated all predictors high. This means that the pay-for-performance system is designed well. But still, however, this did not lead to employees indicating that they changed their behaviour or focus on results.
Overall, there seems to be some correlation between the predictors of pay-for-performance and the change in behaviour. Four companies did not fulfil the predictor requirements of a pay-for-performance system, and it might be that therefore the employees did not change their behaviour. However, Company D did fulfil the predictor requirements, but their employees did not indicate that they changed their behaviour. In order to test the correlations and relations between the predictors of pay-for-performance and the outcome variable ‘change in behaviour’, exploratory research has been conducted. In the next section, the results of the correlation analysis and the regression analysis are discussed.
4.3 Exploratory Results
TABLE 4.3:
Intercorrelations Between the Variables in the Study
Behavioural
change Transparency Outcome fairness Process fairness Controllability Participation
Motivating potential of size of pay out amount Motivating potential of frequency of pay out Behavioural change 1 ,42(**) ,35(**) ,28(**) ,50(**) ,47(**) ,52(**) ,43(**) Transparency 1 ,69(**) ,77(**) ,69(**) ,65(**) ,58(**) ,46(**) Outcome fairness 1 ,73(**) ,68(**) ,60(**) ,54(**) ,63(**) Process fairness 1 ,70(**) ,59(**) ,56(**) ,41(**) Controllability 1 ,69(**) ,54(**) ,53(**) Participation 1 ,52(**) ,56(**) Motivating potential of
size of pay out amount 1 ,58(**)
Motivating potential of
frequency of pay out 1
Since the purpose of this research is to understand the relationship between de predictors (transparency, outcome fairness, process fairness, controllability, participation, size of pay-out amount and frequency of pay-out) and the dependent variable (change in behaviour), the high correlation between the predictors may cause collinearity. The concept of collinearity has been explained in the methods section. Hence, we look for the variance inflation factor (see column ‘VIF’ in table 4.4) to determine if collinearity might bias the regression analysis.
TABLE 4.4:
Collinearity Statistics of Predictors
Tolerance VIF Transparency ,30 3,30 Outcome fairness ,29 3,46 Process fairness ,27 3,66 Controllability ,31 3,23 Participation ,41 2,42 Motivating potential of
size of pay out amount ,52 1,91 Motivating potential of
frequency of pay out ,45 2,23 Dependent Variable: change in behaviour
TABLE 4.5: Eigenvalues
Dependent Variable: behavioural change
As has been explained in the methodology section, a principal component analysis can be conducted in order to determine whether the predictors are independent. The objective is to identify those items that most clearly represent the content domain of the underlying construct. An eigenvalue of 1 has been the criterion for composing the components.
Table 4.6 shows the results of the principal component analysis. There are no hard-and-fast rules for this, but the 0.4 criterion level appears to be most commonly used in judging factor loadings as meaningful (Ford, MacCallum & Tait, 1986). Additionally, a loading that is much stronger on the appropriate factor than on any other factor is considered to be meaningful (Hinkin, 1998).The items in boldface will therefore be excluded from the regression analysis. The items in italic will be merged with the factor ‘process fairness’. Since reasonable performance goals and an organization that gives their employees the capabilities to obtain a bonus can also be seen as making the process to obtain a bonus more fair, these items will, from now on, also represent the construct ‘process fairness’. If the process to obtain incentive pay is not perceived as fair, employees can not reasonably expect that their efforts and contributions will produce worth-while rewards their effort. Therefore it is reasonable to merge these factors. Unfortunately, this means that the variable ‘controllability’ is excluded from further research and the hypothesis concerning this factor can not be analysed.
TABLE 4.6:
Principal Component Analysis of Predictors
Component
Item 1 2 3 4 5
T1.clarity of pay-for-performance ,10 ,87 ,18 ,11 ,09 T2.understands pay-for-performance system ,22 ,74 ,35 ,27 ,01 T3.procedures are well communicated ,19 ,62 ,35 ,35 ,00 T4.individual results are communicated well ,24 ,59 ,49 ,32 ,25
T5.expects questions about incentive pay are
answered ,27 ,15 ,69 ,28 ,18
OF1.pay in line with performance
,84 ,18 ,31 ,09 ,14
OF2.pay in line with devotion ,85 ,22 ,27 ,09 ,13 OF3.pay in line with meaning for org
,86 ,14 ,26 ,17 ,09
OF4.feels recognized by incentive pay
,46 ,35 ,41 ,11 ,20
OF5.good performers earn more
,42 ,28 ,15 ,33 ,40
OF6.collegues who get incentive pay, earn this
,46 ,22 ,35 ,35 -,03 PF1.Pay-for-performance is regulated well ,34 ,34 ,53 ,41 ,28 PF2.accurate assessments ,32 ,41 ,69 ,21 ,15 PF3.objective assessments ,35 ,33 ,63 ,09 ,16
PF4.partly subjective assessment (reco)
,13 ,04 ,10 -,06 ,72 PF5.strict dedication to procedures by management ,20 ,54 ,53 ,25 ,11
C1.good criteria ,30 ,39 ,38 ,47 ,24
C3.I have influence on incentive pay by working
hard ,40 ,19 ,14 ,62 ,05
C4.obtaining a bonus depends on the self
,45 ,42 ,23 ,01 -,10
C5.performance goals are reasonable
,36 -,03 ,50 ,18 -,41
C6.organization made sure I had the capabilities to get
a bonus ,10 ,28 ,69 ,22 -,18
P1.understanding of change in pay system ,38 ,46 ,04 ,46 -,35 P2.understanding of goal pay-for-performance ,45 ,45 ,11 ,42 -,21 P3.participation in payment decisions ,35 ,28 ,17
,73 ,07
P4.opportunities to give opinion ,05 ,02 ,38 ,74 -,02 P5.degree of participation -,01 ,24 ,15 ,71 -,11 Motivating potential of size of pay out amount ,38 ,50 ,10 ,33 ,14 Motivating potential of frequency of pay out
,67 ,14 ,06 ,41 -,02 Rotation Method: Varimax with Kaiser Normalization.
Table 4.7:
Principal Component Analysis of Predictors After Changes in Constructs
Component
Item 1 2 3 4
T1.clarity incentive pay
,85 ,14 ,23 -,06 T2.understands incentive pay
system ,78 ,23 ,34 ,16
T3.procedures are well
communicated ,66 ,19 ,43 ,19 T4.individual results are
communicated well ,76 ,30 ,28 ,27 OF1.pay in line with
performance ,24 ,90 ,15 ,17 OF2.pay in line with devotion
,25 ,91 ,18 ,12 OF3.pay in line with meaning
for org ,19 ,87 ,22 ,17
OF4.feels recognized by
incentive pay ,52 ,52 ,07 ,27 PF1.incentive pay is regulated
well ,51 ,39 ,29 ,45
PF2.accurate assessments
,57 ,38 ,19 ,52
PF3.objective assessments
,51 ,42 ,05 ,50
PF6.performance goals are
reasonable -,09 ,28 ,27 ,63 PF7.organization made sure I
had the capabilities to get a bonus ,39 ,06 ,20 ,72 P1.understanding of change in pay system ,24 ,24 ,72 ,06 P2.understanding of goal pfp ,28 ,41 ,66 ,05 P3.participation in payment decisions ,26 ,32 ,70 ,23 P4.opportunities to give opinion ,13 ,01 ,62 ,45 P5.degree of participation ,20 -,03 ,73 ,18 Rotation Method: Varimax with Kaiser Normalization.
a Rotation converged in 6 iterations.
TABLE 4.8:
Collinearity Statistics After Changes in Constructs
Dependent Variable: Behavioural Change
Collinearity distorts the significance level of coefficients, making it difficult to determine whether any of the predictors is linearly related to the dependent variable. It also makes interpreting the coefficients difficult. When the coefficients are tested the significance will be low, which leads to the interference that there is no linear relationship between the predictors and the dependent variable. In some cases, however, this inference will be wrong. Fortunately, collinearity does not affect the F-test of the analysis of variance. The results of the multiple regression analysis are shown in tables 4.9 and 4.10.
TABLE 4.9: F-test Anova
Sum of
Squares df Mean Square F Sig. Regression 21,98 6 3,66 7,03 ,00 Residual 46,90 90 ,52 Total 68,87 96 Tolerance VIF Transparency ,40 2,51 Outcome Fairness ,41 2,44 Process Fairness ,36 2,82 Participation ,46 2,17 Motivating potential of
size of pay out amount ,53 1,90 Motivating potential of
TABLE 4.10: Multiple Regression Analysis
Standardized Coefficients Sig. Beta Transparency ,10 ,44 Outcome Fairness -,11 ,48 Process Fairness -,02 ,89 Participation ,24 ,06 Motivating potential of
size of pay out amount ,29 ,02 Motivating potential of
frequency of pay out ,16 ,23
R Square 0,32
Dependent Variable: change in behaviour
The regression output reveals that only the ‘motivating potential of size of the pay-out amount’ is significantly related to a change in behaviour (α = .02). However, the F-test (F = 7.03 and α = .00) indicates that the complete model is valid. Summarizing, the model is valid, but only one of the predictors that makes up the model is linearly related to change in behaviour. Simple regressions analyses are performed to find the reason for this. Simple regression analyses determine the relation between each of the predictors and the dependent variable (see table 4.11).
TABLE 4.11: Simple Regressions Unstandardized Coefficients Standardized Coefficients t Sig. Beta Std. Error Beta Transparency ,32 ,07 ,42 4,61 ,00 Outcome Fairness ,34 ,09 ,35 3,85 ,00 Process Fairness ,38 ,10 ,36 3,84 ,00 Participation ,47 ,09 ,47 5,40 ,00 Motivating potential of
Size of pay out amount ,38 ,06 ,52 6,36 ,00 Motivating potential
The significance levels in the multiple regression model lead to the inference that only one predictor surfaces in itself as a predictor of the change in behaviour. The six simple regression models qualify this conclusion. They tell us that the transparency, outcome fairness, process fairness, participation, and the motivating potential of size of pay-out amount and frequency of pay-out are all linearly related to the dependent variable ‘change in behaviour’.
