A bulletproof performance appraisal : quantitative research about the relation between structured performance appraisals, the individual performance, and performance appraisal satisfaction of employees

(1)

A BULLETPROOF PERFORMANCE APPRAISAL

Quantitative research about the relation between structured performance appraisals, the

individual performance, and performance appraisal satisfaction of employees

 Author: C.M. Wienema

 Student number: 10217851

 Master Thesis in Business Administration

 Leadership and Management Track

 The Amsterdam Business School

 University of Amsterdam

 Supervisor: dhr. dr. S.T. (Stefan) Mol

(2)

STATEMENT OF ORIGINALITY

This document is written by Claudia Wienema, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is 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.

(3)

A

BSTRACT

Despite a lot of research, still little is known about the direct influence of performance

appraisals on employees’ individual performance and performance appraisal satisfaction. The

aim of this study is to investigate if structured performance appraisals can improve

employees’ individual performance and performance appraisal satisfaction. Within this

research, 217 performance appraisal forms and survey questions about performance appraisal

satisfaction of employees are combined. The sample consists of employees from a big

international performance-focused organization, who all have a mandatory quarterly

performance appraisal meeting. The performance appraisals are coded on degree of structure

by counting the number of goals per action plan. The individual performance is measured by

actual result per goal, and appraisal satisfaction is measured by the employees’ improvement

value and the employees’ satisfaction with the performance appraisal. After controlling for

earlier individual performance, gender, age, and service years, research findings show that

there seems to be a positive correlation between the structure of the performance appraisal

and the individual performance in the next quarter. Hence, the first hypothesis that a higher

degree of structure will result in higher individual performance was supported. Research

findings do not indicate a correlation or a mediated relationship between the degree of

structure of the performance appraisal, and the satisfaction with the appraisal interview of

employees. This study contributes knowledge to the scientific literature of why structuring

performance appraisals matters, which can be used by organizations to obtain superior

(4)

A

CKNOWLEDGEMENTS

This thesis could not be written without the help of some wonderful people. I wrote this thesis during an internship at a multinational company. Therefore I would like to thank all

employees of this company who took part in this research and who gave me the opportunity to do research within such an interesting Human Resources environment.

Also, this thesis journey would never have been completed without the honest feedback of my supervisor, dr. S.T. Mol. Hereby I would like to thank you for providing me this feedback and professional guidance. I learned a lot in the last couple of months, and you even gave me insights into other interesting research topics.

But, I also have to be honest. During writing this thesis I experienced some difficult times and multiple challenges, but at the end I even start enjoy writing it. Sometimes, this thesis drove me crazy and therefore I was not the best (girl) friend ever. However, some people still helped me personally through this time, motivated me, and even read multiple pieces of this research to help me completing this piece of work. So, this complete thesis would never been finished without their love, advice and help, and I am really grateful for that.

At least, I would like to thank you as a reader. I hope that this thesis contains information that will teach you something new, and above, I hope that you will enjoy reading it!

(5)

4. RESULTS ... 27 4.1MEASURES ... 27 4.1.1INDIVIDUAL PERFORMANCE IN Q2 ... 27 4.1.2INDIVIDUAL PERFORMANCE IN Q3 ... 28 4.1.3RELIABILITY ... 28 4.2DESCRIPTIVE STATISTICS ... 29 4.3EMPIRICAL RESULTS ... 30 4.3.1HYPOTHESIS 1 ... 30 4.3.2HYPOTHESIS 2 ... 32 4.3.2.1HYPOTHESIS 2A ... 32 4.3.2.2HYPOTHESIS 2B ... 33 4.3.3HYPOTHESIS 3 ... 34 5. DISCUSSION ... 35

5.1FINDINGS AND PRACTICAL IMPLICATIONS ... 35

5.2LIMITATIONS ... 37 5.3FUTURE RESEARCH ... 39 6. CONCLUSION ... 40 8. REFERENCES ... 41 APPENDIX A APPENDIX B APPENDIX C APPENDIX D

(7)

1. I

NTRODUCTION

In 1989, Pooyan and Eberhardt concluded: “Despite the growing number of studies investigating employee attitudes toward performance appraisal, relatively little is known about differences in appraisal studies between supervisory and nonsupervisory employees” (p. 9). At the moment, this research conclusion is more than 25 years old and still little is known about performance appraisals and their influence on employees.

Although research findings are still limited, multiple studies in the area of

performance appraisals are done and the emphasis of this research area has changed over the years (Den Hartog, Boselie, & Paauwe, 2004). Before, performance appraisal research used to focus on performance measurement issues and the truthfulness of performance ratings by supervisors. However, over the years, performance appraisal research has been broadened to motivational and social aspects of the performance appraisal process (Fletcher, 2001). According to Fletcher (2001), today performance appraisals are an important practice of performance management, and used by organizations as a tool to evaluate and develop

employees’ competencies, administer rewards, and boost employees’ individual performance. Which factors ought to influence the performance appraisal process has been explored in several studies (e.g. Boachie-Mensah & Dogbe, 2011; Fletcher, 2001; Lawler, Benson, & McDermott, 2012). Previous research on performance appraisals has mainly focused on the (in)direct influence of gender (Grund & Sliwka, 2007; Lyness & Heilman, 2006), age (Brown & Heywood, 2005; Grund & Sliwka, 2007), and employee motivation (Nerdrum & Erikson, 2001) on performance. But most research on performance appraisal has focused on the link between the performance appraisal process and organizational performance (Boselie, Paauwe & Jansen, 2001; DeNisi & Kluger, 2000; Selvarajan & Cloninger, 2011; van Scotter,

Moustafa, Burnett, & Michael, 2007; Wright & Snell, 1998).

1.1 O

RGANIZATIONAL PERFORMANCE

According to Den Hartog et al. (2004) and Guest (1999), a common mistake made in

performance appraisal research focused on organizational performance, is that mostly human resource (HR) managers are used as the only source of information. HR managers focus particularly on the managerial view instead of the employee perspective (Den Hartog et al., 2004). This managerial view on performance influences the relationship between HR

(8)

use multiple sources of information, managers should be involved in the performance

management process. A key performance management tool that communicates standards and evaluates performance is the performance appraisal (Den Hartog et al., 2004). However, research findings of Lawler, Benson, and McDermott (2012) show that both managers and employees usually are poor at conducting performance appraisals.

1.2 I

NDIVIDUAL PERFORMANCE

So, some of the most interesting challenges for future performance appraisal research would be to determine what constitutes effective individual performance, and what should be measured and stimulated to enhance individual performance (Den Hartog et al., 2004). By focusing on performance appraisal research in the direction of individual performance

improvement, the gap between performance management research and performance appraisal practice can be diminished, which will help organizations to improve understanding how to boost individual performance of employees (DeNisi & Pritchard, 2006).

1.3 I

MPLEMENTATION

There are clear parallels between the structure that is imposed on performance appraisals and the structure of the selection interview process. Although it has been shown that structuring selection practices are valuable for organizations to improve the quality of the process (McDaniel, Whetzel, Schmidt, & Maurer, 1994), many organizations fail to implement and use a structured selection process. Instead, organizations still rely on intuitive, highly subjective approaches that increase biases in the selection process and lower the quality of decisions (Dipboye, 1994). Highly subjective approaches are not measurable, although a key factor for implementing performance management practices is measurability (van Dooren, 2005). According to van Dooren (2005), bad implementation of performance measurement is often a result from a lack of resources. Hence, sufficient performance management resources, such as the performance appraisal, are a critical factor for implementing, measuring,

maintaining and extending performance management practices (van Dooren, 2005) and to boost individual employee performance (DeNisi & Pritchard, 2006).

(9)

1.4 C

ONTRIBUTION AND OBJECTIVE

Boachie-Mensah and Dogbe (2011) did study performance appraisals, and especially the link between performance appraisals, individual performance and rewards, but they only studied the situation in a manufacturing company in Ghana. One of their recommendations is that future research should focus on situations within other organizations. This research can extend the findings of Boachie-Mensah and Dogbe (2011) who note that individual

performance will increase through improved goal setting. Increasing individual performance through goal setting means that the first phase of goal setting should be one of the most important practices of performance management. In the case of performance appraisals, by classifying performance appraisals as a goal setting process, the performance appraisal will be structured by goals. This means that employees will work harder if they have goals and believe that performance improves as a result from their effort (Lawler, 2010; Lock and Latham, 1990).

Subsequently, receiving more feedback pertaining to the goal and its achievement during the performance appraisal process motivates employees to increase their individual performance, since employees will be more satisfied with the performance appraisal when they are engaged in the process, and when it is based on predetermined goals (Bargh & Ferguson, 2000; Pooyan & Eberhardt, 1989). Some researchers, however, still observe the area ofemployees’ appraisal satisfaction, structured and unstructured performance appraisals, and the influence on performance as a “relatively young profession” for which theory

building is vital (Katou, 2009). Therefore, the aim of this study is to investigate how performance appraisals can improve individual performance and performance appraisal satisfaction. Moreover, it will be investigated how performance appraisals can be effective, and whether or not the performance appraisal will be more effective when the performance appraisal is structured by goals. Accordingly, the main research question that will guide this study is:

To what extent do structured performance appraisals influence individual

performance, and are there differences in performance appraisal satisfaction between employees who had a structured or unstructured performance appraisal?

(10)

1.5 S

TRUCTURE OF THE RESEARCH

The various studies of performance management, performance management systems, and the performance appraisal process altogether provide a theoretical basis for this study. As

Fletcher (2001) posits, support for the effectiveness of performance appraisals is theoretically grounded in performance management theory. Therefore, performance management theory will be discussed first in the literature review. Second, performance appraisals methods and their effectiveness will be discussed, followed by a review of the literature about performance appraisal satisfaction. The last part of the literature review will focus on the hypotheses, and the conceptual model of this research will be introduced. In the chapters thereafter, the research design will be explained, followed by the results. Finally, this research will be completed with a discussion and conclusion.

2. L

ITERATURE REVIEW

2.1 P

ERFORMANCE MANAGEMENT

Performance management is an interactive process between employees and supervisors about current and future performance (Fernandez, 2005). This means that performance management is an extensive and on-going process of monitoring and enhancing employee performance that includes multiple performance management tools including job design, leadership, training and development, reward systems, and performance appraisal. This on-going process of performance management helps an organization to achieve its goals and improve service delivery (Boxall, Purcell & Wright, 2007; Grobler, 2004). To realize performance

management, a performance management system has to be implemented. A performance management system, which provides an accurate framework for managing employee performance, helps to manage and direct performance by planning, monitoring, reviewing and controlling, moderating, and managing performance appraisals (Bititci, Garengo, Dörfler, & Nudurupati, 2012; Shammot, 2014; Williams, 1998). According to Lawler (2010), the existence of a performance management system is ‘the major differentiator between organizations that produce adequate results and those that excel’ (p. 2).

(11)

According to Erasmus, Swanepoel, Schenk, van der Westhuizen and Wessels (2005), two possible results can follow from a performance management system: satisfactory

performance of employees and unsatisfactory performance of employees. A performance management system is effective when both employees who perform satisfactorily and unsatisfactorily can be managed. Unsatisfactory performance is when, comparing to

predefined standards and goals, an employee underperforms (Erasmus et al., 2005). When an employee underperforms, the supervisor should decide to counteract, and has to support employee development by offer (re-) training, setting (new) clear work performance standards, job rotation, mentoring or coaching, designing a personal development plan or temporary assignments (Aguinis, 2009; Erasmus et al., 2005). When the unsatisfactory performance is not capable to improve by offering developmental support, other remedies such as replacement or termination should be considered (Erasmus et al., 2005). Conversely, Erasmus et al. (2005) state that satisfactory performance is when, comparing to predefined standards and goals, an employee performs well. Satisfactory performance can be managed by recognizing the individual performance, offering (non-financial) rewards, salary increase or a (financial) performance reward (Erasmus et al., 2005). Such rewards will encourage effectively performing employees to strive for superior individual performance (Erasmus et al., 2005).

Additionally, research carried out by Den Hartog et al. (2004) indicates that

performance management is generally an integrated process, which involves the day-to-day management and where managers work together with employees by setting expectations, measuring, and reviewing results, supplying feedback and training and development, and rewarding individual performance. The ultimate aim of this whole performance management process is to improve organizational success by means of individual performance (Den Hartog et al., 2004). But to attain superior individual performance, a performance

management system should communicate clear performance standards, in order to facilitate the comparison of the actual to the desired performance of employees (Williams, 1998). In addition, Brown and Benson (2005) note that actual performance has to be evaluated. An important management tool that helps to communicate clear performance standards and to evaluate actual and desired performance is the performance appraisal. Improving individual employee performance is based on measured performance outcomes that derive from the use of a performance appraisal system (Den Hartog et al., 2004), and the impact of the

(12)

2.2 P

ERFORMANCE APPRAISALS

According to Kirkpatrick (2006), performance appraisals are the ‘cornerstone of the performance management process’ and performance appraisals are ‘vital for the on-going development of employees’ (p.166). In most organizations, performance appraisal methods are used to make the performance management process known and standard within the entire organization (Leonard & Hilgert, 2007). Determining what creates superior individual performance and how different high performance practices can be measured has been extensively researched in the area of performance appraisals (Den Hartog et al., 2004; Williams, 1998).

Hence, a performance appraisal can be defined as a performance management tool that: “provides information that is relevant for many personnel decisions, including salary increases, recommendations for promotion, transfers, and training programs, as well as for employee development and performance feedback” (Cleveland, Murphy, & Williams, 1989, p. 130). In addition, Boxall, Purcell, and Wright (2007) state that performance appraisals can be seen as an important part of the performance management process with the objective of evaluating an employees’ individual performance and developing instructive plans.

According to Fletcher (2001), a performance appraisal plays a varied, but important role within the performance management model and therewith, to manage employee performance. Whereas performance management is an on-going process, performance appraisal is

conducted at discrete time intervals (Boxall, Purcell, & Wright, 2007).

A number of researchers have reported that performance appraisals should be conducted multiple times a year (Cascio, 2006; Fisher, 1995; Kirkpatrick, 2006;

Spangenberg, 1994). According to Kirkpatrick (2006), at least twice a year the employee and supervisor should discuss the employee’s individual performance during a performance appraisal. Fisher (1995) point out that performance appraisals should be held on quarterly basis. Cascio (2006) and Spangenberg (1994) even state that performance appraisals should take place as frequently as possible to correct poor performance and unwanted behaviour. Also Lawler (2010) notes that it is important to conduct performance appraisals frequently, because there are multiple examples where employees thought that they were performing well, however during their annual performance appraisals they found out that they were not performing well in the eyes of their supervisor. Lawler (2010) also points out that the implementation of a performance appraisal process is one of the key principles of effective performance appraisals. So, not only the frequency of the performance appraisals is

(13)

important, but also, or even more important, is effective implementation of the performance appraisal process.

2.3 P

ERFORMANCE

A

PPRAISAL

M

ETHOD

But before performance can be appraised, a measurement instrument has to be established and communicated to all employees (Erasmus et al., 2005). A number of researchers have sought to determine how performance appraisals can be communicated, implemented and used effectively (Lawler, 2010; Lock and Latham, 1990; Marquis & Huston, 2012). Marquis and Huston (2012) explain that for an effective performance appraisal system, employees must be well informed about their tasks and work standards. Communication is a crucial part of the performance appraisal process. During a performance appraisal, supervisors and employees share information about current and potential work-progress, possible solutions for problems and how supervisors can assist or coach employees (Boninelli & Meyer, 2004). When this communication does not take place or is misunderstood by the manager or

employee, conflict can occur. Erasmus et al. (2005) subsequently stress that performance appraisal effectiveness depends on sufficient communication.

Another important point of critique about performance appraisal systems concerns the subjectivity inherent in performance appraisal tools (Pazargadi, Afzali, Javadzadeh, & Alavi Majd, 2005). The manager has the best position to observe and judge the individual

performance of an employee, so it is the manager who should establish performance

standards, provide regular feedback on employee performance, and keep accurate record of the employees’ individual performance (Leonard & Hilgert, 2007). Worryingly, Pazargadi et al. (2005) researched performance appraisal methods and discovered that performance appraisals are usually not based on pre-determined criteria, but on the relation between employees and their supervisors.

However, Nikpeyma et al. (2014) conclude in their research that standard processes, implemented by a formal performance appraisal system, are required to determine

satisfactory individual performance. Nikpeyma et al. (2014) researched performance

appraisal methods across multiple hospital units in a large metropolitan teaching hospital and argued that individual performance should be ranked and compared by a formal and

systematic measurement system in order to gain greater individual employee effectiveness. Research of Erasmus et al. (2005) is in line with the findings of Nikpeyma et al. (2014), but argues that individual performance can also be determined by measuring the outcome of

(14)

work activities. According to Erasmus et al. (2005), one form of assessing individual performance is the on-going task of managers to supervise how well individual employees are doing their job. This on-going task of managers is a form of informal individual performance measurement and often results in useable information about employee performance. However, informal individual performance measurement usually happens in small companies where the manager directly works with all employees. For small companies, an overall impression of employee performance will be enough to manage individual

performance, but does not generate valid information to inform legitimate HR management decisions (Erasmus et al., 2005). To generate valid information about each individual

employee, accurate performance data obtained through standard processes, implemented by a formal performance appraisal system, is required (Erasmus et al., 2005).

So, in order to achieve the determined goals, employee performance needs to be measured with a methodical system (Erasmus et al., 2005; Marquis & Houston, 2012; Nikpeyma et al., 2014). This is in alignment with one of the most relevant findings about effective performance appraisals of Lawler (2010) and Lock and Latham (1990) who state that specific individual goals will facilitate effective performance appraisals. To produce effective performance appraisals, measurable goals have to be set and the business strategy and its goals should guide the performance appraisals (Lawler, 2010). This notion is also explicated in the goal-setting theory of Lock and Latham (1990), who argue that goals need to be specific and challenging, since goal setting theory suggest “that specific and challenging goals result in a higher performance than moderate or easily attainable goals, vague goals, and no goals at all” (p. 252-261). This means that specific goals not only lead to higher job satisfaction and higher employee motivation (Latham & Pinder, 2005), but also to higher individual performance (Lock & Latham, 1990).

2.4 E

FFECTIVE

P

ERFORMANCE

A

PPRAISALS

To make the performance appraisal system more methodical and structured, earlier research about structured and unstructured selection interview methods distinguished two potential approaches as to how performance may be measured: “the structured approach which is formal and research guided, and the unstructured approach, which is informal and guided by intuition” (Dipboye, 1994, p. 81). According to Leonard and Hilgert (2007), specific

judgements to support salary increases, promotion or demotion, relocation, and terminations can be provided with a structured performance appraisal system. Also, a structured

(15)

performance appraisal system can help managers to coach and counsel employees, since it helps clarifying in what area the employees’ problems are. Another important characteristic of structured systems is that the individual performance of one employee can be compared to the individual performance of another employee. Above all, with a structured performance appraisal system, employees will be well informed about how they are performing and suggestions about changes in performance, behaviour or skills can emerge (Leonard & Hilgert, 2007). This is supplementary to the research findings of Erasmus et al. (2005), Marquis and Huston (2012), and Nikpeyma et al. (2014), who clarify that a systematic and structured appraisal method is needed to generate valid information about individual employee performance.

Most of the information about individual employee performance will be measured quantitatively. According to van Dooren (2005), a key factor of an organizations performance

appraisal process is that it is based on quantitative performance information. An advantage of quantitative performance information is that it is tangible which is more precise compared to non-tangible performance information, since tangible performance information provides sufficient information about the actual performance of employees (Berman & Wang, 2000;

van Dooren, 2005). All this performance information should be well documented, since it can be used for future performance appraisals. For a substantial amount, past performance

appraisal information predicts future performance (van Dooren, 2005). This means that the

information of the performance appraisal in the past can be used to predict the individual performance in the future.

Based on the literature about structured performance appraisals that are guided by formal rules (Dipboye, 1994; Erasmus et al., 2005; Leonard & Hilgert, 2007; Marquis and Huston, 2012; Nikpeyma et al, 2014), combined with the literature of Lock and Latham (1990), who assert that goals will lead to higher individual performance, and the research findings of van Dooren (2005) and Berman and Wang (2000), who indicate that past performance appraisal information can be used to predict future performance, it can be expected that when the current performance appraisal is more methodical and structured, the future individual performance is higher than when the performance appraisal is unstructured. From this, the first hypothesis can be derived:

H1: The degree of structure of the performance appraisal at one measurement period is positively related to the individual performance at the subsequent measurement period.

(16)

2.5 A

PPRAISAL SATISFACTION

As stated by Fisher (1995), performance appraisal is a form of measuring performance, with the aim of identifying things that are going well and things that are not. But, to create an effective performance appraisal, employees and managers need to be trained about what is expected before, during, and after the performance appraisal (Lawler, 2010). In addition, a performance appraisal is only effective when it satisfies both the employee and the manager (Lawler, 2010).

Russell and Goode (1988) explain that performance appraisal satisfaction can be determined by assessing the satisfaction with the performance appraisal interview. For both managers and employees, important predictors of performance appraisal satisfaction are the employee’s skills and abilities, job-related results, and the match between the skills, abilities and results (Pooyan & Eberhardt, 1989). Yet, the research of Pooyan and Eberhardt (1989) also shows that in general managers are significantly more satisfied with performance appraisals and described performance appraisals in more favourable terms compared to employees. More recent research of Bargh and Ferguson (2000) specifies that if the employee receives feedback towards the stated goal during the performance appraisal, it will motivate the employee to increase effort in relation to goal pursuit, and will give the employee a better overall feeling about the performance appraisal.

In addition, according to Leonard and Hilgert (2007), an effective performance appraisal is focused on performance related to the stated goals. For both managers and employees, to accomplish the goals, it is important that constructive feedback, which shows progress towards the goals, is provided (Margolis & McCabe, 2006). This is in addition to research of Korsgaard and Roberson (1995), who note that employees would like to participate in performance improvement during the performance appraisal. Performance improvement will be maintained when managers also discuss the attitudes towards the performance appraisal with their employees directly during the meeting (Korsgaard & Roberson, 1995). In addition, Korsgaard and Roberson (1995) present that performance appraisals, which are both connected to goals and discuss the attitude towards the performance appraisal, are positively related to employees’ performance appraisal satisfaction.

So, to obtain an effective performance appraisal, the connection with the

organizational goal setting process is often considered to be crucial (Moynihan & Ingraham, 2003). However, an effective performance appraisal is not directly a satisfying performance

(17)

appraisal. This link between performance appraisal satisfaction and individual performance will be discussed later. Initially, findings from Pooyan and Eberhardt (1989) indicate that employees will have a higher performance appraisal satisfaction when the performance appraisal is a more unstructured and personal chat, while managers will have a higher performance appraisal satisfaction when the performance appraisal is more structured, formal, and based on goals (Pooyan & Eberhardt, 1989). However, Pooyan and Eberhardt (1989) also explain that when performance appraisals are based on predetermined goals, the manager is more motivated to give feedback.

After combining the literature of Pooyan and Eberhardt (1989) about higher

performance appraisal satisfaction of managers when the performance appraisal is based on predetermined goals, and findings from the literature of Bargh and Ferguson (2000) who clarified that receiving more feedback towards the goal during the performance appraisal will motivate employees to increase their individual performance, it can be expected that the degree of structure of the performance appraisal will be positively related to appraisal satisfaction. Performance appraisal satisfaction is a multidimensional construct that can be measured by multiple variables. According Korsgaard and Roberson (1995), an important indicator to assess performance appraisal satisfaction is improvement value. Improvement value is an indicator to measure the attitude towards the appraisal interview and guidance of the manager during the performance appraisal meeting (Korsgaard & Roberson, 1995). Additionally, according to Russell and Goode (1988), performance appraisal satisfaction can also be determined by assessing the employee satisfaction with the appraisal interview itself. Indicated that performance appraisal satisfaction can be split up in improvement value and satisfaction with the appraisal interview, the following hypotheses are conducted:

H2a: There is a positive relationship between the degree of structure of the performance appraisal and the degree of the employees’ improvement value. H2b: There is a positive relationship between the degree of structure of the performance appraisal and the degree of employee satisfaction with the appraisal interview.

(18)

2.6 I

NDIVIDUAL PERFORMANCE AS MEDIATOR

As stated above, when employees and managers are trained about what is expected before, during, and after the performance appraisal, and when the performance appraisal will be structured, it satisfies both the employee and manager. Therefore the performance appraisal will be obtained as effective (Lawler, 2010). But, according to literature of van Dooren (2005), a performance appraisal will be effective when it can be used to point out current performance, and to indicate how superior performance can be accounted in the near future.

Another previous claim that has been made is that employees will have a high performance appraisal satisfaction if the employee’s skills and abilities, and job-related results are aligned (Pooyan & Eberhardt, 1989). High results will be determined when the performance appraisal is effective. A performance appraisal is effective when it is methodical and structured (Berman & Wang, 2000). This means that the employees’ individual results are a mediating factor between the structure of the performance appraisal and performance appraisal satisfaction. From this, the third hypothesis can be conducted:

H3: Individual performance mediates the relationship between performance appraisal structure and performance appraisal satisfaction, so this relationship is stronger for employees who have a high individual performance.

2.7 C

ONTROL VARIABLES

Finally, several studies explored that gender (Grund & Sliwka, 2007; Lyness & Heilman, 2006), age (Brown & Heywood, 2005; Grund & Sliwka, 2007), and service years (Grund & Sliwka, 2007) can influence the performance appraisal process. In addition, based on research from Nerdrum and Erikson (2001) there can be expected that employees who perform well, are more motivated to perform even better in the future. Therefore, in order to examine the influence of other variables on individual performance and performance appraisal

satisfaction, the following control variables are included in hypotheses testing: previous individual performance, gender, age, and service years.

(19)

2.8 C

ONCEPTUAL MODEL

The literature and hypotheses above are reviewed and conducted in order to answer the main question of this research. The following research model shows the hypotheses and directions of the relationships within this research:

Figure 1: Conceptual model of this research

- Hypothesis 1 (H1) will be tested with data set1;

- Hypotheses 2 (H2a; H2b) will be tested with data set 2; - Hypothesis 3 (H3) will be tested with data set 2.

(20)

3. M

ETHODOLOGY

The main focus of this research is about a possible correlation between the degree of structure of the performance appraisal and the individual performance of employees. Within this study, quantitative data that has been collected from different databases within a big international company was explored in the first phase of data collection. Important quantitative data from company databases contained the demographic information of all selected employees. Also, there was historical data accessible about the individual performance appraisals and the increase of individual performance over time. Besides, to research the performance appraisal satisfaction of employees, an internal survey was sent out.

For this research, the data on performance appraisals and performance appraisal satisfaction data was combined. This chapter presents the combined data, which resulted in two datasets that were used to test the hypothesis. But first, the company where the datasets were obtained will be introduced. Then, the different data, the dependent variable, the independent variable and the control variables will be explained. All data analyses were completed using IBM SPSS Statistics Version 22.0.0 (Mac). As both missing data and outliers have the potential to impact the distribution of the results, all data was checked for missing values and outliers (Pallant, 2011; Tabachnick & Fidell, 2007).

3.1 S

AMPLE

:

I

NTRODUCING THE

C

OMPANY

In order to test the hypotheses and answer the research question, a dataset of a big

international company was used. The European headquarter of this company is located in the Netherlands. In Europe, the company has nearly 5000 employees who are located in six different regions around Europe and who are working in fourteen different departments. The company is an international, performance-focused organization with Key Performance Indicator (KPI) plans. KPI’s are the goals, translated in measurable objectives. The performance management process within this company is based on Management by

Objectives (MBO). In short, MBO is a system that tells employees what the company expects from them, including their goals and expected deliverables, in a clear and measurable way. This MBO process, which is focused on discipline, accountability and methodology,

ultimately results in a timely and objective performance appraisal (Karkoulian, 2002). Within this company, MBO is used to achieve superior individual performance. Initiatives are driven by cascaded goals and aligned throughout the whole organization.

(21)

The highlight of the MBO program is the use of performance appraisals. Within this company, performance appraisals are one-on-one quarterly meetings between the direct supervisor and the employee, to review and discuss the individual performance. A

performance appraisal consists of two parts: first self-evaluation takes place, followed by the creation of action plans through progress monitoring. The performance appraisal meeting must be set in advance of the period of appraisal (and formally tracked by performance) every quarter. The quarterly performance appraisals must be done using the official individual MBO plan, the online monitoring dashboard, and the standard performance appraisal form issued from the companies’ system.

Last appraisal period, 99.60% of the requested performance appraisals were completed in the system. This was the highest percentage since the implementation of the performance appraisals in 2013. But, in order to improve performance appraisal process and continuous improvement of the performance management process of this company, also performance appraisal quality had to be measured.

3.2 P

ROCEDURE

3.2.1.CODED PERFORMANCE APPRAISAL FORMS

For this research, historical company data were collected containing 434 individual

performance appraisal forms of 217 employees with individual performance assessed at two points in time (the second and third quarter of 2014). The performance appraisal forms consisted of personalized feedback on individual results. Moreover, it contained a self-made action plan to improve individual results in the next quarter. Each performance appraisal was coded on the degree of structure, the individual performance at the first measurement period, and the individual performance at the subsequent measurement period, performance appraisal satisfaction, age, gender, and service years. The complete explanation of these measures will be described in the next section. All coded variables were very clear and recognizable in the performance appraisal forms, and all the performance appraisals forms within the sample included all the information about the measured variables.

Most employees were male, 48.70% (SD = 0.48), with an average age of 43.41 years (SD = 8.60) and an average of 9.21 service years (SD = 7.87). Most employees were working in sales (27.20%), marketing (17.10%), finance (12.00%), and research and development (12.00%) departments. Overall, Europe was the biggest region (53.00%) and most employees were having functions in low management (65.90%) or middle management (25.30%).

(22)

3.2.2SURVEY

Besides using the coded performance appraisals, also a survey was sent out internally for this research. The survey is undertaken in the period November 25, to December 9, 2015 and was sent out to 588 employees who had their performance appraisal in the third quarter of 2015 (i.e. in the first three weeks of November). This amount of employees was bigger than the amount of employees where the performance appraisal forms were coded from, since the MBO program within this company has been cascaded to lower levels also in 2015. Therefore, the sample of this survey contained the employees wherefrom the performance appraisal forms of 2014 were coded, expanded with employees who were having

performance appraisals since 2015 only. The survey was open for two weeks. Even though there were responses every day, most responses were collected in the first days. Out of the targeted 588 respondents, 269 respondents answered the questionnaire. The effective response rate of this survey was 45.75%.

Of the 269 respondents who completed the survey, most employees were working in sales (19.70%), finance (14.50%) and marketing (13.40%) departments. Overall, most employees were working in Europe (34.90%), and were having a function in middle management (43.50%) or low management (35.30%).

3.2.3COMBINED DATA

Data set 1 will be useful for testing the first hypothesis empirically in the next chapter. Before the second and third hypotheses can be tested empirically, the survey data had to be coupled with the data from the coded performance appraisals and the demographic data from

company databases. Since the survey contained questions only about the performance appraisal satisfaction, region, department, and level, it was hard to couple all this data with data from the coded performance appraisals. Therefore, within both the survey data and the coded performance appraisals data, all duplicates were removed. First, within both the survey data set and the performance appraisal data set, department, level, and work region were combined. If there were two or more employees with exactly the same values on department, level, and work region, data of all these employees was removed. After removing all

duplicates (department * level * work region) within the survey data, 54 filled in surveys remained. Next, the 54 employees within the survey data set were combined with the de-duplicated performance appraisal data by using Structured Query Language (SQL).

(23)

and work region data from the surveys with the department, level and work region data from the coded performance appraisals. So, after combining all de-duplicated employees within both data sets (coded performance appraisals and survey responses), an exact and unique combination was created by matching the variables. Figure 2 explains how the variables from the two different data sets were de-duplicated and coupled. All the de-duplicated and coupled data created a table with coupled survey- and performance appraisal data from twenty

employees, which is data set 2.

Of these twenty employees, most employees, 75.00%, were male (SD = 0.44) with an average age of 42.35 years (SD = 7.99) and an average of 5.50 service years (SD = 4.85). The biggest departments, where most employees were working, were finance (25.00%) and general management (20.00%). Most employees were working the region Europe (45.00%), and in functions within middle management (35.00%).

(24)

3.3 M

EASURES

From the two data sets, data set 1 and data set 2, the dependent, independent and control variables could be researched. In order to test the hypotheses and answer the research question, it was important to understand how the dependent, independent and control variables were measured.

3.3.1STRUCTURE OF THE PERFORMANCE APPRAISAL

To measure the independent variable ‘structure of the performance appraisal’, all the

performance appraisal forms were coded on the number of goals listed under the topic ‘action plan per KPI’. The number of goals per action plan was measured at the end of the second quarter of 2014. To measure the number of goals, all different goals were counted by the coder. The overall amount of goals within the action plans was the value on structure of the performance appraisal.

According to Erasmus et al. (2005), Latham and Pinder (2005), Lawler (2010), Lock and Latham (1990), Marquis and Houston (2012), and Nikpeyma et al. (2014), specific, challenging, and structured goals will results in high individual performance. As stated before, KPI’s are goals, translated in measurable objectives. Therefore, every performance appraisal was divided into targeted and actual results per KPI. To maintain superior individual performance in the next performance appraisal period, every employee had to write an action plan per KPI. To structure the action plan, every part of the action plan had to be structured by a goal. By measuring the amount of goals, there was indicated how

structured the action plan for future individual performance was. This means that a higher amount of goals within the action plan was conceived as a higher degree of structure. By coding the total number of goals within the action plans per KPI, the degree of structure was measured. See Figure 3 in A for an example of a performance appraisal form, structured by amount of goals per action plan.

3.3.2INDIVIDUAL PERFORMANCE

One of the dependent variables is the employees’ individual performance. The measured score on individual performance is present in the results part of the performance appraisal as ‘result score per KPI’. The companies’ system measured the scores on individual

performance per KPI by a complex algorithm, based on e.g. the employees’ working level, final individual results, and a multiplier that is dependent of multiple variables. This way of

(25)

measuring individual performance gave the employee the feeling that they know exactly how they perform, however because of the use of a ‘secret’ complex algorithm, the employee itself cannot predict his or her overall individual performance.

The intent was to measure the actual individual performance by coding the individual results per KPI. The performance appraisals contained results on KPI 1, KPI 2, KPI 3, KPI 4 and KPI 5. In every performance appraisal, the KPI measured a different underlying main-goal, which had been cascaded trough the different management levels. The results per KPI were visualised in colours in the result part of every performance appraisal. See Figure 4 in Appendix B for an example of a performance appraisal form with targeted KPI’s and their actual individual performance. When the actual individual performance was above or equal to target, the actual result of the KPI was visualised as green (coded as 1). When the actual individual performance was at least 10% below target, the actual result of the KPI was

visualised as yellow (coded as 2). Finally, when the actual individual performance was below hurdle, the actual result of the KPI was visualised as red (coded as 3). This coding process created a list with individual performance per KPI. To measure the overall individual

performance per quarter, the scores on individual performance per KPI were computed into a scale. The reliability of this scale will be discussed later.

Individual performance was measured at two points in time: in the second quarter of 2014 (Q2) and in the third quarter of 2014 (Q3). When analysing the missing data within data set 1, it was highlighted that from both measurement periods, the variables that measure the results on KPI 4 and KPI 5, contained a lot of missing variables (missing values on KPI 4: 82.03%, missing values on KPI 5: 97.24%). When analysing the missing data within data set 2, the same problem with missing data as within data set 1 occurred. The analysis of data set 2 showed that at measurement period one (Q2), the variables that measured the results on KPI 4 and KPI 5, contained a lot of missing values (missing values on KPI 4: 45.00%, missing values on KPI 5: 80.00%). This amount of missing data was above the recommended 5 per cent threshold acceptable for missing data (Tabachnick & Fidell, 2007). In addition, since the little amount of variables within both data sets that measured the individual performance in Q2 on KPI 5 all had the same value, the variable individual performance in Q2 on KPI 5 had zero variance and no further statistics with this variable could be computed. Therefore results on KPI 5 were not be included in the further analysis of individual performance in Q2.

The above-mentioned problems did not occur when analysing the variables that measured the results on individual performance at measurement period two (Q3). Therefore,

(26)

to measure individual performance at measurement period two (Q3), the results off all KPI’s (KPI 1, KPI 2, KPI 3, KPI 4 and KPI 5) were used.

3.3.3PERFORMANCE APPRAISAL SATISFACTION

Performance appraisal satisfaction in this study was the second dependent variable and was measured by a survey. According to literature of Russell and Goode (1988), and Korsgaard and Roberson (1995), performance appraisal satisfaction was measured in two ways: improvement value and satisfaction with the appraisal interview. Improvement value and satisfaction with the appraisal interview are related. More improvement value usually implies more satisfaction with the appraisal interview.

The scope of the survey was to investigate the overall performance appraisal satisfaction of the employees. The survey requested the evaluation of five questions and statements. All responses could be given on a Likert measurement scale, ranging from poor (1) to excellent (5). Examples of these items were: ‘To what extent did your last performance appraisal increased your understanding of the job?’ (item 3) and ‘In general, how satisfied are you with the performance appraisal system?’ (item 5). The first survey question was

commissioned by the quality team of the company and therefore not further used within this research. The other survey-questions were based on academic literature and therefore within the scope of this research. Some items were modified to fit this study. See Appendix C for the complete survey that has been sent.

3.3.3.1IMPROVEMENT VALUE

From Russell and Goode (1988), a two-item scale to measure improvement value was used ( = 0.85). To measure the improvement value of employees, respondents were asked to give a response on the items ‘To what extent did your last performance appraisal increase your understanding of the job?’ (item 3) and ‘To what extent do you think the feedback interview helped you learn to do a better job?’ (item 4). Based on the findings of the survey, the overall score on improvement value of the employees was slightly above average (M = 3.33, SD = .843, min = 1.00, max = 5.00).

(27)

3.3.3.2 SATISFACTION WITH THE INTERVIEW

Two items of Korsgaard and Roberson (1995) were used to measure satisfaction with the appraisal interview ( = .85). To measure the employees’ satisfaction with the appraisal interview, respondents were asked to give response on the items ‘From you perspective, to what extent was the performance appraisal meeting a satisfying experience?’ (item 2) and ‘In general, how satisfied are you with the performance appraisal system?’ (item 5). Based on the findings of the survey, the overall score on satisfaction with the appraisal interview was slightly above average (M = 3.32, SD = .737, min = 1.50, max = 5.00).

3.3.4CONTROL VARIABLES

Besides the independent variable, other variables could also influence the dependent variable. Therefore, adding control variables to the analyses would make the results of the regression more powerful. Demographic variables that were not measured within the performance appraisal forms or survey were added by using historical company data. The datasets

contained employee information on age, gender, service years and the performance in Q2. All these variables were used as control variables in the analyses, since the variables contained information that could influence the effect of the structured performance appraisals on the individual performance and performance appraisal satisfaction.

3.4. R

ESEARCH DESIGN

In order to test the hypotheses, separate multiple regression analyses were conducted. First, a hierarchical multiple regression analysis was conducted to see if the degree of structure of the performance appraisal in Q2 predicted the individual performance in Q3.

Second, two hierarchical multiple regressions were conducted to see if the degree of structure of the performance appraisal in Q2 predicted the a) improvement value of the employees; b) satisfaction with the appraisal interview of the employees.

To test if the individual performance in Q3 mediated the relationship between the degree of structure of the performance appraisal in Q2 and the a) improvement value of the employees; b) satisfaction with the appraisal interview of the employees, two process analyses were conducted to estimate direct and indirect effects in the model and to find a mediating effect (Hayes, 2013). According to Montoya and Hayes (2015), “The primary goal of this statistical mediation analysis was to estimate the pathways of influence from X to Y,

(28)

one that operated through a mediator M—the indirect effect of X—and the other bypassing M— the direct effect of X” (p. 9). Two process analyses were conducted with structure of the performance appraisal as independent variable (X), and with a) improvement value; b) satisfaction with the appraisal interview as the dependent variables (Y), and individual performance in Q3 as mediator (M).

3.4.1INFERENTIAL ANALYSIS

In order to generalize findings from the regression analyses in a correct way, the regression data had to meet multiple assumptions since regression analyses that do not meet the

assumptions could result in difficult to interpret results. Multiple regressions require that the data meet the assumptions of normality, linearity and homoscedasticity (Pallant, 2011; Tabachnick & Fidell, 2007). To check whether the data of data set 1 and data set 2 met these assumptions, all the variables were analysed using boxplots, scatterplots, and histograms with normal distribution.

3.4.1.1NORMALITY, LINEARITY AND HOMOSCEDASTICITY

The histogram and the normal plot of standardized predicted values showed that both the regression data of data set 1 and data set 2 contained almost normally distributed errors. The points on the plot were not completely on the line, but close. The plot of data set 1 was a bit more normally distributed than the plot of data set 2,

In addition, the scatterplot of standardized predicted values showed that the regression data of data set 1 met the assumptions of homogeneity of variance (homoscedasticity) and linearity. This means that the variance of the performance in the third quarter was linear and would be stable at all levels of the predictor variables. The same applied for data set 2, although the variance of the performance in the third quarter of data set 2 was a bit less linear and stable than the variance of the performance in the third quarter of data set 1.

3.4.1.2INDEPENDENT VARIABLES AND ERRORS

To check if the residual terms were uncorrelated, the Durbin-Watson value was computed. The Durbin-Watson statistic is always between 0 and 4. A value of 0 indicates positive autocorrelation; a value of 2 means that there is no autocorrelation in the sample, and values towards 4 indicates that there is negative autocorrelation (Rutledge & Barros, 2002). When the Durbin-Watson value was computed with the data sets of this research, it showed that both the data of data set 1 (Durbin-Watson value = 1.98) and data set 2 met the assumption of

(29)

independent errors (Durbin-Watson value = 2.71). This means that there was almost no autocorrelation in the sample.

3.4.1.3MULTICOLLINEARITY

There occurs a multicollinearity problem when a Tolerance score of less than .20 or .10 and/or a VIF of 5 or 10 and above is calculated (O’brien, 2007). The test that investigated if the research data met the assumption of collinearity within these data sets showed that multicollinearity was not a concern for data set 1 (Structure, Tolerance = .98, VIF = 1.02, Performance in Q2, Tolerance = .96, VIF = 1.04; Gender, Tolerance = .93, VIF = 1.07, Age: Tolerance = .81, VIF = 1.23; Service Years, Tolerance = .83, VIF = 1.20). Also for data set 2, multicollinearity was not a concern (Structure, Tolerance = .32, VIF = 3.10, Performance in Q2, Tolerance = .56, VIF = 1.80; Gender, Tolerance = .31, VIF = 3.27, Age: Tolerance = .43, VIF = 2.35; Service Years, Tolerance = .49, VIF = 2.05). This means that, within the sample of this research, none of the predictor variables were highly correlated with one of the others.

3.4.1.4OUTLIERS

Finally, an analysis of standard residuals was conducted to check the regression data on outliers. Data set 1 was bigger than data set 2. Therefore, the value of the outliers in data set 1 was allowed to be a bit higher than the value of the outliers within data set 2. However, the analysis showed that the regression data of data set 1 contained no outliers (Std. Residual Min = -2.55, Std. Residual Max = 3.16). A second analysis showed that the regression data of data set 2 also contained no outliers (Std. Residual Min = -1.50, Std. Residual Max = 1.29).

Accordingly, these tests showed that the regression data of data set 1 and data set 2 met all the assumptions to conduct generalizable regression analyses. This means that regression analyses could be conducted and the results of the regression analyses could be generalized to the complete population.

(30)

4. R

ESULTS

In this chapter, the empirical results of this research will be presented. To test the hypotheses empirically, multiple regression analyses were conducted. The descriptive statistics and correlation of the tested variables will be analysed first. Second, the main results for testing the hypotheses will be presented.

4.1 I

NDIVIDUAL PERFORMANCE ANALYSES

Within this research, individual performance was measured at two measurement periods by the scores on individual performance per KPI. This construct of individual performance was a multidimensional performance construct (Murphy & Shiarella, 1997). This means that

multiple variables and constructs could measure individual performance.

4.1.1INDIVIDUAL PERFORMANCE IN Q2

Multidimensionality of the performance construct also appeared when Principle Components Analysis (PCA) with Varimax rotation was performed with data from data set 1. The plot of the Eigenvalues presented that two components were measuring individual performance in Q2. The initial eigenvalues showed that the first factor (containing the score of individual performance on KPI 1, KPI 2 and KPI 3) explained 42.34% of the variance (Eigenvalue = 1.69). The second factor (containing the score of individual performance on KPI 4) explained 30.94% of the variance (Eigenvalue = 1.24). KPI 1, KPI 2 and KPI 3 all loaded highly on one factor that measured individual performance. Therefore, within data set 1, individual

performance on KPI 1, KPI 2 and KPI 3 were used to create the variable ‘individual performance in Q2’ and to measure individual performance ( = .28).

When PCA with Varimax rotation was created with the variables of data set 2 that measured the results on KPI 1, KPI 2, KPI 3 and KPI 4 in Q2, it appeared that two

components were measuring individual performance. Also the plot of the Eigenvalues presented that two components were measuring individual performance. The initial

eigenvalues showed that the first factor (containing the score of individual performance on KPI 3 and KPI 4) explained 50.83% of the variance (Eigenvalue = 2.03). The second factor (containing the score of individual performance on KPI 1 and KPI 2) explained 27.37% of the variance (Eigenvalue = 1.10). KPI 3 and KPI 4 loaded highly on one factor that measured individual performance. Therefore, within data set 2, individual performance on KPI 3 and

(31)

KPI 4 were used to create the variable ‘individual performance in Q2’ and to measure individual performance.

4.1.2INDIVIDUAL PERFORMANCE IN Q3

To create the variable ‘individual performance in Q3’ another PCA with Varimax rotation was conducted with data set 1. The PCA to create the variable ‘individual performance in Q3’ basically replicated the results from the PCA that has been established to create the variable ‘individual performance in Q2’. The analysis to create the variable ‘individual performance in Q3’ showed that two components were measuring individual performance in Q3. The plot of the Eigenvalues presented that two components were measuring individual performance in Q3. The initial eigenvalues showed that the first factor (containing the score of individual performance on KPI 1, KPI 2 and KPI 3) explained 34.33% of the variance (Eigenvalue = 1.37). The second factor (containing the score of individual performance on KPI 4) explained 26.45% of the variance (Eigenvalue = 1.06). KPI 1, KPI 2 and KPI 3 all loaded highly on one factor that measured individual performance. Therefore, within data set 1, individual

performance on KPI 1, KPI 2 and KPI 3 were used to create the variable ‘individual performance in Q3’ and to measure individual performance ( = .60).

When another PCA with Varimax rotation was created with the variables within data set 2 that measured the results on KPI 1, KPI 2, KPI 3 and KPI 4 in Q3, it appeared that two components were measuring individual performance. The plot of the Eigenvalues presented that also within this data set two components were measuring individual performance in Q3. The initial eigenvalues showed that the first factor (containing the score of individual

performance on KPI 3, KPI 4 and KPI 5) explained 39.86% of the variance (Eigenvalue = 1.59). The second factor (containing the score of individual performance on KPI 1 and KPI 2) explained 27.00% of the variance (Eigenvalue = 1.08). KPI 3, KPI 4 and KPI 5 all loaded highly on one factor that measured individual performance. Therefore, within data set 2, individual performance on KPI 3, KPI 4 and KPI 5 were used to create the variable ‘individual performance in Q3’ and to measure individual performance.

4.1.3RELIABILITY

It emerged that created variables ‘individual performance in Q2’ and ‘individual performance in Q3’ within data set 1 were not reliable ( = .28;  = .60), which threatened the validity of this research. Murphy and Shiarella, (1997) discovered that the way in which independent

(32)

variables are combined could have a considerable impact on the validity of the measured variables. This construct of individual performance is a multidimensional performance construct (Murphy & Shiarella, 1997), which implies that within this study, a combination of measures from results of all KPI’s would capture variance that was not sufficiently captured by the measures of individual job performance studied alone (Murphy & Shiarella, 1997). Every KPI measured a different underlying goal, but overall, it measured the individual performance on all goals that were cascaded from one overall main goal. So, using a combination of measures to predict individual performance would create higher validities than the validity obtained when the measures on individual performance per KPI were used alone (Murphy & Shiarella, 1997). This combination of KPI’s would be used, because one KPI alone did not measure the complete performance construct. Because of the

multidimensional performance construct, the scores on KPI 1, KPI 2, and KPI 3 all differ for each of the individual employees. Therefore, within data set 1, the variable ‘individual performance in Q2’ and ‘individual performance in Q3’ had been created with al three variables to measure the individual performance of employees, even though the reliability scores were not sufficient.

Additionally, for data set 2, no reliability problems occur. Within data set 2, the created variables ‘individual performance in Q2’ and ‘individual performance in Q3’ were reliable ( = .77;  = .68).

4.2 D

ESCRIPTIVE STATISTICS

Data set 1 consisted of 217 coded performance appraisals. The data set showed that the average goal per action plan was 2.3 (M = 2.29, SD = 1.12, min = 0.00, max = 7.50). From analysing the descriptive statistics, it appeared that the average performance in Q3 is higher than the performance in Q2 (Performance in Q2: M = 1.45, SD = .47, min = 1.00, max = 2.67; Performance in Q3: M = 1.54, SD = .41, min = 1.00, max = 3.00).

Data set 2 consisted of combined information from the coded performance appraisals and the performance appraisal satisfaction of 20 employees. In data set 2, the average goal per action plan was 2.7 (M = 2.67, SD = 1.16, min = 1.00, max = 5.50). Within data set 2, the average performance in the Q3 appeared to be lower than in Q2 (Performance in Q2: M = 1.40, SD = .55, min = 1.00, max = 2.67; Performance in Q3: M = 1.03, SD = .60, min = 0.50, max = 2.50). The performance appraisal satisfaction of employees was divided in the

improvement value and the satisfaction of the appraisal interview. Most employees were satisfied with their performance appraisal. The improvement value was 3.23, which is more

(33)

than average on a scale from 1 till 5 (M = 3.23, SD = .90, min = 1.00, max = 4.50). The satisfaction of employees with the appraisal interview was 3.43, which is also more than average on a scale from 1 till 5 (M = 3.43, SD = .67, min = 2.00, max = 5.00).

In addition, results from the bivariate correlations analysis showed that within data set 1, the structure of the performance appraisal was significant positively correlated with

individual performance in Q3 (rs = .17), but not with individual performance in Q2 (rs = .12). Individual performance in Q3 was also significant positively correlated with individual performance in Q2 (rs = .66). Further, more significant findings showed that age and gender were negatively correlated (rs = -.22) and age and service years were positively correlated (rs = .38). See Table 1 in Appendix D for a complete overview of the statistics, correlations and reliabilities of data set 1.

Also for data set 2, a bivariate correlation matrix was conducted. Results from this bivariate correlations analysis showed that the structure of the performance appraisal was significant negatively correlated with age (rs = -.48) and service years (rs = -.59).

Additionally, service years was significant positively correlated with gender (rs = .50). Further, more significant results from the bivariate correlations analysis showed that the appraisal satisfaction of the employees was positively correlated with individual performance in Q2 (rs = .48) and improvement value of the employees rs = .73). See Table 2 in Appendix D for a complete overview of the statistics, correlations and reliabilities of data set 2.

4.3 E

MPIRICAL RESULTS

4.3.1HYPOTHESIS 1

In order to test the first hypothesis, data from data set 1 was used. A two stage hierarchical multiple regression has been conducted with individual performance in Q3 as the dependent variable. Individual performance in Q2, gender, age and service years were entered at the first stage of the regression to control for influence of these variables. The variable Structure of performance appraisal in Q2 was entered at the second stage. The regression statistics are reported in Table 3.

(34)

Table 3: Summary of hierarchical regression analysis for variables predicting individual performance in Q3 (N = 127 ) Variable β t SE R R2 R2 Change Step 1 .62 .38 .38 Individual performance in Q2 .59 8.21* .08 Gender -.07 -0.89 .08 Age .06 0.70 .01 Service years -.13 -1.69 .01 Step 2 .64 .41 .03 Individual performance in Q2 .58 8.08* .08 Gender -.07 -1.01 .08 Age .06 0.79 .01 Service years -.12 -1.59 .01

Structure of performance appraisal in Q2 .17 2.46* .03

*Note: p < .05

The hierarchical multiple regression revealed that at stage one, individual performance in Q2, gender, age and service years contributed significantly to the regression model, F (4,122) = 18.89, p < .05) and accounted for 38.20% of the variation in performance in Q3. Adding structure of the performance appraisal in Q2 to the regression model, explained a significant additional 2.90% of the variation in individual performance in Q3, F (5,121) = 16.94, p < .05. Together, the five independent variables accounted for 41.20% of the variance in individual performance in Q3. This means that this regression model fits the data significantly and can be used to predict 41.20% of the individual performance in Q3 by using individual

performance in Q2, gender, age, service years and structure of the performance appraisal in Q2 (R2 = .412). This results were the same when the analysis was conducted again without including the control variables.

The analysis showed that gender (β = -.07, t(121) = -1.01, p = .313), age (β = .06, t(121) = 0.79, p = .429) and service years (β = -.12, t(121) = -1.59 p = .114) would not

significantly predict individual performance in Q3. However individual performance in Q2 (β = .58, t(121) = 8.08, p < .000) and structure of the performance appraisal in Q2 (β = .17, t(121) = 2.46, p = .015) significantly predicted individual performance in Q3. This means that Hypothesis 1 was accepted: there was a positive relationship between the degree of structure of the performance appraisal at one measurement period and the individual performance at the subsequent measurement period. This means that a higher degree of structure resulted in higher individual performance.

A bulletproof performance appraisal : quantitative research about the relation between structured performance appraisals, the individual performance, and performance appraisal satisfaction of employees