Input factors for team development in a
shift-work organization
A study conducted at Philips Lighting Winschoten
Master’s Thesis in Human Resource Management Steven Marshall
Student nr. 1004611 June 2006
Rijksuniversiteit Groningen
Faculty of Management and Organization
Department of Human Resource Management and Organizational Behavior Supervisor: dr. B.S. Kuipers
Co-assessor: prof. dr. E. Molleman
Abstract
Executive Summary
The benefits of production philosophies such as Lean Production and Sociotechnical Systems theory are hard to compare because the debate between the supporters of each tradition mainly focuses on design instead of developmental processes or their outcomes. To fill this gap, Kuipers (2005) conducted a longitudinal study at Volvo Trucks Umeå in Sweden into the developmental processes concerned with teamwork, showing investing in team development can be a fruitful approach in enhancing employee motivation and team performance. Kuipers developed a model of team development – the 3R-model – in which he relates three dimensions of team responsiveness (job management, joint management and boundary management) to a set of eight dependent variables, both in the domain of business performance (BP) and quality of working life (QWL). All three responsiveness dimensions showed significant effects – cross-sectional and longitudinal – on differentiated sets of results, thus providing insights about which team responsiveness dimensions should be developed to enhance specific results.
Although Kuipers studied the effect of team development on output variables thoroughly, his investigation into the inputs for team development only focused on the location and proximity of responsibilities. Other factors may provide higher leverage solutions for organizations wanting to enhance their team responsiveness. For this reason, the present study elaborates on the input side of Kuipers 3R-model, hoping to provide useful recommendations for organizations on how they can improve team development in their production organization. This study took place at Philips Lighting Winschoten (PLW), manufacturer of high-grade glass components for the lighting industry. As the production department at PLW is organized in a rotating five-shift structure, conditions for team development may significantly differ from day-work organizations, because responsibility for the process and one’s workplace is shared with employees in other shifts. Based on the above, the research question of this study is: Which input factors have a positive effect on team development in a shift-work production
organization?
In this study, the effect of three input factors on the three team responsiveness dimensions is studied: workers’ perceptions of autonomy, feedback and leadership behaviour. Because no performance data were available at the individual level at PLW, the study into output side of the model is less comprehensive than the work of Kuipers, though the effect of the team responsiveness dimensions on two outcome variables - commitment to the organization and sickness absence – is researched. The results are based on questionnaire research conducted twice in all five-shift production teams at PLW. At the input side of the model, a distinction is made between the direct effect of the three input factors, and their effect on short-term team development.
The results show that especially feedback provides a high leverage solution for teams wanting to enhance team responsiveness, as it has a direct relationship with both joint- and boundary management, and has a short-term effect on the development on all three responsiveness dimensions. Autonomy and leadership behaviour also proved to be important input factors for improved team responsiveness, although in both cases it seems advisable that they are tailored to the specific characteristics of the team. At the output side of the model, boundary
Table of Contents
Abstract 2
Executive Summary 3
1. Introduction 5
Trends in work organization 5
The role of team development 7
Overview of content 8
2. The organization 9
Products and production process 9
Control structure of the production process 9
Teamwork at Philips Lighting Winschoten 10
3. Theoretical framework 11
Teams in organizations 11
The 3R-model of team development 11
The current study 13
Input factors for team responsiveness 14
Outcome variables for team responsiveness 17
4. Method 20
Participants and procedure 20
Statistical analysis 20
Measures 20
5. Results 22
Descriptive statistics 22
Testing the model: Input factors for team responsiveness 23
Outcome variables for team responsiveness 25
6. Discussion 27
Review of the results 27
Limitations 28
Suggestions for further research 29
7. Conclusions 31
1. Introduction
Many texts in business science start with the observation that organizations face huge challenges posed by their rapidly changing environment (e.g. Appelbaum & Batt, 1994), and this thesis will not form an exception to this. Markets are globalizing at a dizzying pace, as boundaries are being pulled down and countries full of potential customers are opening up, welcoming western companies and their products. Many of these companies are moving production facilities to these areas because of their cheap labour force. Western economies are shifting from production based to knowledge based value creation, and innovations are following upon each other faster than ever. The evolution of information and communication techniques over the last 25 years has sped up the globalization process, making the world a truly smaller place (Hinds & Kiesler, 2002). Not surprisingly, the impact for companies and the way they operate is huge.
The observations made above may sound as clichés, but they do pose interesting questions. What choices do companies make to compete in this new ever-changing environment? Or more specifically: How do companies organize their primary process in a way that is viable in the long run? Business scholars are continuously trying to answer questions like these, and have given rise to a vast body of literature, often containing contrasting visions on the best way to organize one’s workforce. Boxall & Purcell (2003) provide an excellent overview on the strategic issues concerning labour allocation.
Trends in work organization
Over the past hundred years, numerous philosophies on organizing have made their mark on practice. The first half of the 20th century the Scientific Management principles of F.W. Taylor dominated mass production, with a strict distinction between management and labor in the planning and execution of work (Boxall & Purcell, 2003). ‘In this form of work organization, jobs were designed to minimize discretionary behaviour among workers, with precise routines enforced on how and when work was to be performed and little or no opportunities provided (or expected) for employees to make suggestions for process improvements or technical innovations. The organizational logic for the design of work is typified as low discretion, with routine tasks undertaken in a low trust work environment’ (Fox, 1974, cited from Boxall & Purcell, 2003). Henry Ford pioneered the principles of division of labour in the mass production of automobiles. ‘If worker resistance to the factory regime became problematic then the answer was to pay higher wages in return for compliance and discipline, as Ford did, nicely stimulating demand for his own cars’ (Boxall & Purcell, 2003).
Starting with the famous Hawthorne-studies in the fifties of the last century, new ideas evolved about worker motivation and satisfaction as being beneficial for productivity and efficiency in what has become known as the Human Relations Movement. Nevertheless, ‘while increasing concern was expressed about working conditions and worker motivation, the fundamental principles of work organisation and job design were questioned by few’ (Boxall & Purcell, 2003). However, this steadily changed during the second half of the 20th
century. A growing rejection of the principles of Scientific Management and its alienating effects on workers lead to new ideas about work organization. One approach that received much attention is sociotechnical systems (STS) theory, which strives to enhance quality of working life as well as organizational effectiveness. Sociotechnical work teams were first introduced in the 1950s in British coalmines (Trist & Bamforth, 1951), and gained popularity throughout the decades that followed.
division of labour, STS suggests minimal work division, recommending small organization units or teams with internal coordination and autonomous or semi-autonomous control.’ Following Ashby’s Law (Ashby, 1958), the control capacity of work teams should be larger than the control need, which is determined by the amount of variety in the production process. STS theory suggests this control capacity can best be realized through the use of (semi-) autonomous and multifunctional teams, which are responsible for a whole task, including job regulation (Kuipers & de Witte, 2005). The self-regulating ability of teams is increased using the principle of ‘minimal critical specification’, which dictates that there should be as few restrictions as possible on the ways in which tasks are performed and just enough direction to ensure the task is performed properly. For this reason, STS require a different control structure and style of leadership than in traditional production organizations. De Sitter, Den Hertog & Dankbaar (1997) provide an extensive overview of the principles behind STS theory.
Although several large organizations experimented with sociotechnical systems in the decades after its inception, its popularity reached a peak in the early 1990s of the last decade. ‘Now, ten years later, the word ‘team’ is still left and the words ‘empowered’, ‘self-directed’ or ‘self-managing’ have lost their glitter and glamour’ (Kuipers, 2005). Increased economic pressure and competition have lead to a shift back toward centralization and standardization, and the popularisation of the Japanese production philosophy ‘Lean Production’ (LP), sometimes also referred to by its critics as ‘neo-Taylorism’.
The central idea behind Lean Production is that it tries to achieve a perfectly balanced production system, with the absence of buffers and the goal of zero defects (Niepce & Molleman, 1998). LP seeks to reduce process variability through process controls such as JIT and the development of exact standards for each process. Just as in STS theory, people are clearly viewed as a resource to be developed. The focus in LP is quite different though, as human resource management in LP will mainly be aimed at increasing productivity and efficiency: Workers are expected to deliver a certain amount of work within a predetermined time-span and the work relies heavily on coordination through procedures and direct supervision (Niepce & Molleman, 1996). A main difference with Taylorism is that workers are actually involved in improving the efficiency of the process, for example through participation in quality circles. Niepce & Molleman (1998) provide a thorough comparison between the principles of STS and LP.
Although STS and LP production philosophies are often regarded as being contrary to each other, it is questionable if companies really mind this debate. There are many combinations in business practice, which bring the traditions closer together (Adler & Docherty, 1998; De Leede & Looise, 1999; both cited from Kuipers, 2005). However, the use of elements from both principles in one production system is not without risks. ‘The introduction of principles of self-management will raise high expectations among the workforce. It is important that the results will actually come up to expectations and if this is not the case, introducing STS may be counter-effective’ (Niepce & Molleman, 1996). This may be the case if policies based on the different approaches are not properly aligned. Nevertheless, Niepce & Molleman (1998) demonstrate that it is very well possible to creatively use elements from both approaches in order to maximize the success of a production organization. Instead of maximizing self-management in teams, motivation can also be enhanced by introducing STS principles to a certain degree, without compromising on production efficiency.
development can be a fruitful approach in enhancing employee motivation and team performance, as will be made clear later on in this thesis.
The role of team development
This study starts from the assumption that – if properly embedded in the production policies – team development can be beneficial in many production organizations, both in achieving the company’s business objectives as in improving the quality of working life for employees. Kuipers (2005) reports on the effects of team development on a total of seven variables, in the domain of both business performance and quality of working life. A more detailed description of Kuipers’ 3R model of team development will be provided in the third chapter of this thesis.
Although Kuipers studied the effect of team development on output variables thoroughly, his investigation into the inputs for team development only focused on the location and proximity of responsibilities. Other factors may provide higher leverage solutions for organizations wanting to enhance their team responsiveness. For this reason, the present study will elaborate on the input side of Kuipers 3R model, hoping to provide useful recommendations for organizations on how they can improve team development in their production organization.
This study will take place at Philips Lighting Winschoten, manufacturer of high-grade glass components for the lighting industry. This organization is an interesting investigation subject because, after having employed a socio-technical team organization, several principles from lean production have been adopted over the last couple of years. This trend is by no means unique, as is described above. The value of this study may lay in showing that an organization striving for lean processes can achieve benefits by investing in team development and how it could do so.
Philips Lighting Winschoten has an interesting production organization for a second reason: its production department is organized in a rotating five-shift structure. Conditions for team development may significantly differ for employees in shift-work, as responsibility for the process and one’s workplace is shared with employees in other shifts. And although a large body of literature exists on predictors of success in production organizations (e.g. Appelbaum & Batt, 1994), research on shift-work production systems is scarce and mainly concentrates on the effects for employees (see Blau & Lunz, 1999). This is strange if one considers the large amount of research into – for instance – virtual teams (see Hinds & Kiesler, 2002, for an extensive overview). However, if one compares shift-work teams with virtual teams, one might see some interesting similarities. Virtual teams are characterized by the fact that its members – often located geographically apart – lack face-to-face contact and sometimes even work at completely different moments due to time-zone differences (Kerber & Buono, 2004). Shift-workers have short moments of contact with members of other teams at the beginning and end of their shift, but the fact remains that – for any particular process - a large part of the work is performed by others at different times, and most communication is asynchronous, which makes proper coordination of the process much more difficult.
The results found in the current study are based on one company and therefore might not be generalizable to all other shift-work production organizations – let alone those producing in daytime hours – but if this study can provide some practical insights into the role of team development and its determinants in a shift-work production organization, it will have succeeded in its aim. In sum, the main research question in this study is: Which input factors
Overview of content
2. The Organization
Products and production process
Philips Lighting Winschoten (PLW) is part of the Philips business group Lighting Components and is manufacturer of high-grade glass components. These components are utilized in head- and signal lights for automobiles, halogen lamps, solaria, road lighting, disinfection installations, and many other applications. In total, the plant produces around twenty different types of glass. The glass is processed into roughly 2500 different products, mainly tubes in a wide range of varieties. The plant has circa 225 permanent employees, about half of which are directly involved with the production process.
The glass types that are produced by PLW can be divided into two categories, which are both manufactured in a different part of the factory. In ‘hall A’ quartz glass is produced, this is a very pure sort of glass which has quartz sand as its main ingredient. In some cases small amounts of chemicals are added to the sand to enhance pellucidity. To melt the quartz sand into glass, higher temperatures are needed than for production of normal glass, 2200°C
as compared to 1500°C. The sand is molten in the department’s eight different ovens, from
each of which a long and continuous tube of glass is ‘pulled’. By varying the speed of the pulling process, diameter and wall thickness of the tube can be changed. After the continuous flow of glass tubing has sufficiently cooled down, it is cut off automatically into fixed-sized units, some of which are treated with acid subsequently. Part of the tubes is packed directly for shipment; the other tubes move to the finishing department for final processing.
In ‘hall B’ special glass is made; various glass types with special features. The glass produced in hall B is manufactured using a mixture of raw materials, for this reason it is also called ‘composed glass’. The raw materials are first made into pellets by a designated sub department (the ‘pelletizing plant’ or ‘hall Y’). These pellets are made because they are easily storable, melt faster and guarantee a constant dosage. Hall B has four different ovens in which the pellets are melted. The glass tubing that is pulled from these ovens is cut automatically, after which some tubes are further processed, but the largest part is packed for shipment directly.
In both production halls more types of glass are made than the number of ovens available, so the different glass types are made in production campaigns. However, within a certain campaign the production process is of a continuous nature. For this reason, production operations are organized in a five-shift structure. The different shifts rotate in a ten-day work cycle: All employees consecutively work two morning shifts, two afternoon shifts and two night shifts, after which they have four days off. The pelletizing plant and the finishing department however do not operate a continuous process and are therefore excluded from the five-shift structure.
Control structure of the production process
The production manager carries responsibility for the total production process, and reports directly to the plant manager. He presides over two unit managers; one is in charge of the production of quartz glass in hall A, the other of special glass in hall B/Y. Because the finishing department in hall A and the pelletizing plant in hall Y operate fairly autonomously from the pulling units, they both have their own production leader.
monitor the production process and do the bunkering of raw materials for the ovens. In hall A the shift-coordinators also help out in production when needed.
Where the shift-coordinators are primarily responsible for direct supervision over the production personnel, the responsibility for the process itself lies with the ‘operational group’ (OG). This group consists of the unit manager, a process engineer, and one or two quality engineers. Hall A has five ‘technical specialists’ who are part of the production teams, but in hall B the two technical specialists are also part of the operational group in stead of working in the five-shift production teams. The OG runs through process matters with shift-coordinator and technical specialist every morning on weekdays during the so-called ‘morning prayer’ meeting.
Because production continues 168 hours a week, but the OG and the supporting departments are only present during office hours, proper communication between shifts is essential. The operators working daytime shifts must also act for their colleagues working at night or during the weekend. Communication between shifts takes place through the ‘small and large shift transfer’. The small shift transfer happens before and after every shift, the large shift transfer takes place between the employees finishing their last night shift, and those who are starting a new cycle. Because the employees starting with their first morning shift are returning from four days off, a more extensive transfer is needed.
Teamwork at Philips Lighting Winschoten
The history of team development at PLW goes back to the middle of the 1990s, a period in which the market for quartz glass was growing rapidly. The large increase in volume in hall A had led to a rather chaotically organized production process, and the machine park had become outdated. For these reasons, a project was started in 1994 in which the production process was modernized and restructured in a more logical manner, combining the processes belonging together. Alongside this reorganization a pilot study was started introducing one production team based on sociotechnical principles in the finishing department. This team achieved good results, after which self-managing teams were first implemented in the rest of the finishing department, later in all production departments.
These self-managing teams were based on parallelization of processes, and consisted of the employees in all five shifts involved with one process. A mentor and a coach supervised the self-managing teams, and the old control structure with shift-coordinators was abolished. Communication between employees from different shifts took place at periodical team meetings and through the ‘glass walls’ based on the principles of Suzaki (1993) on which the teams made their performance visible. The development of the production teams was measured using an instrument based on the principles of Van Amelsfoort and Scholtes (1994) who propose that self-managing teams develop in four phases. Hut and Molleman (1998) report on a study they conducted at PLW testing this phased development theory, finding mixed results.
3. Theoretical Framework
Teams in organizations
Teams are everywhere in organizations. If asked, almost everyone with a job will report being part of a work group in some way or another. But what exactly makes a work group a team? It is difficult to provide a single answer to this question. People may do group work without being part of a formal work group or – vice versa – can be member of a work group but not working as a group (Kuipers, 2005). In management literature, the concept of teamwork can also have different meanings. Many modern organizational design and management approaches – like LP and STS theory – have team-based work as an essential ingredient. However, the role work teams fulfil in these approaches may differ substantially.
In this study, the definition for teams – being groups in organizations – by Hackman is used. He defines a team as ‘a collection of individuals who are interdependent in their tasks, who share responsibility for outcomes, who see themselves and who are seen by others as an intact social entity embedded in one or more larger social systems (for example business unit or the corporation), and who manage their relationships across organizational boundaries (Hackman, 1987; cited from Cohen & Baily, 1997). A further distinction can be made into work teams, parallel teams, project teams and management teams (Cohen & Baily, 1997). This study will focus exclusively on the first sort of teams.
The 3R-model of team development
As mentioned in the first chapter, Kuipers (2005) utters in his dissertation that the benefits of LP and STS principles are hard to compare because the debate between the supporters of each tradition mainly focuses on design instead of developmental processes or their outcomes. Besides, he suggests the empirical basis for claims about the effectiveness of different models of designing production is thin. To overcome these shortcomings, he developed a model regarding team development as the whole of inputs, processes and outputs of work teams: the 3R-model. The process of team development, which Kuipers calls Responsiveness (the second ‘R’) and is the core of the model, is defined as ‘the group process of self-management in terms of actions and behaviour in relation to given responsibilities (tasks, goals and challenges, desired outcomes)’. Important elements of this definition are that it focuses on action and behaviour in teams instead of design and conditions, and regards team responsiveness as a process of self-management, which changes over time depending on the (changes of) inputs to teams. (Kuipers, 2005).
such as those with other teams, customers and suppliers. On the other hand, they are challenging virtual boundaries by initiating new solutions to improve products and processes and to master and perform advanced tasks that previously belonged to management’ (Kuipers, 2005).
In his longitudinal study of team development at Volvo Trucks Umeå in Sweden, Kuipers (2005) related these three dimensions of team responsiveness to a set of eight dependent variables, both in the domain of business performance (BP) and quality of working life (QWL). These outcome variables Kuipers calls Results, the third ‘R’ in the 3R-model. All three responsiveness dimensions showed significant effects – cross-sectional and longitudinal – on differentiated sets of results, thus providing insights about which team responsiveness dimensions should be developed to enhance specific results. The relationships between the three responsiveness dimensions and the different outcome variables are depicted in figure 1.
Figure 1: The 3R model: Overview of the most important relationships as found by Kuipers (2005)
The current study
First of all, a major difference with Kuipers’ (2005) approach is that in this study all variables are measured and analysed at the individual level, whereas Kuipers analysed all data at the team level (although his scores on the team development dimensions, as well as the self-report data on QWL outcome variables were aggregated from workers’ individual scores). This is done due to the small number of teams studied (10). Aggregating the individual data to the team level will make statistical analysis impossible.
As stated in the first chapter, this study’s primary aim is to elaborate on the input side of the 3R model, providing recommendations on how (shift-work) production organizations can improve their team responsiveness. Though the number of regulation tasks located at the team level clearly relates positively to the team’s responsiveness, the fact that not all regulation tasks qualify for locating lower in the organization suggests that simply delegating as much responsibilities as possible is not the best solution. To provide a more nuanced insight into the question which input factors enhance team responsiveness, this study will use a different approach. Instead of focusing on the actual distribution of responsibilities in the production organization, the effect of the autonomy perceived by the workers will be studied. The assumption behind this approach is that delegation of responsibilities may increase team development, but only if they are perceived to enhance the members’ sense of autonomy.
The distribution of responsibilities – operationalized in this study through perceived autonomy – is surely one of the factors that may enhance team responsiveness. It is very likely though that more factors may play a role, and simply delegating regulation tasks may not be enough to build responsive teams. This thesis strives to offer a broader insight into the workings of input factors for team development. The choice of input factors in this study is based on the work of Campion, Medsker & Higgs (1993). They provide a conceptual model in which four categories of inputs – and one category with process variables – are of influence on work group effectiveness. In the current study three input variables, each from a different category in Campion et al. (1993), are assumed to be conditional for enhancing team responsiveness. As Campion et al. (1993) mention in their article, autonomy can be seen as the individual level equivalent of their variable ‘Self-Management’ in the Job Design category. From the Interdependence category, the influence of the feedback system in place will be investigated. The third input variable in this study – leadership behaviour – is not directly derived from the Context category in Campion et al. (1993), but is assumed to be of crucial influence on team responsiveness based on later studies (e.g. Stoker, 1999). The fourth input category in Campion et al. (1993) – Composition – is not represented in this study due to the impracticability to operationalize team composition at the individual level. In the next section, the role of the three input variables will be elaborated further and hypotheses about their effect on the three dimensions of responsiveness, as well as a hypothesis about the temporal nature – cross-sectional versus longitudinal – of these effects, will be developed.
Kuipers (2005) reports thoroughly on the relationships between the three dimensions of team responsiveness and the eight BP and QWL outcome variables. In this study, the influence of these variables will not be studied as elaborately, this is partly due to the fact that the shift-work structure in place at Philips Lighting Winschoten highly constrains the possibility of acquiring objective team performance data. Because the company currently measures key performance indicators only at the process level, and every process involves five rotating teams, it is impossible to distinguish the individual contributions of these teams or their members. For this reason, the relationships Kuipers found with BP variables are assumed, and the main focus will lie on predicting QWL outcomes. In the subsequent section, the relationships between responsiveness and the outcome variables will be hypothesized.
Figure 2: Research model
Input factors for team responsiveness Autonomy
Hackman & Oldham (1980) define autonomy as ‘the degree to which a job provides substantial freedom, independence and discretion to the individual in scheduling the work and in determining the procedures in carrying it out’. In their classic article, they state that when a job provides substantial autonomy to the persons performing it, work outcomes will be viewed by those individuals as depending substantially on their own efforts, initiatives and decisions (Hackman & Oldham, 1980). They argue that autonomy leads to experienced responsibility for outcomes of the work, which is one of the critical psychological states needed for achieving high internal work motivation in ones workforce, which in turn leads to effective performance and job satisfaction (Hackman & Oldham, 1980). Many other studies linked autonomy to different outcome variables related to quality of work life. For instance, Bakker, Demerouti, De Boer & Schaufeli (2003) showed that autonomy serves as a predictor of both organizational commitment and short-term absenteeism.
According to Campion et al. (1993), self-management is the group level analogy to autonomy at the individual job level. At the core of sociotechnical systems theory lays the assumption that teams should be given the amount of self-management needed in the light of the characteristics of the products and services made and the work processes needed to realize these outputs (Molleman, 2000). In other words: control capacity should match control need (De Sitter, den Hertog & Dankbaar, 1997). According to Dunphy & Bryant (1996) though, the enthusiasm for empowered operational teams is not matched by strong empirical evidence of their impact in generating significant performance benefits. However, they do suggest that as a result of self-management, team members develop a more informed and holistic understanding of the task and are increasingly committed to it. Greater goal clarity and increased commitment to such goals are major determinants of performance, provided the relevant knowledge needed to make decisions is available (Osburn et al., 1990, cited from Dunphy & Bryant, 1996). Self-management also impacts innovation by increasing the speed with which decisions can be made. Without the need to involve the hierarchy in many decisions, team members can respond to external exigencies in a more timely way (Dunphy & Bryant, 1996).
Hypothesis 1A: There is a positive relationship between perceived autonomy and workers
perception of the team’s job management.
Hypothesis 1B: There is a positive relationship between perceived autonomy and workers
perception of the team’s joint management.
Hypothesis 1C: There is a positive relationship between perceived autonomy and workers
perception of the team’s boundary management. Feedback
Providing feedback to employees is believed to be essential for maintaining employee motivation and satisfaction. According to Hackman & Oldham (1980), feedback can be defined as ‘the degree to which carrying out the work activities required by the job provides the individual with direct and clear information about the effectiveness of his or her performance’. This definition focuses on feedback obtained directly from the job, but the authors note that feedback from others (such as co-workers or supervisors) on how well a job is done can be equally informative for an employee about the results of his or her work. Assuming that the source of the feedback (either the job itself or others) is not of influence on the performance effect feedback has, this broader definition of feedback is used in the present study.
Feedback plays an important role in goal-setting theory as formulated by Locke & Latham (for an overview see Locke & Latham, 2002). They state that one’s performance can be enhanced by providing goals that are both difficult and specific. Goals affect performance in four ways. First, goals serve a directive function; they direct attention and effort toward goal-relevant activities and away from goal irrelevant activities. Second, goals have an energizing function. High goals lead to greater effort than low goals. Third, goals affect persistence. Someone pursuing a goal is more likely to work through setbacks or work harder. Fourth, goals affect action indirectly by leading to the arousal, discovery and/or use of task-relevant knowledge and strategies. For goals to be effective however, people need feedback that reveals progress in relation to their goals. ‘If they do not know how they are doing, it is difficult or impossible for them to adjust the level or direction of their effort or to adjust their performance strategies to match what the goal requires’ (Locke & Latham, 2002). Hollenbeck et al. (1998) advise to complement outcome-type feedback (expressed in terms of bottom-line results) by process feedback (expressed in terms of intervening steps that lead to bottom line results. Feedback promotes both effort and learning and the effectiveness of feedback is enhanced when outcome feedback is coupled with process feedback (Hollenbeck et al. 1998).
Hypothesis 2A: There is a positive relationship between perceived feedback and workers
perception of the team’s job management.
Hypothesis 2B: There is a positive relationship between perceived feedback and workers
perception of the team’s joint management.
Hypothesis 2C: There is a positive relationship between perceived feedback and workers
perception of the team’s boundary management. Leadership behaviour
It has long been argued that leaders play a key role in individual and organizational performance. In the 1940s and 50s, two factors in effective leadership behaviour were uncovered by a group of researchers at Ohio State University, which they labeled ‘consideration’ and ‘initiating structure’ (Stogdill, 1950; see Judge, Piccolo & Ilies, 2004). Consideration can be defined as ‘the degree to which a leader shows concern and respect for followers, looks after their welfare, and expresses appreciation and support’ (Bass, 1990; cited from Judge et al., 2004). Initiating structure is the degree to which a leader defines and organizes his role and the role of followers, is oriented towards goal attainment, and establishes well defined patterns and channels of communication (Fleishman, 1973; cited from Judge et al., 2004).
Although the two behavioural indicators seem to have fallen out of favour in leadership research, recent meta-analyses show that both consideration and initiating structure have strong relations with outcome variables (Kuipers, Stoker & van der Vegt, 2007). Judge, Piccolo & Ilies (2004) showed that consideration was more strongly related to follower satisfaction and motivation, whereas initiating structure was more related to leader job performance and group-organization performance. Burke et al. (2006) showed the relationship between initiating structure and consideration and team performance outcomes.
Kuipers et al. (2007) point to the paradoxical fact that self-managing work teams, which are supposed to take as much responsibility as possible for their work, may fail to perform or be successful in the long run without adequate supervision. The notion that self-managing teams should be lead in a more relationship-oriented and considerate way instead of directive, task-oriented leadership, does not get strong support by empirical evidence. Stoker (1999) states that leadership behaviour should be contingent with the characteristics and developmental state of a team, and different leadership behaviours are needed in different situations. Kuipers et al. (2007) show that consideration has a strong effect on the social integration in self-managing teams, and team performance is enhanced by initiating structure, while consideration can be detrimental.
In this study, the idea that consideration and initiating structure are related to different dimensions of team development is tested. In line with Judge et al. (2004) and Kuipers et al. (2007), consideration is assumed to have a positive effect on joint management, as this responsiveness dimension – in which cooperation is essential – is most likely to be enhanced by a team’s social integration. On the other hand, job management and boundary management are assumed to be enhanced by initiating structure, as job management is more task-oriented in nature and boundary management involves more complex activities which may be performed best in a clear structure. This leads to the following hypotheses:
Hypothesis 3A: There is a positive relationship between the perception of the team-leader’s
Hypothesis 3B: There is a positive relationship between the perception of the team-leader’s
consideration and workers perception of the team’s joint management.
Hypothesis 3C: There is a positive relationship between the perception of the team-leader’s
initiating structure and workers perception of the team’s boundary management. Cross-sectional versus longitudinal effects for the input factors
Above hypotheses about the effects of the input factors on team responsiveness are all formulated cross-sectionally. It is very likely though that these factors also have an effect on the development of team responsiveness over time. Because this study is based on two measurement moments – although with only two months in between – it is possible to investigate the (short-term) longitudinal effect of the input factors on team responsiveness.
Feedback is believed to have a relatively direct, short-cycle effect on the development in team responsiveness. This will especially be the case when proximal rather than distant outcome goals are used (Locke & Latham, 2002) and outcome feedback is complemented by process feedback (Hollenbeck et al., 1998). A feedback loop with these characteristics is assumed not only to facilitate performance (Latham & Seijts, 1999; cited from Locke & Latham, 2002) but will also promote learning and development (Hollenbeck et al. 1998). On the other hand, autonomy and leadership behaviour are assumed to have a much more stable effect on team responsiveness. One’s sense of autonomy is assumed to be relatively stable because the distribution of responsibilities is unlikely to change very often and in some cases can only take place after the transfer of the relevant knowledge needed to fulfill these responsibilities. Leadership behaviour is assumed to be dispositional to one’s manager and – although changeable through training – relatively stable.
Because of the short interval between the two measures (two months), it is expected that in this study feedback will have much stronger longitudinal effect on team responsiveness than autonomy or leadership behaviour, this due to its direct and short-cylce effect on team development. This leads to the following hypothesis:
Hypothesis 4A: In the longitudinal analysis, the positive effect of perceived feedback on
perception of the team’s job management will be stronger than the effect of perceived autonomy and the perception of the team-leader’s initiating structure.
Hypothesis 4B: In the longitudinal analysis, the positive effect of perceived feedback on
perception of the team’s joint management will be stronger than the effect of perceived autonomy and the perception of the team-leader’s consideration.
Hypothesis 4C: In the longitudinal analysis, the positive effect of perceived feedback on
perception of the team’s boundary management will be stronger than the effect of perceived autonomy and the perception of the team-leader’s initiating structure.
Outcome variables for team responsiveness Commitment
or felt obligation. Finally, continuance commitment refers to a general awareness of the cost of leaving the organization or to the perceived number of employment alternatives and degree of sacrifice (Meyer & Allen, 1991). This study will focus exclusively on affective commitment.
Boxall & Purcell (2003) state that teamworking, if properly designed, allows for four necessary types of task integration to occur. One of these is the integration of workers ‘into the cultural or organizational fabric of the firm’ in order to gain higher levels of commitment to the work process, the end user and the organization as a whole. When employees wish to remain part of the firm and see membership as important to them and their sense of worth, a basis is created for the development of ‘social capital’ which can be fundamental to the achievement of organizational advantage.
STS theory scholars generally assume that self-management has a large positive effect on workers’ commitment, but as mentioned earlier in this chapter, their claim has not been backed up by much empirical evidence. Kuipers (2005) did find significant relations between responsiveness and involvement (his term for the concept of commitment). In his cross-sectional analysis at the team level, job management and boundary management relate positively to involvement, and the latter also in the longitudinal model. The analysis for commitment in this study is only performed cross-sectionally and at the individual level. To explore all possible relationships between responsiveness and commitment at the individual level, all three dimensions are hypothesized to have a positive effect on commitment. This leads to the following hypotheses:
Hypothesis 5A: There is a positive relationship between workers perception of the team’s job
management and their commitment to the organization.
Hypothesis 5B: There is a positive relationship between workers perception of the team’s job
management and their commitment to the organization.
Hypothesis 5C: There is a positive relationship between workers perception of the team’s
boundary management and their commitment to the organization. Absence duration and –frequency
In this study, two different absence measures are studied: absence duration and frequency (Hensing, Alexanderson, Alleback & Bjurulf, 1998). According to Bakker et al. (2003), ‘absence duration is the total length of time over a specified period of time, regardless of the number of absence spells’. Absence frequency ‘is the number of spells or times an individual has been absent during a particular period, regardless of the length of each of those spells’ (Bakker et al., 2003). The correlation between absence duration and frequency ranges between a low -.05 and a moderately high .60 (Farrell & Stamm, 1988); Absence duration is generally considered to be an indicator of ‘involuntary absenteeism’ that results from the inability rather than the unwillingness to come to work, for example as a result of ill health. On the other hand, absence frequency is considered to be indicative of ‘voluntary absence’ and a function of employees motivation. (Bakker et al., 2003).
increased motivation and attachment to work and organization, and showed a negative relationship with absence frequency.
Assuming that there is a relationship between job demands and –resources on the one hand, and team responsiveness on the other hand, some parallels can be drawn. In this study, job management is expected to be the prime predictor of absence duration due to the determining effect it can have on job demands. Joint management and boundary management are more likely to be associated with job resources (e.g. social support or decision making possibilities) and thus are expected to be of influence on absence frequency.
Hypothesis 6A: There is a negative relationship between workers perception of the team’s job
management and absence duration.
Hypothesis 6B: There is a negative relationship between workers perception of the team’s
joint management and absence frequency.
Hypothesis 6C: There is a negative relationship between workers perception of the team’s
4. Method
Participants and procedure
A questionnaire was distributed twice among all employees of Philips Lighting Winschoten working in five-shift teams, in November 2006 (T1) and again in February 2007 (T2). Participation was voluntary, and the confidentiality of the answers was emphasized. By means of a unique code, the data of both questionnaires could be linked with each other and with absenteeism data retrieved from the company’s registration system. At T1, a total of 98 employees filled out and returned the questionnaire (response rate = 81,6%). At T2, the questionnaire was returned by 93 respondents (response rate = 79,5%). In total, 73 employees participated in both the first and second measure.
The questionnaires filled in by the shift-coordinators and technical specialists were ultimately removed to make sure their different role in the team would not be of influence in hypothesis testing. After this removal, the sample was (n = 86) at T1 and (n = 79) at T2; 62 of these respondents returned both questionnaires.
The sample at T2 – being the main sample used in hypothesis testing – included 74 males (93.6%) and 5 females (6.4%). For privacy reasons, age was measured using a five-category scale. Of the respondents, 3.8% were aged 15-24, 22.8% were aged 25-34, 51.9% were aged 35-44, 17.7% were aged 45-54 and 2.5% were aged 55-64. The average organizational tenure was 9.8 years (SD = 8.1), the average team tenure was 5.3 years (SD = 5.0). In the five-shift structure, all employees are employed on a full-time basis, working six 8-hour shifts every ten days. The demographics at T1 are not significantly different from those at T2.
Statistical analysis
In testing the hypotheses, multiple regression analysis was used. The regression analyses testing the relation between input factors and the three responsiveness dimensions were each performed in three steps, this was done in order to provide insight into the cross-sectional effects on the dependent variables, as well as the effects on the development of these variables over time. In all analyses, the dependent variables measured at T2 were predicted. In model 0, the control variables gender and team tenure were entered (age and tenure were ultimately left out because of their high correlation with team tenure). The independent variables as measured at T2 were entered in model 1, thus providing a cross-sectional analysis for the second measure. In model 2, the dependent variable at T1 is added as an extra control variable. This last step provides insight into the effect the predictors have on the development of the dependent variable in the months between the two measurements.
The relationship between the responsiveness dimensions and the outcome variables commitment and absence duration and –frequency were only tested cross-sectionally. This was done because absence data were used for the year leading up to T2, and absence data for the year up to T1 would have had too much overlap to control for. Commitment data at T1 were available, but controlling for this data would have decreased the power in hypothesis testing too much, as the n of the combined data is much lower.
Measures
Input factors
Autonomy and feedback were both measured using a dutch translation of items from the Job Diagnostic Survey by Hackman & Oldham (1980). Three items were used measuring autonomy; ‘I can completely determine how I perform my job’, ‘In this job it is not possible to choose how I perform my work’ and ‘The job gives me the possibility to be free in the way I do my work’. Cronbach’s α was low at .50, but rose to .65 after removal of the second item.
Feedback was assessed using six items (α = .63), including ‘Managers and colleagues often
inform me about how well I perform my work’ and ‘The work itself scarcely provides me with clues if I’m performing well or not’.
The leadership behaviours consideration and initiating structure were assessed using the 10-item scale developed by Stoker (1999). This scale originally distinguishes five different leadership behaviours, but factor analysis performed by Kuipers, Stoker & van der Vegt (2007) reveiled a two factor structure for this scale, with five items indicating consideration and five items for initiating structure. This two factor structure was replicated for the current data set, though with six items indicating consideration (items 1, 2, 6, 7, 8 and 9; Cronbach’s α = .92) and four items for initiating structure (items 3, 4, 5 and 10; Cronbach’s α = .84). Exemplary items are ‘My manager is friendly and open’ (consideration) and ‘My
manager determines in detail what has to be done, and how it has to be done’ (initiating structure).
Team responsiveness
Team responsiveness was measured using the 46-item scale developed by Kuipers (2005). Job management was measured with 20 items, such as ‘All team members act as substitutes for each other if needed’. Crobach’s α for job management was .89 at T1 and .92
at T2. The 13 items used to measure joint management included ‘team members motivate each other to perform their jobs properly’. For this variable, Crobach’s α was .87 at T1 and
.89 at T2. Finally, boundary management was assessed with 13 items, for which Crobach’s α
was .85 at T1 and .86 at T2. An example item for boundary management is ‘Contacts with internal customers and suppliers (other teams or departments) regularly lead to improvement initiatives for the team’.
Outcome variables
Commitment was measured using the Dutch version of Allen & Meyer’s (1990) affective commitment scale, translated and validated by De Gilder, van den Heuvel & Ellemers (1997). This scale consists of 8 items, including ‘I feel emotionally attached to the organization’. Cronbach’s α was .88.
5. Results
Descriptive statistics
Two primary analyses were conducted for the data collected. First of all, the data gathered at both times of measurement were compared to provide insight into the stability of the scores on the variables over time. Secondly, the variables that were finally included in the analyses – i.e. all variables at T2, and the responsiveness variables at T1 – were correlated. Please note that the n in both analyses differs. This is due to the fact that in the first analysis only the respondents who participated at both measures could be included. The variables in the correlation table were compared pairwise, which for most correlations between T2 variables leads to higher n’s. This difference is also present in the analyses for testing the model. The longitudinal input – process models have the lower combined n, and the (purely cross-sectional) analysis for commitment has the higher n for variables at T2. The analysis for absence is also cross-sectional, but has a much lower n because PLW only registers absence data for employees who have a contract with PLW and not for the circa 30 respondents who are either on a long-term hire from a local temp agency, or are on a year contract taking part in Philips’ working and learning project.
Table 1: Averages on the variables at T1 and T2 and t-values for the changes over time.
Variable Mean at T1 Mean at T2 t-value
Autonomy 3.51 3.51 .000 Feedback 3.21 3.15 .673 Consideration 3.91 3.73 1.431 Initiating structure 3.47 3.25 2.191* Job management 3.95 3.84 1.839 Joint managemenent 3.33 3.30 .557 Boundary management 2.98 3.09 -1.423 Commitment 3.21 3.30 -1.309 *p<.05, **p<.01, ***p<.001, combined n at T1 and T2 = 62.
Table 1 provides an overview of the t-tests which were performed to compare the means of the variables at T1 and T2. Only the average perception of leaders’ initiating structure fell significantly. This is possibly due to the fact that in one part of the factory some team leaders swapped places between the two measurements, so part of the respondents had a new manager they were less acquainted with at the second measure.
Table 2: Correlations for all variables used in model testing.
Table 2 depicts the correlations between all variables used in model testing. The many significant correlations between the input factors, responsiveness dimensions and commitment suggest there are plenty of relationships between these variables. However, absence duration and –frequency only correlate significantly with each other. Almost all correlations between the independent variables are also significant, but only the correlation between the leadership behaviors ‘consideration’ and ‘initiating structure’ is high enough to suggest collinearity. Joint management also shows high correlations with job- and boundary management. It must be noted that in both cases factor analysis did suggest the current differentiation between the variables. Further attention to this matter is paid in the discussion.
Testing the model: Input factors for team responsiveness Job management
In this first analysis, the relationship between input factors and job management was tested. The input factors together account for 40% of the variance of job management in the cross-sectional model. The results as shown in table 3 indicate that both perceived autonomy and initiating structure have rather strong positive effects on job management, thus supporting hypothesis 1A and 3A. Hypothesis 2A is not supported by the data, as perceived feedback does not show a significant cross-sectional relationship with job management. In the longitudinal analysis however, the three input factors all show a roughly equally strong relationship with job management. Although this does not provide univocal support for hypothesis 4A, the fact that in longitudinal analysis the effect of feedback becomes stronger, and the effects of autonomy and initiating structure become weaker does indicate in its direction.
Table 3: Regression analysis results for job management (n=61)
Standardized coefficient Beta
Joint management
The cross-sectional analysis for joint management explains an impressive 44% of variance. Table 4 shows that all three input factors have a significant positive relationship with joint management, thus providing support for hypotheses 1B, 2B and 3B. In the longitudinal model, the effect of perceived feedback is the strongest, and despite of controlling for earlier level joint management its Beta hardly decreases. This provides strong support for hypothesis 4B. For the other two input factors, the effect is much smaller than in cross-sectional analysis, and the effect of autonomy is even no longer significant.
Table 4: Regression analysis results for joint management (n=60)
Standardized coefficient Beta
Model 0 Model 1 Model 2 Model 0 Gender -.086 -.025 -.060 Team tenure -.138 -.036 .052 Model 1 Autonomy .255* .059 Feedback .288* .265** Consideration .326** .193* Model 2 Joint management at T1 .542*** R2 .030 .472 .679 R2 change .030 .441*** .208*** *p<.05, **p<.01, ***p<.001 Boundary management
Table 5: Regression analysis results for boundary management (n=60)
Standardized coefficient Beta
Model 0 Model 1 Model 2 Model 0 Gender -.133 -.103 -.059 Team tenure -.100 .024 .199 Model 1 Autonomy .210 .106 Feedback .335* .259* Initiating structure .257* .169 Model 2 Boundary management at T1 .453*** R2 .032 .363 .516 R2 change .032 .331*** .153*** *p<.05, **p<.01, ***p<.001
Outcome variables for team responsiveness Commitment
Workers perception of a team’s job management (hypothesis 5A), joint management (hypothesis 5B) and boundary management (hypothesis 5C) were expected to show significant positive relationships with worker commitment to the organization. As is visible in table 6, these first two hypotheses should be rejected, but the latter is supported. In the analysis for commitment only boundary management has a significant effect on commitment. Nevertheless, a neat 32% of variance is explained by the input factors.
Table 6: Regression analysis results for commitment (n=76)
Standardized coefficient Beta
Absence duration and –frequency
Two analyses were conducted for absenteeism, one for duration and one for frequency. Table 6 indicates that hypothesis 6A – in which job management is expected to have a significant negative effect on absence duration – is confirmed, although the effect is not very strong. Table 7 shows that none of the predictors have a significant relationship with absence frequency, its percentage of explained variance hardly changes after adding the three responsiveness dimensions. Therefore, the expected effects of joint management (Hypothesis 6B) and boundary management (Hypothesis 6C) are not confirmed in this study.
Table 7: Regression analysis results for absence duration (n=47)
Standardized coefficient Beta
Model 0 Model 1 Model 0 Gender .094 .080 Team tenure -.057 -.076 Model 1 Job management -.492* Joint management .503 Boundary management -.156 R2 0.009 0.122 R2 change 0.009 0.113 *p<.05, **p<.01, ***p<.001
Table 8: Regression analysis results for absence frequency (n=47)
Standardized coefficient Beta
6. Discussion
This study started off with the following research question: Which input factors have a
positive effect on team development in a shift-work production organization? To answer this
question, a number of input factors were derived from the conceptual framework of Campion et al. (1993), and entered at the input side of Kuipers (2005) model of team development. In the next section, the results of this study will be reviewed in the light of the overall model. Besides, some special attention will be paid to the fact that this study took place in a shift-work production organization. As this study does not make any comparison with ‘normal’ production work, it is impossible to draw conclusions about differences for shift-work. Nevertheless, when theoretical reasons exist to expect that certain findings could be specifically relevant for shift-work organizations, these will be mentioned.
Review of the results
In the current study, autonomy was considered to be an essential ingredient for team responsiveness, and thus was hypothesized to have a positive relationship with all three responsiveness dimensions. The analyses indeed show this relationship for both job management and joint management. However – at least at the individual level – autonomy does not have a significant effect on boundary management. Simply giving teams more autonomy does not seem to improve a team’s capability for relationship management or innovation in this study. This may be related to the fact that of all three responsiveness dimensions boundary management is most ‘outward focused’ and involves having contacts with other parties in or outside the plant. Boundary management could prove to be too complex to enhance by maximizing self-management – especially in a shift-work context, due to the irregular working hours. It might be more fruitful to – for instance - provide teams with clear structures defining one’s responsibilities.
Having autonomous teams can even have detrimental effects for boundary management. For instance, Van der Zwaan & Molleman (1998) point to the possible occurrence of ‘negative group dynamics’ in self-managing teams. ‘Strong intra-group relations and a subsequent internal orientation often result in the development of strong group norms, which can foster group conservatism and group introversion. This may reduce or even block the innovative capacity of the task group. A second potential drawback arises when the commitment of group members to their group grows so strong that the group (gradually) loses its capacity and willingness to communicate with the outside world and to critically asses its own functioning’ (Van der Zwaan & Molleman, 1998). Organizations wanting to give their teams more autonomy should be aware of the possibility of these negative group dynamics, and should try to prevent their occurrence through proper management and definition of responsibilities.
Unlike with the cross-sectional analyses, the effects found in the longitudinal analyses are unlikely to be reversed, this because they control for the scores on the responsiveness dimensions as measured two months earlier. Based on the results from these analyses, it is safe to say that feedback does have a – slightly delayed – positive effect on the development of all three responsiveness dimensions. Of the three input factors studied, feedback even shows the strongest effect in these analyses (at least, it has the highest Beta’s in all three cases). All in all, one can justify the conclusion that feedback acts as a high leverage input factor for team development.
Of the two leadership behaviors, consideration was assumed to have a positive effect on joint management, and initiating structure to positively relate with both job management and boundary management. These hypotheses are all confirmed in the cross-sectional analyses. This provides strong support for the notion that a coaching style of leadership is not a panacea for team development and performance (cf. Stoker, 1999; Kuipers et al., 2007), but but a mix of leadership styles should be used which is compatible with the characteristics and developmental state of a team (Stoker, 1999). This study shows that initiating structure can also be of vital importance for a team’s success. A shift-work production structure is an important contingency factor one should consider. As one process is performed by five different teams, proper coordination of the activities is much more difficult. However, the collinearity between the two factors calls for some caution when distinguishing the effects of both behaviours. This point will be elaborated in the study’s limitations section. For now it suffices to conclude that leadership behaviour clearly plays an important role for team development.
At the output side of the model, the effects of team responsiveness on commitment and absenteeism were studied. Commitment to the organization was expected to be enhanced by all three responsiveness dimensions. The positive relationship between boundary management and commitment is confirmed by the data, but no relationship is found for job- and joint management. Therefore, one can conclude that to successfully involve team members with the organization, job enlargement and enrichment and delegating management of internal team processes do not suffice. Especially in a shift-work organization, work is complicated by the fact that one has to deal with other parties that work different hours, including – and not in the last place – the teams working other shifts. Only when a team successfully manages these boundaries, one only feels truly committed to the organization.
In the analyses for absenteeism, job management was expected to be negatively related with absence duration and joint- and boundary management with absence frequency. The hypothesis that job management reduces absence duration was supported. This is not very surprising if one considers the many studies on the relationship between job characteristics, stress and health effects (e.g. Bakker et al., 2003; Demerouti et al., 2001; Karasek, 1979; see also Karasek & Theorell, 1990). None of the responsiveness dimensions acts as a predictor of absence frequency though. This is probably due to the lack of power caused by the low n in the analyses for absenteeism in this study, combined with the fact that the amount of variance explained in absenteeism research is often low, especially for absence frequency (cf. Bakker et al., 2003).
Limitations
