Comparing Structural and Institutional Contingencies
Influencing Organizational Design
A Theoretical and Empirical Study in Work Autonomy
Master Thesis International Business & Management
A.M. BALTES University of Groningen Faculty of Economics and Business
Duisenberg Building
9747AE Groningen, the Netherlands Tel: (06) 303 649 83
e-mail: a.m.baltes@student.rug.nl
2 ABSTRACT
3 TABLE OF CONTENT
1. INTRODUCTION 7
2. BACKGROUND AND HYPOTHESES 11
2.1. Work Autonomy 11
2.1.1. Definition 11
2.1.2. Autonomy at work – put in practice at different levels 12
2.1.2.1. Employee level 12
2.1.2.2. Managerial level 12
2.1.2.3. Team level 13
2.1.2.4. Organizational level 13
2.1.3. Autonomy at work - clarifying positive outcomes 13
2.1.4. Determinants of the level and effectiveness of autonomy 14
2.2. Structural Contingency Theory 15
2.2.1. Structural contingency theory 16
2.2.2. Critiques on structural contingency theory 16
2.2.3. Hypotheses deriving from the structural contingency theory 17
2.2.3.1. Organizational size 17
2.2.3.2. Education level 17
2.2.3.3. Type of industry 18
2.3. Institutional Theory 18
2.3.1. Institutional theory 19
2.3.2. Scott’s typology of institutional pillars 19
2.3.2.1. Institutions as regulative systems 20
2.3.2.2. Institutions as normative systems 21
2.3.2.3. Institutions as cultural-cognitive systems 21
2.3.3. Critiques of institutional theory 21
2.3.4. Hypotheses deriving from the institutional theory 22
2.3.4.1. Regulative dimensions 22
2.3.4.2. Normative and cultural-cognitive dimensions 23 2.4. Summary of Hypotheses and Conceptual Model 25
2.4.1. Summary of hypotheses 25
4
3. DATA & METHOD 27
3.1. Sample 27
3.2. Data Description 27
3.2.1. Dependent variable 28
3.2.2 Independent structural contingency variables 29
3.2.3. Independent institutional variables 30
3.2.4. Control variables 32
3.3. Method 33
4. RESULTS 36
4.1. Descriptive Statistics 36
4.2. Results Multiple Regression Analysis 38
4.3. Extension Multiple Regression Analysis 42
5. DISCUSSION & CONSLUSION 48
6. REFERENCES 51
5 LIST OF TABLES
Table 1: Scott’s Typology of Institutional Pillars and Carriers 20
Table 2: Minimum, Maximum, Mean and Standard Deviation 37
Table 3: Results Regression Analysis Structural Contingencies 39 Table 4: Results Regression Analysis Institutional Contingencies 41 Table 5: Results Extension Multiple Regression Analysis 45
LIST OF FIGURES:
Table 1: Conceptual model 26
LIST OF TABLES APPENDIX
Table A1: Overview Sample 60
Table A2: Overview Included Industries 61
LIST OF FIGURES APPENDIX
Figure A1: Residuals Histogram 63
6 ACKNOWLEDGEMENT
When I started to write my thesis, I thought I would never be able to do it in two and a half months. So much information I had to collect and read, selecting the appropriate information and rewrite it all over again to make it a good and grounded research. After ten weeks of hard working I can finally say that I did do it and in my opinion quite successful.
Writing this thesis gave me the opportunity to challenge myself and to look and go beyond my own borders. Writing a thesis has a lot to do with perseverance. Even though you are fed up with it, have been non-productive or just don’t know what to do anymore? You have to get yourself together and move on. Ask the help of your supervisor or fellow student and keep on reading and writing because in the end you’ll find a way to do it and you will be able to continue working. That is actually the most important thing I have learned while writing my thesis. Moreover, I have learned to use SPSS syntax with some help from my supervisor. I am very thankful for that, since the word SPSS scared me off in the first place. Now I know how I can use it and I know that SPSS isn’t that scary at all, you just have to do it!
I wouldn’t be able to write my thesis without the unconditional support of my family and friends. Especially I would like to give of word of thanks to my sister Mariëlle, who told me she was proud of me every day. Moreover I would like to thank my friends Holmer Kok and Dirk-Jan Palland for giving me feedback and being my sparring partners in times of difficulty. Last but not least I would like to thank my supervisor Andre van Hoorn for the pleasant cooperation and giving me the support and useful feedback whenever I needed it. In the end I can genuinely say that I enjoyed writing my thesis and everything that belonged to it. I would like to thank the reader for its interest and hope that he or she gets inspired by the topic and enjoys reading it as well.
7 1. INTRODUCTION
In the past five decades, organizational studies have been developed that explain one or more aspect of organizations (Scott, 1992). Academic writing about organizational design has been dominated by the classical management school up until the late 1950’s. This theory states that there is one single organizational design which is suitable and highly effective in all kinds of organizations. This classical organizational design was characterized by a high degree of decision-making and planning at the top of the hierarchy. Therefore empowerment or work autonomy were not a concern and only present at higher levels in the organization, since behavior of lower hierarchical levels and operations were specified in advance (Brech, 1957).
However, as the field of management matures and more companies report that they expand their existing international activities combined with increasing competition, many researchers have started to question the organizational value of the diffusion of ‘homogenized’ management practices, that ignore national cultural nuances as well as other institutional and even firm-level differences (Gamble, 2003; Tsui and Jiaotung, 2007).
This is in line with Erez (2010) who argues that job design – the structure, content and configuration of a person’s work tasks and roles (Parker & Ohly, 2008)- is nested within organizations and nations. At national-level, cultural values are being internalized through the process of socialization. In this way, cultural values serve as criteria for evaluating whether a certain job design provides the opportunity for experiencing a sense of self-worth and well-being (Erez and Earley, 1993) Therefore, different job designs emerge in different cultures.
At organizational level however, Erez (2010) argues that job design is the direct representation of the person’s work environment. It is said by Donaldson (2008) that the structural contingency is the main theory of organizational design. More recently however, it has been argued that institutional theory also provides insight about organizational design. The theories are similar in a way that both theories produce outcomes that are beneficial for the organization. However, they are different in the outcomes that they theorize.
8 to be too static since organizations are seen as being forced to adjust its design to contingencies in order to be most effective (Donaldson, 2006). As a consequence, the analysis is depersonalized and there is an absence of analysis at the level of human actors (Pennings, 1992), institutions and their associated values, beliefs and power.
Including institutions, human actors and their national culture is important because to operate effectively in a particular country, it is commonly argued that organizations must be sensitive to the prevailing attitudes and values in that country (Edwards and Kuruvilla, 2005). Newman and Nollen (1996, p.755) very well explain the probable consequences when the socio-cultural context is not taken into consideration. “When management practices are inconsistent with deeply held values and expectations, employees are likely to feel dissatisfied, distracted, uncomfortable, confused and uncommitted, and this will result in a lowered ability and willingness to perform well”.
While national culture differences can be important and must be understood, their role needs to be put in the context of other contextual factors (Gerhart and Fang, 2005). Therefore a second source of pressure towards decentralized decision-making about organizational design is the need to abide by national-level regulations and institutions in the labour market (Boxall, 1999).
A theory that connects both pressures, culture and national level regulations, is the institutional theory (e.g. Scott, 1995). This theory is said to be the most popular organizational theory at the moment (Donaldson, 2006). The theory differentiates itself from the structural contingency theory by taking the institutional environment into consideration, including regulative, normative and cultural-cognitive pillars. Moreover, the theory holds that organizations adapt to their institutional environment, thereby maximizing legitimacy and external support of the organization. This adaption may be without implementing characteristics that produce higher operational effectiveness, which is in contrast to the structural contingency theory (Dimaggio and Powell, 1991).
In sum: the institutional theory differs from the structural contingency theory in stressing the symbolic more than the real, and in adoption as adapting to institutional norms rather than organizational task contingencies which generally increases effectiveness (Donaldson, 2006, p.28).
9 contingency theory, the contingency theory remains the explanatory theory of organizational design. Furthermore, because design descriptions of both theories usually diverge and might be in conflict, it can be difficult to select the most optimal design for an organization.
To make a contribution in the academic literature of job design, I will analyze theoretically and test empirically to what extent the level of autonomy is determined by structural and/or institutional factors. In this way it can be identified which fit is more superior over the other and which theory is the most appropriate for determining the level of autonomy at work. The reason why I would like to concentrate on the phenomenon of work autonomy is because autonomy has drawn considerable attention in the last three decades in the context of job characteristics and job design by e.g. Hackman and Oldman (1975) and Sims, Szilagyi, and Keller (1976). According to Chandra Sekhar (2011) work autonomy has been identified as one of the most important variables in many research domains. This might be due to the fact that, in his and in other studies (Gardell, 1977; Griffin, 1981; Hackman and Lawler, 1971) autonomy has been positively linked to important variables and organizational outcomes such as employee performance, work satisfaction, job involvement, supervision and reduced employee absenteeism.
Nevertheless, most of these previous studies (Hackman and Oldman, 1975; Sims, Szilagyi and Keller, 1976; and Decotiis and Koys, 1980) treated work autonomy as a global concept, and solely focused on the uni-dimensional nature of autonomy. Conversely, Breaugh (1989) suggests that exploring the multi-dimensional nature of work autonomy, not only leads to improved theory development, but also helps to design effective organizational strategies. This is important, because as long as studies of autonomy at work rely exclusively on micro thinking, the politics of ‘empowerment’ in any given culture at any given time will probably be limited to regimes of illusory control (Jermier and Micheals, 2001, p 1009).
A small number of studies (e.g. Dobbin and Boychuk, 1999; Foley and Polanyi, 2006) have put autonomy in a wider environmental context. However none of them focused on institutional as well as structural contingencies influencing the level of autonomy. To take a step in that direction, this study examines the institutional and structural environment of organizations, affecting the level of autonomy at work through the lens of the structural contingency theory of Donaldson (1996) and the institutional theory of Scott (1995).
10 “To what extent is the level of work autonomy determined by structural- and/ or institutional contingencies?”
11
2. BACKGROUND AND HYPOTHESES
The following section will first present some background information concerning work autonomy. After that the structural contingency theory will be discussed and is subsequently linked to autonomy by some theoretically grounded hypotheses. Finally, the institutional theory will be explained and linked to autonomy by some theoretically grounded hypotheses. 2.1. Work Autonomy
Jermier and Michaels (2001) argue that the theme of autonomy has originally been a central theory of human growth and development in society for several decades. More recently however, the concept of work autonomy has drawn considerable attention in the context of job characteristics and job design (Hackman and Oldman, 1975: Sims, Szilagyi and Keller, 1976). In order to get a better understanding of the concept of autonomy at work, the next section will first define autonomy at work. After that, autonomy at work will be discussed at different levels within the organization and finally, several determinants that have previously been linked to autonomy across countries will briefly be discussed.
2.1.1 Definition
Hackman and Oldman (1975, p.162) 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 to be used in carrying out.” This definition is quite similar to the definition of Hackman and Lawler (197, p.265), which define work autonomy as “the extent to which employees have a major say in scheduling their work, selecting the equipment they like and deciding on the procedure to be followed”.
Chung (1977) on the other hand is somewhat broader in its approach by including the factor ‘have some say over goal setting’. He discussed autonomy in terms of the degree to which an individual is able to (1) determine his own work methods, (2) pace himself and have control over work schedules and (3) have some say over goal setting. This feeling of personal control over one’s job environment is associated with a variety of personally and organizationally valued outcomes (Loher, Noe, Moeller and Fitzgerald, 1985).
12
2.1.2 Autonomy at work – put in practice at different levels
Autonomy at work can take place at different levels: employee-, managerial-, team- and organizational level. In general, work autonomy has been positively linked to important organizational variables such as employee performance, work satisfaction, job involvement, supervision, reduced employee absenteeism (Gardell, 1977; Griffin, 1981; Hackman and Lawler, 1971; Hackman and Oldman, 1975; Sims, Szilagyi, and Keller, 1976), participation in decision making and organizational commitment (Chandra Sekhar, 2011). It is thus assumed that, all these outcomes will partly be enhanced when jobs are high on work autonomy. The study of Helms and Editor (2006) determines work autonomy at four different organizational levels, these levels will be discussed below.
2.1.2.1. Employee level
Autonomy at employee level; it is argued by Helms and Editor (2006) that by giving employees more autonomy, more responsibility is felt by the employee for their own work outcomes. Moreover they are better able to use their personal attributes to contribute to job performance. This is supported by Niessen and Volmer (2010) who found support in their experiment to the assumption that participants with poor performance had low autonomy as opposed to their counterparts who had high autonomy.
Moreover, it is argued by Helms and Editor (2006) that too much autonomy, on the other hand may lead to employee dissatisfaction because each employee has a different need for autonomy in his or her job. Some employees prefer more direction from a manager than another. As a result, some employees may feel uncomfortable with autonomy and their association with a particular job or project. In addition, when employees are not well equipped in terms of training or in personality (e.g. conscientiousness, extraversion (Barrick and Mount, 199) and a pro-activity (Fuller, Hester and Cox, 2010)) to exercise autonomy, it may result in workplace tension and poor performance (Helms and Editor, 2006).
2.1.2.2. Managerial level
13 feelings of motivation and satisfaction for the manager. Conversely, if managers make poor decisions due to their higher levels of autonomy, this may be harmful for employees as well as the organization as a whole (Helms and Editor, 2006).
2.1.2.3. Team level
Autonomy at team level; Jermier and Michaels (2001) argue that self-organizing teams possess a certain degree of self-management and collaborative teamwork. In most of these situations, members experience relatively high levels of autonomy by collectively controlling, for example, the pace of work, task allocation, recruitment and selection of new members and so on. Moreover, it is said by Helms and Editor (2006) that autonomy at team level may lead to increased creativity, flexibility and innovation within an organization. However, it must be said that, the success of a self-managed work-team, greatly depends on the capabilities of the employees to work together. Too much autonomy may lead to reduced communication due to individuals who work too independently. As a result this may lead to poor team performance. Nonetheless, despite substantial variance in research findings regarding the consequences of autonomous work-teams, positive impacts were found on motivation and productivity, attitudes and perceived quality of work life (Guzzo and Dickson, 1996).
2.1.2.4. Organizational level
Finally, autonomy at organizational level. Helms and Editor (2006) argue that this type of autonomy is often incorporated in the structure and culture of the organization. Both authors argue that bureaucratic organizations often have little autonomy, where more organic and new organizations can rely more often on empowerment, autonomy and employee participation to succeed their organizational operations. It is said that reduced barriers to cooperate can positively influence workplace functions and organizational performance. A possible drawback of organizational autonomy is disparity among units, due to differences in work practices and rules. Moreover, in worst case scenario’s, work autonomy may lead to unethical behaviour. It is therefore essential to have some supervision in organizations to prevent wrongdoing (Helms and Editor, 2006).
2.1.3. Autonomy at work - clarifying positive outcomes
14 because it is the foundation of human dignity and the source of morality (Hill, 1991). In most Western cultures, an adult person’s dignity and self-respect is usually based on possessing some degree of control in making decisions that deeply affect their life (Jermier and Michaels, 2001). This is in line with Breaughs’ (1985, p.557) argument who suggests the following; when autonomy needs are satisfied, feelings of responsibility are experienced, work activities can be carried out in different ways which provides variety, and such autonomy may be interpreted as recognition.
2.1.4. Determinants of the level and effectiveness of autonomy
As discussed in the previous section autonomy is an important variable when deciding for organizational design due to its positive personal and organizational outcomes. Nevertheless, differences in the level- and effectiveness of autonomy across countries have mainly been unexplained. The studies who did explain these differences are briefly discussed in this section.
Jermier and Michaels (2001) describe organizational democracy as a stream of research on autonomy. According to Jermier and Michaels, different authors who have published their work in the Economic and Industrial Democracy Journal (e.g. Stymne, 1980: Moss, 1991; Foley and Polanyi, 2006) conclude that self-directed teams tend to be more effective in nations other than the USA. The standpoint of organizational democracy provide two explanations to this finding. First, they suggest how individualistic- and collectivist countries shape attitudes and other reactions towards team-based organizing. Second, the authors suggest a need to develop programs of inquiry. These programs should go beyond the psychology of autonomy and explain different structures of control. The structures should be based on a countries institutional environment (Jermier and Michaels, 2001) which are anchored in political, legislative and cultural factors (Forbes and Jermier, 1995). According to these arguments the institutional environment influences the appropriate level and effectiveness of autonomy.
15 United States, Canada and Australia, workers would have lowers levels of autonomy. Moreover, because work control is a zero-sum game in countries where managers have substantial autonomy, workers will have subsequently relative low autonomy.
Second, the collective bargaining thesis suggests that union bargaining strategy is key. Where on the one hand unionist will have high levels of autonomy in ‘co-determination’ countries. In these countries unions negotiate for ongoing work participation in decisions about the organization or work through work councils and other means. On the other hand, unionist will have low autonomy in ‘job control’ countries. In these countries, autonomy is restricted by contractual rules that specified work routines and job boundaries (Kochan, 1980). Finally, the authors argue that, national management, training, bargaining and unemployment systems operate according to different logics (Fligstein and Byrkejflot, 1996; Scott, 1994). For example, when work is controlled to a greater extent by rules, management practices tend to be oriented towards routinization of work, which is known as direct control. Conversely, when work is controlled by skill, management practices tend to be oriented towards employee control of the work process, also known as responsible autonomy. It can be said that in the study of Dobbin and Boychuck, both organizational as well as institutional factors influence the level of work autonomy.
This section clearly reflects the dilemma for organizational design. There are several explanations, why levels and the effectiveness of work autonomy differ across countries. But which theory explains most of the variance and is therefore most important to take into consideration? This question is unanswered. To take a step in that direction, this study examines the institutional as well as the structural environment of organizations, affecting the level of autonomy of employees. This will be done through the lens of the institutional theory (Scott, 1995) and the structural contingency theory (Donaldson, 1996). The following section first explains the structural contingency theory of Donaldson. After that, the institutional theory of Scott will be explained.
2.2. Structural Contingency Theory
16
2.2.1 Structural contingency theory
The contingency theory of organizational structure provides a major framework for the study of organizational design (Donaldson, 1995a, 2001). The theory relates variations in organizational design to characteristics of the organization (its contingencies).
There are several contingency factors coming from the organization, such as: strategy, size, task uncertainty and technology. Nevertheless, these organizational factors in sum reflect the influence of the environment in which the organization is located. Consequently, the structure is caused directly by the internal organizational factors and only indirectly by the environment (Donaldson, 1996). In other words: fitting the organization to its contingencies, leads to receiving benefits of internal effectiveness (Donaldson, 1996, 2008).
It can therefore be said that at the heart of the theory is the relationship between misfit with the organization and performance. This misfit provides the explanation of why organizations adopt the structures that they do and thereby explaining the associations between structural and contingency variables (Donaldson, 2006, p.29).
2.2.2 Critiques on structural contingency theory
The structural contingency theory is criticized because it is said to be too static and fails to deal with organizational change and adaptation (Galunic and Eisenhardt, 1994). This is because organizations are seen as being forced to adjust its structure to contingencies such as size and technology in order to be most effective (Donaldson, 2006).
Donaldson (1987, 2001) disagrees with this critique since he developed the theoretical model of Structural Adaption to Regain Fit (SARFIT). In this model Donaldson argues that an organizational design’s fit only remains temporary and that fit and misfit alternate each other. Moreover it is argued by (Burns and Stalker, 1962) that organizations can change from a mechanistic to organic structure in response to technological and market changes in the environment.
17
2.2.3 Hypotheses deriving from the structural contingency theory
According to the structural contingency theory, organizational design is influenced by several organizational contingencies like, organizational size, task uncertainty, technology and strategy which all directly influence organizational design. However, it is also said that the organizational environment indirectly influences organizational design. It is impossible to test all structural contingencies which possibly influence the level of autonomy due to the limited availability of time and data. For this reason I decided to test one direct contingency, namely organizational size, and two indirect contingencies, namely education level and type of industry.
2.2.3.1. Organizational size
According to Child (1975) several studies have found a high degree of association between organizational size and the extent to which they reflect particular structural characteristics. This is in line with Donaldson (1996) who argues that as size increases, organizational structure will be characterized by a tall hierarchy and extensive specialization. It is generally known that bureaucracy brings drawbacks through rigidity, dysfunctions and some loss of control. However these dysfunctions are often out-weighed by an increase of control and coordination due to the formal structure of job duties as well as well defined job descriptions (Aycan, 2005). This is in accordance with Blau and Schoenherr (1971) who find that large organizations are most likely to adopt formal work rules and written job descriptions that reduces the level of job autonomy. It is therefore predicted that employees working for larger organizations will have lower levels of autonomy opposed to those who work for smaller organizations.
Hypothesis 1: The greater the organizational size, the smaller the level of work autonomy available to employees.
2.2.3.2. Education level
18 Moreover, Donaldson (1996) argues that as organizations seek to innovate, tasks arise which cannot be pre-specified. As a result, more educated and trained employees are allowed to have higher levels of discretion and autonomy in order to encourage employees to use their own initiative, rather than being imposed by hierarchical superiors. It is thus expected that employees that acquired a higher level of education have greater levels of autonomy than employees that acquired a lower level of education.
Hypothesis 2: The higher the level of education, the greater the level of work autonomy available to employees.
2.2.3.3. Type of industry
The level of autonomy at work should vary as a result of predictability of the work process (March and Herbert, 1958). On the one hand, industries where the work process is predictable, the ability to program decision making will reduce the level of autonomous decision making. On the other hand, in industries where the work process is unpredictable, the level of autonomy will increase due to the inability to program decision making (Dobbin and Boychuk, 1999). Moreover, jobs that require innovation frequently have job descriptions that allow a certain degree of freedom to employees to increase creativity (Jackson Schuler and Rivero, 1989). This indicates that the level of autonomy for employees, that work in unpredictable and innovative industries (e.g. professional services, construction, high-tech services), will be higher than the level of autonomy given to employees that work in predictable industries (e.g. agriculture, mining, manufacturing and retail trade).
Hypothesis 3: The less predictable the industry, the greater the level of work autonomy available to employees.
2.3. Institutional Theory
19
2.3.1 Institutional theory
In contrast to the earlier structural contingency theory, where the importance of materialist and organizational forces were emphasized (e.g. organizational size, industry and educational level), the institutional theory calls attention to the role of ideational forces of organizations (e.g. knowledge systems, beliefs and rules in the structure of the organization and operation) (Scott and Christensen, 1995). Organizations are seen as open systems, affected by- and affecting the environment in which they exist (Scott, 1992). In other words, it can be said that the ground in which organizations are rooted lies in the importance of the wider social and cultural environment. For these reasons, the institutional theory provides an alternative framework for decision making and structural design (Scott and Christensen, 1995).
The main purpose of institutions is to reduce the uncertainty actors experience about the behavior of others and allow them to make credible commitments to each other (Hall and Soskice, 2001). Additionally, Ostrom (1990) suggests that institutions provide the ability to (1) exchange information among actors, (2) monitor the behavior of these actors, and (3) sanction inappropriate behavior by cooperative effort.
Institutions are defined by Scott (2008, p.48) as “Institutions are comprised of regulative, normative and cultural-cognitive elements that, together with associated activities and resources, provide stability and meaning to social life”.
As a result, adapting organizational design to institutions will lead to greater legitimacy and external support (Scott, 1995). Moreover, when management practices are consistent with deeply held values and expectations, employees are likely to feel focused, satisfied, comfortable, and committed. This will result in an increased ability and willingness to perform well.
2.3.2. Scott’s typology of institutional pillars and carriers
20 Moreover, the pillars exercise three forms of pressures on organizations to conform to their expectations. According to DiMaggio & Powell’s (1983) typology, coercive pressures arise from societal expectations and inter-organization interdependence; normative pressures arise from professionalization and finally, mimetic pressures are obtained from experienced uncertainty in the environment. When organizations in the same field are subject to similar coercive, normative, and mimetic pressures, it is said by Yang and Konrad (2011) that these organizations tend to develop similar sets of organizational structures. For this reason, institutional pressures result in organizational homogeneity (Scott, 1995). Table 1 gives an overview of Scott’s typology of institutional pillars and carriers.
TABLE 1:
Scott’s Typology of Institutional Pillars and Carriers
Source: Scott (2001, p. 77)
2.3.2.1. Institutions as regulative systems
21 2.3.2.2. Institutions as normative systems
Normative systems consists of both norms and values, where values, are conceptions of the preferred and desirable and norms, specify how things should be done; they define legitimate means to pursue ends (Scott, 2008, p.54). Normative systems are typically viewed as imposing constraints on social behavior. At the same time, they empower and enable social action (Scott, 2008). Moreover, there is a great awareness of one’s role in society and the concern to behave appropriately. In addition, behavior is in accordance with other people’s expectations and internalized standards of conduct (Scott and Christensen, 1995). According to DiMaggio and Powell’s (1983) typology, this mechanism is a normative process and is characterized by social- and professional norms and standards (Scott, 1995).
2.3.2.3. Institutions as cultural-cognitive systems
Cultural-cognitive systems represent the shared conceptions that appoints the nature of social reality and the frames in which meaning is made (Scott, 2008). The meaning we attribute to objects and activities are shaped by symbols, words, signs and gestures. Meanings arise in daily interaction and are maintained and transformed as they are used to make sense of the ongoing stream of happenings and events (Scott, 2008). It can be said that the cultural-cognitive systems are taken for granted because in this particular system, institutions are not bundles of regulations or collections of norms, but knowledge systems (Scott and Christensen, 1995). This is also reflected in Hofstede’s (1991) graphic metaphor, where he argues that culture provides patterns of thinking, feeling and acting, which can also be described as mental programs or as ‘software of the mind’. According to DiMaggio and Powell (1983), this mechanism is a mimetic process in which the wider belief system and cultural frames are imposed on and/ or adopted by individual actors and organizations.
Misalignments of the pillars and carriers may lead to unintended consequences within the organizational field, such as (organizational) conflicts, goal confusion and institutional change (Scott, 2008).
2.3.3. Critiques of institutional theory
22 institutional theory refutes that organizational movements which regain fit, raise organizational performance (Donaldson, 1987). Finally, the institutional theory emphasizes institutional isomorphism. However, some of the longitudinal evidence show the opposite (Meyer, Scott and Creighton, 1988; Donaldson, 1995b).
It is indisputable that critiques of both theories should be taken into consideration. However, the interesting thing about this study is that I will investigate differences in the explanation of the level of autonomy by both theories. As a result it might be shown that complementarities of both theories are found in explaining organizational design. In order to empirically test to what extent both theories explain differences in levels of work autonomy, the next section will formulate theoretically grounded hypotheses derived from the institutional theory.
2.3.4. Hypotheses deriving from the institutional theory
According to the institutional theory, organizational design is influenced by regulative, normative and cultural-cognitive pillars. Possible variables which influence the level of autonomy at regulative level are, for example, legal system, ownership dispersion, the level of shareholder protection and the amount of rules and laws applying to organizations. At cultural-cognitive/ normative level possible variables which may influence the level of autonomy are for example, cultural values ( Hofstede, 2001; Trompenaars, 1993; Aycan et al., 2000 and House et al.,1999), the level of trust or the level of responsibility given to employees. It is impossible to test all variables influencing the level of autonomy, considering the limited availability of time and data. For this reason I decided to test one regulative dimension- legal system, and three cultural cognitive dimensions- individualism/ collectivism, acceptance of uncertainty and the distribution of power.
2.3.4.1. Regulative dimensions
Legal system
23 be less constrained than socialist law. Civil law is mainly focussed on discovering a just solution to a particular problem instead of following a strict procedure (La Porta, et al. 1999). Finally, common law countries are characterized by their intent to limit the power of the sovereign (Finer, 1997). As a result common law ‘put their emphasis on the private rights of individuals and especially their property rights’ (Finer, 1997, p.1348). In sum it can be said that on the one hand, socialist law give very little opportunity to organizations as well as employees to operate freely and make individual decisions due to the dominant presence of the State. While common law countries on the other hand, highly value individual freedom and allow for individual participation and decision making. Civil law is more in between both legal systems, the ability for individuals to make individual decisions is dependent on the situation. It is therefore expected that employees in socialist law and civil law countries have
lower levels of autonomy compared to employees situated in common law countries. Hypothesis 4: In common law countries, employees have higher levels of autonomy than they
have in civil or socialist law countries.
2.3.4.2. Normative and Cultural- cognitive dimensions
Research suggests that the effect of work autonomy on work satisfaction and performance differ across cultures. Intrinsic motives for autonomy are important for well being across cultures (Ryan and Deci, 2000) yet antecedents to such motivation vary cross culturally (Gelfand, Erez, Aycan, 2007). To give an example Iyengar and Lepper (1999) found that, while personal choice was critical for intrinsic motivation among Anglo-Americans. Asian Americans on the other hand were more intrinsically motivated when authority figures or fellow colleagues made choices for them. To test the underlying cultural values leading to differences in the level of work autonomy, the underlying hypotheses are formulated.
24 higher levels of autonomy. On the other hand, in collectivistic cultures, members of society as well as members of the organizations will have strong expectations that decisions must be made in a consultative process (Crossland and Hambrick, 2011). It is thus expected that in countries with low levels of individualism but high on collectivism, employees will have low levels of autonomy.
Hypothesis 5: The more individualistic a country, the greater the level of work autonomy of employees in that country.
Acceptance of Uncertainty
Erez (2010) argues that cultures that are known for their low levels of uncertainty avoidance, organizations should be more tolerant to relaxed rules and regulations. As a consequence, task design should be less structured and fixed but more flexible and open. For this reason, it is expected that the level of work autonomy would be high in countries with low levels of uncertainty avoidance because employees are allowed to determine their own work methods, pace and schedule.
On the other hand, in countries with high levels of uncertainty avoidance, there should be avoidance of deviant persons and ideas. Moreover, high levels of uncertainty avoidance should increase the level of bureaucracy, and the adherence to rules and regulations which would decrease the chaos in the organization (Brown and Eisenhardt,1998). Finally, in high uncertainty avoidance cultures, jobs are defined in specific rather than in broad terms in order to reduce role ambiguities (Wong and Birnbaum- More, 1994). As a result organizations use tools such as technology, rules and rituals to cope with uncertainty and unpredictability. It is therefore expected that the level and acceptance towards autonomy at work decreases when uncertainty avoidance is high.
Hypothesis 6: The greater the avoidance to uncertainty in a country, the smaller the level of work autonomy of employees in that country.
Distribution of Power
25 argue that in countries with high acceptance of unequal power distribution, superiors are expected to act autocratically without consulting subordinates. Hofstede (2001, p.108) interprets this as a conflict between ‘values as the desirable and values as the desired’ because in countries with high acceptance of power distribution, there can be a strong ideological push toward formal participation (e.g. union representation, workers councils) but little functional participation of subordinates in daily issues. It is thus, unlikely that a high level of autonomy will be given to employees in countries where unequal power distribution is accepted because high levels of unequal power distribution are linked to the concentration of authority.
On the other hand, it is argued by Vitell, Nwachukwu and Barnes (1993) that in countries with low acceptance of unequal power distribution, individuals might look more to both their fellow colleagues and informal norms for guidance on appropriate behavior. As a result, countries with low acceptance of unequal power distribution are more likely to have greater levels of autonomy because it is expected that employees are allowed more frequently to determine their own work method, pace and that they can participate in decision making within the organization.
Hypothesis 7: The greater the acceptance of power distribution in a country, the smaller the level of work autonomy of employees in that country.
2.4. Summary of Hypotheses and Conceptual Model
To give a clear overview of the hypotheses to be tested, the following page provides a conceptual framework together with an overview of all hypotheses formulated in the previous sections.
2.4.1. Summary of hypotheses
Hypothesis 1: The greater the organization, the smaller the level of work autonomy available to employees.
Hypothesis 2: The higher the level of education, the greater the level of work autonomy available to employees.
26 Hypothesis 4: In common law countries employees have higher levels of autonomy than they have in civil or socialist law countries.
Hypothesis 5: The more individualistic a country, the greater the level of work autonomy of employees in that country.
Hypothesis 6: The greater the avoidance to uncertainty in a country, the smaller the level of work autonomy of employees in that country.
Hypothesis 7: The greater the acceptance of power distribution in a country, the smaller the level of work autonomy of employees in that country.
2.4.2. Conceptual model
In order to give a graphical overview of the hypotheses and their expected relationships to work autonomy, a conceptual model is presented in figure 1.
FIGURE 1: Conceptual Model Structural Contingencies (H1) Organizational size - (H2) Educational level + (H3) Type of industry + make (Unpredictable or not)
Institutional Contingencies (H4) Legal System +
(Common law or not) (H5) Individualism +
(H6) Avoidance of Uncertainty -
(H7) Distribution of Power -
27 3. DATA & METHOD
This section first describes the sample and datasets employed for the empirical analysis. Second, the measurement of the dependent and independent variables are explained. Finally, the methodology applied to the analysis is discussed.
3.1. Sample
To determine whether work autonomy and indirectly job design is influenced more at organizational- or at national level, the structural contingency theory and institutional theory are being used. In order to conduct the research, a sample is needed which includes individuals employed at different organizations as well as living within different institutions and cultures. For this reason I choose to investigate individuals living in Europe, because even though political-, trade-, and financial integration have been prominent pillars in the past two decades (Friedrich, Schnabel and Zettelmeyer, 2013), cultural differences as well as historically determined legal origins remain.
The sample includes all persons aged 15 and over, resident within private households regardless of their nationality, citizenship, language or legal status. No additional criteria are employed to determine the sample. In total, 122.577 individuals who experienced some degree of work autonomy have been included in this research. Table A1 in the appendix gives an overview of the number of participants per country and the average level of autonomy experienced in that country. Germany is best represented in this research with 8.644 individuals and Italy least with 1.339 individuals. On average, Nordic countries like Sweden, Denmark and Finland experienced the highest level of autonomy. Whereas more Eastern European countries like Russia, Czech Republic and Slovakia experienced on average the lowest levels of autonomy.
3.2. Data Description
The main data source employed in this research is the secondary dataset of the European Social Survey (ESS), cumulative file rounds 1-5 (ESS, 2012). This dataset provides academically driven, multi-country survey data of over 230.000 individuals living in 30 European countries. All country samples are equivalent national probability samples.
28 elderly persons. Moreover, the disproportionate representation of geographic coverage and difficulties in considering the right sample size might be a drawback using this sample strategy (Ezattti-Rise and Murphy, 1995). However, because the sample size is over 120.000 individuals and for each country a general weight is computed, I expect a well proportionate representation of the geographic area and an appropriate sample size.
The dataset represents the cumulative results of five rounds of ESS. The first round was conducted in 2002-2003 and the fifth round in 2010-2011. In this way, the cumulative dataset is a cross-sectional representation of the previous survey rounds. In each round, questions have been asked- or could have been answered in different ways. Especially the fifth round contains dissimilar answer possibilities. In order to create homogeneous results I decided to limit myself to rounds 1-4 covering the survey rounds between 2002-2008.
The countries incorporated in the dataset are: Austria, Belgium, Bulgaria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Israel, Italy, Luxembourg, Netherlands, Norway, Poland, Portugal, the Russian Federation, Slovakia, Slovenia, Spain, Sweden, Switzerland, Turkey, Ukraine and the United Kingdom. The variables used from this dataset are the level of autonomy, type of industry, level of education and organizational size.
The ESS dataset has several advantages. First, the study of ESS is funded by the European Science Foundation and is designed and carried out at exceptionally high standards (ESS, 2013), this increases the reliability of the data and indirectly the results. Moreover the sample size is very extensive, which increases the desired precision of the estimate and generalization of the findings.
Other datasets employed in this study are Hofstede’s website (2013) for determining the values of individualism, power distance and uncertainty avoidance at national level. Finally the dataset of La Porta, Lopez-de-Silanes, Scheifler and Vishny (1999) is being used to determine the legal systen for each country.
3.2.1 Dependent variable
29
3.2.2. Independent structural contingency variables
In this research several independent variables are being used. In order to determine to what extent structural contingency variables influence the level of work autonomy, organizational size, educational level and type of industry are being operationalized. Table 2 gives an overview of the descriptive statistics of all structural contingency variables and is presented and analyzed in section 4.1.
Organizational size
Blau (1972, p.3) defines organizational size as “the scope of an organization and its responsibilities’. Size is usually operationally defined as the number of employees. However, in this study organizational size is determined by establishment size, using the survey question from the ESS “Including yourself, about how many people are/were employed at the place where you usually work/worked?”. Each individual was able to choose between five categories where (1) indicates < 10, (2) 10-24, (3) 25-99, (4) 100-500 and (5) > 500 employees. The categories are operationalized through four dummy variables, where > 500 is the base case. Data for each individual can be found in the ESS.
Type of industry
Type of industry is operationalized by using the ESS question, what does/ did the firm/ organization you work/worked for make or do? To cover survey rounds 1-4, I combine two variables: nacer 1 and nacer 11.
Moreover, in my hypothesis I argue that the level of autonomy at work is dependent on the predictability of work processes within an industry. Stankovic and Ramamritham (1990) state that when a work process is predictable, it means that it should be possible to show, demonstrate or prove that requirements are met, taken the underlying assumptions into consideration. For example, concerning failures or workloads.
30 An overview of the incorporated industries can be found in Table A2 included in the appendix.
Education level
Educational level is determined by using the ESS question, what is the highest level of education you have achieved? Seven different levels can be distinguished, ranging from less than secondary to master level. I codify educational level through six dummy variables, where (1) indicates a particular level and (0) something else. I use the lowest level of education as base case, which covers the people who completed less than lower secondary. The other dummy variables are ranged from the second lowest to the highest level of education.
3.2.3. Independent institutional variables
In order to determine to what extent institutional variables influence the level of work autonomy, legal system is being operationalized as well as three cultural dimensions of Hofstede. It is noteworthy that Hofstede’s research on national culture has been criticized by e.g. McSweeney (2002) for several reasons. For example, the analysis of Hofstede was based on primary data of employee-attitude surveys undertaken in 66 subsidiaries of the large multinational enterprise IBM. Moreover the uni-level analysis excludes the interplay between macro- and micro-scopic cultural levels and finally, culture has been characterized by a limited number of dimensions. Despite the fact that these points of critique can be important and must be understood, Hofstede’s research has several advantages as well. For example, Hofstede’s values are most widely used in the measurement of national culture (Haxhi and van Ees, 2010; Chui and Kwok, 2008; Kirkman, Lowe and Gibson, 2006), this helps facilitating acceptance and systemic comparison. Moreover, Hofstede’s data allowed for the largest sample of countries to be included in this research and finally, the institutional analysis is carried out at national level, which matches the national values of Hofstede. For these reasons I find Hofstede’s dimensions sufficient enough measuring normative- cultural cognitive values. Again, Table 2 gives an overview of the descriptive statistics of all institutional contingency variables and is presented and analyzed in section 4.1.
Legal system
31 codes; (3) German commercial code; (4) Scandinavian commercial code and (5) Socialist/ Communist laws. Because distinctions between French, German and Scandinavian commercial codes are very subtle (La Porta et al., 1999) I bundle them together and rename them as civil law. I codify legal system using two dummy variables, where civil law is the base case and common law and socialist law are represented by the two dummy variables. Legal origin for each country can be found in the article of La Porta, et al. (1999).
Individualism/ Collectivism
Individualism/ collectivism is operationalized by using Hofstede’s cultural dimension power distance. Hofstede (2001, p.225) defines individualism as “a society in which ties between individuals are loose: Everyone is expected to look after him/herself and his/her immediate family only”. As opposed to collectivism, which is defined by Hofstede (2001, p.225) as “a society in which people from birth onwards are integrated into strong, cohesive in-groups, which throughout people’s lifetime continue to protect them in exchange for unquestioning loyalty”. All values of individualism can be found on Hofstede’s website except for Cyprus and Ukraine. Missing values were excluded from the sample. The higher the number on individualism, the more individualistic a country is.
Avoidance to uncertainty
Avoidance to uncertainty is operationalized by using Hofstede’s cultural dimension uncertainty avoidance. Uncertainty avoidance is defined by Hofstede (2001, p.161) as “the extent to which the members of a culture feel uncomfortable by uncertain and unknown situations”. All values of uncertainty avoidance can be found on Hofstede’s website except for Cyprus and Ukraine. Missing values were excluded from the sample. The higher the number on uncertainty avoidance is, the more avoiding a country is to uncertainty.
Power distribution
32
3.2.4 Control variables
Gender
Research has shown that gender affects job outcomes, such as wage, power, status and autonomy (Brown, 1975). Discrimination in job outcomes between men and women have been linked by research in two ways. First, it has been shown that gender discrimination has led to lower wages, status power and autonomy for jobs traditionally held by woman (Jacobs, 1990). Moreover, research has shown that discrimination is present when hiring and promoting women in less attractive jobs than similarly qualified men (Rosenfield and Kalleberg, 1990). For these reasons it might be expected that the level of work autonomy is higher for men than women. The data is obtained from the cumulative ESS. I codify gender through a dummy variable, where (1) indicates men (0) women. Table 2 gives an overview of the descriptive statistics of all control variables and is presented and analyzed in section 4.1.
Age
A higher age is often associated with a greater level of experience and subsequently the improved ability to make decisions independently. Therefore it might be expected that the level of autonomy increases when age does as well. The data is obtained from the cumulative ESS. I operationalize age by recoding the year in which each individual is born to the numeric variable in years.
Employment relation
It might be expected that, when somebody is self-employed, he or she might experience high levels of work autonomy because the person is its own supervisor. Whereas employees or people working for their own family business probably experience lower levels of work autonomy because they are often more controlled by their executive. The data is obtained from the cumulative ESS. I codify employment relation through two dummy variables where employee is taken as base case. People working for their own family business and self-employed people are represented by the dummy variables where (1) indicates a particular work relation and (0) indicates something else.
Country
33 explained variance at national level. Doing so will help me illustrate to what extent the institutional variables explained the total variance possible at country level. The data is obtained from the cumulative ESS. I codify country through dummy variables, where the United Kingdom is base case. The remaining 29 countries are reflected by dummy variables where (1) indicates that an individual belongs to a specific country and (0) indicates that an individual is not.
3.3. Method
For this research I made a database through SPSS syntax. The dataset includes the level of work autonomy, structural contingency,- institutional,- and control variables. In order to test the hypotheses relating structural contingency variables at individual level, and institutional pillars at national level to work autonomy, I use a multiple regression analysis with dummy variables. I select this type of analysis because I use one metric dependent variable as well as several other metric and non-metric independent variables. The reason why I frequently use dummy variables is because it’s a way to transform non-metric to metric variables. Moreover categorizing particular variables would lead to loss of information. Using dummy variables, prevents me from doing so and gives me the opportunity to capture valuable information.
The primary motive for choosing a regression analysis is generally forecasting. The regression analysis enables me to develop an estimation equation by which I predict values for the dependent variable from the values of several other independent variables. Moreover I am able to evaluate the key drivers of work autonomy and indirectly work design by comparing the model’s fit for two different sets of contingencies.
The multiple regression analysis contains several assumptions; normality, linearity and homoscedastisity. In order to test the validity of the model, the descriptive statistics reflect the accuracy of the data by examining whether all variables have a valid value. In addition, normality is visually reflected through a residual histogram. Linearity and homoscedastisity are tested trough a residual scatterplot.
34 compare the variance explained by different contingencies. I’ll do so by executing the following equations in the regression analysis:
1. Work autonomy = β0 + β1 SIZE<10 + β2 SIZE10-24+ β3 SIZE25-99 + β4 SIZE100-499 + β5 LOWER SECONDARY + β6 LOWER UPPER SECONDARY + β7 UPPER SECONDARY + β8 ADVANCED DEGREE + β9 BACHELOR DEGREE+ β10 MASTER DEGREE + β11 UNPREDICTABLE INDUSTRY+ ε
This equation explains to what extent structural contingencies influence the level of work autonomy and whether or not the variables are statistically significant.
Where:
Work autonomy: is the dependent variable
β0, β1, β2, β3: are the coefficients influencing the level of autonomy. The value of the regression coefficient states the amount that the dependent variable varies, with each unit change of the associated independent variable (Blumberg, Cooper and Schindler, 2008).
SIZE, INDUSTRY, etcetera: are the independent variables.
ε: the error variable, accounting for all the variables, measurable and immeasurable, that are not part of the model.
2. Work autonomy = β0 + β1INDIVIDUALISM + β2UNCERTAINTY AVOIDANCE + β3POWER DISTANCE + β4COMMON LAW+ β5SOCIALST LAW+ ε
This equation explains to what extent institutional contingencies influence the level of work autonomy and whether or not the variables are statistically significant.
Work autonomy = β0 + β1 SIZE<10 + β2 SIZE10-24+ β3 SIZE25-99 + β4 SIZE100-499 + β5 LOWER SECONDARY + β6 LOWER UPPER SECONDARY + β7 UPPER SECONDARY + β8 ADVANCED DEGREE + β9 BACHELOR DEGREE+ β10 MASTER DEGREE + β11 UNPREDICTABLE INDUSTRY+ β12INDVIDUALISM + β13UNCERTAINTY AVOIDANCE + β14POWER DISTANCE + β15COMMON LAW + β16CIVIL LAW+ ε
36 4. RESULTS
In the following section the results of the data analysis will be presented. First, the descriptive statistics will be addressed. After that the results of the multiple regression analysis will be presented and analyzed, followed up by the results from the extended multiple regression analysis.
4.1. Descriptive Statistics
Table 2 gives an overview of the descriptive statistics. As mentioned before the analysis was carried out over 139486 individuals of which 122577 individuals experienced some degree of work autonomy. Approximately 49% of the respondents is men and 51% women, which gives an equal distribution on gender. The average age is 47 years.
Looking at the minimum and maximum values, it can be concluded that all variables have a valid value. This confirms the accuracy of the data. The average level of individualism, uncertainty avoidance and power distance are respectively 62.55, 66.64 and 45.47.
Many of the variables have been operationalized through a dummy variable, for example, organizational size, education level, legal origin, industry and employee relation. Their means multiplied by hundred, gives the percentage of people who worked for a particular organizational size, completed a particular level of education as highest degree, belong to a particular legal system or unpredictable industry.
Moreover, as mentioned before the multiple regression analysis assumes that the variables included in the analysis satisfy the assumptions of normality, linearity and homoscedasticity. I performed one strategy for checking the conformity of the analysis, namely a post-analysis. In order to give a graphical impression of the models fit, I employed a histogram of residuals as well as a scatterplot, which checks for linearity and homoscedasticity. Residuals reflect the errors or variability that the regression analysis was unable to predict accurately (University of Texas, 2003). Figure A1 in the appendix reflects the residuals histogram, which is non-normal. Figure A2 in the appendix represents the scatterplot of residuals. The variance of the error variable is required to be constant, which means that no apparent change in the residuals should be visible (Keller, 2009). Unfortunately the error variable is not constant which means that some variables do not conform the linearity and homoscedasticity assumption.
37 three different ways, namely; logarithmic, square root and an inverse transformation. However, these transformations did not have any influence. The inability to satisfy the assumptions of the multiple regression analysis does not mean that the results from analysis are invalid, however they are less strong.
TABLE 2:
Minimum, Maximum, Mean and Standard Deviation
Variables Min Max Mean & St. Dev.
1. SIZE < 10 0 1 0.3441 (0.4750 ) 2. SIZE 10-24 0 1 0.1807 (0.3847 ) 3. SIZE 25-99 0 1 0.2131 (0.4095) 4. SIZE 100-499 0 1 0.1497 (0.3567) 5. LOW SECONDARY 0 1 0.0986 (0.2981 ) 6. LOW UPPER SECONDARY 0 1 0.1564 (0.3632) 7. UPPER SECONDARY 0 1 0.1109 (0.3140 ) 8. ADVANCED DEGREE 0 1 0.0556 (0.2292 ) 9. BACHELOR DEGREE 0 1 0.0608 (0.2389 ) 10. MASTER DEGREE 0 1 0.0638 (0.2444 ) 11. UNPREDICTABLE INDUSTRY 0 1 0.4385 (0.4962 ) 12. INDIVIDUALISM 00.27 0.089 0.6255 (0.1533) 13. UNCERTAINTY AVOIDANCE 00.23 01.12 0.6664 (0.2398) 14. POWER DISTANCE 00.11 01.04 0.4547 (0.2113) 15. COMMON LAW 0 1 0.1300 (0.3400 ) 16. SOCIALIST LAW 0 1 0.2000 (0.4020 ) 17. AGE 00.13 01.23 0.4766 (0.1730 ) 18. GENDER 0 1 0.4900 (0.5000) 19. SELF-EMPLOYED 0 1 0.1178 (0.3223) 20. FAMILY BUSINESS 0 1 0.0142 (0.1183)
38 4.2. Results Multiple Regression Analysis
In each model I report a standardized regression coefficient along with its t-value in parentheses. I quoted the regression coefficients as standardized coefficients, because in this way a standard unit of analysis is being used for all variables which make comparison easier. The standardized coefficients are called beta weights (β) and their values indicate relative importance of the associated independent variables (Blumberg, Cooper and Schindler, 2008). Table 3 exhibits the results of the multiple regression models deriving from the structural contingency theory. The models are designed to test the hypotheses formulated previously. Model four is the full model.
The first and fourth model are used to test hypothesis 1. A negative and significant relationship has been found in model one and four for most organizational sizes except for those who had less than 10 employees employed. This negative correlation supports hypothesis 1 which expects a decrease in the level of autonomy available to employees when organizational size increases.
The second and fourth model are used to test hypothesis 2. A negative and significant relationship has been found for those individuals whose highest degree of education was lower secondary until upper secondary. However, an upward trend is analyzed and a positive and significant relationship is found for those individuals whose highest degree of education completed is advanced or higher. This positive correlation supports hypothesis 2, which expects an increase in the level of autonomy when the level of education increases.
39 TABLE 3: Results Structural Contingencies
Variables Model 1 Model 2 Model 3 Model 4
SIZE < 10 0.104*** 0.139*** (22.666) (30.984) SIZE 10-24 -0.026*** -0.009*** (-6.198) (-2.213) SIZE 25-99 -0.018*** -0.014*** (-4.093) (-3.434) SIZE 99-499 -0.022*** -0.014*** (-5.563) (-3.497) LOW SECONDARY -0.106*** -0.097*** (-36.156) (-33.320) LOW UPPER SECONDARY -0.068*** -0.055***
(-22.779) (-18.696) UPPER SECONDARY -0.023*** -0.017*** (-7.893) (-5.923) ADVANCED DEGREE 0.037*** 0.041*** (12.722) (14.374) BACHELOR DEGREE 0.091*** 0.088*** (31.310) (30.391) MASTER DEGREE 0.097*** 0.095*** (33.439) (32.591) UNPREDICTABLE INDUSTRY 0.131*** 0.105*** (46.377) (36.848) R2 0.016 0.039 0.017 0.069
