The influence of hierarchy in organizations on employee turnover intentions

(1)

The Influence of Hierarchy in

Organizations on Employee

Turnover Intentions

Bas Lensen (10018972)

Master Thesis Business Economics

Specialization: Organization Economics

Date: January 8, 2016

Supervisor: Thomas Buser

(2)

1

Abstract

I used a vignette study to examine the effect of hierarchy on one’s turnover intentions. Among 178 respondents I found that employees’ turnover intentions are lower when senior colleagues do not act as if they are above others in the hierarchy and when the general workplace atmosphere is informal. I call this informal hierarchy. The deployment of a flat/non-hierarchical organization, characterized by few organizational layers and lines of command, leads to higher expectations about job satisfaction and is perceived to be more attractive. However, this does not lead to lower turnover intentions. Two job aspects following from having a flat structure, higher job complexity and more empowerment do not show a significant effect. These results suggest that people are less likely to quit firms that are not hierarchical.

1. Introduction

One of the major challenges for a firm is retaining their talented employees. According to research by the Economist, the war for talent is getting harder. Businesses have increasing difficulties in recruiting and retaining skilled employees. Many executives indicated that employee recruitment and retention would become tougher (66%) or even ‘significantly harder’ (20%) (Murray, 2008). In 2012, four years and an economic crisis later, the war for talent was predicted to remain intense even though unemployment increased significantly (Bishop, 2011). The term “war for talent” was first used by researchers at McKinsey (Chambers et al, 1998). They argue that it is important to shape an environment that is attractive for talented employees to work in. According to them:

‘’You can win the war for talent, but first you must elevate talent management to a burning corporate priority. Then, to attract and retain the people you need, you must create and perpetually refine an employee value proposition: senior management's answer to why a smart, energetic, ambitious individual would want to come and work with you rather than with the team next door. That done, you must turn your attention to how you are going to recruit great talent, and finally develop, develop, develop!’’ (p. 44)

There are loads of angles for firms in creating the value proposition as mentioned by the McKinsey research. For instance a firm could look at the (financial) incentives, on-the-job training, and growth opportunities they offer their talented employees. These are just examples of many factors in the value proposition that can be offered to employees to attract them, and to make sure that they want to keep working for your organization. Therefore,

(4)

3

research that tries to get insights in what elements people are attracted by in a job and what influences people’s decision to quit their job is of crucial importance for organizations.

Since merely attracting the right people does not mean much if you are not able to retain them, this thesis will focus on employees’ decision to quit working for a given firm.

One of the factors that might be a predictor of employees’ quit behavior, is whether their organization is a hierarchical one. Since hierarchy can exist formally, characterized by the amount of layers in the organizational chart, or informally, by recognizing that some people are above you and respecting them and their decisions, the following research question will be answered:

To what extent does an organization’s level of hierarchy, both formal and informal, influence an employee’s intention to quit?

A lot of research on quit behavior of employees has already been done. In an early paper on the matter, Weiss (1984) finds that tenure, education, gender, race, and job complexity influence the likelihood that an individual will quit his job. Other authors found evidence that general training, work group cohesion, pay, alternative pay, and bonus structure influence quit behavior (Mueller & Price, 1990; Brewer, 1996; Blakemore et al, 1987).

One of the issues that is discussed quite often is what kind of organizational structure firms should employ. According to an article on the Forbes Magazine website, although still relatively rare, flat organizations are gaining in popularity (Wirtham, 2014). Some potential benefits of a flat organizational structure have been discussed. In fact, there is a growing amount of evidence that flat organizations outperform hierarchical organizations (Kastelle, 2013). Quinn (1985) argues, among others, that motivated employees are crucial to the success of a project, and that large bureaucratic organizations often suffer from motivation problems. Bureaucracy is more likely to occur when an organization consists of many layers and lines to top-management are long. Therefore, to the extent that the motivational issues lead to employees leaving the firm, there may be an effect of organizational hierarchy on employee turnover.

Some job aspects associated with flatter structures and their effect on voluntary quits have been examined: for instance, empowerment, responsibility and job complexity (Weiss, 1984; Dow Scott et al, 2003).

(5)

4

Formal hierarchy is likely a factor in employees’ work-related decisions. However, an organization’s structure might not be the only way in which an organization can be hierarchical. An organization can be quite flat on paper, but still require employees to recognize their superiors as people well above them in hierarchy or treat people with higher seniority as if they stand above them in hierarchy. In other words, hierarchy can exist not only on paper, but it can also be present in the general atmosphere at the workplace. This is only partially recognized by the current literature. Perhaps it is not merely the presence of hierarchy by organizational design that is relevant for an employee’s quit intentions, but also his perception of the presence or absence of superiority of certain individuals. A research that examines the overall effect of hierarchy, both through organizational structure and organizational formality, on quit behavior is lacking.

Since data on quit behavior is scarcely available and organizational structure is hard to measure, I will focus on the intention to quit and I will use a vignette study to measure this. Researchers already argued that intention to quit is a good predictor of actual quit behavior (Firth et al, 2004) and some empirical support for this has been offered (Mobley et al, 1978; Miller et al, 1979).

This study shows that individuals are less likely to quit if their boss or other senior employees treat them as equals and the general workplace atmosphere is informal. Suggesting that subjects care more about whether they are treated as if they are lower in the hierarchy than whether or not they actually are. Furthermore, there is a weakly significant effect of the presence of hierarchy by design on how attractive individuals perceive their job to be, and on how satisfied they expect to be from the job. Finally, two job aspects associated with organizational hierarchy were examined: centralization/lack of authority and the job complexity that stems from the fact that the individual performs various tasks that sometimes conflict with other people’s tasks. No effect of these job aspects on intention to quit, job satisfaction and job attractiveness was found.

This paper is structured as follows. The next section provides an overview of the literature related to this thesis’ research question. This is followed by a discussion of the methodology. After that, the results will be presented. Then, a discussion of the results follows. The last section will draw conclusions, address limitations and give recommendations for future research.

(6)

5

2. Related Literature

This chapter reviews the existing literature. Papers relevant to this thesis’ research question can be split in three categories: literature on organizational structure related to hierarchy, literature on organizational formality (atmosphere, social ties, attitudes towards colleagues), and literature on voluntary employee turnover. I will treat each category separately. After the theoretical part, I will review the existing empirical work in all of these fields.

2.1 Organizational Structure

The first dimension across which hierarchy can exists, is through the organizational structure of a firm.

James and Jones (1976) look at various dimensions of organizational structure and propose seven dimensions of organizational structure. The following three dimensions will be examined in this thesis:

I. Configuration - "shape" of the role structure including subordinate ratios (span of control), height of workflow hierarchy, and various percentage measures including percent of direct workers and nonworkflow personnel.

This indicates whether the organization is organized hierarchically or not. Generally an organization that consists of many management layers and narrow span of control is considered a ‘tall’ or ‘hierarchical’ organization, while an organization that has few management layers and broad span of control is called a ‘flat’ organization (Ghiselli & Siegel, 1972).

This ‘flat’ versus ‘tall’ dimension of hierarchy is the first dimension that will be examined in this thesis. This can be seen as the formal presence of hierarchy by definition. The effect of the two organizations on employees’ intention to quit can be positive or negative. On the one hand, it is possible that an employee will get a dissatisfactory feeling from having a lot of superiors working above him or her. On the other hand, an ambitious and talented employee might also like the idea of being able to grow to the top of the firm.

Clark et al (2009), examine the effect of an employee’s wage relative to his colleagues’ on his or her job satisfaction. They differentiate between two aspects of income inequality with

(7)

6

respect to co-workers: a status effect and a signaling effect. The idea behind the status effect is that negative externalities arise from an individual earning or consuming less than others in his or her reference group. In an organizational context, earning less than your colleagues gives you lower job satisfaction. However, also positive externalities could arise from earning less than your co-workers. This is labeled the signaling effect: the idea that co-workers earning a higher wage serves as a signal of possible higher future earnings for the individual. The authors develop a model in which an individual’s utility depends on his own earnings and on his co-workers, where they allow for both a positive and a negative effect of members of the reference group’s earnings.

Clark et al (2009) focus on wage, but it is possible that having many people above you in the hierarchy serves the same function. Being at the very bottom of the organization might impact an individual’s confidence, jealousy, and dissatisfactory feelings. As with relative wages, being not at the top of an organization with many layers might also contain a signaling effect. By definition, a more hierarchical organization means more positions on a higher level an individual can grow into. If an employee is confident about his abilities and about promotion opportunities within the organization, having a hierarchical organizational structure can serve as a signal of possible future promotions. Blau (1968) finds that taller organizations have more explicit promotion regulations.

In Lazear and Rosen’s (1981) tournament model, employees in an organization are considered competition for promotion. The best-performing worker receives a price. Contestants exert more effort as the difference between winners’ and losers’ earnings is larger. High earnings at the top of the hierarchy serve as incentives for workers at lower job levels. Dissatisfaction with promotion opportunities have been shown to make individuals more likely to quit (Shields & Ward, 2001). Therefore, promotion opportunities are likely important for respondents.

II. Specialization- division of labor according to functional specialization.

More hierarchical firms usually have more division of labor (Marengo & Dosi, 2005). If the lines between low or high levels become shorter or disappear altogether, it might become more blurry who performs which task. As a consequence, less hierarchy might make one’s job more complex.

(8)

7

There is a relation between hierarchy and task complexity. Ouchi and Maguire (1975) show that people with higher positions in the organization perceive their tasks to be more complex, routine tasks to be less prevalent, and their tasks to be more interdependent on other employees. Ilgen & Hollenbeck (1991) argue that more complex jobs exists at higher levels in the hierarchy. Therefore, to the extent that a flatter organization means that some managerial tasks and responsibilities are placed to people lower at the organization, flatter organizations will have more complex jobs for non-managerial staff.

According to Marengo and Dosi (2005), the process of division of labor takes usually place in hierarchical firms. Furthermore, less hierarchical organizations tent to have less well defined jobs (Murphy & Jackson, 1999). This implies that the division of labor that is more prevalent in hierarchical firms, leads to employees having to focus on fewer tasks and that the boundaries of their job are well defined.

As to whether employees value working on one specific set of tasks, there are different views. On the one hand, researchers have claimed that individuals value enrichened jobs. There is a large amount of articles that claim that individuals’ job satisfaction is negatively related to job specification and positively to job enlargement (Shepard, 1970). However, as boundaries between responsibilities become blurrier, role ambiguity and conflict might increase, since flat organizations have fewer managerial posts, which increases competition between co-workers (Baillien & De Witte, 2009), and since flat organizations take more time to resolve conflicts and coordinate effort (Carzo & Yanouzas, 1969). There is a vast body of research on the effect of role ambiguity and role conflict on job satisfaction, commitment, and performance (Tubre & Collins, 2000). Tubre & Collins find a significant negative relationship between role ambiguity and job satisfaction and a minor negative relationship between role conflict and job satisfaction. Role ambiguity is furthermore more prevalent in higher levels of the organizational hierarchy (Singh & Rhoads, 1991).

The second tested effect of hierarchy on intention to quit, is mediated by the job complexity that arises from having a wide range of tasks that sometimes conflict with other employees’ tasks.

III. Centralization of decision making and authority.

Baker et al (1999) argue that authority is the defining feature of hierarchy. According to Fredrickson (1986), centralization refers to the degree to which the right to make decisions is

(9)

8

concentrated. Higher levels of centralization make it easier to coordinate organizational decision making. However, the downsides are that it places large cognitive demands on the managers with authority (Fredrickson, 1986) and decreases agents’ effort due to lower payoffs (Aghion & Tirole, 1997) and lower motivation (Zábojník, 2002). Zábojník (2002) develops a model based on Aghion & Tirole’s (1997) in which the employee works on his own idea and therefore is more optimistic about the outcome. This will increase his incentive to exert more effort. The motivational effects of decentralization are also discussed by Sliwka (2001). He argues that the ability to make decisions improves job satisfaction, which will improve motivation. Giving more discretion to employees on how to do their job leads to higher motivation and better organizational performance. Child (1984), states: ‘most people are

willing to give more to their jobs when they have a high degree of individual freedom, discretion and control over their work’ (p. 148).

Caroli and Van Reenen (2001) also argue for a positive effect of more empowerment on job satisfaction. According to them, delegation of responsibility means that employees are more involved, share more information and that lower level staff participates more. If this leads to job enrichment, more delegation might lead to higher job satisfaction. But, they also suggest an opposite effect of decentralization on job satisfaction. If the increase of responsibility that comes with decentralization rises employees’ stress levels, this may lead to lower job satisfaction.

Besides these three points, James and Jones (1976) also propose the following four points, followed by the relation to hierarchy.

IV. Formalization- the extent to which rules, procedures, instructions, and communications are written and the degree to which roles are defined.

This aspect relates to how hierarchy is formalized by rules and procedures. A hierarchical organization might be characterized by certain rules (e.g. regarding how to come to a decision), procedures (e.g. how one should contact his or her superior face to face vs. email), and communications (e.g. addressing superiors with sir/madam instead of using his or her first name).

(10)

9

Related to hierarchy, since larger organizations usually have more hierarchical levels (Blau, 1972).

VI. Standardization- degree to which procedures (e.g., selection, advancement, workflow control) are standardized.

Especially standardization of procedures related to promotion of an employee and workflows regarding approval of plans are a promising area for how hierarchy might impact intention to quit. More hierarchy means more positions to be promoted to, but if there is an ambiguous and non-standardized promotion process this means less. This relates very much to the first dimension of hierarchy that will be examined. In this thesis, promotion opportunities were assumed to be present in the tall organization only.

VII. Interdependence of Organizational Components- degree of task interdependence and autonomy with respect to intraorganizational functions.

If less hierarchy leads to less division of labor, and more ambiguity about responsibilities, then one’s tasks performance are likely to depend largely on other people.

Altogether, looking merely at the structure of the organization, I propose the following channels through which hierarchy might influence voluntary employee turnover: general presence of hierarchy by design, job complexity that arises from having many tasks that can also be seen as other employees’ tasks, and decentralization/empowerment.

2.2 Organizational Formality

Morand (1995) refers to informality as ‘social situations or gatherings that are

generally characterized by behavioral spontaneity, casualness, and interpersonal familiarity’.

As opposed to: ‘situations and social relations that are more regimented, deliberate, and

impersonal in nature’ for the term formality. More concretely, he sketches two kinds of

workplaces. Formal organizations are those that have groups of individuals wearing business suits, sitting upright in chairs, seated in a symmetrical arrangement around a table, exhibiting sober facial expressions and paying concerted attention to the subject matter. In contrast, a looser meeting of a group of people, people with more relax posture, drinking soda out of cans, and less formal ways of addressing each other is called informal.

(11)

10

Reif et al (1973) propose eight concepts that represent an informal organization: 1) Voluntary teamwork, 2) Clique, 3) Personal Influence, 4) Co-worker evaluation, 5) Social interaction, 6) Group cohesion, 7) Social group membership, and 8) Grapevine.

These are not hard-guided rules by which one can define (in)formal organizations, but the use of vignettes, as will be explained in the methodology section, allows for capturing a sense of hierarchy that arises from interactions between superiors and subordinates and a formal atmosphere.

One thing that is highly recognized by the literature is that social connections and informal relationships in the workplace matter for employees’ job satisfaction. Roy (1959) found evidence that supports the hypothesis that informal interaction between workers has an influence on job satisfaction. Dur and Sol (2010) argue that social interaction with colleagues is one of the most important factors in determining job satisfaction. This view is recognized by managers as most state that they strongly encourage friendships in the workplace (Berman et al., 2002) and many firms organize drinks and social events to promote social connections in the workplace (Cohen & Prusak, 2001). Related to hierarchy social connections are mainly relevant with respect to an employee’s relation with his superiors. How approachable your colleagues are, including those who have a higher position than you, might be relevant for an employee’s intention to quit.

Therefore, the final aspect of hierarchy is the feeling that arises from the interaction between an employee and his superiors, and whether the workplace atmosphere is considered (in)formal.

2.3 Voluntary Employee Turnover

Weiss (1984) develops a model with channels through which variables can influence one’s decision to quit. According to this model, there are three categories through which factors that possibly influence quit behavior are channeled. These are: the employee’s outside option, job satisfaction at the current job, and the cost of quitting (i.e. the general disutility of quitting a job). Weiss’ model assumes that an individual will only quit his job if the utility he or she expects to get from working somewhere else or being unemployed minus the general disutility of quitting a job is higher than his or her expected utility at the present job. Formally worker i quits a job if:

(12)

11

𝑉(𝐴𝑂_𝑖) > 𝑉(𝑃𝐽_𝑖) + 𝐶_𝑖 (1)

Where 𝑉(𝐴𝑂𝑖) is the worker’s expected lifetime utility if he decides to take his best alternative

option, 𝑉(𝑃𝐽𝑖) is the worker’s expected lifetime utility if he decides to keep working at the

current firm and 𝐶𝑖 the worker’s disutility of quitting a job.

The most likely channel through which the four factors that this thesis examines can influence one’s intention to quit is: job satisfaction. In fact, the focus has been on this channel throughout this paper so far

Neither of these aspects have an influence on one’s outside option, since they are all related to one’s current job rather than characteristics that increase or decrease one’s outside option. Furthermore, the individual’s outside option is controlled for by mentioning alternative jobs in the vignette study and by asking respondents about their gender, age, and employment status.

Regarding the cost of quitting, it is possible that individuals find it more difficult to quit a job when they are closely connected to their co-workers and superior. The presence of informality can thus also have an effect through the cost of quitting channel as mentioned by Weiss (1984). For the three other determinants this channel is less likely.

Weiss’ (1984) model is mainly an economical one, in which employees make an elaborate cost-benefit analysis for their lifetime/future utility. Of course this might not be a realistic assumption. Many factors are likely to play a role in one’s decision to quit working for a firm. Psychologists and sociologists have also tried to build a theoretical model that might explain an employees’ intention to quit. Psychologists’ view is that employees enter an organizations expecting certain things about its operations and how they are treated. It is the extent to which these expectations are (not) met, that influences an employee’s job satisfaction and organizational commitment. This will lead the employee to leave the firm or stay at the firm (Mueller & Price, 1990). Sociologists also view job satisfaction and commitment as important variables that intervene between structural conditions and the decision to leave (Mueller & Price, 1990).

Hence, if an individual comes into a job expecting to have a lot of responsibility and to be actively involved in the decision making process, he or she might be disappointed that his expectation is not met if he or she is not involved in the decision making process at all. Consequently this might be a reason for him or her to leave. The same reasoning applies to the other factors.

(13)

12

2.4 Empirical Work

An early empirical contribution was made by Weiss (1984). Weiss used a sample of 2431 semiskilled manufacturing workers at two manufacturing firms that were all hired in two specific years. He matches these workers’ personal and job characteristics with whether they were still employed at the firm and, if not, at what time they quit. He finds that: workers who quit a job to start working at these firms were less likely to quit than workers who were unemployed, workers that work on more complex tasks are more likely to quit than those that work on easier tasks, the older the worker the less likely he is to quit, better-educated workers are less likely to quit than their less educated counterparts, white employees are more likely to quit than black employees, and that female employees are less likely to quit than male employees. By looking at actual quit behavior of employees that differ only in certain aspects, there is high external validity in his paper.

One of the aspects that this thesis and Weiss’ paper have in common, is job complexity. For which he finds a positive impact on quit behavior. As he argues, this is against common belief. Folk wisdom, as Weiss calls it, argues that more complex jobs increase job satisfaction. Among psychologists, there has been a debate about job enrichment programs.

One might expect though that employees perceive more complex jobs to be more attractive beforehand. When asked directly, it is possible that employees perceive that they prefer to work on complicated tasks. Thus, perhaps the results of this thesis will differ from Weiss’. Weiss measures job complexity by looking at the estimates made by engineers at the firm about how long it should take to learn the skills required at the job. This seems like a good measure, but perhaps it is not only the difficulty of the tasks the worker is required to do per se. Possibly employees have different preferences regarding the number of tasks they have to perform. Hence, there might be a difference between pure job complexity of having to do complex tasks and job complexity due to the fact that you need to do several different tasks. Clark et al (2009) merge data from eight waves of survey data from the Danish sample of the European Community Household Panel with administrative records. Clark et al use one question of this Household panel as their dependent variable, namely: “How satisfied are you with your work or other main activity?” The responses to this question are matched with data from the administrative Integrated Database for Labour Market Statistics. This dataset

(14)

13

contains identifiers for the individual’s earnings and the establishment where the individual works for all individuals aged 15 to 74. They consider other workers in the same firm as the reference group. Thus, they use data on job satisfaction, own earnings, and reference group’s earnings. When regressing an individual’s job satisfaction on his or her own earnings and average earnings within his or her establishment they find a positive effect for both own and establishment earnings on job satisfaction. They argue that the signaling effect of others’ earnings dominates the status effect. A similar approach to test whether hierarchy has a similar effect on job satisfaction as reference group wage would be problematic, since organizations do not report on how hierarchical they are. Therefore, Clark et al’s (2009) status and signaling effect of an organization’s hierarchical layers will be examined by using vignettes rather than real life data.

Mueller & Price (1990) examine voluntary employee turnover by analyzing data obtained from nurses at a university hospital. In their model, work conditions, personnel characteristics, and environmental conditions have an impact on voluntary employee turnover mediated by job satisfaction, organizational commitment, and intent to stay. Mueller & Price use intent to stay as a mediator, since they measure actual quit behavior. However, I will use this as a proxy for actual quit behavior. The positive influence from intention on actual quit behavior suggests that this is a fair proxy.

Morrisson (2004) argues that intention to leave is one of the best antecedents of actual employee turnover. She examines the impact of workplace friendships on job satisfaction and intention to quit. She finds that the presence of friendship and a friendly environment at the workplace lead to more job satisfaction, organizational commitment, and lower intention to quit. These findings are related to my proposed atmosphere of informality. If a workplace feels highly formal, and there is a general hierarchy that exists informally, by the way superiors treat subordinates, it may very well be perceived as less friendly. Morrison uses questionnaires to test her hypotheses. Nurses at a hospital in New-Zeeland were asked questions related to friendship in the workplace, workgroup cohesion, organizational commitment, job satisfaction, and intention to leave.

This approach was also used by Riordan and Griffeth (1995). In a small electric utility firm they asked employees about their friendship opportunities, job satisfaction, and intention to leave the firm. Among others, Riordan and Griffeth find that the presence of friendship opportunities has a positive effect on job satisfaction and a negative effect on employee

(15)

14

turnover. In their conclusion they argue that: ‘’These findings suggest that further exploration

of informal relations within organizations, such as friendships, may be productive for increasing our understanding of the dynamics of organizational behavior’’(p. 150).Therefore, a next step is to see to what the degree the formality of employee-boss relations matters for organizational outcomes.

Morrisson (2004), and Riordan and Griffeth (1995), focus on employee-employee relations rather than boss-employee relations. The relation between an employee and his or her superior has a clear relation to how hierarchical a workplace is. It is likely that an employee values a good relationship with his boss in which he is valued to be equal. Therefore, there might be an effect on how one is treated by his superior, and how he should address him, on the employees’ intention to quit. The current literature on social workplace connections has a relation to this thesis and helps to construct the vignettes used for this thesis. New insights arise with respect to the dynamics of the boss-employee relation and its effect on employee’s job satisfaction and turnover intentions.

Reif et al (1973) asked respondents to rate the value of eight concepts related to the formal organization and eight concepts of the informal organization. Among 341 managers and white-collar employees of various organizations, they found that individuals perceive the formal organization to be more valuable in satisfying their needs. Therefore, it is possible that there is no effect of having an informal atmosphere at the workplace on intention to quit in this thesis' study.

Fehr et al (2013) conduct an experiment in which some subjects are principals and others are agents. Each principal is matched with an agent. The principal chooses to keep or delegate authority and both principal and agent choose their effort levels. One of their results is that the party that does not have control provides substantially less effort than the Nash equilibrium would predict. In fact, a large fraction of them provide zero effort even though this is never optimal in the experimental setting. The authors argue that this phenomenon can be explained by a demotivating effect of a lack of authority.

Dominguez-Martinez et al (2014) find a similar result in their experimental setting. In this case the worker does not choose an effort level, but whether to look for information about certain projects or not. Also the principal first decides whether he monitors or does not monitor the agent, then he or she observes whether the agent collects information and decides to follow the agent’s recommendation or not. The authors find that monitoring

(16)

15

significantly discourages information gathering when the interests of the principal and agent are weakly aligned, whereas it encourages information gathering when the interests are fully aligned. Regarding the discouraging effect under weak alignment, the authors argue that the boss’ decision to exert control via monitoring has a negative effect on the worker’s performance through a loss of feelings of autonomy.

Dow Scott et al (2003) examine the effect of participation in the decision-making process using a questionnaire. They find that participation in decision-making reduces one’s intention to quit. They asked employees at a US-invested company in China about, among others, their job satisfaction, intention to quit, and participation in the decision-making process. In their research the effect of participation in the decision-making process on intention to leave is mediated by job satisfaction. However, Bluedorn (1982) does not find this effect. In his paper, centralization was related to neither job satisfaction nor intend to leave. Bluedorn conducts a survey among the personnel of in the operations division of a large insurance firm. The population of the survey consisted almost exclusively of women, suggesting that perhaps women have less preference for authority than men.

Falk and Kosfeld (2006) use an experiment to demonstrate the hidden cost of control. They show that when the principal controls the agent, the agent exerts less effort and his performance is lower. In addition to this, the authors also present vignettes to their subjects. They present five workplace scenarios for which there are two conditions. One in which the principal does not control the worker and one in which he does. Then they asked each respondent to rate their work motivation. In all five scenarios work motivation was significantly reduced by control. I will make use of workplace scenarios as well, as will be discussed more elaborately in the methodology section.

2.5 Hypotheses

There are four hypothesis that follow from the literature discussed.

First, the effect of a hierarchical design of an organization, which exists from several layers, will be examined. Based on the fact that individuals value promotion opportunities (Shields & Ward, 2001), and on the fact that individuals view colleagues’ wages as a positive signal for own future wages (Clark et al, 2009), one might expect that people are more likely to quit flat organizations. However, Clark et al also state that the composition of the reference

(17)

16

group is important. It might be somewhat difficult for individuals to view their hypothetical colleagues above them in hierarchy as a reference group. Therefore, the signaling effect might be less prevalent in this thesis than in Clark et al’s. I suspect that any status effect will outweigh the signaling effect. Therefore, I hypothesize a positive effect of having an organization that is characterized by a lot of layers and hierarchy on intention to quit. That is, people will be less likely to quit flat organizations.

Hypothesis 1: The use of an organization that consists of a lot of layers and hierarchy levels has a positive effect on intention to quit.

Second, there are a lot of articles that claim a negative relation between job specialization and job satisfaction and a positive relation between job enlargement and job satisfaction (Shepard, 1970), indicating that people like to work on different tasks. However, Weiss (1984) finds a positive impact of job complexity on intention to quit. Furthermore, role conflict and ambiguity have been shown to negatively impact job satisfaction and organizational commitment (Tubre & Collins, 2000). Therefore, I hypothesize that as the range of tasks increases and responsibilities overlap, intention to quit increases.

Hypothesis 2: People are more likely to quit jobs that are characterized by a wide range of tasks and unclear boundaries between who performs which tasks.

Third, decentralization, delegating authority, and empowerment have been shown to increase employees’ motivation and job satisfaction (Dow Scott et al, 2003; Dominguez-Martinez et al, 2014; Fehr et al, 2013). Hence, the third hypothesis is that giving more responsibility and authority to employees lower in the organizational hierarchy makes an employee less eager to leave the organization.

Hypothesis 3: People are less likely to quit a job if they feel like they are involved in the decision-making process and that they have some authority and responsibility.

Finally, social workplace connections and informal relationships are considered as an important factor in job satisfaction (Dur & Sol, 2010). Research has already shown that social

(18)

17

ties and informal relationships relates to job satisfaction positively and intention to quit negatively (Roy, 1959; Morrison, 2004; Riordan & Griffeth, 1995). Therefore, I hypothesize that the boss’/manager’s ties with his or her subordinates and whether he positions him-or herself as an unapproachable superior leader have an impact on one’s decision to quit.

Hypothesis 4: People are less likely to quit a job when their boss is approachable and does not act like he is superior to his subordinates, and when the atmosphere at the workplace is informal.

3. Methodology

3.1 Method

These hypotheses will be answered by having respondents fill out a vignette questionnaire. According to Alexander and Becker (1978, p.94), vignettes are ‘’short descriptions of a person

or social situation that contain precise references to what are thought to be the most important factors in the decision-making or judgment-making processes of respondents”.

In Falk and Kosfeld (2006), each respondent gets a different version of a workplace scenario, after which they answer particular questions. Any significant differences between the responses for the different versions can only be attributed to the aspect that was different for both scenarios.

For this thesis, a hypothetical workplace was described five times. Each of the five versions differed in one aspect compared to the control version. Respondents were randomly shown one of the five scenarios. After this, they were asked to indicate on a five point Likert scale how attractive they perceived this job to be, how satisfied they would expect to be at such a job, and whether they would consider quitting this job likely. In the control version of the workplace description, the workplace was characterized by: 1) a hierarchical structure in organizational design, 2) specialized and not extremely complex job, 3) centralized decision making and low responsibility, and 4) a highly formal atmosphere and high formality in contact with superiors. Each of the four treatment versions replaced one of the aspects of this job with the opposite: 1) flat organizational structure, 2) wide range of tasks and larger more complex jobs with unclear boundaries, 3) involvement in the decision-making process and high responsibility, and 4) informal atmosphere with approachable superiors. Table 1 shows the

(19)

18

crucial paragraph of each treatment. Appendix A1 shows the full workplace description and Appendix A2 shows the questions.

Any differences between the treatments and the control scenario, must be caused by the small differences between them. Other aspects of the job and the firm are described as well, in order for respondents to not be able to figure out the purpose of the survey. Furthermore, some information about other firms that might want to hire them was mentioned in order to keep the outside option fixed. The outside option is further controlled for by asking respondents questions about their age, gender, education and employment status. All respondents were invited to read one of the scenarios, which was assigned randomly to them, and to fill out the questions after.

(20)

19 Table 1. Defining paragraph of each treatment.

Control Treatment

At your current firm, you report back to your supervisor, who in his turn has to report to his superior. There are various layers within the organization and a lot of people working above you. There is a clear hierarchical structure. If you are competent enough you might move up the ladder within the organization. This will require a lot of hard work and persistence, and you have to compete with others.

At this firm you have a supervisor, but lines to top-management are quite short. There are not that many people working above you since there are few layers. The structure is not hierarchical at all. Therefore, there are not many positions above you that you can grow into. (Treatment 1)

You have one clear task and you are expected to focus on this task. This task is quite demanding, but because it is the only task, overall job complexity is not extremely high.

You have a wide range of tasks that compete for your attention. This means that you will not focus on one task only but on multiple tasks, which can also be performed by your colleagues. Though these tasks are quite interesting, it makes your job a lot more complex. (Treatment 2)

Decisions are taken by your superiors. You get only limited say in this, of course your opinion will be taken into account, but in the end authority to make decisions lies with the management. Due to this, there is only limited responsibility.

You are actively involved in the decision-making process. You have some authority to make decisions that in your eyes are the best. Being the one that has authority to take decisions also means that you have quite a responsibility. (Treatment 3)

The atmosphere at your job is formal. You only know the co-workers who are working at the same level as you and your direct supervisor. If contact with management occurs at all, it is mainly through email and limited to comments/feedback about your assignments. It is appropriate to address more senior employees and employees with a higher position than you with sir or madam. You are expected to be highly professional. The main topics of conversation are work-related issues. Professionalism is an important value at this company.

The atmosphere at your job is fairly informal. You can just walk in your boss’ office to ask for advice and all co-workers are familiar to each other. You can easily offer ideas to co-workers whose job description is more senior than yours. You use first names to address people. At your job you and your co-workers, including your boss, discuss issues not related to the job. You know a great deal about their personal lives. You meet with them outside office hours on several occasions. (Treatment 4)

Notes: This table shows the crucial paragraph in which each treatment differs from the control version of the

workplace description. Only this paragraph was altered leaving the other paragraphs the same as in the control version.

This thesis measures individuals’ intention to quit, which is something different than actual quit behavior. However, according to several researchers, intention to quit is significantly related to actual quit behavior (Mobley et al, 1978; Miller et al, 1979; Firth et al, 2004).

(21)

20

There are limitations to this vignette approach. With measuring responses to a hypothetical workplace we lose external validity. The fact that intention to quit is related to actual quit behavior does not mean that an individual that indicates that he will quit a hypothetical job, would actually quit in real life. It is possible that respondents will have difficulties relating the hypothetical job to a real life setting. Furthermore, in a real life setting older people are more likely to have both higher positions and higher salaries. In this vignette study this was not necessarily the case. What was mentioned, however, was that the respondent received a market-conform wage. To the extent that older employees have a higher perception of market-conform wage for people their age, older respondents will assume that they earn a higher wage than younger respondents.

The use of vignettes also offers some benefits compared to other methods. A lot of research has been conducted by using questionnaires in which respondents answer

questions about aspects of their job and their job satisfaction and turnover intentions. The issue with this method is that it measures correlation rather than causality. Since one cannot exclude that if a person is already unsatisfied with his job and considers quitting, his

perceptions of other aspects of his job change. Hence, reverse causality cannot be excluded. The vignettes only change one aspect and thus all variation between the different versions has to be attributed to the difference between the versions. Atzmüller and Steiner (2010) argue that vignettes allow for causal investigations of respondents judgements.

Looking at actual quit behavior, like Weiss (1984) does, gives the highest external validity. However, with this method internal validity might be lower, since external factors might compromise the data. Furthermore, the specific variables of which this thesis aims to examine the effect are not all measurable in real life since they largely depend on individuals’ perception. The degree to which a firm is formal is hard to measure but can be captured by sketching a hypothetical workplace.

Hence, despite its limitations, a vignette study allows for measuring variables that otherwise would be hard to measure and allows for assigning causal interpretation to the effects. However, it is important to keep in mind that individuals answered questions about a hypothetical job. Therefore interferences about employees’ decisions in real life should be made with caution.

(22)

21

3.2 Sample

Via an online survey distributed through various channels (161 respondents) and some paper versions handed out in the train from Amsterdam to Brussels (19 respondents), a sample size of 180 respondents was obtained (36 for each version). A build in function in Qualtrics made sure that one of the versions was randomly shown to a person who clicked the link. Most respondents came from various online channels: Facebook, Email, LinkedIn, and WhatsApp. Five respondents filled out the survey on my personal computer. The offline versions handed out in the train were randomly given to passengers. I tried to give passengers who were clearly traveling together the same versions to prevent them from figuring out the differences between them. A tradeoff was made there, since one could suspect people traveling together as either friends or partners to attach greater value to social connections. However, only one couple filled out the Treatment 4 version and it seems safe to assume that most people value friendships or relationships in a real life setting. The largest amount of respondents is either Dutch (74%) or Belgian (13%). 52% of the respondents is female. Average age is 31 years old, ranging from 18 till 65. Out of these responses, 51% is employed and 41% are students. The education levels are: Bachelor degree (43%), Master’s degree (36%), MBO (12%), and high school (8%). The characteristics by treatment are depicted in Table 2. No significance differences between the five treatments were found. A χ2_{test cannot reject that}

gender, employment, education, and nationality are equal for each treatment. That is all p-values in the last column of table 2 are high. For age, an F-test was used, yielding insignificant results. This means that the randomization mechanism was successful.

The average responses to the three key questions are depicted in table 3.

(23)

22 Table 2. Background characteristics by Group

Variable Total Sample Control T1 T2 T3 T4 P-value

Gender (Woman) 51.7% 47.2% 44.4% 58.3% 50.0% 58.3% 0.670 Age 31.0 28.6 31.4 31.3 29.0 34.6 0.288 Employment 0.957 Studying 40.6% 41.7% 41.7% 38.9% 41.7% 38.9% Working 51.1% 50.0% 50.0% 50.0% 50.0% 55.6% Unemployed 6.1% 8.3% 5.6% 8.3% 2.8% 5.6% Other 2.2% 0.0% 2.8% 2.8% 5.6% 0.0% Education 0.341 High School 8.3% 5.6% 13.9% 5.6% 11.1% 5.6% MBO 11.7% 8.3% 8.3% 13.9% 8.3% 19.4% Bachelor’s Degree 43.4% 41.7% 33.3% 41.7% 41.7% 58.3% Master’s Degree 36.1% 44.4% 44.4% 36.1% 38.9% 16.7% Other 0.6% 0.0% 0.0% 2.8% 0.0% 0.0% Nationality 0.429 Dutch 73.9% 72.2% 66.7% 72.2% 72.2% 86.1% Belgian 13.3% 16.7% 22.2% 13.9% 5.6% 8.3% Other 12.8% 11.1% 11.1% 13.9% 22.2% 5.6% N 180 36 36 36 36 36

Notes: This table shows all background characteristics for the sample as a whole and per treatment. The P-values

in the right column indicate that these characteristics do not differ between treatments, i.e. randomization was successful.

Table 3. Means responses to Key questions

Variable Scale Mean Standard Deviation

Job Attractiveness 1(very unattractive )-5(very attractive) 3.344 0.999 Job Satisfaction 1(very dissatisfied)-5(very satisfied) 3.422 0.933 Intention to quit 1(very unlikely)-5(very likely) 2.905 1.018

Notes: This table shows the response to the three questions that form the dependent variables on a 5 point

(24)

23

4. Results

Figures 1-3 show the mean responses for intention to quit, job satisfaction, and job attractiveness, by treatment.

Figure 1 offers support for hypothesis 4. Respondents in the fourth treatment, significantly report a lower intention to quit. The effect is not only statistically significant, but also economically significant. The difference between the Control group and the Treatment 4 group, is more than one point on a five point Likert scale. This means that an organization in which formal hierarchy exists, but is not felt due to the fact that superiors do not act as such, will have a much lower quit rate among their employees. Furthermore, as shown in Figure 3, respondents undergoing Treatment 4 also significantly indicate that the job would make them more satisfied. The difference between means of the control group and the Treatment 4 group is 0.8 on a five point scale. The effect on perceived attractiveness of the job is more or less the same.

With multiple comparisons of means, the probability that one is significant just by chance is higher. That is there is an increased possibility of making a type 1 error. One way to correct for this is to divide the alpha by the number of comparisons, four in this case. This is called the Bonferroni method (Keller, 2009). To cope with the issue of multiple comparisons, also Bonferroni-corrected alphas were calculated. The results of Treatment 4 survive Bonferroni-corrected alphas (p-value<0.01/4=0.0025).

Figure 1 does not show a significant effect of the other treatments. Intention to quit is slightly lower for Treatment 1 and 2, Treatment 3 shows the same intention to quit as the control group. However, the differences are not significant. Therefore, the data does not show lower quit intention for flatter organizations, more complex jobs, and empowerment. Hence, hypothesis 3 cannot be confirmed. There is no significant difference between Treatment 1-3 and the control group in terms of job satisfaction and job attractiveness either.

(25)

24

Figure 1. Comparison of mean responses to Intention to Quit question

Notes: This graph shows that respondents in T4 report significantly lower Intention to quit than the control group.

Scale 1-5. Standard deviations: C=0.941, T1=1.000, T2=0.910, T3=0.996, T4=0.899. P-values generated from comparison of treatments versus control group: T1=0.387, T2=1.000, T3=0.710, T4=0.000. ***Significant at 1%.

Figure 2. Comparison of mean responses to Job Attractiveness question

Notes: This graph shows that respondents in T4 report significantly higher perceived job attractiveness than the

control group. Scale 1-5. Standard deviations: C=1.040, T1=0.871, T2=0.971, T3=0.937, T4=1.018. P-values generated from comparison of treatments versus control group: T1=0.146, T2=0.627, T3=0.369, T4=0.001. ***Significant at 1%. 3.17 2.97 3.17 3.08 2.14 Control T1 T2 T3 T4 3.06 3.39 3.17 3.25 3.86 Control T1 T2 T3 T4 *** ***

(26)

25

Figure 3. Comparison of mean responses to Job Satisfaction question

Notes: This graph shows that respondents report significantly higher expected job satisfaction than respondents

in the control group. Scale 1-5. Standard deviations: C=0.899, T1=0.994, T2=0.862, T3=0. 0.874, T4=0.941. P-values generated from comparison of treatments versus control group:: T1=0.248, T2=0.369, T3=0.0.200, T4=0.002. ***Significant at 1%.

Table 4 shows the OLS regressions for the three dependent variables on the treatments. The signs for Treatment 1, 3, and 4 are as expected, for Treatment 2 the signs are the opposite of what was hypothesized. Only Treatment 4 shows a significant effect on the three dependent variables. The coefficient of Treatment 4 on intention to quit is -1.028. This means that respondents in this group show 1.028 lower intention to quit on the five point scale. Likewise, respondents in Treatment 4 show 0.806 higher perceived job attractiveness and 0.694 higher expected job satisfaction compared to the control group. As with the comparison of means, the p-value of the coefficient of Treatment 4 was large enough to survive the Bonferroni correction, (p-value=0.000<0.0025). For all of the other treatments, no significant effect was found against nominal and Bonferroni-corrected alphas.

3.14 3.39 3.33 3.42

3.83

Control T1 T2 T3 T4

(27)

26

Table 4. OLS regressions with Intention to Quit, Job Attractiveness, and Job Satisfaction as

dependent variables.

(1) Intention to Quit (2) Job Attractiveness (3) Job Satisfaction Treatment 1 -0.194 (0.224) 0.333 (0.229) 0.250 (0.216) Treatment 2 -0.000 (0.224) 0.111 (0.229) 0.194 (0.216) Treatment 3 -0.083 (0.224) 0.194 (0.229) 0.278 (0.216) Treatment 4 -1.028*** (0.224) 0.806*** (0.229) 0.694*** (0.216) N 180 180 180 R-squared 0.148 0.079 0.060

Notes: The regression shows that there is a significant effect of Treatment 4 on all dependent variables.

Standard errors in parentheses. ***=Significant at 1%.

Table 5 shows the OLS regressions for the three dependent variables on the treatments and the control variables: gender, age, employment, and education. All treatments have the expected signs, except for Treatment 2. After inclusion of the control variables, magnitude of the effect for all treatments increased. For intention to quit the coefficient of Treatment 4 is -1.123, significant at the one percent level. This means that respondents in the fourth treatment have a 1.123 lower intention to quit on the five point scale compared to the control group. For job attractiveness and job satisfaction this is 1.001 and 0.834 respectively. Also significant at the one percent level, against both nominal alpha and Bonferroni-corrected alpha.

For the other three treatments the effect on intention to quit was not significant. Treatment 1 shows a positive effect on perceived job attractiveness, significant at ten percent and a positive effect on expected job satisfaction on the edge of being significant at the ten percent level (p-value=0.109). Individuals in Treatment 1 report 0.403 higher on job attractiveness and 0.336 higher on expected job satisfaction. This finding is economically significant, since on a five point scale, employees find the job 0.4 points more attractive and expect to be 0.34 points more satisfied. These findings offer some support in favor of hypothesis 1. Although subjects do not quit as a consequence of it, flat organizations might

(28)

27

have an effect on job attractiveness and job satisfaction. There is some indication that individuals have a preference for flatter organizations. However, the effects are only weakly significant and when applying Bonferroni-corrected alphas both effects were insignificant. Therefore, the evidence for an effect is weak. The coefficients of Treatment 2 and 3 are insignificant for both job satisfaction and job attractiveness.

The fourth regression in Table 5 shows that after inclusion of job satisfaction as independent variable in the regression with intention to quit as dependent variable, the effect of Treatment 4 is still significant, even against Bonferroni-corrected alpha. This indicates that the effect of Treatment 4 on intention to quit goes beyond job satisfaction. There is a direct effect of Treatment 4 that leads to lower intention to quit.

Of the control variables, age was significant for regressions 2 and 3 at the one percent level, and for regression 1 at the five percent level. Older people are more likely to quit. Older people furthermore expect to be less satisfied from the hypothetical job, and they perceive the job as less attractive. For each year of age, an individual reports 0.016 higher intention to quit, 0.022 lower perceived attractiveness, and 0.020 lower job satisfaction. When a Bonferroni correction was made, the effect on intention to quit remained significant only at the ten percent level (p-value<0.025). The effect on job attractiveness and satisfaction remained significant at the one percent level after correction (p-value<0.0025).

The final finding from table 2, is that men perceive the job to be more attractive than women. The effect is quite large, man report 0.303 higher perceived attractiveness than women, significant at 5%. There is also an effect on job satisfaction, this is nearly significant at 10%. The magnitude of this effect is 0.224 higher compared to women. Both results did not survive the Bonferroni correction.

(29)

28

Table 5. OLS regressions with Intention to Quit, Job Attractiveness, and Job Satisfaction as

dependent variables, including control variables.

(1) Intention to Quit (2) Job Attractiveness (3) Job Satisfaction (4) Intention to Quit Treatment 1 -0.253 (0.220) 0.403* (0.218) 0.336 (0.208) -0.006 (0.160) Treatment 2 -0.063 (0.218) 0.212 (0.217) 0.271 (0.207) 0.136 (0.158) Treatment 3 -0.114 (0.220) 0.236 (0.218) 0.275 (0.208) 0.088 (0.159) Treatment 4 -1.123*** (0.223) 1.001*** (0.222) 0.834*** (0.211) -0.511*** (0.168) Gender -0.159 (0.145) 0.303** (0.144) 0.224 (0.138) 0.005 (0.106) Age 0.016** (0.007) -0.022*** (0.007) -0.020*** (0.006) 0.001 (0.005) Employment 0.065 (0.129) -0.023 (0.128) 0.034 (0.122) 0.090 (0.093) Education 0.042 (0.077) 0.071 (0.076) -0.020 (0.073) 0.027 (0.056) Job Satisfaction - - - -0.734*** (0.059) N 178 178 178 178 R-squared 0.214 0.208 0.163 0.5929

Notes: Regressions 1-3 show that Treatment 4 and age have a significant effect on all dependent variables.

Gender and Treatment 1 have a significant effect on job attractiveness. Regression 4 shows that Treatment 4 remains significant after controlling for Job Satisfaction. Standard errors in parentheses. Gender: 1=Man, 0=Woman. *** Significant at 1%, ** Significant at 5%, * Significant at 10%.

A final set of regressions on the dependent variables was run with an interaction term for age and Treatment 1. The idea is that for younger individuals promotion opportunities are likely to be more important since they are at the very beginning of their career. Older respondents might see the signaling effect of possible promotions as less important and might focus on the status effect. Perhaps older individuals do not like the idea of being at the very bottom of an organization, since they already have years of experience. However, for all dependent variables no effect was found. These regressions can be found in Appendix A3.

(30)

29

5. Discussion

The results from the mean comparison and regressions are in favor of hypothesis 4. It is not possible to pinpoint exactly what aspect of the treatment appeals so much to the respondents, but it seems plausible that employees value to be treated as equal by their superiors. Not having to call him sir or madam, being able to approach and offer ideas to him or her, and being able to discuss issues that are not related to work, creates a feeling that the workplace is not so hierarchical all together, despite the many layers on paper. Hence, this thesis shows that an individual’s perception of being regarded as someone that is low in the hierarchy is more important than being positioned at a lower level in the organization per se. This is in line with previous research showing that social connections at the workplace are important (Morrison, 2004; Riordan & Griffeth, 1995).

The results show only marginal support in favor of hypothesis 1. No significant different intention to quit was found among subjects in this treatment. However, there is some (weak) evidence that subjects do perceive organizations that are characterized as flat and non-hierarchical as more attractive, and that they expect to be more satisfied working for such a firm. However, since these results were significant only at ten percent or close to being significant at ten percent, and given that they were not resistant against using Bonferroni-corrected alphas, we should not overlook the possibility that these results were generated by chance. These results have a couple of implications if they are actually true. First, although in this sample the higher job attractiveness and job satisfaction were not translated in lower intention to quit, this may be the case in a different setting since this is merely a hypothetical situation. Perhaps respondents react differently when they actually have to work in such an environment. Second, the higher perceived job attractiveness implies that flatter firms can more easily attract new employees than more hierarchical firms. Attracting employees is also a crucial activity for many organizations. Third, the results suggest that employees expect to be more satisfied working for the less hierarchical firm. Given that job satisfaction is positively related to organizational citizenship behavior (Organ & Ryan, 1995), negatively to absenteeism (Wegge et al, 2007), satisfied employees are important for organizations.

The fact that Treatment 4 remained significant after controlling for job satisfaction in a regression, suggests that people experience a general cost of quitting their job when they

(31)

30

are close to their colleagues. This means that the effect of an informal organization and good relations to their boss is not only mediated by job satisfaction.

This study does not show any support for the second hypothesis. Some previous research has claimed a positive impact of job enrichment (Shepard, 1970) and some research has shown a negative impact of job complexity (Weiss, 1984), and role conflict/ambiguity (Tubre & Collins, 2000) on job satisfaction. The job described in Treatment 2 touched upon all these aspects. It described an enrichened (the tasks were explicitly described as interesting), and complex job, and the potential for role conflict and ambiguity was mentioned as well. The fact that no effect was found might be caused by all these factors balancing each other out. Furthermore, this finding implies that it depends on what aspects of job complexity are highlighted whether or not there is a positive or negative effect. This potentially explains the differences in findings by previous papers.

No results in favor of the third hypothesis were found. This is not in line with previous research since, being able to impact the decision-making process, and being empowered to take decisions, have been shown to be important for employees’ job satisfaction and motivation (Falk & Kosfeld, 2006; Dominguez-Martinez et al, 2014; Dow Scott et al, 2003; Fehr et al, 2013). Caroli and van Reenen (2001) argue that the increased responsibility following from decentralization might lead to higher stress levels and lower job satisfaction. Since Treatment 3 explicitly mentioned that responsibility was quite high, the potential positive effects of being able to take decisions might be offset by a distaste for responsibility.

The fact that this study shows that people seem to care more about the informal aspect of the organization contradicts with Reif et al’s (1973) finding that people find the formal organization more important in satisfying their needs. However, as they indicate themselves, their finding was already contradicting previous research. This study offers evidence in favor of the importance of the informal organization.

One additional finding was the lower quit intentions among younger employees. This conflicts with Weiss’ (1984) finding that older employees are less likely to quit. It seems logical that from a certain age, individuals refrain from quitting any job, since they might expect to have greater difficulties in finding another job. A possible explanation is that individuals who are at the very beginning of their professional career might be highly satisfied that they have a job to begin with. Some of the subjects were still students during this research, so perhaps the idea of having a job appeals that much to them that they do not consider quitting. Another

(32)

31

possible explanation is that there are aspects of the job that appeal more to younger subjects than to older subjects. For instance, learning opportunities were present across all versions of the job description. It is possible that younger respondents value these learning opportunities more than older respondents do, since younger respondents are at the beginning of their career.

Finally, the fact that men perceive the job as more attractive and expect to be more satisfied from it, is something that conflicts with previous research. Various authors have found the opposite (Blanchflower & Oswald, 1999; Clark, 1996; 1997). Poza & Sousa-Poza (2000) found women are more satisfied with their job in some countries, but not in others, including the Netherlands. The fact that the vast majority of respondends to this thesis’ questionnaire were Dutch might explain why an opposite effect was found in this thesis. Furthermore, the evidence is not highly convincing given the low significance level.

6. Conclusion, Limitations and Future Research

This thesis aimed to examine the relationship between the level of hierarchy and an employee’s intention to quit working for a given firm. Based on the literature various aspects that come with hierarchy in a firm were identified and four hypotheses were tested. I found evidence that the informal way in which hierarchy exists in a firm has an impact on one’s decision to quit working for the organization. Specifically, when the atmosphere is informal, senior managers are easy to approach and treat you like they are equal to you, employees will be less likely to quit. Furthermore, the mere presence of less hierarchical layers and lines of command, has a positive impact on perceived job attractiveness and expected job satisfaction. However, this effect is not highly convincing since it was only marginally significant and not significant against Bonferroni-corrected alphas. Therefore, it should be interpreted with caution. Both findings indicate that the level of hierarchy is a factor in employees’ quitting intentions. This thesis does not find an impact on job complexity, arising from having to do many tasks overlapping with others’ responsibility, and of empowerment on turnover intentions.

There are policy implications that follow from these results. First, firms that see retaining their employees as top priority should look at the hierarchical structure of the firm, and more importantly should make sure that employees in the organization are treated as

The influence of hierarchy in organizations on employee turnover intentions