Work design, role demands, supportive work context : their relationship to Job Crafting and how this is moderated by proactive personality

Work design, role demands, supportive work context

Their relationship to Job Crafting and how this is moderated by

proactive personality

Bachelor Thesis

Bachelor in Economics & Business Specialization: Business Administration University of Amsterdam

Date of submission: 29-06-2016 Version: final version

Author: Nikki Tsiftis

Student ID 10591370 Word count:

Supervisor: mw. dr. C.T. (Corine) Boon Second supervisor: Ms E. Federici MSc

2

Statement of Originality

This document is written by Student Nikki Tsiftis who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

3

Abstract

The relationships between work characteristics and self-reported job crafting behavior, and the moderating role of proactive personality, were investigated in a cross sectional study of employees in The Netherlands (N = 126). The work characteristics included job autonomy, role overload and LMX, representing work design, role demands and supportive work context. Hierarchical multiple regression analysis has showed that job autonomy, role overload and LMX are positively associated with job crafting. Additional analyses showed that proactive personality did not positively moderate any of the hypothesized relationships (job autonomy – job crafting, role overload – job crafting, LMX – job crafting). However, a positive relationship between proactive personality and job crafting was found. The study suggests that job crafting can be influenced by certain job resources available to the employees and certain job demands, so it offers the organization an opportunity to influence job crafting behavior.

4

Table of Contents

Abstract ………. 2

1. Introduction ………... 5

2. Theoretical framework and hypotheses ………. 8

2.1 Job Crafting as dependent variable ……… 8

2.2 Work design, role demands and supportive work context as predictors of job crafting 9 2.2.1 Work design and job crafting …..………... 10

2.2.2 Role demands and job crafting .………... 10

2.2.3 Supportive work context and job crafting ….………... 11

2.3 Proactive personality as a moderator ………..……….. 12

2.4 Conceptual Model ………. 13

3. Methodology ………. 15

3.1 Research design ……….. 15

3.2 Sample and data collection ………. 16

3.3 Measures ………. 16 3.3.1 Job autonomy ……… 17 3.3.2 Role overload ………. 17 3.3.3 LMX ……….. 18 3.3.4 Proactive personality ……… 18 3.3.5 Job crafting ………. 18 3.3.6 Control variables ………. 18 3.4 Data analysis ……….. 18 3.4.1 Procedures ……….………. 18

3.4.2 Assumptions of the model ……..……… 19

4. Results ………. 21 4.1 Descriptive statistics ………. 21 4.2 Correlations ……… 21 4.3 Regression ……… 22 4.3.1 Direct relationships ……… 18 4.3.2 Conditional effects ………. 24

5. Discussion and Conclusion ……… 27

5.1 Analysis results and theoretical implications ……….……… 27

5.2 Limitations and directions for further research ……… 28

5.3 Practical implications ……….. 30

5.4 Conclusion ………... 31

Reference List ……… 32

5

Introduction

For decades, research about job design focused mostly on the top-down approach which entails that managers design the jobs for employees (Wrzesniewski, Berg & Dutton, 2010). Contrary, job crafting refers to a bottom-up approach where employees take initiative to proactively craft their own jobs and change task, relational and cognitive boundaries of their jobs (Wrzesniewski., 2010; Demerouti, 2014). So with job crafting employees take personal initiative in shaping their job characteristics to let their job better fit with their personal needs and abilities. This leads to higher engagement, higher job satisfaction and higher job performance which makes job crafting important for organizations to consider (Tims, Bakker & Derks, 2013; Bakker, Tims & Derks, 2012).

Therefore, it’s interesting to look at factors that predict job crafting in order to improve employee well-being and performances. For example, previous research shows that factors like person-job misfit can be a predictor of job crafting (Tims & Bakker, 2010). However, it’s reasonable to assume that there are more predictors of job crafting, which this paper is focused on. In this research, job autonomy, role overload and Leader Member Exchange (LMX) are the predictors of job crafting. Job autonomy and role overload are both related to the job, and LMX is related to the supervisor. These predictors give organizations the opportunity to influence these work characteristics at any time. Therefore, it is possible for organizations to stimulate job crafting (Tims & Bakker, 2010).

Job autonomy, role overload and LMX represent work design, role demands and supportive work context. These 3 constructs have been used before in a research conducted by Parker, Axtell, & Turner (2001), in which the importance of work design, role demands and supportive work context in designing a safer workplace, was measured. Job autonomy has been identified as one of the important features of work design for employee outcomes, like motivation and job satisfaction (Hackman & Oldham, 1980; Parker & Wall, 1998). Role overload has been found to affect important employee outcomes as well, such as well-being and commitment (Kahn, Wolfe, Quinn, Snoek & Rosenthal, 1964). LMX is chosen since it represents a supporting context within work is carried out through supportive supervision (Kahn et al., 1964). The next paragraphs explain the theoretical explanation for choosing these constructs in a job crafting study. By choosing these three constructs, a diverse group of predictors of job crafting can be analyzed by looking at theories from different fields (e.g. job demands-resources theory, LMX theory, conservation of resources theory). First of all, job autonomy is addressed as a predictor of job crafting. When autonomy is enhanced, employees report acquiring new skills and experiencing more responsibility for problems at work. Therefore, the opportunity to decide for oneself what and how to do the job may

6 be a predictor for job crafting (Tims et al., 2012). Wrzesniewski and Dutton (2001) found that certain job characteristics can encourage job crafting, like autonomy in the job for example, which may lead to more perceived opportunities for job crafting.

Secondly, role overload is hypothesized to have a positive relationship with job crafting. Traditionally, role overload is classified as a hindering job demand (Crawford et al., 2010; LePine et al., 2005), which would mean that employees may not undertake job crafting in role overload situations without sufficient facilitation (Solberg, 2016). However, research suggests that views and responses of employees towards role overload could be more complicated in nature (Eatough et al., 2011; Gilboa et al., 2008). Role overload is positively associated with job crafting, when employees have proactively taken on more tasks and responsibilities (Gilboa et al., 2008). On the other hand, job demands such as work overload, can lead to a person-job misfit (Tims et al., 2012), which in turn can lead to job crafting behavior. According to Tims & Bakker (2010) person-job misfit leads to job crafting behaviors.

Thirdly, LMX is expected to have a positive relationship with job crafting. LMX focuses on dyadic relationships between the leader and the follower where negotiation takes place (Graen & Uhl-Bien, 1995). LMX begins as transactional social exchange and evolves into transformational social exchange. Social exchange of psychological benefits exchanges, which can be trust, approval and support. LMX can enhance the willingness of employees to engage in job crafting, because when the LMX is positively perceived by the employee, there is a strong relationship, mutual learning and mutual trust (Graen and Uhl-Bien, 1995). This makes it for the employee less scary and more convenient to change different job characteristics when the manager supports its choices.

The aim of this paper is to investigate the impact of job autonomy, role overload and LMX on job crafting and if proactive personality could moderate these relationships. If proactive personality would moderate these relationships, it would imply that the impact of work design, role demands and supportive work context on job crafting is stronger for people who score high on proactive personality than for people who score lower on proactive personality. In this way, this paper would be a contribution to the existing literature in the field of job crafting.

Furthermore, studies show that a positive relationship between proactive personality and job crafting exists (Tims et al., 2012; Van Wingerden, Derks & Bakker, 2015). So proactive personality can be seen as a predictor of job crafting as well. However, proactive personality as a moderator between predictors of job crafting and job crafting is a new topic of interest in the field of job crafting. A study of Tims & Bakker (2010) showed that proactive personality positively moderates the relationship between person-job misfit and job crafting. So according to Tims et al. people who have a person-job misfit and

7 score high on proactive personality tend to engage more often in job crafting behavior than people who score lower on proactive personality. Therefore, it’s reasonable to assume that proactive personality could be a moderator in other relationships between predictors of job crafting and job crafting as well.

If it turns out job autonomy, role overload and LMX predict job crafting, and proactive personality is a moderator in the above mentioned relationships, it would imply practical relevance as well. These are all factors that can be influenced by the organization, which makes the study relevant for organizations that are willing to create contexts in which job crafting is more favored. Enhanced job crafting behavior, in turn, has a positive impact on employee well-being (Tims et al., 2013). According to their study, job crafting increases engagement, job satisfaction and decreases burnout.

8

2. Theoretical framework and hypotheses

2.1 Job crafting as dependent variable

According to Wrzesniewski, Berg & Dutton (2010), job crafting involves altering one or more of the following core aspects of work: tasks, relationships and/or perceptions. In this study, the dependent variable job crafting, is specified in these different forms, adding skill crafting as well (Bindl et al, 2014). First of all, enhancing or limiting task crafting means changing the boundaries of your job by taking on more or fewer tasks, expanding or diminishing their scope, or changing how they are performed. Secondly, enhancing relationship crafting means enhancing the nature or extent of your interactions with other people. Thirdly, enhancing or limiting cognitive crafting means changing how you think about the purpose of certain aspects of your job; or you can reframe the job as whole. Enhancing skill crafting means seeking out opportunities for extending the overall skills, trying to learn new things that went beyond your core skills and actively trying to develop wider capabilities in your job.

Wrzesniewski et al. mention the fact that: ‘Employees, at all levels, in all kinds of occupations, who try job crafting often end up more engaged and satisfied with their work lives, achieve higher levels of performance in their organizations, and report greater personal resilience’ (p. 115). These work-related benefits are associated with meaningfulness, which captures the amount or degree of significance employees believe their work possesses (Rosso, Dekas, & Wrzesniewski,, 2010). So job crafting can enhance the meaningfulness of work for the employee. However, according to the trend in literature, meaningfulness has a generally positive or beneficial outcome for individuals as well as organizations (Rosso et al., 2010). Therefore, it’s interesting for organizations to know which job resources and which job demands result in job crafting behavior, in order to engage in beneficial outcomes for both employees and the organization itself.

Job crafting is often classified as a proactive behavior, as it reflects a self-initiated effort to bring about change (Bakker et al., 2012; Wrzesniewski & Dutton, 2001; Petrou et al., 2012; Tims & Bakker, 2010). In fact, a lot of concepts about job crafting refer to it as a response to one’s present work situation. Other researchers view employees’ self-initiated efforts to alter aspects of their task and relational responsibilities at work as a means to deal more effectively with current job demands (Tims et al, 2012; Petrou et al., 2012). Last mentioned conceptualization is specifically targeted at job characteristics, which is based on the established framework of Job Demands-Resources (JD-R) theory (Demerouti, Bakker, Nachreiner & Schaufeli, 2001). According to JD-R theory, job characteristics can vary widely across occupations but can always be classified into two categories: job demands and job resources. In this

9 conceptualization, job crafting is defined as ‘the self-initiated behaviors of employees to make changes in their level of job demands or job resources’ (Wang, Demerouti, Bakker, 2015, p. 4).

Different types of job crafting exists, which are focused on increasing or decreasing job demands and job resources. As a result, employees can shape the job to better fit their personal abilities, preferences, and needs, helping them to maintain motivation and protect well-being (Wang et al., 2015). Petrou et al., (2012, p. 501) describe job crafting as ‘proactive employee behavior consisting of seeking resources, seeking challenges, and reducing demands’’. Expansion-oriented job crafting is about increasing the number or complexity of tasks and interactions with others, so seeking resources and seeking challenges (Laurence, 2010). Contraction-oriented job crafting refers to reducing complexity of the tasks or limiting the number of relationships at work (reducing demands).

2.2 Work design (resources), role demands and supportive work context as predictors of job crafting 2.2.1 Work design and job crafting

Job autonomy is an important job resource and is defined as the extent to which a job allows one the freedom to schedule work, make decisions and select the methods used to perform tasks (Hackman & Oldham, 1976). Job resources refer to those physical, psychological, social, or organizational aspects of the job that may be functional in achieving work goals, reduce job demands or stimulate personal growth and development (Demerouti, Bakker, Nachreiner & Schaufeli, 2001). Autonomy at work is often essential for employee health, because a high level of discretion at work is associated with more opportunities to cope with stressful work situations. When autonomy is enhanced, employees also report acquiring new skills and experiencing more responsibility for problems at work. Therefore, the opportunity to decide for oneself what and how to do the job may be a predictor of job crafting (Tims et al., 2012). Tims & Bakker (2010) and Wrzesniewski & Dutton (2001) both found that higher perceived autonomy accommodates job crafting due to higher perceived latitude to craft the job. Moreover, Leana et al. (2009) found in their study that discretion is positively associated with individual job crafting.

According to the Job Demands-Resources (JD-R) model, every job contains job characteristics that can be divided into job demands and job resources. Job resources, such as job autonomy can lead to higher job satisfaction and work engagement (Tims et al., 2014). Higher work engagement in turn, may leads to job crafting behavior as well. Wrzesniewski & Dutton (2001) explained in their model that motivation to job craft is more likely to be enhanced when employees perceive that opportunities for job crafting exist. They assert that autonomy in the job leads to perceived opportunities for job crafting and encourages employees to alter the task and relational boundaries of their jobs.

10 Therefore, I expect the following:

Hypothesis 1: There is a positive association between work design (job autonomy) and job crafting. 2.2.2 Role demands and job crafting

Role demands such as burnout, work overload and stress are the organizational, physical or social aspects of a job that require sustained effort and cause physiological and psychological costs (Kinnunen, Felft, Siltaloppi, & Sonnentag, 2011). These job demands, such as work overload, can lead to a person-job misfit (Tims et al., 2012), which in turn can lead to job crafting behavior. According to Tims & Bakker (2010) person-job misfit leads to job crafting behaviors. Person-job misfit focuses on the match between personal characteristics and those of the jobs or tasks that are performed at work, which can be differentiated into demands-abilities fit and needs-supplies fit (Edwards, 1991; Kristof, Brown, Zimmerman & Johnson, 2005). Not only the current misfit between job demands and resources, and personal attributes predicts job crafting. Also the future misfit between job demands and resources may predicts job crafting.

Earlier studies have found that certain unfavorable work characteristics, such as work overload, can lead to stress and burnout (Lee & Ashforth, 1996). Job crafting can lead to higher well-being of employees (Tims et al., 2013). Job crafting may buffers the impact of role overload on job strain. In this case, employees experiencing role overload, could engage in job crafting, increase their person-job fit and reduce the amount of unfavorable work characteristics.

Therefore, I expect the following:

Hypothesis 2: There is a positive association between role demands (role overload) and job crafting. 2.2.3 Supportive work context and job crafting

LMX (Leader-Member Exchange) is a relational concept of leadership and LMX theory differs from other leadership approaches, because it is explicitly focused on unique and dyadic relationships between the leader and the follower where negotiation takes place (Graen & Uhl-Bien, 1995). Identifying characteristics of dyadic relationships is one of the aspects of investigation on LMX. These characteristics could be trust, respect and mutual obligation, which generate influence between parties. LMX results in: building strong relationships with followers, mutual learning and accommodation. Graen & Uhl-Bien mention the advantage of LMX, which is that LMX accommodates different needs of employees and it can provoke superior work from different types of people (1995). They also mention the disadvantages; LMX is time-consuming, and relies on long-term relationships between specific leaders and members. LMX is both transactional and transformational, it begins as transactional social exchange and evolves into transformational social exchange (Graen & Uhl-Bien, 1995). The exchange can be a material transaction

11 (e.g. material compensation) or a social exchange of psychological benefits and favors (e.g. trust, approval, support).

Especially the social exchange of psychological benefits is important. First of all, employees will need the potential support of their supervisor to engage in job crafting. Berg et al. (2010) found that employees are sometimes required to shape expectations of their supervisors beforehand in order to be able to job craft. They described a case in which a supervisor’s attitude restricted an employee from crafting her job altogether (Berg et al., 2010, p. 170). This highlights the importance of the supervisor’s support and approval in order to be able to engage in job crafting as an employee. Supervisors should support job crafting or at least should not forbid employees to job craft. So supervisory support is important in order to facilitate job crafting, which is higher when high LMX relationships are perceived. Secondly, mutual trust between employees and leaders may play an important role. Mutual trust can help in establishing a open and supportive climate that stimulates personal initiative. (Berg et al., 2008). Wang, Demerouti & Bakker state that a healthy and trusting work climate accepts mistakes as learning experiences, which is likely to encourage employees to try new things. Employees are likely to have more confidence in their own abilities when they feel trusted by their leaders.

Gerstner & Day (1997) found that positive LMX relationships can lead to supervisors rating an employee more favorably and can affect actual employee performance. These LMX perceptions affect the rating of an employee and its performance, since it causes a supervisor to form positive or negative expectations about that employee. So when LMX relationships are perceived high, positive expectations about the employee are formed, which may increases supervisor’s support and approval to let the employee engage in job crafting behavior. In conclusion, LMX can enhance the willingness and possibility of employees to engage in job crafting, since high perceived LMX relationships lead to strong relationships, mutual learning, mutual trust and an increase in supervisor’s support and approval. This makes it for the employee less scary and more convenient to change different job characteristics when the manager supports your choices. As a consequence, when employees perceive high LMX relationships, risk-taking and involvement – which are associated with job crafting (Wrzesniewski & Dutton, 2001) – are more likely to take place. Therefore, I expect the following:

Hypothesis 3: There is a positive association between supportive work context (LMX) and job crafting.

12 Proactive personality is defined by Bateman & Grant (1993) as one who is relatively unconstrained by situational forces and who effects environmental change. People with a proactive personality identify opportunities and act on them, show initiative, take action and persevere until they bring about meaningful change (Crant, 2000, p. 439). People who do not have a proactive personality show the opposite patterns; they fail to identify, do not take opportunities to change things and are likely to passively wait and react to instructions of leaders. Previous research already found that a positive relationship between proactive personality and job crafting exists (Tims et al., 2012; Van Wingerden, Derks & Bakker, 2015). So this means that worker’s job performance can be positively influenced by proactive behavior (Crant, 1995; Porath & Bateman, 2006), since job crafting is a strong indicator for self-improvement behavior and in-role performance according to Major, Turner & Fletcher (2006). However, proactive personality as a moderator between job autonomy, role overload and LMX is a new topic of interest. In the study of Tims & Bakker (2010), proactive personality positively moderates the relationship between person-job misfit and job crafting. So according to Tims et al. people who have a person-job misfit and score high on proactive personality tend to engage more often in job crafting behavior than people who score lower on proactive personality. Therefore, it’s reasonable to assume that proactive personality could be a moderator in other relationships between predictors of job crafting and job crafting as well.

2.3.1 Proactive personality as a moderator between job autonomy and job crafting

Job crafting has been seen as a form of proactive behavior (Bakker et al., 2012; Wrzesniewski & Dutton, 2001; Petrou et al., 2012; Tims & Bakker, 2010), so it would be logical that an employee with a proactive personality engages easier in job crafting behavior (Tims et al., 2012; Van Wingerden, Derks & Bakker, 2015). Conservation of Resources theory (COR) describes that people value their resources and that they always try to maintain them (Hobfoll, 2001). It mentions that resources strengthen each other and that resources generate other resources, which would imply that proactive personality and job autonomy could strengthen each other. This is called resource caravans (Hobfoll, 2002). So the relationship between job autonomy and job crafting is strengthened by proactive personality.

Therefore, I hypothesize the following:

Hypothesis 4: Proactive personality positively moderates the relationship between of work design (job autonomy) and job crafting, such that it is stronger for people who score high on proactive personality than for people who score lower on proactive personality.

13 When experiencing role overload, or in other terms experiencing a job demand, it’s easier for people with a proactive personality to see role overload as a challenging job demand, instead of a hindering job demand. People with a proactive personality may experience role overload as a challenge to get all the work done every time. Proactive people ‘actively seek information and opportunities for improving things’ (Crant, 2000, p. 437). Probably they enjoy the high pressure more than people who do not have a proactive personality. Furthermore, people with a proactive personality tend to occur more often in proactive behavior. If employees have proactively taken on more tasks and responsibilities, role overload could be viewed as a challenge (Gilboa et al., 2008). Therefore, I hypothesize the following:

Hypothesis 5: Proactive personality positively moderates the relationship between role demands (role overload) and job crafting, such that it is stronger for people who score high on proactive personality than for people who score lower on proactive personality.

2.3.3 Proactive personality as a moderator between LMX and job crafting When employees experience high LMX relationships, it is found that supervisory support, mutual learning and mutual trust are likely to exist, which may increase job crafting behavior as stated in H3. As already mentioned, high perceived LMX relationships lead to more risk-taking and involvement – which are associated with job crafting (Wrzesniewski & Dutton, 2001). The study of (N. Li, Liang, & Crant, 2010) revealed that people with a proactive personality establish positive relationships with supervisors. It requires the support of others to implement positive changes, since it’s risky (Grant & Ashford, 2008). Therefore, people who score high on proactive personality are assumed to have more positive relationships with supervisors, are likely to implement positive changes more often and take risks more often than people who score low on proactive personality. Because of the positive relationships with supervisors, people with a proactive personality easier possess resources and sponsorships (Crant, 2000), since supervisors usually possess more resources than coworkers (Morrison & Phelps, 1999). Therefore, an employee with a high LMX relationship has more job resources. So in order to gain the benefits of having good LMX relations, the employee should have a proactive personality. Since high perceived LMX relations lead to more risk-taking and involvement – which are associated with job crafting (Wrzesniewski & Dutton, 2001) – a proactive personality is desirable. People high in proactive personality tend to ‘identify opportunities and act on them, show initiative, take action and persevere until meaningful changes occur’ (Crant, 2000, p. 439). This supposes that those people high in proactive personality make more use of the good relationships with their supervisors, the mutual trust and the supervisory support since they probably take more risk.

14 Work design

Role demands

Supportive work context

H1

H2

H3

H4 _{H5 H6}

things happen when they believe in it, whatever the expectations are. (Bateman & Crant, 1993). So the expectations of the supervisor do not matter that much compared to employees scoring low on proactive personality. So the expectation is that having a proactive personality stimulates employees to engage in job crafting, it strengthens the relationship between LMX and job crafting. Also in this case, according to COR theory, people value resources and try to maintain them (Hobfoll, 2001). Proactive personality and LMX strengthen each other. Therefore, I hypothesize the following:

Hypothesis 6: Proactive personality positively moderates the relationship between supportive work context (LMX) and job crafting, such that it is stronger for people who score high on proactive personality than for people who score lower on proactive personality.

2.4 Conceptual model

The hypotheses discussed in the paragraphs above are visualized in this conceptual model below.

Enhancing Task Crafting Limiting Task Crafting

Enhancing Relationship Crafting Enhancing Skill Crafting

Enhancing Cognitive Crafting Limiting Cognitive Crafting Job autonomy

Role overload

LMX

15

3. Methodology

This section explains how the hypotheses of this research will be tested. First of all, the method of research will be discussed. Secondly, a description of the sample will follow and the data collection will be explained. After that, a description of measurement of the independent variables, dependent variable, moderator, and control variables will follow. Finally, the data analysis will be explained.

3.1 Research design

This study aims to investigate the relation that role overload, job autonomy and LMX have with job crafting, and the way proactive personality moderates these relationships. To conduct this research and collect data, an online-questionnaire based survey is used. The advantage of approaching respondents with an online-questionnaire based survey, is that it costs little effort to gather a large group of respondents against a low cost (Saunders et al., 2009). So online surveys allow a researcher to gather a large sample in a short period of time, despite the distance. Since this thesis had to be conducted in a short period of time, online surveys were the best option. A large sample is desirable in order to have accurate measurements and draw reliable and valid conclusions. Furthermore, an online-questionnaire based survey ensures that questions can be asked in a consistent and standardized way, which enables a researcher to compare answers from respondents easily (Wright, 2006). On the other hand, some disadvantages of online-questionnaire based surveys also exist. For example, the answer options are predetermined, which results in limited answers for respondents. However, for this research the emphasis is on generating a large amount of data in a limited amount of time in order to draw reliable and valid conclusions. Therefore, the online survey was the best option. According to the differences among respondents, the influence of the predictors on job crafting can be tested, as well as the influence of proactive personality as a moderator.

The questionnaires are filled out by dyads. A dyad consists of a pair of two persons, concerning one employee and his/her direct manager. A manager is allowed to fill out the questionnaire for a maximum of 2 employees, which counts for 2 dyads in that case. Two types of questionnaires are used in order to gather the information: one for the employee and one for the manager. The purpose of the manager questionnaire, is evaluating the employee on different aspects, as well as self-rating the degree to which he or she has a proactive personality. The purpose of the employee questionnaire, is self-rating the degree to which he or she believes to be proactively job-crafting his/her job, self-rating the degree to which he/she has a proactive personality and self-rating the degree to which he or she has job autonomy,

16 experiences role overload and is satisfied with the LMX. For this research, only the employee questionnaires are being analyzed, since these questionnaires contain all the necessary variables to test the hypotheses of this research.

3.2 Sample and data collection

The population consists of every person in the Netherlands that has a job and that has at least one manager and preferably also a colleague. Since the hypotheses tested in this paper do not necessarily relate to any specific variables in the population, no further information on the population is disclosed.

The sample size was aimed to be as large as possible within the time frame of this research. A large sample size allows for better generalization of results found in the study (Saunders et al., 2009, pp. 217-218). The data collection is done by 5 students, but each student conducts its own research. Based on an excel sheet containing all the invited respondents, being held up-to-date by all 5 students, the estimated response rate was 75%. To gather respondents, the online questionnaires were send by e-mail to the managers and employees of organizations. A personal code was send to each of the respondents, as well as the link to the online questionnaire on the website Qualtrics. The different codes match the questionnaires of the managers and their employee(s), so they could be analyzed together. In the questionnaire it was ensured that the responses would be treated carefully and anonymously. After two weeks, reminders were sent to all the managers and employees who were willing to participate, but did not respond yet. One week after the first reminder, the second reminder was sent. In the end, the total sample size was N=126 employees, with a response rate of 88% (the response rate of the manager questionnaire was 83%). From the respondents, 47.7% is male and 52.3% is female. The mean age is 31.80 (SD 12.00) ranging from 17 to 63. The mean organizational tenure is 5.98 years (SD 8.12) ranging from 0 to 40 years. The education level is divided into High School (19.2%), Lower vocational education (MBO) (21.5%), Higher vocational education (HBO) (30.8%), University bachelor degree (13.8%), University master degree (13.1%), PhD (0.8%) and Other (0.8%). The mean LMX tenure is 2.15 years (SD 2.47) ranging from 0 to 13 years.

3.3 Measures

The control variables (gender, age, educational level, organizational tenure and LMX tenure) as well as the predictors (role overload, job autonomy and LMX), the moderating variable (proactive personality) and the outcome variable (job crafting) were measured at one point in time. The predictors, the moderating variable and the outcome variable were measured from the employee perspective using Likert scales.

17

3.3.1 Role overload

The independent variable role overload was measured from the employee perspective, using 3 items from a scale developed by Bolino & Turnley (2005). The scale role overload measures the degree to which the respondents experience that they have too much work or not enough time to complete the tasks. Responses were made on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). High scores on the items are indicative for a high level of role overload. The internal consistency was high with a Cronbach’s α of 0,910. More sample items can be found in Appendix A at the end of this paper. Example of item: It often seems like I have too much work for one person to do.

3.3.2 Job autonomy

The independent variable job autonomy was measured from the employee perspective, using 3 items from a scale developed by Spreitzer (1995). The scale measures psychological empowerment, which can be divided into meaning, competence, self-determination, and impact. Self-determination is an individual’s sense of having choice in initiating and regulating actions (Deci, Connell, & Ryan, 1989). Self-determination reflects autonomy in the initiation and continuation of work behaviors and processes; examples are making decisions about work methods, pace, and effort (Bell & Staw, 1989; Spector, 1986). Job autonomy measures how much influence the workplace offers over sequence, time frame, method, and means of one’s work tasks. All items were measured on a 7-point Likert scale measuring from 1 (strongly disagree) to 7 (strongly agree). High scores on the items are indicative for a high level of job autonomy. The internal consistency was high with a Cronbach’s α of 0,928. More sample items can be found in Appendix A at the end of this paper. Example of item: I have significant autonomy in determining how I do my job.

3.3.3 LMX

The independent variable LMX was measured from the employee perspective, using the LMX 7 scale from Graen and Uhl-Bien (1995). This scale features 7 items regarding employees’ perceptions of the supervisor-subordinate relationship. The original scale of Graen and Uhl-Bien consists of questions, so these were rewritten towards statements suitable for a Likert scale. The respondent is given the option to indicate the LMX on a 1 (strongly disagree) to 7 (strongly agree) Likert scale. High scores on the items are indicative for a good relationship between the supervisor and subordinate. The internal consistency was high with a Cronbach’s α of 0,898. More sample items can be found in Appendix A at the end of this paper. Example of item: My supervisor understands my problems and needs.

18

3.3.4 Proactive personality

The moderator proactive personality was measured from the employee perspective, using the 6-item adaptation of Bateman and Crant’s (1993) Proactive Personality Scale (Claes et al., 2005; Parker, 1998). This scale features 6 statements in which respondents self-rate the degree to which they have a proactive personality. Responses were made on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). High scores on the items are indicative for a strong proactive personality. The internal consistency was good with a Cronbach’s α of 0,774. More sample items can be found in Appendix A at the end of this paper. Example of item: I am always looking for better ways to do things.

3.3.5 Job crafting

The dependent variable job crafting was measured in the employee questionnaire using the scale designed by Bindl et al. (2014). This scale features 20 statements regarding typical behavior that is associated with job crafting. The respondent is then given the option to indicate the extent to which he or she participated in such behavior over the past week on a 1 (not at all) to 5 (a great deal) Likert scale. High scores on the items are indicative for a high level of job crafting behavior. The internal consistency was good with a Cronbach’s α of 0,874. More sample items can be found in Appendix A at the end of this paper. Example of item: I changed my tasks so that they were more challenging.

3.3.6 Control variables

Age (in years), gender (1= male, 2=female), highest educational level (1=High School, 2=Lower vocational education (MBO), 3=Higher vocational education, 4=University bachelor degree, 5=University master degree, 6=PhD, 7=Other), organizational tenure (in years) and LMX tenure (in years) were used as control variables in all the analyses.

3.4 Data analysis 3.4.1 Procedures

After collecting all the data, the data had to be prepared before further analysis could be done in SPSS. First of all, the data had to be cleaned up, which was done by manually omitting responses that were unfinished or unlikely. This screening of data was done by using frequency tables of all the variables in order to examine errors in data entry. The amount of unlikely responses was limited; no responses were deleted because of unlikeliness. Furthermore, missing values were reported in order to not let them

19 significantly influence the outcomes of this research. Then the sample characteristics could be derived, such as gender, age, education and tenure, as well as the descriptive statistics containing N, mean, standard deviation and range of all the items.

Further data analysis was also done using SPSS. The employee questionnaire provided 164 items, of which 44 are relevant to this research. Scale means were computed by merging 39 of the 44 items into 5 constructs. These 5 constructs correspond to the 5 variables discussed in the conceptual model and hypotheses of this paper. Cronbach’s Alpha was measured for each construct in order to test the reliability of each construct. Reliability is the extent to which data collection techniques or analysis procedures yield consistent findings (Saunders et al., 2009). The reliability scores of each construct can be found in paragraph 3.3 and in table 2 in the results section. All merged construct scales were found to be reliable as Cronbach’s Alpha for each construct was > 0,8. After that, the correlations between the items were measured and can be found in table 2. To rule differences between for example 5 and 7 point scales out, all the constructs were standardized.

Hierarchical multiple regression was used to test all the hypotheses regarding suspected moderating effects of proactive personality. In order to calculate the interaction effect, new variables of the correlation between the moderator and the determinants (role overload, job autonomy and LMX) had to be created (Baron & Kenny, 1986). Three items were created, which are: Z score role overload * Z score

proactive personality, Z score job autonomy * proactive personality and Z score LMX * proactive

personality. First of all, it was checked if the control variables add information and explain variance of the model. Then, the direct relationships between the predictors and job crafting will be measured, while controlling for age, gender, education and tenure. Lastly, the moderating effects of proactive personality will be measured by adding the standardized predictor * proactive personality, by using Process in SPSS.

3.4.2 Assumptions of the model

When using hierarchical multiple regression analysis in order to test the hypotheses, a couple of assumptions have to be met. Therefore, the sample size, linearity, non-collinearity, independence, homoscedasticity and normality are being looked at.

The sample size should have a certain size in order to preserve the generalizability of the results (Pallant, 2005).

Linearity of the relationship between the dependent variable and each of the independent variables was measured through scatterplots of all the relationships. All the variables of the model had a linear relationship with job crafting. Also homoscedasticity was checked by using scatterplots.

20 Homoscedasticity means that for each independent variable or combination of independent variables, the variance of the residuals is constant. Homoscedasticity was met for all variables, except for the control variables organizational tenure and LMX tenure.

Non-collinearity means that no exact linear relationship between any of the independent variables exists. Table 2 shows that the assumption of non-collinearity is met (r<.80 between all independent variables).

Independence of observations can be checked by using the Durbin-Watson statistic, which tests for autocorrelation in the residuals from a regression analysis. The values should be between 1.5 and 2.2, or more precisely it should be around 2, which indicates that the residuals are uncorrelated (Field, 2009). This assumption was met, since the DW was 1.949.

Finally, the normality check was done by evaluating two components of normality, namely skewness and kurtosis. The values for skewness (asymmetry) and kurtosis should fall within the range from ± 2 SE of either skewness or kurtosis. Skewness refers to symmetry of the distribution, while kurtosis refers to the peakedness of the distribution (Tabachnick & Fidell, 1996). Role overload and age slightly surpassed this assumption. The skewness of organizational tenure (skewness = 2.122, SE = .212), LMX tenure (skewness = 2.078, SE = .212), job autonomy (skewness = -1.107, SE = .213) and LMX (skewness = -1.265, SE = .215) weren’t between these values as well. The kurtosis of age (kurtosis = -2.023, SE = .422), organizational tenure (kurtosis = 4.463, SE = .422), LMX tenure (kurtosis = 4.396, SE = .422) and LMX (kurtosis = 3.080, SE = .427) were also not between these values. When removing the outliers from the dataset, the skewness and kurtosis went significantly down. Therefore, investigated was the influence of the outliers on the significance of the results. Outliers were defined being ±3 SD from the mean. Hierarchical multiple regression analysis was done before and after removing outliers. In the end it appeared that the outliers didn’t have a significant impact on the results, so they were left in the dataset. Furthermore, Likert scale types of variables don’t really have outliers, which applies to the predictors, moderator and outcome variable. Also, when data points are suspected of being legitimate, the data is more likely to be representative of the population as a whole if outliers are not removed (Orr, Sackett, & Dubois, 1991). The 6 outliers found in job autonomy, LMX and proactive personality were all legitimate outliers. Moreover, 5 out of the 6 outliers from the control variables were legitimate as well, while one outlier was an error. So the normality check was partly met for some variables, which has to be kept in mind by interpreting the conclusion. In conclusion it can be said that the assumptions of sample size, linearity, non-collinearity and independence of observations are met. The assumptions of

21 homoscedasticity and normality were partly met, which has to be taken into consideration when interpreting the results.

4. Results

4.1 Descriptive statistics

The first part of the analysis contains the descriptive statistics of the used constructs. Job autonomy, LMX and proactive personality are perceived high with scores of respectively 78%, 78% and 74% of the maximum item score. Job crafting and role overload are perceived moderate with a score of 60% and 54% of the maximum item score.

Table 1. Descriptive Statistics

Variable Mean Std. Dev. Range

1. Job autonomy 5.49 1.26 1-7 2. Role overload 3.77 1.54 1-7 3. LMX 5.48 0.94 1-7 4. Proactive personality 5.17 0.81 1-7 5. Job crafting 2.98 0.52 1-5 4.2 Correlations

When looking at the Parsons correlations (table 2), it can be concluded that job autonomy has the highest correlation with job crafting (r.362 p<0.01). Furthermore, role overload has a correlation with job crafting (r.295 p<0.01) and LMX has a correlation with job crafting (r.337 p<0.01). Proactive personality has a significant correlation with job crafting as well (r.327 p<0.01). Both job autonomy (r.356 p<0.01) and LMX (r.355 p<0.01) have a correlation with proactive personality. When looking at the control variables, education does not have any correlation with another item or construct. Age positively correlates with job autonomy (r.299 p<0.01) and role overload (r.182 p<0.05). Gender only has a correlation with proactive personality (r-.236 p<0.01). LMX tenure is the control variable with the most correlations. The highest significant correlations are between organizational tenure and age (r.627 p<0.01) and between job

22 autonomy and LMX (r.518 p<0.01).

The assumption of no perfect multicollinearity is also met, since there is no exact linear relationship between any of the independent variables (r<.80).

As can be seen in table 2, the scales all have good to high reliability, with Cronbach’s Alpha of ≥ 0.774. For every scale, the corrected item-total correlations indicated that all the items have a good correlation with the total score of the scale (all above 0.30, except for 3 out of the 20 items of the job crafting scale). Also, none of the items would substantially affect reliability if they were deleted.

Table 2: Means, Standard Deviations, Correlations

Variables _{M SD 1 2 3 4 5 6 7 8 9 10} 1. Job autonomy _{5.49 1.26 (.928)} 2. Role overload _{3.77 1.54 .141 (.910)} 3. LMX _{5.48 0.94 .518** -.047 (.898)} 4. Proactive personality 5.17 0.81 .356** .029 .355** (.774) 5. Job crafting _{2.98 0.52 .362** .295** .337** .327** (.874)} 6. Gender _{1.52 0,50 .018 .088 -.117 -.236** .003 -} 7. Age _{31.80 12.00 .299** .182* .134 .030 -.087 -.086 -} 8. Educational level _{2.85 1.35 .095 .047 .073 .037 .075 -.058 .125 -} 9. Organizational tenure 5.98 8.13 .169 .194* .071 -.037 -.101 .075 .627** .085 - 10. LMX tenure _{2.15 2.47 .229** -.025 .194* .153 -.039 .036 .339** -.123 .349** -}

**. Correlation is significant at the 0.01 level (2-tailed) **. Correlation is significant at the 0.05 level (2-tailed)

4.3 Regression

23 Hierarchical multiple regression was performed to investigate the ability of job autonomy, role overload and LMX to predict levels of job crafting, after controlling for gender, age, education, organizational tenure and LMX tenure.

In the first step of hierarchical multiple regression, 5 predictors were entered: gender, age, education, organizational tenure and LMX tenure. This model was not statistically significant F (5,120) = .543; p > .05), and explained 2.2% of variance in job crafting. After entry of job autonomy, role overload and LMX at step 2, the total variance explained by the model as a whole was 30.8% F (8,117) = 6.519; p < .001. The introduction of job autonomy, role overload and LMX explained additional 28.6% variance in job crafting, after controlling for gender, age, education, organizational tenure and LMX tenure (R² change = 0.286; F (3,117) = 16.137; p < .001. In the final model, 3 out of the 8 predictor variables were statistically significant, with role overload recording a higher Beta value (β .325, p < .001) than job autonomy (β .267 p < .01) and LMX (β .152 p < .05).

Table 3: Hierarchical Multiple Regression Model of Job crafting

R R² R² Change B SE β t Step 1 .149 0.022 Gender .034 .092 .034 .366 Age -.067 .121 -.067 -.556 Education -.079 .118 -.080 -.671 Organizational tenure .104 .094 .103 1.111 LMX tenure .036 .099 .036 .361 Step 2 .555 .308*** .286** Gender .015 .080 .015 .185 Age -.204 .106 -.203 -1.926 Education .057 .081 .056 .704 Organizational tenure -.097 .102 -.098 -.958 LMX tenure -.016 .087 -.016 -.178 Job autonomy .266 .095 .267** 2.799

24

Role overload .326 .081 .325*** 4.036

LMX .239 .098 .152* 1.543

Note. Statistical significance: *p <.05, **p<0.01, ***p<0.001

The relationship between job autonomy and job crafting is significant (β .325, p < .001), as well as the relationship between role overload and job crafting (β .325, p < .001), and the relationship between LMX and job crafting (β .152 p < .05). Therefore, H1, H2, and H3 are accepted.

4.3.2 Interaction effects

PROCESS was used in order to test hypotheses 4, 5, and 6 of this study. These hypotheses test whether proactive personality moderates the relationship between job autonomy, role overload, LMX and job crafting. Model 1 tests the interaction effect of proactive personality on the relationship between job autonomy and job crafting. Model 2 tests the interaction effect of proactive personality on the relationship between role overload and job crafting. Finally, model 3 tests the interaction effect of proactive personality on the relationship between LMX and job crafting.

The regression coefficient for XM in model 1 is c3 = .1710 and is not statistically significant from zero, t (119) = 1.66, p > .05. Thus, the effect of job autonomy on job crafting does not depend on whether an employee has a proactive personality. Model 1 accounts for 25.42% of variance in job crafting. So H4 is rejected. The regression coefficient for XM in model 2 is c3 = .1447 and is not statistically significant from zero as well, t (119) = 1.62, p > .05. Therefore, the effect of role overload on job crafting does not depend on whether an employee has a proactive personality. Model 2 accounts for 25.71% of variance in job crafting. So H5 is rejected. The regression coefficient for XM in model 3 is c3 = .1705 and is not statistically significant from zero, t (117) = 1.31, p > .05. Thus, the effect of LMX on job crafting does not depend on whether an employee has a proactive personality. Model 3 accounts for 21.61% of the variance in job

crafting. So H6 is rejected.

The interaction effects are not visualized in a graph, since proactive personality was not a significant moderator in any of the models, see table 4. However, proactive personality is in all the models positively associated with job crafting, which shows that proactive personality predicts job crafting behavior. So employees having a proactive personality don’t engage more often in job crafting, but there is a positive relationship between proactive personality and job crafting.

25 Coefficient SE t p Model 1 Intercept i1 -.0577 .0856 -.6745 .5013 Job autonomy (X) c1 .3475 .1116 3.1129 .0023 Proactive personality (M) c2 .2472 .1064 2.3229 .0219 Job autonomy*Proactive personality (XM) c3 .1710 .1032 1.6568 .1002 Gender .0656 .0946 .6933 .4895 Age -.1495 .1076 -1.3894 .1673 Educational level .0521 .1125 .4630 .6442 Organizational tenure -.0598 .0946 -.6327 .5282 LMX tenure -.0763 .1404 -.5439 .5875 Model 2 Intercept i1 -.0067 .0841 -.0793 .9369 Role overload (X) c1 .3182 .1023 3.1113 .0023 Proactive personality (M) c2 .3057 .1024 2.9859 .0034 Role overload*Proactive personality (XM) c3 .1447 .0891 1.6249 .1068 Gender .0602 .0954 .6304 .5296 Age -.1316 .1061 -1.2401 .2174 Educational level .1026 .1116 .9197 .3596 Organizational tenure -.0541 .0949 -.5706 .5693 LMX tenure .0342 .1254 .2729 .7854 Model 3 Intercept i1 -.0582 .0882 -.6597 .5108 LMX (X) c1 .3045 .0856 3.5561 .0005 Proactive personality (M) c2 .2678 .1069 2.5049 .0136 LMX*Proactive personality (XM) c3 .1705 .1299 1.3124 .1919 Gender .1368 .0906 1.5094 1.339

26 Age -.0629 .0974 -.6453 .5200 Educational level .0520 .1274 .4084 .6837 Organizational tenure -.0939 .1010 -.9294 .3546 LMX tenure -.0792 .1210 -.6551 .5137 Model 1: R²=0.2542 p<0.01 F(8,119)=3.4460 Model 2: R²=0.2571 p<0.001 F(8,119)=4.9730 Model 3: R²=0.2161 p<0.001 F(8,117)=3.8011

27

5. Discussion and Conclusion

5.1 Analysis results and theoretical implications

In the past decades, due to rapid changes in technology and heavy competition, organizations need to adopt various adaptive strategies like acquisitions, outsourcing and mergers, and restructuring in order to survive and remain flexible (Sparks, Faragher, & Cooper, 2001). Job crafting is the solution, since it facilitates employees in adapting to change while their well-being and work motivation is maintained (Petrou et al., 2012). Therefore, it’s interesting to look at factors predicting job crafting, in order for organizations to create contexts in which job crafting is more favored. In this way, organizations stay successful in times of uncertainty (Wang, Demerouti & Bakker, 2015). In this study, relationships between job autonomy, role overload and LMX on job crafting has been researched. These are all factors that can be influenced by the organization, which makes the study relevant for organizations that are willing to create contexts in which job crafting is more favored. Furthermore, the moderating role of proactive personality on the mentioned relationships has been looked at. The first hypothesis of this study was:

Hypothesis 1: There is a positive association between work design (job autonomy) and job crafting.

This hypothesis was supported. The results showed that job autonomy is a positive and significant predictor of job crafting. The positive relationship between job autonomy and job crafting was already expected, based on the existing literature. Wrzesniewski & Dutton (2001) explained in their model that motivation to job craft is more likely to be enhanced when employees perceive that opportunities for job crafting exist. They assert that autonomy in the job leads to perceived opportunities for job crafting and encourages employees to alter the task and relational boundaries of their jobs.

Hypothesis 2: There is a positive association between role demands (role overload) and job crafting.

This hypothesis was accepted as well. The results showed that role overload is positively associated with job crafting. This positive relationship was expected as well, based on the existing literature. Job demands, such as work overload, can lead to a person-job misfit (Tims et al., 2012), which in turn can lead to job crafting behavior. According to Tims & Bakker (2010) person-job misfit leads to job crafting behaviors. Furthermore, role demands lead to undesired outcomes, such as stress. When engaging into job crafting, the person-job misfit can be reduced, which leads to less undesired outcomes as well.

28 The hypothesis was supported. The results showed a positive relationship between LMX and job crafting. This conclusion was expected as well, since the importance of potential support of the supervisor was marked, as well as the importance of approval of the supervisor and mutual trust. Mutual trust, mutual learning and accommodation helps stimulating employees to engage in job crafting behavior. Mutual trust can help in establishing an open and supportive climate that stimulates personal initiative. (Berg et al., 2008). Wang, Demerouti & Bakker state that a healthy and trusting work climate accepts mistakes as learning experiences, which is likely to encourage employees to try new things. Employees are likely to have more confidence in their own abilities when they feel trusted by their leaders.

Hypothesis 4: Proactive personality positively moderates the relationship between work design (job autonomy) and job crafting, such that it is stronger for people who score high on proactive personality than for people who score lower on proactive personality.

Hypothesis 5: Proactive personality positively moderates the relationship between role demands (role overload) and job crafting, such that it is stronger for people who score high on proactive personality than for people who score lower on proactive personality.

Hypothesis 6: Proactive personality positively moderates the relationship between supportive work context (LMX) and job crafting, such that it is stronger for people who score high on proactive personality than for people who score lower on proactive personality.

Hypotheses 4, 5, and 6 were rejected. The results did not show an interaction effect of proactive personality on the relationships between job autonomy, role overload, LMX and job crafting. However, positive relationships between proactive personality and job crafting existed. This was expected as well, according to the literature. Different studies had shown that a positive relationship between proactive personality and job crafting exists (Tims et al., 2012; Van Wingerden, Derks & Bakker, 2015).

5.2 Limitations and directions for further research

The cross sectional design of the research is a limitation, since the data are collected at a single point in time and consist of the whole study population. This means that change over time cannot be measured that well due to limited time. This implicates that mutual/reciprocal relationships cannot be measured as well, for example between the variables role overload, job autonomy, LMX and job crafting. So the causality between the variables cannot be tested. Since causal relationships cannot be concluded, the

29 possibility of having reverse causality between variables cannot be ruled out as well.

Non-probability convenience sampling technique is used to gather the respondents (N=126), which were working employees in the Netherlands. This is a limitation, since it hinders the generalizability of the findings of the study, which is due to lower external validity. Moreover, sample used is a heterogeneous sample, since different sectors and positions are included. The sample heterogeneity lowers the validity and makes it impossible to make generalizations, since it is not possible to control for all the changing variables.

The questionnaires were based on self-rating, which may have caused self-overestimation and social desirability bias. These biases may have reduced the reliability and validity of measurements. Especially for the items measuring proactive personality, these biases could have had influence. Only 6 respondents rated themselves lower than 4.00 on proactive personality, which indicates that overall, the respondents rated themselves high on proactive personality. Thus, over-reporting ‘good behavior’ or underreporting ‘bad behavior’ could have happened. Remarkable was the high score on the constructs LMX and job autonomy. Some employees asked if the questionnaires were really anonymous. This may have had influence on the high scores on LMX and job autonomy, since people may have had fears that the questionnaire was not totally anonymous. As a result, these people probably did not want to give too negative scores on their manager and work characteristics. This may have influenced the measurement of

role overload as well.

What also has to be taken into account, is the fact that not all the assumptions for using hierarchical linear regression were met (homoscedasticity and normality were partly met). This may have had influence on the significance of the results. So further analysis using a sample which meets all the assumptions, may result in more significant relationships and would show a moderating effect of proactive personality on certain work characteristics predicting job crafting.

The dependent variable, job crafting, is measured as one variable and not divided into different forms of job crafting. As a result, it is not known whether, for example, the positive relationship between role overload and job crafting is only significant for a special type of job crafting, for example limiting versus enhancing task crafting/relationship crafting/skill crafting/cognitive crafting (Bindl et al., 2014). Furthermore, the scale of job crafting used, was a relatively new one invented by Bindl et al., which may

reduced the reliability of the study.

For further research, it may be useful to execute a longitudinal study in order to detect the causality of effects between job autonomy, role overload, LMX and job crafting. Such a study could detect any reciprocal relationships between for example predictors of job crafting. As already mentioned, one of

30 the highest correlations was between job autonomy and LMX (r.518, p < .01). It would be interesting to know if job autonomy and LMX are inter-related, in a way that more supervisory support would lead to more autonomy for the employee. This is called a positive synergy effect and means in this case that multiple predictors of job crafting could result in a positive impact that is greater than the sum of the isolated predictors (Delery, 1998). So future research can focus on identifying combinations of predictors that complement each other. If organizations want to stimulate job crafting behavior, they can focus on those predictors that generate the largest effect together. Furthermore, it may be useful to investigate if job autonomy, role overload, LMX and other possible predictors positively associate with specific types of job crafting (e.g. enhancing or limiting task crafting, relationship crafting, skill crafting, cognitive crafting). Some researchers have linked job crafting with specific outcomes related to for example performance (Berg et al., 2010). However, a need exists for research that link specific forms of job crafting to particular individual and organizational outcomes. Furthermore, concerning the positive relationship between proactive personality and job crafting, it would be interesting to know which certain personality traits are related to specific forms of job crafting. The final suggestion would be to use a larger sample size in order to make the results more generalizable to the population and to reduce the biases.

5.3 Practical implications

Job crafting is an individual behavior, but it has beneficial outcomes for both the individual and the organization. Therefore, it’s interesting for organizations to know which job resources and which job demands result in job crafting. Job autonomy and role overload are both work characteristics related to the job, that can be influenced by the organization. LMX is a work characteristic related to the supervisor, which can also be influenced by the organization. So all the predictors in this research can be influenced by the company. Since positive relationships were found between job autonomy, role overload, LMX and job crafting, organizations can influence job crafting behavior among its employees. This means that organizations have indirectly the possibility to increase meaningfulness of work – which is associated with numerous work-related benefits, such as increased motivation, job satisfaction and performance – by increasing job crafting behavior among employees.

First of all, job autonomy can be promoted by the organization, by providing a job position that enables substantial freedom, provides independence and provides flexible working, in a way that the employee can schedule its work and can determine its procedures in carrying the work out. A company-culture that encourages autonomy and provides supervisory support and social support help in achieving this. Secondly, LMX can be influenced by the organization through encouraging supervisory support.

31 Supervisors should be carefully selected and sufficient training should be given to supervisors in order to develop and encourage supervisory support. So supervisors should adopt a coaching-oriented style and be supportive in order to enhance LMX.

Proactive personality was also found to be a predictor of job crafting behavior in all the interaction models. Since proactive personality is a personal characteristic, it can only be influenced by the company during the initial selection process. So managers can enhance job crafting behavior by selecting and hiring those people who score high on proactive personality.

5.4 Conclusion

The aim of this study is to investigate the impact of work design (job autonomy), role demands (role overload) and supportive work context (LMX) on job crafting and if proactive personality moderates these relationships. It is interesting to look at factors predicting job crafting, since organizations can create contexts in which job crafting is more favored. In this way, organizations can stay successful in times of uncertainty and receive positive outcomes for both the employee and the organization. Positive relationships between job autonomy, role overload, LMX and job crafting were found. No significant interaction effects of proactive personality were found. However, proactive personality did predict job crafting in all the interaction models. Job crafting can be promoted by the organization by creating a company culture that encourages autonomy and provides supervisory support. Furthermore, organizations should carefully select and train their supervisors, as well as selecting carefully their employees, in order to develop supervisory support and attract proactive employees.