Strong social mentoring networks : a research on the relationship between tie strength and mentor types on job satisfaction, and the moderating effect of proactive personality

(1)

Strong social mentoring networks

A research on the relationship between tie strength and mentor types

on job satisfaction, and the moderating effect of proactive personality.

MASTER’S THESIS

Willemijn Kuitenbrouwer (10019545) Final version, 23-06-2016

MSc, in Business Administration – Strategy Track ABS, UvA

(2)

1

STATEMENT OF ORIGINALITY

This document is written by Student Willemijn Kuitenbrouwer 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 nog 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)

2

This study investigates the effects of a mentoring network on job satisfaction and specifies the moderating influence of proactive personality. By illustrating the separate effects of the strength of ties (weak versus strong), the types of mentoring ties (instrumental, psychosocial or both) and the combined effects of strength and type of mentoring ties, this research integrates two different streams within the field of social network theory. Multiple hierarchical analyses demonstrated a general positive effect for strong ties, largely confirming the hypotheses. An interesting notion was pointed out by the findings for the combined effect of the strength of ties and the type of mentoring ties. It seemed that strong ties are most beneficial for all types of ties, suggesting that the strength of a tie is an inherent characteristic of a mentoring tie. This provides new insights and proposes boundary effects to the use of mentoring ties. Further, even though evidence was found for the moderating effect of proactive personality, these effects were inconsistent with the expectations and did not provide one unambiguous explanation. Since proactive personality did prove to have a moderating effect, further investigation is needed to offer clarification for the occurrence of this effect and to focus on underlying mechanisms which can explain this unambiguity, implying for the necessity of future research.

Key words: strength of ties, mentor tie type, instrumental/psychosocial mentoring ties, proactive personality, job satisfaction.

(6)

5

1. Introduction

After decades of research, job satisfaction has proven to be an important antecedent for a successful business (Harter, Schmidt, & Hayes, 2002). Research demonstrates that job satisfaction is inherent to higher employee performance, more organizational commitment and lower turnover rates (Yücel, 2012) - a scenario that is positive both for employer and employee. Locke (1976, in: Brown & Peterson, 1993, p. 63) defined being satisfied about one’s job as “a pleasurable or positive emotional state resulting from the appraisal of one’s job or job experiences”. For managers, this implies that they should create an atmosphere that makes employees satisfied with their jobs; therefore it is crucial to understand what influences satisfaction and how managers can stimulate this. Determinants of job satisfaction have been a topic that many scholars have tried to cover (e.g. Brown & Peterson, 1993; Flap & Völker, 2001; Judge, Heller, & Mount, 2002; Martins, Eddleston, & Veiga, 2002; Yang, 2010). Although a large amount of literature exists on this subject, there are still some questions remaining.

One of the antecedents of job satisfaction is the presence of a mentor. Individuals that are mentored report more job satisfaction than individuals that are not mentored (Allen, Eby, Poteet, Lentz, & Lima, 2004; Ragins, Cotton, & Miller, 2000). Mentors can provide their mentees with learning opportunities, can give constructive feedback and can offer career advice, which can improve a mentees career chances (Janasz & Godshalk, 2013). However, mentor-mentee relationships arise within social networks and can therefore occur in many different forms. These forms can affect the feeling of job satisfaction in different ways. A better understanding of the influence of a mentor on job satisfaction can therefore be obtained from a network perspective.

Job satisfaction is determined by the appraisal of one’s job experiences (Locke, 1976, in: Brown & Peterson, 1993, p. 63); it is the combined effect of all factors that are influencing

(7)

6 an individual at work. The network perspective takes these influences into account since this approach moves away from the view that an employee acts as an isolated individual and emphasizes that his actions are influenced by interactions with others (Brass, Galaskiewicz, Greve & Tsai, 2004). These interactions flow through ties, which are the building blocks of the social network and determine how actors are connected to each other (Borgatti & Foster, 2003). An actor can be an individual but also a team or organization. This research will focus on the individual level and more specifically on the dyadic relations among individuals (Brass et al., 2004) – with a dyad implying the relationship between two actors (Field, 2013). All dyadic ties in the social network can contribute to job satisfaction in a different way, depending on the characteristics of the tie.

Research has looked into different characteristics, such as the strength of the tie (Granovetter, 1973) and the type of tie (Eby et al., 2013). Granovetter (1973) states that the type of information that is flowing through a ties depends on the strength of the tie; weak ties seem to spread new information and reach a larger amount of different actors, whereas strong ties seem to spread similar information. Eby et al. (2013) argue that besides having social ties with different strengths, an actor can have a mentoring tie which indicates a specific developmental relationship between an experienced mentor and a less experienced protégé, where the mentor is providing support and assistance to the latter. Mentoring relationships appear to have a positive influence on the protégé’s career success and increase his or her feeling of well-being (Fagenson, 1989; Sorcinelli & Yun, 2007). An actor can have multiple mentors and these can offer different types of support, namely psychosocial or instrumental support (Eby et al., 2013). Psychosocial support ties enhance the feeling of satisfaction with the social part of the job, whereas instrumental support ties increase satisfaction with the functional part of the job (Flap & Völker, 2001). Nowadays, a network needs to have a variety of

(8)

7 mentoring relationships containing both instrumental and psychosocial mentoring support ties to keep up with the current dynamic career environment (Emmerik, 2004).

The strength of a tie and the type of mentoring support that can be provided through a tie are two research fields in social network theory that have so far been studied separately, while they might have related characteristics. Psychosocial mentoring ties provide mental support in the form of acceptance and confirmation and can be seen as friendship ties (Higgins, 2000). This might be related to the characteristic of a strong tie which communicates similar information and will therefore be useful to confirm existing thoughts (Granovetter, 1973). Instrumental mentoring ties, on the other hand, provide information and advice about work-related subjects (Higgins, 2000), which might be most useful when communicated through a weak tie since these provide new information (Granovetter,1973).

Theorizing about the effect of mentoring networks on job satisfaction should be done with consideration for personality traits of an individual. The proactive component of a personality appears to influence career prospects (Seibert, Crant, & Kraimer, 1999), such that individuals with these personality traits improve their personal circumstances (Crant, 2000) and, as a consequence, report greater job satisfaction (Seibert et al., 1999). Moreover, proactive individuals seek more interaction (Bateman & Crant, 1999) and have better relationships with their mentors (Li, Liang, & Crant, 2010), which suggests that the effect of mentoring networks can be different for individuals with a proactive personality compared to individuals with a non-proactive personality.

This study builds further on the literature about the content of mentoring networks by investigating the combined effects of the strength and type of mentoring ties on job satisfaction. Additionally, the moderating effect of proactive personality on this relationship is studied. Therefore, this research will examine the following question:

(9)

8 What is the relationship between the content of a mentoring network and job satisfaction, and to what extent does proactive personality moderate this relationship?

This research makes a contribution to the existing literature by integrating two research streams of social network theory, namely tie strength and types of mentoring ties. These research areas have so far been treated separately, while the strength of ties and mentoring types have corresponding characteristics. By specifying the combined effect of the strength of ties and mentoring tie types this research will be able to investigate if the ‘strength of weak ties’ (Granovetter, 1973) also applies for mentoring ties. Besides, it is of importance to managers to have an understanding about how to increase their employees’ job satisfaction. Before investing in an extensive network, knowledge should be gained about the effects of mentoring relationships and the conditions these relationships should meet.

The second contribution of this research is made by investigating if proactive personality is a moderating factor on the relationship between mentoring networks and job satisfaction. Previous research studied the direct effects of proactive personality on diverse outcomes, but has not examined the possibility of indirect effects. Considering that someone’s personality influences his or her social network, it is reasonable to expect that proactive personality also moderates the feeling of job satisfaction that is gained from different ties. Knowledge about these effects of personality characteristics is important for managers, especially with regard to non-proactive employees. Proactive employees will take advantage of their mentoring network and create opportunities for themselves, whereas non-proactive employees will not make optimal use of their network. Managers could either stimulate the non-proactive employees to take more initiative or inform the mentors of these non-non-proactive employees and create circumstances that encourage the employee to behave more proactively.

(10)

9

2. Literature review

2.1 Social networks

The literature on social networks is comprehensive; therefore an explanation about the meaning of a social network is discussed first. This concept implies the presence of multiple relationships among actors. These actors can be seen as separate ‘nodes’ that are connected to each other through ‘ties’. These ties enable resources and knowledge to flow from one node to the other. If the network is seen from the viewpoint of one of the nodes, this one is called the ‘ego’, whereas the other nodes are called ‘alters’ (Brass et al., 2004). The ego can have ties with different alters, and these alters can be interconnected as well. This research focuses on ties from an ego’s perspective and therefore takes a dyadic approach. Studying relationships on the dyadic level provides insights into the contribution of each dyadic tie on job satisfaction and allows the results to show specific effects of different ties (Sosa, 2011). The term ego will hereafter be referred to as the individual.

Moving on from terminology to research, one can conclude that the effects of social networks have been studied extensively. A lot is known about how social ties can influence an individual’s satisfaction at work, performance, commitment to the organization, career prospects, and probability of resignation (e.g. Flap & Völker, 2001; Markiewicz, Devine, & Kausilas, 2000; Robert & O’Reilly, 1979; Seibert, Kraimer, & Crant, 2001; Seibert, Kraimer, & Liden, 2001). Upon this moment, reviews show that social networks seem to be correlated with positive outcomes, such as career satisfaction, and decrease negative outcomes, such as employee turnover, (Borgatti & Foster, 2003). Besides, studies have deepen the insights on the antecedents and consequences of social networks from different levels of analysis (Brass et al., 2004). Of interest for this research are the findings on the individual level of analysis. Evidence has shown that social networks often consist of similar actors, such that individuals tend to interact with others that are similar to themselves (Reagans, 2005). As a consequence of this

(11)

10 relationship both actors become more similar. When an individual has more of these ties with similar others, this leads to more satisfaction, especially in small organizations (Roberts & O’Reilly, 1979). However, not all ties exist out of voluntary choice; due to the organizational structure, an individual can have ties with dissimilar others they preferably do not interact with, implying that not all ties necessarily lead to positive outcomes. Negative interactions appear to have a bigger influence on feelings, such as satisfaction, than positive interactions, negative ties can therefore overshadow the effect of positive ties (Labianca & Brass, 2006). Due to the dyadic approach, this study is able to separate the effects of the negative ties from the positive ties and prevents that the effects of negative ties can overshadow the effects of positive ties.

Besides focusing on social network theory about the general existence of a tie between two individuals, the literature has also directed much attention to the role these individuals have in the relationship, referring to the roles of a mentor and mentee (Eby et al., 2013; Underhill, 2005).

2.2 Mentoring networks

Traditional research on mentoring relationships describes this concept as the dyadic relationship between a mentor and a protégé, where the mentor has experience and holds a senior position and can thereby provide the protégé with developmental assistance (Higgins & Kram, 2001). Later research dissociated from the thought of mentors being higher ranked and showed that mentoring relationships could occur on any level (Higgins, 2000). Individuals with mentors seem to have better career prospects with higher promotion rates, higher salaries, more career and job satisfaction, and lower turnover rates (Allen et al., 2004; Craig, Allen, Reid, Riemenschneider, & Armstrong, 2013; Emmerik, 2004; Seibert et al., 2001). Although a mentoring relationship exists between two individuals, a protégé can have multiple different mentoring ties. Kram (1985, in: Higgins & Kram, 2001) was the first to suggest that individuals

(12)

11 could benefit from having an entire mentoring network and receiving support through multiple ties. By surrounding oneself with more developmental relationships in the form of mentors, individuals feel more supported and will feel more intrinsically motivated to develop and pursue their ideas, which leads to a feeling of job satisfaction (Sosa, 2011). Evidence proves that an individual should get as much mentors as possible; the bigger the social mentoring network, the higher the level of satisfaction (Emmerik, 2004; Higgins, 2000). Moreover, this mentoring network can be created over a period of time, not all ties have to be made at once as networks can evolve. It is probable that an individual’s network at the beginning of employment consists of a few mentors (Baugh & Scandura, 1999) and over time this will evolve to more and diverse mentors (Higgins & Kram, 2001). Furthermore, over time new mentoring ties can arise while some others will fade away, for example due to a mentor leaving the organization. However, the information gained through these mentoring ties goes beyond the existence of the tie (Kram, 1985, in: Higgins & Kram, 2001).

Besides the positive effects of mentoring networks itself, the literature provides insights on the comparison of individuals with and without mentoring ties (Allen et al., 2004; Fagenson, 1989; Underhill, 2006). Outcomes as job satisfaction, career chances, promotions, and opportunities showed significantly stronger positive effect for protégés with a mentor, than for others without a mentor. The job level in the organization was also taken into account and it seemed as if both employees with low-level positions and high-level positions in the organizations were more satisfied if they reported to have a mentor. This means that employees on every level of the organization can benefit from having a mentor (Fagenson, 1989). Higgins and Nohria (1999, in: Higgins, 2000) conducted a similar research on the effects of career stages. They assumed that employees start in a low-level job position in the early career stage and move on to a high-level position in the later career stage and found evidence that mentors are meaningful both in earlier and in later career stages.

(13)

12 The findings in the literature about social network ties are unambiguous; career prospects are more positive for individuals with a social network (Borgatti & Foster, 2003) and even more for individuals with mentors in their network (Allen et al., 2004; Underhill 2006). To consolidate this knowledge further, the foundation of this research will be based on the assumptions that the general presence of ties as well as the presence of mentoring ties will have positive effects on job satisfaction.

2.3 Strength of ties

So far the strength of ties has not come to discussion, although it is an important part when it comes to investigating social networks. The strength of ties is defined by Granovetter (1973, p. 1361) as: “The strength of a tie is a (probably linear) combination of the amount of time, the emotional intensity, the intimacy (mutual confiding), and the reciprocal services which characterize the tie”. Based on these characteristics, different ties can be divided into the groups strong, weak and absent. The research of Granovetter (1973) was the first to emphasize on the possibilities and effects of different strengths of ties, paradoxically stating that weak ties open new doors of information and that strong ties are characterized by an information overlap. Considering that individuals tend to interact with others who are similar to them (Reagans, 2005; Roberts & O’Reilly, 1979), these individuals will spend a great amount of time with each other and, therefore, a strong relational tie will arise. Choosing similar connections and becoming even more similar rises the expectation that these individuals have similar information or that this is at least to a large extent overlapping. Granovetter (1973) therefore argued that strong ties can be used to intensify the thoughts one already had, but there is a small chance that new information will flow through these ties. He also states that weak ties have the opposite effect. Since a weak tie implies that the individuals in the particular relationship do not interact often, there is an increased chance that they are owners of dissimilar information.

(14)

13 Therefore, weak ties are a source of new information (Granovetter, 1973). After this proposition, scholars have tried to explore the consequences of having strong and weak ties (Higgins & Kram, 2001; Podolny & Baron, 1997; Reagans, 2005; Seibert et al., 2001; Sosa, 2011; Zagenczyk, Gibney, Murrell, & Boss, 2008). The strength of weak ties and the implication about new information was proven to occur in a wide variety of situations. Increased creativity was found at individuals with weak ties, explained by the reasoning that new insights and information flowing through weak ties can lead to creativity (Sosa, 2011). On the other hand, the same research showed that strong ties can be a source of creativity because individuals with a strong relationship feel positive affect towards one another and have the intrinsic motivation to work together. This implies that strong and weak ties can both have positive effects, although in different manners.

Reagans (2005) examined the combination of strong and weak ties in someone’s social network and argued that the form of a social network is determined by three mechanisms that influence one another: the basic preference for social contact with similar others, the preference for identification, and the preference for competition. The basic preference for social contact and the need for identification are pursued through developing strong ties with similar others, while striving for competition reduces the search for similarity. If competition among similar colleagues increases, they will seek less interaction with each other and more with dissimilar others. This competition aspect will make existing relationships weaker. This suggests that strong ties are developed purposefully to gain positive effects, whereas weak ties are the consequence of the avoidance of a certain relationship. Another way to describe identification and competition is through the need for social acceptance and the need for career growth (Higgins & Kram, 2001). A feeling of recognition and social acceptance arises through confirmation of one’s thoughts and opinions, obtained through a strong tie. Meanwhile, for career growth one needs new information about job opportunities and sponsorship, which can

(15)

14 best be obtained through weak ties. Consequently, this argument assigns positive effects to both weak and strong ties.

Yet, some researchers argue that a social network should not contain both weak and strong ties. Podolny and Baron (1997) state that the strategic decision should be made to choose between investing energy in having weak ties or strong ties, while Seibert et al. (2001) argue that it is better to have weak ties because that gives an individual the advantage to serve as a bridge between workgroups. They specify these weak ties by arguing that it should be ties with individuals having different jobs or higher-level jobs.

Zagenczyk et al. (2008) discuss that the literature on social networks shows differential effects of weak and strong ties on task mastery, role clarity, the transferring of complex and noncomplex knowledge, suggesting that both weak and strong ties have separated beneficial effects. Their research, on the other hand, provides evidence that strong ties had the strongest positive relation with organizational behavior, which in turn led to more job satisfaction. This statement is emphasized by research that indicated a positive effect of strong community ties on career satisfaction, as people derived socioemotional support from these ties (Martins et al., 2002). The context of receiving support from the community might differ from the support one receives from a colleague, but the underlying mechanism of perceived support through a strong tie can be transferred to a work-related context.

Considering the literature on strength of ties, most research shows positive effects of both weak and strong ties and emphasizes that these ties should be used to gain different outcomes. Yet, the effects on job satisfaction lack thorough investigation and since the studies of Zagenczyk et al (2008) and Martins et al. (2002) raise the expectation that strong ties are the strongest predictors of job satisfaction, it is expected that strong ties contribute to more job satisfaction than weak ties.

(16)

15 Hypothesis 1: Strong ties lead to higher amounts of job satisfaction than weak ties.

Research on the strength of ties has not focused a lot of attention on mentoring networks. Higgins and Kram (2001) have specified expectations about the strength of mentoring ties. They argue that mentoring relationships that lead to the best outcomes are characterized by reciprocity (Higgins & Kram, 2001), as such that the mentor is willing to provide the protégé with profound advice or support and the protégé will accept this advice and provide feedback. This reciprocal exchange through mentoring ties is only meaningful with high levels of social interaction. In other words, the willingness to help each other is a characteristic of a strong tie. In addition, weak mentoring ties can be experienced as a signal of a lack of trust in the protégé’s potential and can increase stress and a feeling of insecurity. These observations suggest that positive outcomes are expected for strong mentoring ties and negative outcomes are expected for weak mentoring ties.

Hypothesis 2: Strong mentoring ties lead to higher job satisfaction than weak mentoring ties.

2.4 Types of mentoring ties

Multiple scholars have performed research into mentoring networks and have extended the knowledge about the amount (Higgins, 2000; Oh, Chung & Labianca, 2004), type (Eby et al., 2013), density, and range (Podolny & Baron, 1997; Higgins & Kram, 2001) of mentoring ties. Research indicates that having more mentoring ties is inherent to more job satisfaction (Higgins, 2000). In the existing literature mentoring ties are divided into two segments, however there is no consistency in the conceptualization of these segments. Instrumental versus psychosocial support (Eby et al., 2013), strategic versus trusting ties (Flap & Völker, 2001), career versus psychosocial mentoring (Craig et al., 2013), and career versus psychosocial assistance (Higgins,

(17)

16 2000) are some examples of the terminologies that are used to describe the same separation between work-related and social-related ties. The first describes the need for advice, coaching, and career sponsorship and the second describes the need for social acceptance, recognition, and mental support (Higgins & Kram, 2001).

This is a purposeful separation because actors invest in ties with a specific goal in mind. The work-related ties, further mentioned as instrumental ties, enhance one’s satisfaction about the functional part of the job, whereas social-related ties, further mentioned as psychosocial ties, enhances one’s satisfaction with the emotional part of the job (Flap & Völker, 2001). According to them, these different ties only fulfill their own purpose and do not have an effect on the other part of the job. Oh et al. (2004) suggest in their research on informal social ties that it is not always better to have multiple social ties of one type, but that one should strive for diversity in one’s social network ties. These findings are supported by Higgins (2000) who assumes that it is necessary to satisfy both the functional and the emotional part of the job to achieve the highest degree of job satisfaction and, therefore, argues that both types of mentoring ties need to be present in the individual’s social network. Yet, she specified this further by examining the separated effects of psychosocial and instrumental mentoring ties and showed that psychosocial support predicted job satisfaction to a greater extent.

So far, instrumental and psychosocial mentoring ties have repeatedly been investigated and discussed as separate mentoring ties in a social network. Left out of consideration is the fact that a protégé can receive both types of support from one mentor, while Oh et al. (2004) argue that in practice it is reasonable that mentor relationships develop and become sources of different types of mentoring support. These developed ties were found to be the most robust providers of social capital. Even though the research of Oh et al. (2004) did not focus on job satisfaction, it does give insights into the development of mentoring ties.

(18)

17 In sum, the literature suggests that both instrumental and psychosocial mentoring ties lead to job satisfaction and that the effect of psychosocial mentoring ties accounts for a greater part of this effect (Higgins, 2000). Furthermore, it illustrates that mentoring ties over time can develop into sources of both types of mentoring (Oh et al., 2004) which can cause even higher job satisfaction. Considering these findings, the hypothesis will be:

Hypothesis 3: A tie that provides psychosocial mentoring support leads to higher job

satisfaction than a tie that provides instrumental mentoring support, and a tie that provides both psychosocial and instrumental mentoring support leads to the highest job satisfaction.

2.5 Strength versus type of mentoring tie

In addition to the findings that psychosocial mentoring ties have more effect on job satisfaction than instrumental mentoring ties, Higgins (2000) indicated that psychosocial support had the strongest positive influence on job satisfaction when a mentor provided high amounts of support. Frequent interactions are needed to provide this support, which implies that psychosocial mentoring is most effective when flowing through a strong tie. Although these results are interesting, they disregard the effects of instrumental mentoring ties, which proposes new options for research into the strength of mentoring ties. Psychosocial mentoring might be most useful coming from a strong tie because emotional support is most meaningful when it comes from someone who one interacts with often and can be considered as a friend. In addition, Higgins (2000, p. 290) claims: “It only takes one good friend to make an individual feel satisfied at work”. On the other hand, instrumental mentoring might be most effective when provided through weak ties because these contain new information which can be used as advice- or career sponsorship. (Granovetter, 1973). When both types of mentoring are provided through one tie, this tie is probably strong for two reasons: 1) due to the exchange of both psychosocial

(19)

18 and instrumental support, the two individuals will interact more often, 2) for effective psychosocial support a tie must be strong and, therefore, overrules the preferred weakness of the instrumental part. In the current state of the literature, these two concepts have not been investigated together and, therefore, will be examined in this research to gain deeper insights. Assuming the previous, the hypotheses will be as followed:

Hypothesis 4a: A weak instrumental mentoring tie contributes more to an individual’s job

satisfaction than a strong instrumental mentoring tie.

Hypothesis 4b: A strong psychosocial mentoring tie contributes more to an individual’s job

satisfaction than a weak psychosocial mentoring tie.

Hypothesis 4c: A tie that provides both instrumental and psychosocial mentoring contributes

more to an individual’s job satisfaction when this tie is strong rather than weak.

Hypothesis 4d: A strong tie that provides both instrumental and psychosocial mentoring has a

stronger positive effect on job satisfaction than all other mentoring ties.

2.6 Proactive personality

Personality traits determine an individual’s behavior within an organization. Under the same circumstances, one employee may therefore behave differently than another. A component that determines an employee’s organizational behavior can be the rate of proactivity, or in other words having a proactive personality. Research has shown that the proactive component can significantly affect a career in terms of promotions, increasing salary, satisfaction with one’s career (Seibert et al., 1999), entrepreneurial intentions (Crant, 1996), perceptions of being a charismatic leader (Crant & Bateman, 2000), and job performance (Thompson, 2005). Having a proactive personality describes the urge to take initiative and to make changes to improve personal circumstances (Crant, 2000). Moreover, Bateman and Crant (1993) argue that

(20)

19 proactive personality is related to the internalization of the personality factors extraversion and conscientiousness. Extraversion is defined as the need to be assertive, to feel stimulated and to interact with others, while conscientiousness describes the need for goal-directed behavior. More specifically, people with these characteristics feel the need for achievement and dominance and are willing to participate in extracurricular activities to enable changes to happen. These people are not affected by situational forces but rather are the ones who are shaping the environment (Bateman & Crant, 1993). People who do not have a proactive personality show the opposite behavior. They behave in a passive manner, they respond reactively to situational circumstances and adapt to changes rather than take action (Bateman & Crant, 1993).

As individuals with a proactive personality tend to seek more and intense interaction with others (Bateman & Crant, 1993; Li et al., 2010), they are expected to approach their mentors more often and gain emotional support or knowledge about identifying opportunities through these social interactions. When individuals seek more interaction and therefore will gather information, they will have more knowledge than others who do not proactively seek interaction. This information exchange has been described by Dansereau, Graen and Haga (1975) as the leader-member exchange (LMX). A higher quality relationship between leader and member leads to more and better exchange of social support and information (Joo & Ready, 2012). Research has shown that individuals with a proactive personality have a better relationship with their supervisor than individuals without a proactive personality. Besides, supervisors are willing to provide more support to proactive individuals (Li et al., 2010). In this research these supervisors can be seen as mentors if they provide individuals with developmental assistance, for instance about career opportunities or in the form of emotional support. Career prospects, such as making promotions, as well as social aspects, such as being satisfied, are positively related to having a proactive personality (Seibert et al., 1999).

(21)

20 Considering that, on the one hand, individuals with a proactive personality are seeking more interaction with their mentors (therefore these ties become stronger) (Bateman & Crant, 1993) and that, on the other hand, individuals are more satisfied when they perceive to have a good relationship with their supervisor (or in this case their mentor) (Joo & Ready, 2012), it can be assumed that individuals with a proactive personality will perceive more job satisfaction through their strong mentoring ties. Individuals with a non-proactive personality do not naturally seek for more interaction with their mentors, and therefore, their strong mentoring ties will presumably contribute to even amounts of job satisfaction as their weak mentoring ties. As a result proactive personality seems to have a moderating effect on the relationship between mentoring ties and job satisfaction. Since the literature lacks insights into the effect of proactive personality on different types of mentoring relationships, this research will provide insights on this phenomenon. Summarizing the above mentioned literature, respondents with a proactive personality will gain more job satisfaction from strong mentoring ties than from weak mentoring ties, while respondents with a non-proactive personality will gain even amounts of job satisfaction from strong and weak mentoring ties.

Hypothesis 5: Proactive personality has a moderating effect on the relationship between

mentoring ties and job satisfaction, such that for individuals with a proactive personality, strong mentoring ties lead to higher job satisfaction than weak mentoring ties. While for individuals with a non-proactive personality, strong and weak mentoring ties lead to even amounts of job satisfaction.

(22)

21 In sum, this research focuses on the effects of the strength of ties and type of ties on job satisfaction and the moderating influence of proactive personality on this relationship. The predictions of this study are summarized in the conceptual model as presented in Figure 1.

Figure 1

Conceptual model

METHOD

3. Method

3.1 Participants and procedure

This research was conducted at a Dutch company that runs an insurance and energy comparison website. The company has 32 employees, consisting of full-timers, part-timers, and interns. This specific company was suitable for the research due to a few advantages. First of all, the employees in the company are divided over two separate floors. The first floor is the department of the employees who run the daily operations such as IT, finance, marketing, and business development, and the second floor is the customer service department. It is a small company, but due to this separation in workplaces there is a chance that there is a lack of ties between employees working on both floors. Since the customer service department is smaller than the other departments, this could mean that these employees have a smaller and different social

Mentoring network - Type of ties - Strength of ties Job satisfaction Proactive personality

(23)

22 network than colleagues that are working on the first floor, which can influence their feeling of job satisfaction.

Another reason for choosing this company is the high circulation rate of interns. Especially in the beginning of an internship, interns will have a different social network in terms of size and strength than employees who have been working at the company for a longer period of time. Besides, at the start of the internship, the interns are assigned a supervisor who guides them during their internship, whom they work together with on a regular basis and, therefore, will probably be a mentor to them. At the moment of this study, the new interns were working for the company for 2,5 months. Therefore they have had time to invest in their social network. However, their relationships will presumably not be on the same level of depth as those of the other employees (Baugh & Scandura. 1999).

Data for this research was conducted by means of an online survey (see Appendix 1 and Appendix 2 for Dutch and English example). This measurement was most appropriate for this study for three reasons. First of all, the method provides anonymity and was therefore best suitable to collect reliable data. Second, it allowed respondents to reply at a moment that was most convenient which benefits the quality of the given answers. At last, a reminder could be sent to respondents to increase the response rate.

Before the survey was sent, an informational interview with one of the managers was held to verify if all survey questions were valid and to control if the terminology was understood as it was intended. The final survey was sent by e-mail containing introductory information about the research, a statement of confidentiality and the link to the online questionnaire. Two reminders were sent which resulted in 30 employees to respond to the request of participation, which gave a response rate of 93,75%. This percentage was high enough for research on social networks, what required a minimum response rate of 80% (Sparrowe, Liden, Wayne, & Kraimer, 2001). Of the respondents, 67% were male with an average age of 26 years old.

(24)

23 Considering education level, 63% have participated in HBO and 30% in WO. Respondents worked at the company for an average period of 20 months and worked on average 4 days a week.

3.2 Measures

The survey was based on self-reported measures and all variables that consisted of two levels were made into dummy variables. The constructs that have been used were translated and back translated from English to Dutch, to make sure that the constructs were understood correctly.

Tie presence. To map the social network, all ties between employees needed to be

determined. This was measured using a fixed roster with all the employee names. The roster was presented to the respondents so that they could indicate for every colleague how often they communicated with him or her, varying between a few times a month, once a week, multiple times a week, once a day, multiple times a day, or less (Lee & Kim, 2011). Respondents were instructed not to fill in anything for colleagues they never communicated with or with whom communication occurs less than once a month. This option was treated as an absent tie and the names of these employees were excluded for further questions. Tie presence was an ordinal variable ranging from 0 to 5, with 0 being the absence of a tie and 1 to 5 specifying the frequency of communication through the present tie. This variable was recoded into a dummy variable with the value 0 for absent tie and 1 for present tie, the latter containing all values from 1 to 5. This separation between a tie being present and absent is based on the fact that the company has a small personnel database and the employees are working in the same small area. Communication with one another will therefore occur regularly. In case communication occurs less frequently than once a month, this tie can be seen as non-existent.

Strength of ties. This construct is based on the question about tie presence. If respondents

(25)

24 with a colleague (values 1 or 2), this tie was labeled as weak. When respondents indicated that they communicated multiple times a week, once a day or multiple times a day (values 3 to 5), this tie was labeled as strong, consistent with Lee and Kim (2011). By making this division, the variable became a dummy variable with the levels weak tie (value 0) and strong tie (value 1).

Mentor tie presence. The content of the social network will be specified further by

examining the presence of mentoring relationships. A roster showed the names of the employees, but only those names who were previously indicated as a present tie by the respondent. For all these employee names, respondents could state if they were seen as a mentor and what type of mentor. The type of mentor was not relevant for this construct and will be further explained beneath. All types of mentoring ties were taken together and coded as the mentor tie being present with the value 1 and the option ‘does not apply for this colleague’ was determined as the mentor tie being absent, with the value 0.

Strength of mentor ties. A variable had to be created to measure the effect of the strength

of mentoring ties. Two variables were therefore combined into a new variable: mentor tie presence and strength of ties. Since this variable measures the effects of present mentoring ties, all absent mentoring ties were excluded. The new variable was a dummy variable with the value 0 for weak mentoring ties and 1 for strong mentoring ties.

Mentor tie type. If respondents thought of another colleague as being a mentor to them,

they could specify if this was an instrumental mentoring tie, a psychosocial tie or if the mentor provided both types of mentoring. Instrumental mentoring ties were explained in the survey by stating that these are colleagues that one approaches to gain information or advice about work-related subjects. Psychosocial mentoring ties were explained by stating that these are colleagues that one approaches to discuss personal issues with, to get mental support or because these colleagues are perceived as friends (Higgins, 2000). When both types of support were present, this was determined as a colleague providing both instrumental and psychosocial support. This

(26)

25 is a nominal variable with values ranging from 0 to 3; 0 indicates the absence of a mentor tie and values 1 to 3 indicate the presence of a mentor tie and specify the type of mentoring tie.

Mentor type versus strength. The variables strength of ties and mentor tie type were

combined to create this construct. Absent mentoring ties were excluded. All present mentoring ties hereby contained information about the type of mentoring tie and the strength of the tie. This resulted in a nominal variable ranging from 1 to 6 with the levels: weak instrumental ties, strong instrumental ties, weak psychosocial ties, strong psychosocial ties, weak ties that provide both instrumental and psychosocial support, strong ties that provide both instrumental and psychosocial support.

Proactive personality. This construct was measured using six of the original 17 items of

the Proactive Personality Scale of Bateman and Crant (1993). The items used in this research appeared to have the highest factor loadings (Parker, 1998). An example of one of the items is ‘if I see something I don’t like, I fix it’. All items are measured on a 5-point Likert scale ranging from strongly disagree (1), to strongly agree (5). These items form a reliable scale with a reported Cronbach’s alpha of .85 (Parker, 1998).

Job satisfaction. Most prior research on job satisfaction was focused on overall job

satisfaction; however, for this study it was needed to gain information about how satisfied respondents were about their job due to their relationship with a specific colleague. Respondents were explained that colleagues could influence the way an individual feels about his or her work and could make them more or less satisfied. Respondents were shown a roster and were asked to complete the question ‘dealing with this colleague makes me feel…’ with answer options on a 7-point Likert scale ranging from very unsatisfied about my work (1) to very satisfied about my work (7).

(27)

26

4. Results

The next section will discuss the results in four parts. First, the confirmative factor analysis (CFA) and the reliability analysis for the proactive personality scale will be discussed. Second, the analytical strategy will discuss the preparation of the dataset and the adjustments that have been made to enable the execution of the analyses. Next, descriptive statistics and Pearson’s correlation coefficients are reported for the dependent, control and independent variables to indicate potential correlations between variables. The last part will discuss the analyses that have been conducted to test the hypotheses.

4.1 Scale development

The six items that measure proactive personality were tested for correlation and reliability using a confirmative factor analysis (CFA) and Cronbach’s alpha for reliability analysis. The Cronbach’s Alpha for the proactive personality scale showed high reliability (α = .70) (Field, 2013). The scale could not be improved by deleting one of the items and corrected item-total correlations were above .30 which indicates good correlations with the total score of the scale. The CFA was conducted to examine if the items were measuring the same construct. One factor came forward with an eigenvalue of 2.46 and an explained variance of 41.06%, meaning that the variance in the scale explained 41.06% of the variance in proactive personality. At last, all items clustered together on the factor with factor loadings higher than .40. Therefore, all six questions were combined and one continuous scale was computed for proactive personality.

4.2 Analytical strategy

Before the data could be used to conduct analyses, some preparations were made to clean the data set. Frequencies were run to disclose any missing values within the data. For the variables

(28)

27 strength of ties, mentoring presence, strength of mentoring ties, type of mentoring ties, and mentor type versus strength, a missing value was assigned when a tie was reported as absent.

Proactive personality was measured on a continuous scale, but for the regression analysis on interaction it had to be recoded into a dummy variable with the values non-proactive and proactive personality. Since this item ranged from 1 to 5, the middle and neutral value 3 should normally separate the non-proactive respondents from the proactive ones. In this case, this would not have be an equal division because respondents in this study reported high amounts of proactive personality. Four respondents reported scores beneath 3, opposed to 26 respondents reporting scores above 3. Therefore, it was decided to divide the respondents with a non-proactive and a proactive personality based on the average score (M = 3,56), which led to an equal division of respectively 52% versus 48% of the respondents.

To examine the interaction effect, for all six levels of the variable mentor type versus strength, a separate dummy variable needed to be created. Every dummy indicates with value 1 that the specific tie is present, and with value 0 that one of the other types of ties is present. This produced the six dummy variables instrumental, strong-instrumental, weak-psychosocial, strong-weak-psychosocial, weak-bothtypes and strong-bothtypes. Subsequently, interaction variables could be created to see the effect of proactive personality on each of the six levels by combining each of the six above mentioned dummy variables with the dummy variable proactive personality.

Another adjustment in the dataset was made for the absent ties. If respondents indicated that a tie was absent, the question about job satisfaction was left out for this tie. In this case, no values were given for the variable job satisfaction, arguing that colleagues who one has no tie with cannot influence the feeling of job satisfaction, neither positively nor negatively. For this reason, all absent ties where given the value ‘4’ on job satisfaction which means that the specific colleague did not influence the respondent’s job satisfaction.

(29)

28 At last, the ties were analyzed asymmetrically, meaning that the value of a tie was determined by the opinion of the respondent; these were not confirmed by the party on the other side of the tie. Striving for ties that are symmetrically confirmed emphasizes the existence of the tie. For this study, symmetry in ties was not feasible and partially not required. First, this research focused on a personal topic, namely the influence that a specific colleague has on job satisfaction. Respondents might therefore be hesitant to give truthful answers, worried that these answers become public. To avoid socially desired answers, it was important to guarantee respondent anonymity. Therefore, it was unattainable to find corresponding ties and the analyzed ties were based on an asymmetric opinion. For mentoring ties on the other hand, symmetry was not required. This tie can be perceived differently by two parties. A respondent can perceive a colleague as a mentor while the colleague does not perceive the respondent to be a mentee.

4.3 Correlations

Means, standard deviations and Pearson’s correlations coefficients among the dependent, control and independent variables are presented in Table 1. All control variables age, gender, education level, work duration, and work days a week showed correlations on a statistical significant level with at least one of the dependent variables. Even though the effect sizes were in general relatively small, they were entered into the regression analyses as control variables to ascertain that the effects of the examined relationships could not be assigned to alternative influences. Some correlations between variables could not be computed because one of the variables was constant. For example, strength of ties was only measured for present ties, what made the variable tie presence a constant variable where no correlations could be computed for. These are indicated in the table with a c.

(30)

29 Table 1

Means, standard deviations and correlations

Variables M SD 1 2 3 4 5 6 7 8 9 10 11 12 1. Job satisfaction 4.41 .77 - 2. Age 26.20 4.75 .01 - 3. Gender .67 .47 .02 .14** - 4. Education level 3.20 .65 .08 .08* .00 - 5. Employment duration 20.73 19.65 .04 .55** .29** .03 -

6. Work days a week 4.26 1.15 -.01 .36** .22** .06 .34** -

7. Proactive personality 3.56 .52 .15** .12** -.06 .09* .04 .12* - (.70) 8. Tie presence .73 .44 .32** -.18** -.07* .13** -.18** -.07* -.06 -

9. Strength of ties .54 .49 .34** .04 .11* -.14** .18** .23** .06 c -

10. Mentor tie presence .45 .49 .47** -.08* -.10* .07* -.06 .03 .03 .54** .37** -

11. Strength of mentor ties .68 .46 .28** .04 .17** -.17** .12* .14* .04 c 1.00** c -

12. Mentor tie type 1.88 .93 .34** -.17** .15* .02 .06 -.06 .04 c .27** c .27** -

Note. ** Correlation is significant at the .01 level (2-tailed)

* Correlation is significant at the .05 level (2-tailed)

(31)

30

4.4 Hypotheses testing

The hypotheses are tested using various hierarchical multiple regression analyses and a one-way ANOVA in case the data allowed this analysis to be performed. The control variables that were entered in the regression analyses were age, gender, education level, work duration, and work days a week. It can be expected that these variables have an effect on the emergence and content of a social network. Arguing that the longer an employee works at an organization and the more often this employee is at the office, the bigger the possibility that this employee has an extended social network. Besides, being older enlarges the amount of time an employee can be employed. Since the correlation table showed that these five variables correlated with the dependent variables, they should be taken into account. Controlling for these variables reveals the actual effect of the relationships that are examined in this research.

Before the hypotheses were tested, two regression analyses were conducted to ascertain the general effect of the presence of both common ties and mentoring ties on job satisfaction (Table 2). For the presence of common ties, the five predictors age, gender, education level, work duration, and work days a week were entered in the first step of the hierarchical multiple regression. This model was significant (F (5, 949) = 2.22, p = .04, N = 954). Education level showed a statistical significant effect (p < .01, β = .08), although the effect size was small. When tie presence was entered in step two, the model remained significant (F (6, 948) = 21.24,

p < .000) and explained 11,9% of the variance in job satisfaction. Both tie presence and work

duration were significant predictors of job satisfaction in the second model with a higher reported unstandardized beta for tie presence (p <.000, β = .34) than for work duration (p <.000,

β = .10). Tie presence seemed to have a positive effect on job satisfaction, meaning that if a tie

was present, job satisfaction increased with .60 on a scale of 1 to 7 (b = .60). The same predictors were entered in the next regression model to examine mentor tie presence. The first model was again significant (F (5, 949) = 2.22, p = .04, N = 954), showing statistical significance for

(32)

31 education level (p <.01, β = .08). Once mentor tie presence was entered in the second model, the explained variance of the model increased to such an extent that it explained 24% of the variance in job satisfaction (R2 Change = .23; F (6, 948) = 50.52, p < .000). Three predictors in

the second model showed statistical significance, with mentor tie presence as the strongest predictor (p < .000, β = .48; gender: p = .03, β = .06; work days a week: p < .01, β = -.08). The presence of a mentor tie suggests that job satisfaction increased by .75 compared to the absence of a mentor tie (b = .75). These results confirm the previous findings in the literature about the effect of ties.

Table 2

Results of hierarchical multiple regression for the general effect of ties on job satisfaction

Common ties Mentoring ties

Step 1 Step 2 Step 1 Step 2

Age -.01 .03 -.01 .03

Gender .01 .02 .01 .06*

Education level .08** .00 .08** .04

Work duration .06 .10** .06 .06

Work days a week -.04 -.04 -.04 -.08**

Tie presence .34*** .48*** b .60 .75 R .10 .34 .10 .49 R2 .01 .11 .01 .24 R2 Change .11 .23 Note. N = 954.

Standardized beta’s are reported for the entered predictors.

Statistical significance *p <.05 (two-tailed), **p <.01 (two-tailed), ***p <.000 (two-tailed).

4.4.1 Hypothesis 1

A hierarchical multiple regression analysis was conducted to test the effect of the strength of ties on job satisfaction, entering age, gender, education level, work duration, and work days a week in the first step and strength of ties in the second step, as shown in Table 3. The first

(33)

32 model was significant (F (5, 696) = 5.42, p < .000). Work duration appeared to have a significant positive effect on job satisfaction (p <.000, β = .17) and work days a week a significant negative effect (p =.05, β = -.08), but this model explained merely a small percentage of 3,8% of the variance in job satisfaction. Adding strength of ties in the second step increased the explained variance to 15% and showed a significant model (F (5, 696) = 20.55, p < .000). Small effects were presented for age (p = .04, β = .07), work duration (p < .01, β = .12) and work days a week (p <.000, β = -.14), the latter of which had a negative unstandardized Beta, suggesting that respondents who work more days a week report lower levels of job satisfaction. The strongest effect was presented for strength of ties which proved to be a significant predictor of job satisfaction. The positive effect means an increase in job satisfaction of .60 on a scale of 1 to 7 for strong ties compared to weak ties (p <.000, β = .35, b = .60) when age, gender, education level, work duration, and working days a week are kept constant. This confirms the first hypothesis that strong ties will lead to higher amounts of job satisfaction than weak ties.

Table 3

Results of hierarchical multiple regression analysis for strength of ties on job satisfaction

Step 1 Step 2

Age .06 .07*

Gender .02 .01

Education level -.01 .03

Work duration .17*** .12**

Work days a week -.08* -.14***

Strength of ties .35*** b .60 R .19 .38 R2 .03 .15 R2 Change .14 Note. N = 701.

(34)

33 4.4.2 Hypothesis 2

The second hypothesis describes the expectation of strong mentoring ties leading to higher job satisfaction than weak mentoring ties. In a hierarchical multiple regression, age, gender, education level, work duration, and work days a week are entered in the first step and strength of mentor ties is entered in the second step (Table 4). The first model was statistically significant (F (5, 426) = 4.59, p < .000), but the model explained only 5,1% of the variance in job satisfaction and the effects of the predictors work duration and work days a week were weak (gender: p = .02, β = .11 work duration: p = .02, β = .12; work days a week: p < .01, β = -.15). When strength of mentor ties is entered in step 2, the explained variance increases to 13,3%. This model is significant (F (6, 425) = 10.86, p <.000) with a strong positive effect for strength of mentor ties (p <.000, β = .29, b = .59). This means that reported job satisfaction is .59 higher for strong mentor ties than for weak mentor ties, confirming the second hypothesis.

Table 4

Results of hierarchical multiple regression analysis for strength of mentor tie on job satisfaction

Step 1 Step 2

Age .09 .07

Gender .11* .07

Education level .05 .10*

Work duration .12** .10*

Work days a week -.15** -.18***

Strength of mentor ties .29***

b .59

R .22 .36

R2 .05 .13

R2_Change _.12

Note. N = 431.

(35)

To examine the third hypothesis, a one-way ANOVA analysis is conducted. This hypothesis states that types of mentoring ties will have different effects on job satisfaction. Psychosocial mentoring ties are supposed to lead to more job satisfaction than instrumental mentoring ties, and ties that provide both instrumental and psychosocial support are supposed to have the strongest effect on job satisfaction. The ANOVA could be used since this research is non-experimental and the Levene’s test for homogeneity of variances showed equality of variance between the groups (p = .20). The analysis indicates a significant effect of the mentor tie type on job satisfaction (F (2, 429) = 29,00, p < .000), as presented in Table 5. The Bonferonni post-hoc test specifies that the difference between instrumental and psychosocial mentoring ties is not significant (p = .07), but statistical significant differences do exist between both types of mentoring ties and instrumental ties (p < .000) and between both types of mentoring ties and psychosocial mentoring ties (p = .02). Receiving both types of mentoring support through one tie appears to contribute to the highest feeling of job satisfaction, see Table 6. These results do partially support the hypothesis.

Table 5

Results of the one-way ANOVA for type of mentor tie on job satisfaction

SS DF MS F Sig.

Type of mentor tie 45.01 2 22.50 29.00 .000

Error 332.90 429 .77

Total 377.91 431

Table 6

Means, standard deviations and number of observations for the different types of mentor ties

Type of mentor tie M SD N

No tie 4.08 .34 523

Work-related ties 4.52 .89 219 Psychosocial ties 4.83 .93 48 Both types of ties 5.21 .85 165

(36)

To test the fourth hypothesis, the predictors age, gender, education level, work duration, and work days a week were entered in the first step of the hierarchical multiple regression analysis and six dummy variables were entered in the second step. In this way, the effect of every combination of strength and type of mentoring tie on job satisfaction could be pointed out while controlling for the effects of the other combinations. The first model was significant (F (5, 949) = 2.22, p = .04); education level showed statistical significance (p < .01, β = .08), but the explained variance was only 1,2%. The second model was statistically significant (F (11, 943) = 47.46, p < .000) and demonstrated an explained variance of 35%, meaning that the variance in the six dummy variables together explain 35% of the variance in job satisfaction. All predictor variables that were entered in the second model present significant effects, except for weak psychosocial ties (p = .12, β = .04). Age, education, and work days a week are the predictors with the smallest effect size. Age and education level report a positive effect of respectively .07 (p = .05) and .05 (p = .03), and work days a week reports a negative effect of -.07 (p < .01). The remaining dummy variables demonstrate different effect sizes, presented in Table 7. Weak instrumental ties appeared to be a significant predictor of job satisfaction, but the effect size is negligibly small (p = .05, β = .05, b = .16), meaning that job satisfaction increases .16 on a scale of 1 to 7 for weak instrumental ties. Strong instrumental ties, on the other hand, have a positive statistical significant effect that is relatively strong (p < .000, β = .29, b = .68); these ties account for an increased job satisfaction of .68. Strong psychosocial ties and weak ties that provide both instrumental and psychosocial support have the same positive effect size (strong psychosocial tie: p < .000, β = .24, b = 1.02; weak both types of ties: p < .000, β = .24, b = 1.11). The existence of these ties increases job satisfaction by 1.02 for strong psychosocial ties, and 1.11 for weak ties that provide both instrumental and psychosocial support. The strongest effect size is appointed to strong ties that provide both instrumental and

(37)

36 psychosocial support. This type of tie is the strongest predictor of job satisfaction when all other variables are controlled for (p < .000 β = .51, b = 1.14). The presence of this type of tie increases job satisfaction by 1.14 on a scale of 1 to 7.

Table 7

Results of hierarchical multiple regression analysis for mentor type versus strength on job satisfaction Step 1 Step 2 B Age -.01 .07* Gender .01 .01 Education level .08** .05* Work duration .06 .03

Work days a week -.04 -.07**

Weak instrumental ties .05* .13

Strong instrumental ties .29*** .68

Weak psychosocial ties .04 .26

Strong psychosocial ties .24*** 1.02

Weak both types ties .24*** 1.11

Strong both types ties .51*** 1.14

R .10 .59

R2 .01 .35

R2 Change .34

Note. N = 954.

Statistical significance *p <.05 (two-tailed), **p <.01 (two-tailed), ***p <.000 (two-tailed).

These results support three out of the four sub-hypotheses. H4a predicted that weak instrumental mentoring ties would lead to higher levels of job satisfaction than strong instrumental mentoring ties. The opposite was confirmed, strong instrumental ties were stronger predictors of job satisfaction than weak instrumental ties. Therefore this hypotheses is rejected. H4b assumed that strong psychosocial mentoring ties would contribute to more job satisfaction than weak instrumental mentoring ties, and this hypotheses was confirmed by the regression analysis. Strong psychosocial ties were relatively strong predictors of job satisfaction, whereas weak psychosocial mentoring ties reported no statistical significance as a predictor of job

(38)

37 satisfaction. H4c was supported as well. This hypothesis stated that ties that provide both instrumental and psychosocial mentoring would contribute more to job satisfaction when these ties are strong rather than weak. Although the effects sizes are significant and favorable for both, the results show indeed that strong ties that provide both instrumental and psychosocial mentoring are stronger predictors of job satisfaction than weak ties that provide both types of mentoring. Finally, H4d predicted that strong ties that provide both instrumental and psychosocial mentoring support would have the strongest positive effect on job satisfaction compared to all other mentoring ties. Considering the results, this hypothesis is confirmed. This combination between type of mentoring ties and strength of ties showed a positive effect that was stronger than the effect of all other mentoring ties.

4.4.5 Hypothesis 5

For the fifth hypothesis, the moderating effect of proactive personality was examined by means of a hierarchical multiple regression analysis. Three steps were taken to determine the effect of the interaction variables without influencing effects of other variables. To establish this, the first step controlled for age, gender, education level, work duration, and work days a week. In the second step, proactive personality and all six dummy variables weak instrumental tie, strong instrumental tie, weak psychosocial tie, strong psychosocial tie, weak ties that provide both types of mentoring, and strong ties that provide both types of mentoring were entered. In the third and last step the interaction variables were entered. An overview of the regression results is presented in Table 8.

The first model was significant (F (5, 789) = 9,75, p < .000) and revealed small significant effects for education level (p = .02, β = .08) and work days a week (p = .02, β = -.08) and a relatively strong positive effect of work duration on job satisfaction (p < .000, β = .20), meaning that respondents that have been working at the company for a longer period of

Strong social mentoring networks : a research on the relationship between tie strength and mentor types on job satisfaction, and the moderating effect of proactive personality