The effect of power differences on creative performance in an organizational social network

(1)

The effect of power differences on creative

performance in an organizational social network

Erwin Dijkstra - 6069436

University of Amsterdam - Master Thesis MSc Business Administration Track: Strategy

Academic year: 2014-2015 Supervisor: Nathan Betancourt

(2)

Statement of originality

This document is written by Erwin Dijkstra 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

This thesis investigates the effects of psychological safety and power on the relationship between tie strength and creative performance in the friendship and advice network. Several studies have indicated that tie strength is positively associated with creative performance on a dyadic level, but the effects of psychological safety on this relationship have not been studied yet. Relationships inside organizational contexts are constructed differently compared to relationships outside an organization, implying that strong ties inside an organization are not necessarily friends who trust each other. Power is regarded as a possible driver of psychological safety, so the effect of power will also be examined.

To examine the effects of psychological safety and power on the relationship between tie strength and creative performance, an empirical study was conducted in one organization, consisting of 26 Dutch employees. The following effects were tested: direct effects of tie strength on creative performance, the moderating effect of psychological safety, and finally, the three-way interaction effects of different values of psychological safety and power.

Results show that psychological safety and power significantly affected the relationship between tie strength and creative performance, such that low psychological safety and high power positively affected the relationship. On the other hand, high psychological safety and low power also affected the relationship. The results will be discussed at the end of the thesis.

(4)

List of tables and figures

Tables

Table 1 Means, standard deviations, correlations and reliabilities 21

(friendship network)

Table 2 Means, standard deviations, correlations and reliabilities 22

(advice network)

Table 3 Summary of Regression Analysis Results: Psychological safety as a 26

moderator of the relationship between friendship network tie strength and creative performance

Table 4 Summary of Regression Analysis Results: Psychological safety as a 27

moderator of the relationship between advice network tie strength and creative performance

Table 5 Summary of Regression Analysis Results: Psychological safety 31

and power as moderators of the relationship between friendship network tie strength and creative performance

Table 6 Summary of Regression Analysis Results: Psychological safety 32

and power as moderators of the relationship between advice network tie strength and creative performance

Table 7 Three-way interactions. Coefficients present the conditional 33

effects of tie strength on creative performance with different values of the moderators

Figures

Figure 1 Conceptual model of discussed variables 15

Figure 2 Moderation of psychological safety on the relationship between 29

tie strength in the advice network and creative performance

Figure 3 Moderation of psychological safety and power on the 35

relationship between tie strength in the friendship network and creative performance

Figure 4 Moderation of psychological safety and power on the 36

relationship between tie strength in the advice network and creative performance

Figure 5 Friendship network web 54

(6)

1. Introduction

Resource-based scholars emphasize the importance of knowledge creation and sharing to gain sustained competitive advantage. Scholars argue that firms with valuable, rare, inimitable, and non-substitutable resources have a higher probability of achieving better firm performance and innovative capabilities than its competitors (Wiklund & Shepherd, 2003). Several studies have shown that creative employees are more likely to bring up innovative ideas (Perry-Smith & Shalley, 2003; Baer, 2010). Grant (1996) states that knowledge can be a very valuable resource as it can be hard to imitate or substitute this type of resource. Social network researchers have proven that both weak and strong interpersonal relationships are important contributors to increase individual’s creativity (Baer, 2010; Perry-Smith, 2006; Sosa, 2011; Zhou et al., 2004). Past research has focused on different aspects of the relationship between tie strength and creativity. For instance, many scholars argue that weak ties are more effective in enhancing creativity when the focus lies in the number of ties on a person level (Granovetter, 1973; Friedkin, 1980). Weak ties are expected to boost creativity more than strong ties because the information exchange between individuals is less likely to be redundant (Perry-Smith, 2006). Other studies show that strong ties are more likely to have knowledge of each others’ skills and expertise, have access to each others knowledge, are willing to help each other, and feel psychologically safe, which are all factors that are conducive to creative performance (Sosa, 2011; Gong et al., 2012).

Little research has focused on the effects of psychological safety on creative performance. An individual feels psychologically safe when he feels comfortable asking someone else for advice or shares ideas (Siemsen et al., 2009). Cross et al. (2001) argue that psychological safety promotes creativity and learning, while Siemsen et al. (2009) state that safe relationships are more likely to occur when individuals are strongly connected. These finding imply that the degree of creative performance does not only depend on tie strength, but also on psychological

(7)

safety. No research has examined if the positive relationship between tie strength and creative performance depends on the degree of psychological safety. Strong ties in organizations are not necessarily trusted relationships, because these relationships are not always voluntary, meaning that employees work together, even though it is possible they do not prefer working together. Furthermore, the definition of psychological safety claims that the feeling of safety might depend on who an employee talks to. Judging by the definition of psychological safety, it seems reasonable to assume that employees feel less psychologically safe when they talk to their manager than when they talk to someone who is equal. This thesis aims to get a deeper understanding of the influence of psychological safety and power on the relationship between tie strength and creative performance.

This thesis has important contributions to the existing theory and practice. This thesis outlines the importance of having strong working relations with colleagues regarding the perceptions of creative performance of colleagues, meaning that managers should stimulate their employees to build stronger work relationships with other colleagues. This thesis also sheds light on how psychological safety affects creativity. Psychological safety does not depend on tie strength per se, indicating that strong ties do not necessarily feel comfortable working with each other. Furthermore, even when employees are not close with each other they perceive colleagues they feel comfortable with asking for advice as more creative. This indicates that organizations should adapt a culture in which employees feel the freedom to share ideas and ask questions, especially when employees approach powerful colleagues, since power negatively affects the feeling of psychological safety.

The thesis will be structured as follows. In the next chapter, all relevant theories and related researches will be discussed. In addition, the hypotheses of this thesis will be formulated. In the third chapter, the research method and design will be explained. The results will be discussed in the fourth chapter. The fifth chapter clarifies the most significant findings. In the last chapter a conclusion will be drawn.

(8)

2. Literature Review

In this chapter, all relevant theories and researches that have been conducted in the past will be discussed. The theories are derived from different fields of literature, like psychology, organizational learning, management and innovation, and social network analysis. Furthermore, research methods and results are discussed to show where these theories are based on. The first section will discuss why creativity is important to organizations and how social networks contribute to creativity. The second section discusses how tie strength is defined and how it is usually measured. In the third part, theories about psychological safety will be outlined. In the final section, power will be defined and linked to tie strength and psychological safety. Finally, a conceptual model of the discussed variables will be presented.

2.1 Knowledge flow and creative performance

Resource-based scholars emphasize the importance of knowledge to gain sustained competitive advantage. Scholars argue that firms with valuable, rare, inimitable, and non-substitutable resources have a higher probability of achieving better firm performance and successful innovation (Wiklund & Shepherd, 2003). Innovation basically means that organizations have the ability to successfully implement new ideas. However, new ideas do not arise out of nowhere. While innovation focuses on the implementation of new ideas, creativity is the basis of the production of new ideas. As Gurteen (1998) states, creativity is the first step in the process of implementing new ideas. The second step is innovation, which aims to implement these new ideas. Therefore, creativity is a key driver of innovation (Amabile, 1988). It is crucial for organizations to be dynamic and capable to innovate at high speed, especially in volatile markets (Teece, 2007). Leonard and Sensiper make a distinction between tangible and intangible resources and they emphasize the importance of tacit knowledge in order to gain sustained competitive advantage (1998). Tacit knowledge is defined as complex knowledge,

(9)

which is difficult to transfer to others, while explicit knowledge is easier to transfer to other individuals. Grant (1996) argues that knowledge and the integration of knowledge are important determinants of gaining sustained competitive advantage, because useful knowledge can be valuable, rare, inimitable and non-substitutable. Psychological studies of creativity primarily focused on cognitive and motivational processes inside the brain of an individual to get a better understanding of why some individuals are more creative than others. Social network and innovation researchers conducted studies concerning interpersonal relationships, knowledge exchange, and creativity (Zhou et al., 2004). Gong et al. (2012) state that knowledge exchange is an important antecedent of creativity because it provides cognitive resources for creativity and knowledge.

In social network literature, many researchers use the term ‘ties’ to explain relationships between individuals. Researchers use different factors to determine the strength of ties between individuals. For example, individuals have strong relationships with their friends, family or their neighbours, while they have weak relationships with individuals they do not speak to frequently, like acquaintances (Granovetter, 1983; Montgomery, 1992). The most frequently used definition of tie strength is “the amount of time that has been spend together, communication frequency, emotional intensity, intimacy, and the of the tie” (Granovetter, 1973).

Reagans and McEvily (2003) tested the relationship between tie strength and the ease of knowledge transfer. They collected data on knowledge content, expertise, tie strength, and the ease of knowledge transfer by administering a survey. They also tested whether personal attributes, such as race, sex, level of education and tenure affected the ease of knowledge transfer. Findings show that the four attributes had no effect on the ease of transferring knowledge. On the other hand, results showed a significant relationship between tie strength and ease of knowledge transfer. They used the tie strength characteristics which are proposed

(10)

by Granovetter to define tie strength. Results indicated that strong ties positively affect the ease of knowledge transfer (2003).

Granovetter (1973) states in his articles that weak ties can also be a very effective way to transfer relevant information. He argues that weak ties are more effective sources of novel information than strong ties. Strong ties are more likely to be homophilous, meaning that their way of thinking is often similar. Homophily does not only refer to the way of thinking or thoughts, it also refers to demographic features like gender, age, socials status, race, education, religion, and occupation (McPherson, 2001). It is also more likely that the social worlds will overlap as soon as two individuals have a strong relationship. For example, if person A and person B have a strong tie, and person B and person C have a strong tie as well, person A will most likely have at least a weak tie with person C (Borgatti & Halgin, 2011). According to Granovetter’s article “The strength of weak ties” (1973), weak ties are more effective sources of nonredundant knowledge than strong ties. Weak ties interact less frequently and feel less affection, while strong ties are positioned in overlapping social worlds. Weak ties are not immersed in the same network of social relationships (Granovetter, 1982). As a result, weak ties are more likely to be connected to individuals with different ideas, perspectives, and interests (Perry-Smith & Shalley, 2003). This implies that weaker connections provide nonredundant information, whereas strong connections primarily provide redundant information (Burt, 1997; Granovetter, 1973; Perry-Smith, 2003). This does not mean that strong ties always share the same opinions, thoughts, and ideas. When individuals’ opinion differ from others they perceive as strong ties, they are more likely to experience social pressure, which leads to conformity. This means that individuals will change their opinions and thoughts, rather than discussing and learning from it (Perry-Smith, 2006). Weak ties, on the other hand, are more likely to facilitate autonomy, which means that individuals do not feel the urge to identify themselves with other social groups (2006). They do not feel the social pressure to change their thoughts and ideas about topics.

(11)

Evidence from multiple studies suggests that nonredundant knowledge improves an individual’s creativity (Zhou et al., 2004). “Individuals who have access to a wide variety of alternatives, example solutions, or any potentially relevant ideas are more likely to make connections that could lead to creativity” (Amabile, 1996). In her research, Perry-Smith explores the effects of tie strength and network positions on individual creative contribution. The results of her study provide evidence that weak ties are more effective contributors to creativity than strong ties (2006). The network she constructed was based on the friendship network, in which she measured tie strength by using measurement scales from Granovetter’s article (1973).

This outlined situation explains that weak ties do not have to be the source of novel information per se. Weak ties are more likely to provide novel information when they form a bridge between two social networks. For example, when person A has a tie with person D, while person B and C have no tie with person D, the tie between person A and D will probably be a weak tie, because if it was a strong tie, person B and C would have had at least a weak tie with person D. Bridges are very useful to get access to novel information, because an individual will receive information that has not been circulating within his own social network (Borgatti & Halgin, 2011). Friedkin (1982) conducted a research that was based on Granovetter’s assumptions regarding the strength of weak ties. His findings provide evidence that strong ties are more efficient individually than weak ties, but he emphasizes that the strength of weak ties lies in their numbers (1982). The advantage of weak ties compared to strong ties is that individuals usually have more weak ties, which are easier to maintain. However, multiple researches have shown that the effect of weak ties is curvilinear, which means that the positive relationship between the number of weak ties and creativity will decline at a certain point (Zhou et al., 2009). When individuals have a high amount of weak ties they have less time to interact or discuss relevant topics, resulting in little involvement and meaningless weak ties (Perry-Smith & Shalley, 2003). Second, individuals might gather too

(12)

much novel information, which disables them from processing it effectively. This will result in confusion rather than creativity (Ward et al., 1999). Finally, individuals will focus more on gathering new information rather than developing creative ideas (Perry-Smith & Shalley, 2003). Multiple studies have shown that the strength of weak ties depends on its numbers, but that an excessive can eliminates this strength.

Studies have focused on the relationship between tie strength and creativity on both the person level as the dyadic level. As discussed, weak ties are more efficient when an individual possesses the right amount of weak ties, but on the dyadic level, tie strength and creativity are positively related, meaning that the stronger the tie, the higher the perception of creativity of the alter (Friedkin, 1982; Sosa, 2011). Scholars argue that closeness and communication frequency are not the only determinants of tie strength. The willingness and the intrinsic motivation of an employee to work together with colleagues also plays a role in the strength of a tie (Sosa, 2011). It is also more likely that a close relationship generates positive affect, which enhances creativity. Cross et al. (2001) conducted a qualitative research on how employees relied on their relationships with other employees within an organization to successfully do their work. Their approach was different compared to other research regarding social network analysis. Instead of asking their respondents “why” they received knowledge from others, they asked from “whom” they received knowledge. Results from the interviews indicate that there are four elements that distinguishes effective relationships from ineffective relationships: knowing who knows what, the ability to gain timely access to knowledge, the willingness of others to help or give advice, and the degree of safety in the relationship (2001). As Borgatti and Cross state (2003), before person A turns to person B for information, person A must have some understanding of person B’s skills and expertise. It is more likely that individuals know what others know when they have strong connections (Hansen, 1999).

As Granovetter (1973) states, individuals interact less frequently with weak connections than strong connections, which implies it is more likely that weak connections have less

(13)

knowledge about each other’s skills and expertise. The other two characteristics however, are more complex because willingness and the feeling of safety to ask for advice or knowledge are highly dependent on tie strength (Hansen, 1999; Levin & Cross, 2004; Reagan & McEvily, 2003). Time on the other hand, could depend on someone’s willingness to work together or it could be imposed because employees are expected to work together. Literature often refers to social cohesion to define networks based on strong connections. The reason why individuals are willing to help each other is not because they are “friends” per se, but because they are afraid that other individuals within the strong network will notice that one is not willing to help others, resulting in a damaged reputation (Reagan & McEvily, 2003).

Depending on the perspective, literature regarding tie strength and creative performance provide contradictory results on which level of tie strength is most effective when it comes to increasing creativity. These contradictions are dependent on the two perspectives, namely the person level and the dyadic level.

2.2 Tie strength

The elements introduced by Granovetter (1973) provide a more accurate description of tie strength compared to the distinction between friends and acquaintances, because individuals use different criteria when they call someone a friend (Burt, 1990). Some will call close contacts or people they meet with daily ‘friends’, while others use other criteria. Although Granovetter’s four identifying properties are frequently used in later studies, Krackhardt criticizes these four characteristics. He asserts that emotional intensity, intimacy, and reciprocity are subjective and interpretive, which means they are hard to measure (1992). Communication frequency is the only objective characteristic. Krackhardt introduces three properties to identify the strength of ties, which are interaction, affection, and time. Two individuals must have frequent interaction to have access to each other’s knowledge. The

(14)

second characteristic is based on Heider’s Balance Theory (1958). The tie is balanced when the affection between person A and person B is reciprocal. There will be an imbalance when the relationship between person A and B is asymmetric (Krackhardt, 1992; Scott, 2013). When this asymmetry occurs, either person A or B will try to create a balance in the relationship. For example, when person A likes person B, but person B does not like person A, person A has two options to create a balanced relationship. Person A will try harder and hopes that person B will start to like him, or person A will try to dislike person B as well (Krackhardt, 1999). The third characteristic is time, which indicates that person A and B must have a history of interactions (Krackhardt, 1992).

Other studies investigate the role of personal attributes to determine the strength of ties between colleagues and friends (Lincoln & Miller, 1979). Their findings show that sex and race have greater influence on friendship ties than work related ties, while authority and education are stronger predictors for the strength of work related ties (1979). The reason for these differences is that individuals form ties in different contexts. Friendship ties are built in informal contexts, which means that they can form ties with anyone they like. Work related ties on the other hand, are less ‘voluntary’, which means that individuals do not have the freedom to position themselves in any set of ties they like within an organization (Lincoln & Miller, 1979). This research provides important implications about the role of ties in different environments.

The conceptualization of tie strength has many different approaches, depending on its contexts. In organizational contexts for example, networks can also be based on work relationships. Many researchers have investigated the relationship between tie strength and knowledge sharing or creativity within organizations, in which they focused on advice or idea’s networks rather than friendship networks (Baer, 2010; Levin & Cross, 2004). Advice networks provide more accurate descriptions of how knowledge flows in the organization and therefore how tie strength affects creativity (Zhou et al., 2009). Hansen (1999) introduced a

(15)

work related meaning on the dimensions which are proposed by Granovetter (1973). For instance, closeness is based on the intensity of the working relationship, while communication frequency focuses on task related matters (Levin & Cross, 2004). In Baer’s (2010) study, the network was based on the degree of idea generation. He argues that novel information is key for generating creative ideas, and so he only focused on direct idea ties (2010). This does not mean that friendship networks are not important, because one important factor of asking for advice is trust. Individuals will trust their ‘friends’ more than ‘acquaintances’, which implies that individuals are more comfortable when they ask their strong ties for advice (Krackhardt, 1990).

To summarize, two types of networks will be identified in this thesis. One network will be based on friendship and the other one is based on advice seeking. Tie strength in the friendship network can be best defined by using elements from Granovetter’s and Krackhardt’s perspective. Although Krackhardt suggests that communication frequency is the only element that can be measured accurately, the other elements, which are the time spend together, emotional intensity, intimacy, and reciprocity, are comparable to Krackhardt's dimensions of tie strength. Krackhardt states that affect is important because it shows whether individuals like each other. Granovetter refers to this element by suggesting that emotional intensity and intimacy also determine the strength of a tie. In this article, tie strength will be measured by the degree of interaction frequency, closeness, willingness to maintain the relationship and spending time together during non-working hours. As argued by multiple authors, elements such as communication frequency, closeness and positive affect are all conducive to creative performance. It is more likely that employees who interact more frequently have more knowledge of each other’s skills, while it is also expected that employees who are close trust each other and feel psychologically safe, and are more willing to help each other. Literature focusing on person level tie strength and creativity claims that the number of weak ties are most effective in generating creative ideas, while scholars who focus on the dyadic level, claim

(16)

that strong ties are most effective. This thesis focuses on the dyadic level and, based on the outlined theories, the following hypotheses are proposed.

H1a: (Friendship network) Tie strength is positively associated with creative performance.

H1b: (Advice network) Tie strength is positively associated with creative performance.

2.3 Psychological safety

Psychological safety is defined as the risk of admitting ignorance about a topic. High psychological safety means that employees feel comfortable to ask for advice or share ideas without the fear of harming its own status and career (Siemsen et al., 2009). Cross et al. (2001) state that high psychological safety promotes learning and creativity, because individuals are able to speak up, ask for advice when necessary and admit mistakes without fearing negative consequences. Psychological safety can be affected on three different levels.

First, the organizational culture and structure determines whether people feel safe to speak up or ask for advice (Baer & Frese, 2010). Baer and Frese conducted a research on the relationship between organizational climates, psychological safety and process innovation. They measured the organizational climate by focusing on the management of the organization, because the management plays a crucial role in shaping the culture of the organization (Baer & Frese, 2003). The sample counted over 6000 German companies consisting of 100 to 900 employees. Seven items were used to measure psychological safety. The results show that psychological safety is related to group performances, not only for big companies, but also for mid-sized companies, which are usually less bureaucratic and more open (Baer & Frese, 2003). Furthermore, they conclude that companies with a high degree of innovativeness and a climate in which employees do not feel psychologically safe perform less than organizations which do not innovate at all. Only organizations with high levels of innovativeness and high

(17)

psychological safety performed better than organizations that did not innovate. This implies that organizations innovate inefficiently when employees do not feel psychologically safe (Baer & Frese, 2003). They also state that employees who are feeling psychologically safe are more likely to propose or share their ideas and thoughts, which is crucial for creativity and innovation.

Psychological safety also occurs on the dyadic level when two individuals interact (Cross et al., 2001; Siemsen et al., 2009). In their research, Borgatti and Cross (2003) investigated the relationship between seeking for information from others and the cost it will incur to ask that individual for help. Psychological safety is defined as cost in this study. The sample of the research consisted of 37 information scientists in a global pharmaceutical organization. The results did not support the hypothesis that there is a significant relationship between information seeking and psychological safety (Borgatti & Cross, 2003).

Finally, the degree of psychological safety also depends on personal characteristics of an individual (Zhang et al., 2010). As Siemsen et al. (2009) state, self-consciousness is one psychological determinant of the degree of psychological safety, pointing to the fact that individuals who are more self-conscious will be more confident and therefore feel less fear of harming their own status and reputation when they ask for advice.

One important determinant of having a safe relationship is trust. Levin and Cross (2004) examined how trust mediates the effect of tie strength on knowledge access. Their findings indicate that tie strength has positive impact on trust. The stronger the tie between individuals, the more they trust each other. These findings were also supported by Siemsen et al. (2009). This implies that safe relationships occur more often when ties are strong. Gong et al. (2012) studied the relationship of trust and creativity. They focused on 174 retail stores, which counted 375 employees in total. The results of their research show that trust and creativity are positively associated. Zhang et al. (2010) examined the effect of psychological safety on trust and the individual’s intention to share knowledge. Their results show that trust is positively

(18)

related to psychological safety and psychological safety positively affects individual’s intention to share knowledge.

In short, many scholars argue that psychological safety is linked both to tie strength as creative performance. Based on the assumption that psychological safety enhances creativity and the assumption that psychological safety is positively related to tie strength, the following hypotheses are stated:

H2a (friendship network): Psychological safety moderates the relationship between the tie strength and creative performance in such a way that high levels of psychological safety makes the relationship stronger, while low levels of psychological safety negatively affect the relationship.

H2b (advice network): Psychological safety moderates the relationship between the tie strength and creative performance in such a way that high levels of psychological safety makes the relationship stronger, while low levels of psychological safety negatively affect the relationship.

2.4 Power

Researchers often define power as the ability to order others to do things, even though they are unwilling or resistant to doing it (Krackhardt, 1990). Power is also defined as individuals’ control over unique resources (Magee & Galinsky, 2008). Krackhardt states that the degree of power is not only determined by ones ability to get things done from individuals despite resistance of those individuals (1990). Informal power, like the ability to influence others, is also a determinant of power (Krackhardt, 1990). He states that some actors are influential because they possess traits that reflect qualities of trust and personal charm, which is called charisma.

(19)

The most obvious form of differences in power is structural influences. Some individuals are given privileges and power because they are ranked higher in the organization (Krackhardt, 1990). He states that actors with more authority will have more power. Both potency and charisma were later used as measurements in Anderson’s study (2012).

Centrality in the informal network is also a predictor of power. Individuals that are ranked higher in the organization carry a lot more responsibility over the actions of individuals who are ranked lower. They are required to keep track of how subordinates work together and how they relate to each other. This implicates that higher ranked individuals, who have more power and authority, have a more accurate picture of the construction of the network in the organization (Krackhardt, 1990). Multiple researches have shown that the possession of power has a strong impact on the self-concept of an individual. Individuals will experience more positive and less negative affect, are more assertive, and they enjoy higher self-esteem, physical health, and longevity (Anderson et al., 2012).

Employees tend to fear a manager’s opinion about them more than from their peers, implying that the degree of psychological safety depends on the power of the alter and how the ego respects and appreciates the alter. Individuals tend to value appreciation and respect from a superordinate more than individuals who are located on the same position in the social hierarchy. This implies that employees feel less psychologically safe when they ask a manager for advice, because managers have more power.

In conclusion, power is based on hierarchy and social hierarchy, meaning that power differences are not only based on hierarchy, such as job functions, but also on an employees’ ability to influence others, which does not depend on the job function per se. It is expected that individuals who are ranked lower or are less powerful feel less psychologically safe when they ask a colleague for advice who is ranked higher in the hierarchy.

(20)

H3a: (Friendship network) Power moderates the influence of psychological safety on the relationship between tie strength and creativity, such that higher levels of power will negatively affect psychological safety, which in turn negatively affects the relationship between tie strength and creative performance.

H3b: (Advice network) Power moderates the influence of psychological safety on the relationship between tie strength and creativity, such that higher levels of power will negatively affect psychological safety, which in turn negatively affects the relationship between tie strength and creative performance.

2.5 Conceptual Model

The conceptual model which is presented in figure 1 on the next page depicts the expected effects, which are discussed in this chapter. Tie strength is expected to be positively associated with creative performance, while psychological safety is expected to make this relationship even stronger. Power is expected to negatively affect this relationship because powerful individuals are in a better position to damage less powerful people’s reputation, status and career.

Figure 1: Conceptual model of discussed variables. Positive effects are indicated with ‘+’ and negative effects are indicated with a ‘-‘.

(21)

3. Method

In this chapter the set up of the empirical study is discussed. In the first section, the characteristics of the sample will be outlined. In the second part, the measurements of each variable will be discussed, which includes the number of items, examples of items, and scales which are used for measurements. In the last section, the analytical strategy will be outlined.

3.1 Sample

The sample of this empirical study consists of 26 Dutch employees who are working in the same organization. Only one organization was approached for this research. The organization is specialized in market research, which is the reason why this organization is chosen for this study. Knowledge and creativity are important aspects and therefore this organization fits well to this study. It is very common in the social network analysis field to focus on a small sample since this type of studies focus is on the relationship between individuals rather than the individual itself. The questionnaire link was sent through email to every employee in the organization.

All 26 respondents completed the survey (response rate 100%). Of all respondents, 21 employees were female, which is 80.8%. The age ranged from 25 to 51 year with a mean of 35.27 and a standard deviation of 6.41. All respondents have been working at the company for longer than six months, while 19 employees have been working at the company for longer than three years (73.1%). 15.41% of the respondents reported a tenure of 6 to 12 months, while 11.5% reported a tenure of 1 to 3 years. None of the employees have been working at the organization less than 6 months. The majority of the sample has a university degree (61.5%), while 26.9% completed an education program on HBO level, and 11.5% on MBO level. The part-time/fulltime ratio seems to be equally divided, as 12 respondents (46.5%) work at the

(22)

organization on a part-time basis, while the remaining 14 respondents (53.8%) work on a fulltime basis.

3.2 Measurements

The items were derived from existing scientific literature. Some items were used many times in other researches, which indicates that the validity of these items is high. The items were derived from English written literature, which means the items had to be translated into Dutch to make sure that the respondents have a clear understanding of the questions.

3.2.1 Dependent variable: Creative performance

To measure the degree of creativity per respondent three items were used from the study of Zhou and George (2001). Each respondent is asked to rate his or her colleague on creativity. This type of measurement of creativity has been used in multiple studies in the past (Sosa, 2011). Each question will be constructed as a matrix, which means that a list of names of colleagues is presented vertically, while the question and its scales are presented horizontally. Two example questions are: “Do you think he or she is a source of creative ideas?” And: “Do you think he or she comes up with creative solutions to problems?” The ratings were measured by using a 5 point Likert-scale ranging from 1 (not at all characteristic) to 5 (very characteristic).

3.2.2 Independent variable: Tie strength in friendship networks

Three items were used to measure tie strength in friendship networks. The items were derived from multiple studies. One item was derived from Baer’s research on tie strength and creativity (2010). Another item was derived from Levin and Cross’ study (2004). Respondents will have to answer these questions for all their colleagues so each question will also be constructed as a matrix. One example question is: “Indicate how close you are with each person.” The scale

(23)

ranges from 1 (not at all close) to 4 (very close). Scores of 1 or 2 indicate weak ties, while scores of 3 or 4 indicate strong ties (Baer, 2010). Another question is: “How often do you communicate with each person?” The ranking scale ranges from 1 (once every two months) to 4 (daily).

3.2.3 Independent variable: Tie strength in advice networks

Tie strength in advice networks will be measured by using four items which are taken from articles of Zhou et al. (2009), Levin and Cross (2004), and Yang and Tang (2003). The questions for this variable will also be constructed as a matrix, meaning that the colleagues will be listed vertically, while the questions and the corresponding scales are located horizontally. One example question is: “To what extent is this person an important source of advice when you have a work related problem?” Different scales are used per item. One item ranges from 1 (not at all) to 4 (extremely), while another measurement of an item ranges from 1 (not at all close) to 4 (very close).

3.2.4 Moderating variable: Psychological safety

To measure psychological safety, respondents were asked to answer three questions. One item, which is derived from Borgatti and Cross’ (2003) study, is formulated as follows: “I feel comfortable when I ask this person for advice.” The other questions were taken from Edmondson’s study (2003). A scale ranging from 1 (not at all) to 4 (extremely) is used for all items. Respondents will have to answer how psychologically safe they feel towards every colleague.

3.2.5 Moderating variable: Power

The items to measure power are derived from Krackhardt's article (1990) in which he investigates the influence of power on network centrality in advice networks. In his study,

(24)

respondents were asked to rate each other on each dimension of power, which are power and hierarchy, influence, and job function: “To what extent is each person able to get things done despite resistance of the person in question?” The question regarding influence is: “To what extent is each person able to influence others in their acts and thoughts?” It is highly important that each employee rates all his colleagues on power, instead of dividing the group of colleagues who will be rated into different parts, because each respondent rates from a different perspective, which negatively affects internal consistency. With a larger sample the survey would become too exhaustive, which increases the chance that biases will occur. Respondents were first asked to select their job title, which consists of 1 (Back Office), 2 (Junior Research Managers), 3 (Research Managers), 4 (Senior Research Managers), and 5 (Managing Partners). The other two items were rated on a 5 point Likert-scale ranging from 1 (totally not powerful/influential) to 5 (very powerful/influential).

3.2.7 Control variables

In the last part of the survey respondents are asked to answer questions regarding education and tenure. Job tenure is defined as the length of stay at the organization, which’ scale ranged between 1 (0-6 months) to 4 (3 years and longer). Education was coded as 1 (MBO), 2 (HBO), and 3 (WO). These variables were chosen as control variables as they have been proven to be related to networks and creativity (Zhou et al., 2009).

3.3 Analytical strategy

The surveys were constructed in an online software called Qualtrics. The link of the survey was sent on the 28th of April. All the data was collected on the 11th of May. After retrieving the data from Qualtrics, it had to be transformed in Excel to make it suited for analysis in SPSS. The data had to be transformed into a vertical edge list in which each row presents a relationship between the respondent and his or her colleague. Each row contains ratings

(25)

regarding tie strength in the advice network and friendship network, psychological safety, power, and creative performance, indicated by the respondent, which is the ego, towards the colleague, which is the alter. So for example, the ego indicates how he or she perceives the strength of the tie with his or her colleagues. This means that tie strength does not have to be symmetric, as the alter may perceive the relationship differently. To test the hypotheses, two datasets are used, one for the friendship network and one for the advice network. Respondents were asked to indicate whom they approach for advice and how strong the tie is with the colleagues that are approached. This means that tie strength with colleagues that are not approached is indicated with a zero. To test relationships, analyses are more reliable when the colleagues that form no tie are excluded from the dataset, because the zeros will be taken into account in the analysis, which affects the results. Therefore, two datasets are used.

Variable ratings were calculated by calculating the mean of the total number of items that were used to measure each variable. The means and standard deviations are presented in table 1 (friendship network) and table 2 (advice network). The table also shows the Cronbach’s alpha for each predictor variable between the parentheses at the end of each row. The Cronbach’s alpha is higher than .7 for every variable, which means that the internal consistency of the scales is good.

To reduce the risk of multicollinearity, the predictor variables will be mean-centered, which means that the mean will be subtracted from the scores. This step is crucial in testing moderation, since the chance that the correlation between the interaction of two specific variables (X * Y) and X and Y is very high. To generate reliable estimates of the individual regression coefficients it is important to minimize multicollinearity. Two-way and three-way interactions will be executed to test if psychological safety and power moderate the relationship between tie strength and creative performance. The interacting variables will be calculated by multiplying the centered variables with each other. Hierarchical regressions will be executed to test the hypotheses.

(26)

(27)

(28)

4. Results

This chapter starts with outlining significant correlations regarding creative performance. The first correlation matrix is based on the friendship network (N=650) and the second table is based on the advice network (N=199). This chapter continues by explaining the direct effects of variables on creative performance, two-way interactions between tie strength and psychological safety and its effect on creative performance, and eventually, three-way interactions between tie strength, psychological safety and power.

4.1 Significant correlations

The correlation matrix in table 1 is based on the friendship network, which contains all possible relationships. The second table is based on the advice network, which only contains working relationships that were indicated by the respondents. This section mainly explains the correlations which are presented in the first table. Both tables show the means, standard deviations, correlation coefficients and reliabilities of the variables. Several correlations between variables are significant and strong, which will be outlined in this section.

First, short job tenure (tenure 2) correlates negatively (r = -.195, p < .01) with creative performance, while long job tenure (tenure 4) correlates positively (r = .116, p < .01) with creative performance (Table 1). This could be explained by the fact that employees who have been working at the organization for a longer period of time have more knowledge of tasks, procedures, and more experience. Furthermore, only the highest education level is correlated positively (r = .219, p < .01) with creative performance, while the other two levels, education1 (r = -.194, p < .01), and education2 (r = -.100, p < .05) are correlated negatively.

Tie strength in the advice network (Table 2) (r = .305, p < .01), tie strength in the friendship network (r = .394, p < .01), psychological safety (r = .206, p < .01) and power (r = .523, p < .01) correlate significantly with creative performance. This implies that there is a

(29)

direct relationship between these variables. The positive correlation between tie strength and creative performance could mean that the ego perceives alters they communicate more frequently with and are closer with as more creative than alters with whom they are weakly connected.

Table 1 also shows that the independent and moderator variables correlate significantly with each other. For instance, the correlation matrix shows that power and psychological safety correlate negatively (r = -.202, p < .01), which implies that an increase of psychological safety means that power will decrease, and vice versa. This negative correlation might indicate that the ego feels less psychologically safe towards colleagues who are more powerful or have more influence over the ego. On the other hand, when the alter has less influence over the ego, the ego feels more psychologically safe.

Furthermore, the correlation matrix also shows significant correlations between psychological safety and tie strength in the friendship (r = .492, p < .01) and advice network (r = .378, p < .01) indicating that psychological safety and tie strength are related. This indicates that the ego feels more psychological safe with colleagues he or she is strongly connected to.

Finally, The correlation table also reports a significant correlation between tie strength in the friendship network and tie strength in the advice network (r = .498, p < .01). This could mean that employees work more closely with their friends. It could also mean that colleagues enjoy working together and, as a result, become friends.

(30)

(31)

(32)

4.2 Direct effects on creative performance

Table 3 and 4 show results of the hierarchical regressions and the direct effects of multiple variables on creative performance. The first step tests the effects of control variables job tenure and education levels on creative performance, which shows that tenure2 (β = -.213, p < .01), education1 (β = -.173, p < .01) and education3 (β = .127, p < .01) significantly affect creative performance in the friendship network. In the advice network however, the results only show a significant effect of education1 (β = -.193, p < .05) on creative performance. The variable power is added to the regression in table 5 for the friendship network (β = .447, p < .01) and in table 6 for the advice network (β = .275, p < .01). Both variables have a significant direct effect on creative performance. This could mean that employees perceive powerful colleagues as more creative.

To test the hypotheses, tie strength is added in second step of the hierarchical regression. Results shows that tie strength in the friendship network (β = .341, p < .01) is positively associated with creative performance. A significant amount of variance is explained (𝛥𝑅2₌

.137, p < .01) by adding tie strength and psychological safety (β = .055, p > .05) to the regression. Adding tie strength in the advice network (β = .242, p < .01) and psychological safety (β = .159, p < .05) caused a significant extra amount of variance explained of creative performance (𝛥𝑅2 = .112, p < .01). The results show that tie strength in both networks have a positive relationship with creative performance, which means that hypothesis 1a and hypothesis 1b are supported. Employees who have s strong tie with their colleague, perceive them to be more creative.

4.3 Moderation effects on creative performance

To test hypothesis 2a and 2b, interaction variables are constructed to test the moderating effect of psychological safety on the relationship between tie strength and creative performance. The

(33)

independent variables tie strength and psychological safety are mean-centered to conduct a moderation test. Hypothesis 2a and 2b predicted a positive moderation effect of psychological safety on the relationship between tie strength and psychological safety. The results show that these predictions are not supported; the interaction between tie strength and psychological safety in the friendship network was not significant and also not positive (β = -.008, p > .05). Hypothesis 2a is not supported.

The interaction between tie strength in the advice network and psychological safety has a significant effect on creative performance (β = -.178, p < .05). The interaction added a significant amount of variance explained of creative performance (𝛥𝑅2_{= .028, p < .05).}

However, hypothesis 2b is not supported because psychological safety negatively affects the relationship between tie strength in the advice network and creative performance, while the hypothesis predicted a positive effect. Based on Dawson’s (2013) two-way interaction plot sheets, it is possible to depict slopes that show the conditional effect of tie strength on creative performance, depending on different values of psychological safety. The slopes of high psychological safety and low psychological safety are depicted in figure 2. The figure shows

Figure 2: Moderation of psychological safety on the relationship between tie strength in the advice network and creative performance.

1 1.5 2 2.5 3 3.5 4 4.5 5

Low Tie strength High Tie strength

Cr ea ti v e P erf orm an c e Low Psychological Safety High Psychological Safety

(34)

that psychological safety moderates the relationship between tie strength and creative performance in a way that lower levels of psychological safety result in more creative performance when tie strength is high. Based on the Johnson-Neyman technique, it is possible to estimate the slope of psychological safety and its significance of the effect on the relationship between tie strength and creative performance. The results show that high psychological safety does not significantly affect this relationship (B = .084, 𝑡₍₁₉₅₎ = 0,895, p > .05), while low values of psychological safety do moderate the relationship (B = .386, 𝑡₍₁₉₅₎ = 3,302, p < .01). This means that low values of psychological safety have a significant effect on the relationship between tie strength, meaning that individuals who do not feel psychologically safe perceive individuals as more creative when ties are strong. This could indicate that, even though employees communicate very frequently and work closely with colleagues they perceive as creative, they do not feel psychological safe towards these employees.

(35)

(36)

(37)

(38)

4.4 Three-way interaction effects on creative performance

In this section a three-way interaction is executed to test if power affects the influence of psychological safety on the relationship between tie strength and creative performance. This test is conducted for the friendship network and the advice network. Hypothesis 3a and 3b predict that power moderates the influence of psychological safety on the relationship between tie strength and creative performance in a way that higher levels of power will reduce psychological safety, which in turn will reduce the positive effect of tie strength on creative performance.

Table 5 presents the interaction effect between tie strength in the friendship network, psychological safety, and power (β = -.085, p < .05). A significant amount of variance is explained (𝛥𝑅2 = .004, p < .05). This indicates that there is a significant interaction effect of the three variables on creative performance. Figure 3 presents the slope of each combination between psychological safety (high or low) and power (high or low). The significance of each slope is presented in table 7. A comparison of the slopes indicates that, regardless of tie strength, the level of creative performance is highest when psychological safety and power are high. An assumption that can be made is that employees who feel psychologically safe perceive higher power colleagues as more creative, especially when they have a strong tie with that colleague. Based on the significance of the slopes of each combination of tie strength, psychological safety and power, results show that the effect of the moderators is significant when psychological safety is high and power is low (B = .284, 𝑡(642) = 4,295, p < .01), whereas

the conditional effect is not significant when psychological safety and power are high (B = .117, 𝑡(642) = 1,658, p > .05). Hypothesis 3a predicted that a high level of power and a low

level of psychological safety has a negative effect on the relationship between tie strength and creative performance. Results of the slopes, which are depicted in table 3, show that this hypothesis is not supported (B = .213, 𝑡₍₆₄₂₎ = 2,468, p < .05). When power is high and

(39)

psychological safety is low, tie strength has a more positive effect on creative performance. This effect is slightly stronger compared to the effect of tie strength on creative performance when the values of both moderators are zero.

Figure 3: Moderation of psychological safety and power on the relationship between tie strength in the friendship network and creative performance.

The results in table 6 report a significant interaction effect between tie strength in the advice network, psychological safety and power on creative performance (β = -.190, p < .05). Adding the interaction in the sixth step of the hierarchical regression analysis results in a significant extra amount of variance explained of creative performance (𝛥𝑅2_{= .016, p < .05).}

Figure 4 on the next page shows different results compared to the slopes of the moderators in the friendship network. First, the graph shows that employees who feel less psychologically safe towards less powerful colleagues, perceive their colleague as more creative when the tie between them is weak, rather than strong. However, this conditional effect of the moderators was not significant. Furthermore, the results only report one significant effect of psychological safety and power on the relationship between tie strength and creative performance. A significant effect is reported when the employee does not feel psychologically safe and perceive the colleague as a powerful person. So with low values of psychological safety and

1 1,5 2 2,5 3 3,5 4 4,5 5

Low Tie Strength High Tie Strength

C re at iv e P er fo rm an

ce (1) High Psychological Safety, High Power

(2) High Psychological Safety, Low Power

(3) Low Psychological Safety, High Power

(4) Low Psychological Safety, Low Power

(40)

high values of power the relationship between tie strength and creative performance is strongest (B = .436, 𝑡₍₁₉₁₎ = 3,669, p < .01). Hypothesis 3b predicted that the relationship between tie strength and creative performance is strongest when power is low and psychological safety is high. The slope ‘high psychological safety, low power’ is not significant (B = .059, 𝑡₍₁₉₁₎ = 0,470, p > .05), which means that the hypothesis is not supported. The other two slopes ‘high psychological safety, high power’ (B = .063, 𝑡(191) = 0,388, p > .05) and ‘low psychological

safety, low power’ (B = -.1581, 𝑡₍₁₉₁₎ = -0,571, p > .05) also did not report significant conditional effects. The difference in perceptions of creativity hardly depends on tie strength when psychological safety and power are high.

Figure 4: Moderation of psychological safety and power on the relationship between tie strength in the advice network and creative performance.

1 1,5 2 2,5 3 3,5 4 4,5 5

Low Tie Strength High Tie Strength

C re a ti v e P e rf o rm a n c e

(1) High Psychological Safety, High Power

(2) High Psychological Safety, Low Power

(3) Low Psychological Safety, High Power

(4) Low Psychological Safety, Low Power

(41)

5. Discussion

This chapter will elaborate more on the findings which are presented in the results section. Relevant literature and theories will be used to explain and clarify why and how certain outcomes may have occurred. Subsequently, practical and managerial implications of these findings will be outlined. This chapter continues by explaining the limitations of the findings and the entire research as a whole. Further research topics are discussed in the last section of this chapter.

5.1 Theoretical and practical implications

This thesis aims to provide a deeper understanding of the relationship between tie strength and creative performance in an organizational context and variables that influence this relationship. This thesis contributes to recent research by examining the effects of psychological safety and power on the relationship between tie strength and creative performance. Previous research has not focused on studying the moderating effect of psychological safety on this relationship, while it has been argued that tie strength and psychological safety independently are positively associated with creative performance. This thesis also adds an extra dimension to the existing literature by studying the effects in two different networks, namely the friendship network and the advice network. Both networks are based on different factors in which ties are constructed very differently. As hypothesized, tie strength is positively associated with creative performance, but, contrary to the predictions of this research, results show other effects of psychological safety and power on this relationship. Previous research argues that tie strength and creative performance are positively associated because employees are more willing to work together, which results in more proactive and motivated approaches towards working tasks (Sosa, 2011). However, the results indicate that employees’ perception of their colleagues creative performance is higher when

(42)

tie strength is high, even though psychological safety is low. This questions the assumptions that employees are motivated to work closely with colleagues, because it is less likely that employees who feel psychologically safe are motivated to work together and experience positive affect.

The first section of the results outlines the direct effects of control variables, tie strength, psychological safety, and power on creative performance. Employees who completed lower education levels were perceived as less creative. The significant relationship between a low education level and creative performance can be explained by the fact that intelligent individuals are generally better able to think divergent, which results in higher rates of creativity (Nusbaum & Silva, 2010). Based on this conclusion and the fact that individuals who completed higher levels of education are usually more intelligent, it seems logic why employees with lower levels of education are less creative. Low job tenure (6 to 12 months) also showed a negative effect on creative performance. This implies that employees who have been working at the organization for a relatively short time are learning to do specific tasks and asking for advice rather than assisting and explaining others in their work tasks.

Previous research on social network analysis at the dyadic level has pointed out that there is a relationship between tie strength and creative performance. The results regarding hypothesis 1a and 1b predicted a positive relationship between tie strength and creative performance, which are consistent with results from earlier research. Significant relationships were found in this study regarding tie strength in the friendship and advice network and creative performance. Although many researchers have argued that knowledge is more likely to overlap when employees form strong ties because of a higher interaction frequency, results from previous studies provide evidence that redundancy of knowledge does not necessarily means that strong ties inhibit creative performance (Sosa, 2011). Sosa states that the intrinsic motivation to work together might be a key factor in the construction of strong ties.

(43)

Individuals who are willing to engage in a working relationship are also more likely to experience a positive affect, which enhances creativity (2011). This makes a lot of sense regarding the friendship network, since tie strength is based on communication frequency regarding personal topics, closeness, and the willingness to maintain the friendship (Granovetter, 1973). In the advice network it is harder to clarify why tie strength is positively associated with creative performance, because a strong relationship does not necessarily mean that an employee is willing to work with his colleague. The correlation matrix shows a significant correlation between tie strength in the friendship network and in the advice network, which could mean that employees prefer to engage in a working relationship with colleagues they have strong connections with in the friendship network. It could also mean the opposite way: employees engage in a working relationship and as a result, they get closer on a personal level. Based on the assumptions of Cross et al. (2001), it could also mean that employees gain more knowledge of their colleagues’ skills and expertise, resulting in higher perceptions of the colleagues’ creativity. However, having more knowledge of someone else his skills and expertise does not necessarily mean that he is more creative.

The second hypothesis tested the interaction effect between tie strength and psychological safety on creative performance and did not show significant effects in the friendship network. From a psychological safety perspective, it is argued that employees are more likely to be creative when they operate in an environment which is safe for suggesting new ideas, admitting mistakes, and asking task-related advice (Gong et al., 2012). Levin and Cross (2004) conducted a research on the relationship between tie strength and information seeking, mediated by trust. Trust is an important driver of psychological safety, and their study shows there is a significant relationship between tie strength and trust. Tie strength, especially in the friendship network, already indicates that two individuals are friends who trust each other. Therefore, the positive relationship between tie strength and creative performance does not

The effect of power differences on creative performance in an organizational social network