‘network strength’

(1)

1

THE EFFECT OF NETWORK STRENGTH ON TEAM INFORMATION SHARING: A COMPARISON BETWEEN FRIENDSHIP AND GOSSIP NETWORK

Master thesis, Msc, specialization Human Resource Management University of Groningen, Faculty of Economics and Business

February 25, 2018

LAURA ZSÓFIA GYŐRI Student number: 3296946

Nieuweweg 10B 9711 TD Groningen tel.: +31 (6) 26998102 e-mail: l.z.gyori@student.rug.nl

Supervisor Y. Yuan Co-Supervisor dr. E. Martinescu

Acknowledgments: I would like to thank my supervisor for her guidance and also my boyfriend and fellow students for their support and help.

(2)

2 ABSTRACT

This study compared friendship network strength and gossip network strength at the workplace. The subjects of the comparison were the difference between the two types of network strengths and their effects on information sharing and information elaboration. I examined the information sharing and the information elaboration habits in these two networks, assuming that team energy is mediating the above mentioned relationships.

Therefore, this paper’s main research question is whether there is a difference between information sharing and information elaboration habits in friendship network and in gossip network, and if so, to what extent do they differ. In order to compare the network types, I analyzed the network data with a less developed structural feature; ‘network strength’. Among others, my hypotheses were that there is a negatively skewed curvilinear relationship between friendship network strength and information sharing/information elaboration, while there is an inversed curvilinear relationship between gossip network strength and information sharing/information elaboration. 30 teams from 23 different companies, from two countries (Hungary, Netherlands) served as the sample of my research, which altogther consisted of 139 individuals. My dataset showed no significancy on the predicted relationships. However, it can serve as a complementation to the existing studies in the social network field, because there is no previous study which compares these types of networks, especially with this network property, namely network strength.

Keywords: friendship network strength, gossip network strength, information sharing, information elaboration, team energy

(3)

3

INTRODUCTION

A major issue for managers and executives nowadays is to enable their team or department to reach the appointed goals, by performing the assigned key performance indicators. The reason for this is that it is needed to help the organization to be successful. One of the factors which can influence team performance is information sharing (Mesmer-Magnus & DeChurch, 2009) therefore this paper focuses on this latter mentioned process and on information elaboration in work teams. As the connections between the members in a team play a major role in the willingness of them to share information (van Knippenberg, de Dreu & Homan, 2004), this paper investigates two adverse types of connections, friendly and gossiping. Friendships between colleagues have many positive effects, such as better performance, increased job involvement, higher job satisfaction, higher organizational commitment, and decreased turnover (Riordan & Griffeth, 1995). Gossip at the same time can have both positive and negative effects (Kurland & Pelled, 2000). Positive gossip can enhance the reputation of the subject of the gossip, which in turn results in improved work outcomes (Kurland & Pelled, 2000). Contrary, negative gossip can have the opposite effect, it demotivates people and they become less productive (Shethna, 2017).

While friends and friendship networks at workplace are widely researched, gossip and gossip network are less developed areas of the scientific literature. Although, gossip is also a basic and inevitable phenomenon in every area of life. People not only gossip in private life, but also at the workplace. A recent study showed that people spend altogether 65% of their time with gossiping (Dunbar, 2004). Therefore, it suggests, that people spend quite some time with gossiping at the workplace as well (Ellwardt, Labianca & Wittek, 2012). In this present study, two types of informal networks are distinguished; friendship networks and gossip networks. Both of them are present in the organizational environment. Network properties, characteristics of the networks, can be used to analyze and compare these different social

(4)

4

networks (Tichy & Fombrun, 1979). In this present study, I focus on network strength, which covers the number of the existing one-way connections between the individuals within a working team.

Additionally, I examine the effect of the strength of the above mentioned networks on work related or strategic information sharing and information elaboration processes. It is important to emphasize, that I examine strategic information sharing and not social information sharing. Strategic information sharing makes a team more efficient, while social sharing is not strictly goal-oriented (Talja, 2002). Throughout the whole paper, I refer to strategic information sharing and information elaboration. Moreover, information sharing (work-related) is also positively related to team performance (Mesmer-Magnus & DeChurch, 2009). As a consequence, when information sharing occurs with a great degree in a work team, it predicts good performance outcomes, therefore work-related sharing is beneficial for the organization. Hence, the first contribution of this paper to the existing literature is that it distinguishes between friendship network strength and gossip network strength related to strategic information sharing and information elaboration.

In work teams information sharing is indispensable. It is essential for the successful operation of the team and team members cannot avoid communicating with each other.

During the information sharing process employees can learn from each other. Several factors determine whether team members are willing to share information; one of them is the friendship connections with the colleagues (Haythornthwaite & Wellman, 1998). In the case when there are no friendship connections between the team members, information sharing and elaboration is less likely to occur. When employees consider their fellow team members solely as colleagues or even rivals, they are not likely to engage in information sharing and elaboration. Although, as friendship connections start to appear, people are more willing to share and elaborate on their information with each other (Haythornthwaite & Wellman, 1998).

(5)

5

However, too close friendship connections can distract the employees from work-related topics, and they rather share social information, without serving the interests of the organization (Morrison & Nolan, 2007). Therefore, I assume that the positive effect of friendship connections on information sharing and elaboration only holds until a certain point.

At that point the friendship network becomes too strong, and the individuals in the team find their friendship network more important than the work related information sharing and elaboration.

Regarding gossip, communication is a basic element, since without information sharing there would be no gossip. However, gossip tends to be negative more often than positive (Wert & Salovey, 2004). Thus, as gossip appears in a team, negative gossip always appears with a greater degree than positive gossip. Negative gossip can destroy the reputation of a team member (Kurland & Pelled, 2000) who in turn will lock himself/herself, and will be less willing to share or elaborate on information. Consequently, by an increasing number of gossip connections, the negative effects of gossip will come to the fore, and there might be less information sharing and information elaboration within the team due to the predominant negative gossip. Thus, I predict, that a stronger gossip network can decrease the extent of information sharing and elaboration. Hence, I predict that the overall gossip network strength has a generally negative effect on information sharing and on information elaboration.

Positive interactions among team members have an energizing effect, which results in a more cooperative behavior (Dutton & Ragins, 2007). In contrast, when the interactions among team members are negative, human energy is decreasing (Fritz, Lam & Spreitzer, 2011), therefore one can expect that team members act less cooperatively. Team members might be more hostile against each other due to these negative interactions. Therefore, as previous research indicated, the atmosphere in a working team influences the team energy (Fritz, Lam & Spreitzer, 2011). When there is a good atmosphere in the team, the energy will

(6)

6

be high (Fritz, Lam & Spreitzer, 2011). Additionally, for bad atmosphere the opposite might be true. Friendship and gossip network strength might therefore affect team energy. Following the logic above, when friendship connections increase team energy, the members act more cooperatively, which results in more information sharing and information elaboration among team members. Contrary, gossip connections can decrease the energy of the team, thus making the team members less cooperative. This means, there will be less information sharing and elaboration. Hence, I assume, that friendship network strength can increase the team energy, which in turn encourage the members to share their information with each other and elaborate on them, while gossip network strength can decrease the energy of the team, which leads to less information sharing and elaboration. Therefore, the second contribution of this present study is the examination of the mediating role of team energy on the relationship between friendship/gossip network strength and information sharing/elaboration.

The third contribution of this paper is more practical. This present study can enable executives to categorize the social networks in the organization, and form teams where the energy is sufficient for information sharing and elaboration, due to the friendship connections.

When employers are able to recognize gossip networks, they can work on the process to reformulate the teams, and convert all of them into a friendship network. In recent days, when every company strives for a better performance, it can serve as one possible solution.

In summary, the contribution of this study is threefold. First, it aims to gain insight in the role of friendship network strength and gossip network strength on information sharing and elaboration in working teams. Second, it argues that team energy mediates these effects of network strength (friendship/gossip) on information sharing and information elaboration.

Third, the more practical contribution of this paper is to enable executives to form teams in which the team energy is satisfactory for strategic information sharing and elaboration.

Moreover, as a consequence, a desired team performance can be reached.

(7)

7

THEORY AND HYPOTHESES

Information sharing and information elaboration within a work team

Information sharing within a work team

In order to have an excellent organizational performance and satisfying outcomes, companies are widely leaning on team work, which requires information sharing among the members.

The reason behind is that the total knowledge which is needed to reach this performance is distributed among the employees (Thomas-Hunt, Ogden & Neale, 2003). As Mesmer-Magnus and DeChurch (2009) also revealed; it is indeed more beneficial to use teams since the pool of available information can be expanded and a higher quality of team work can be reached.

Therefore, organizations which operate with work teams assume that information flows freely among the individuals (Thomas-Hunt, Ogden & Neale, 2003). However, in real life it is not always the case, so many studies were initiated to examine this field. Mesmer-Magnus and DeChurch defined information sharing in their study as “a central process through which team members collectively utilize their available informational resources” (2009, p.535).

Existing literature examines both the consequences and the influencing factors of information sharing. However, since the present study only focuses on the conditions under which information sharing occurs, some of these conditions are highlighted. First, there are two basic requirements for making a successful information sharing process; one is that the team members need to be connected to each other, and the other is that they need to speak the same language (Cabrera & Cabrera, 2005). When these are given, there are still a number of factors which can influence the behavior of the source of information. Just to mention the relevant ones; when people trust each other, and they share a strong group identity (they are friends) there is a bigger chance that people are willing to share information (Cabrera &

Cabrera, 2005). Moreover, when people feel they can help others (people help to their friends)

(8)

8

if they share information, it is also more likely that this phenomenon will occur (Cabrera &

Cabrera, 2005). Since the comparison between the effects of friendship and gossip network strength on information sharing is not revealed yet, this paper aims to do so.

Information elaboration within a work team

Turning to the other dependent variable; information elaboration is the process in which relevant ideas, knowledge and insights are exchanged, discussed and integrated in a group (van Knippenberg et al., 2004). Other definitions exist, but the key words are the same;

information elaboration means exchanging and integrating knowledge. This mechanism goes beyond the phenomena of information sharing to the extent that during information elaboration team members give detailed explanations for their ideas, they spend precious time to discuss the different perspectives, they integrate the information and they decide what will be done to handle a particular problem or situation (Hoever, van Knippenberg, van Ginkel &

Barkema, 2012).

There are a few drivers, which leads to information elaboration. These are the following: need for cognition (Kearney, Gebert & Voelpel, 2009), team members’ process accountability (Scholten, Van Knippenberg, Nijstad & De Dreu, 2007), the types of the tasks within a team (van Knippenberg et al., 2004), or teams composed of individuals with high cognitive abilities (Resick, Murase, Randall & DeChurch, 2014). Knippenberg and his colleagues (2004) indirectly argued about the relationship between information elaboration and friendship network, however, there is no study that exactly examines the connection between friendship and gossip network strength and information elaboration.

Social networks

Regarding the definition of social networks, many different ones have been developed over the years. The most applicable regarding the present topic was phrased by Mitchell; he thinks

(9)

9

of a social network as “a specific set of linkages among a defined set of persons, with the additional property that the characteristics of these linkages as a whole may be used to interpret the social behavior of the persons involved" (1969, p.2). The definition states that there are different characteristics of the linkages between the people involved in a network.

This is important because the present study distinguishes between friendship networks and gossip networks. There are different network properties along which social networks can be compared and analyzed. In this present study, social networks are examined regarding two main sets of properties; structural characteristics (network strength) and nature of links (friendship and gossip) (Tichy, Tushman & Fombrun, 1979).

Friendship network strength

In the literature, friendship is often called an informal relationship between colleagues who share friendship connections with each other (Nielsen, Jex & Adams, 2000; Erbe, 1966).

Friendship network covers the friendship connections in a social network, while friendship network strength shows the number of the one-way friendship connections among the members. Several existing studies examine the direct effect of friendship network on turnover (Riordan & Griffeth, 1995), job satisfaction (Riordan & Griffeth, 1995), or organizational choice (Kilduff, 1992), so this study focuses on a less developed field. The present study analyzes the effect of friendship network strength on information sharing and on information elaboration.

There are some papers which examined friendship networks and information sharing.

Based on these studies it is known that a dense friendship network facilitates and increases the information sharing (Ingram & Roberts, 2000), informal connections help colleagues to share strategic information with each other (Katz & Martin, 1997), and there is a positive relationship between knowledge transfer and strong connections (friendship connections) in a

(10)

10

network (Uzzi, 1996, 1997, 1999; Hansen, 1999). Moreover, it has been proved that friendship groups perform better than acquaintance groups due to the greater degree of cooperation, which requires communication and information sharing (Jehn & Shah, 1997). On the contrary, people are less willing to share information and therefore disrupting the possibility of information elaboration, when the others in the team are dissimilar to them (Knippenberg et al., 2004). Prior research has shown that people are more likely to become friends with others who are similar to them (Van Duijn, Zeggelink, Huisman, Stokman &

Wasseur, 2003). Following this line of reasoning, friendship network strength might affect the amount of information shared and elaborated on in the team. In teams where the team members consider themselves friends, there might be more information sharing because friends are more likely to help each other (Clark, Mills, & Corcoran, 1989). Team members that have strong friendship connections in the network are more encouraged to share information among their peers. On the contrary, team members that do not consider their fellow peers as friends are less encouraged to share information among their team members.

These team members might dislike each other. Prior research has shown that team members that do not like their peers are less likely to share strategic information among their team members (Knippenberg et al., 2004). Therefore, I expect that increased friendship network strength leads to more information sharing and information elaboration. Hence, taking together these thoughts, I theorize that in a friendship network people are more likely to engage in information sharing and information elaboration. Consequently, I believe that friendship network strength has a generally positive effect on both information sharing and information elaboration. However, as written in the introduction, too close friendship connections can result in more or only social information sharing in the team, which underplays sharing work related information (Morrison & Nolan, 2007). Therefore, I presume

(11)

11

a curvilinear relationship between friendship network strength and information sharing and information elaboration. Thus, my hypotheses 1a and 1b are the following:

H1a: There is a negatively skewed curvilinear relationship between friendship network strength and information sharing.

H1b: There is a negatively skewed curvilinear relationship between friendship network strength and information elaboration.

Gossip network strength

For management researchers gossip was not an interesting topic for a long time, but recently there are more and more studies regarding this phenomenon (Kniffin & Sloan Wilson, 2010).

One of the reasons why it was neglected for many years is the broad variety of the definitions of gossip. Therefore, I highlight the one which shows why it is interesting to study gossip in a working environment. Kurland and Pelled defined workplace gossip as “informal and evaluative talk in an organization about another member of that organization who is not present” (2000, p.429). Gossip network covers the gossip connections (both positive and negative) between team members, whereas gossip network strength measures the compactness of the one-way gossip connections in a particular network. Although literature often differentiate between positive and negative gossip (Kurland & Pelled, 2000; Elias & Scotson, 1994), this study aims to examine the effect of the overall gossip connections.

In general, people find gossip entertaining (Grosser, Lopez-Kidwell & Labianca, 2010) and they spend a lot of their time by doing it (Dunbar, 2004). However, prior research has shown that gossip tends to be negative more often than positive (Wert & Salovey, 2004).

As a consequence, the degree of negative gossip outweighs the degree of positive gossip in a team. Hence, the overall effects of gossip networks might be negative. Thus, in the present study I assume that gossiping creates a negative atmosphere in a team. Based on the findings

(12)

12

of Knippenberg et al. (2004) it is logical to think that opposed to the positive and friendly atmosphere, in a gossiping (unpleasant) environment people are less open, and less willing to spend their time by discussing things with people who they dislike. Therefore, I predict that in a strong gossip network information sharing and elaboration are less likely to occur. Hence, I presume that gossip network strength has an inverse curvilinear effect on information sharing and on information elaboration. When there are no gossip connections in a team, information sharing and information elaboration is on a certain level. However, as gossip appears and the number of gossip connections increase, the degree of negative gossip also increases.

Consequently, they underplay the positive effects of the few positive gossip, and as the gossip network gets stronger there will be less information sharing and elaboration. Thus, I hypothesize that:

H2a: There is an inversed curvilinear relationship between gossip network strength and information sharing.

H2b: There is an inversed curvilinear relationship between gossip network strength and information elaboration.

Team energy

After reading several sources about team energy, this paper ventures to create an own definition; “team energy is an outcome of the motivation, commitment, and enthusiasm of the group members, which helps employees to keep themselves active in order to achieve the common goal of the team”. In academic articles the term ‘team vitality’ is the most similar concept to the above created definition. Kaiser wrote about this concept in several articles with different co-authors; he defined team vitality as “the degree of morale, engagement, and cohesion among members of the teams for which the managers are responsible” (Kaiser &

Overfield, 2010, p.114 and Kaiser, McGinnis & Overfield, 2012, p.126). Team energy is the

(13)

13

aggregated individual energy from each team member (Cole, Bruch & Vogel, 2012). If each team member is energized, the team will also be energized. Energized teams are more likely to be successful and have good performance, since collective energy is positively related to unit performance (Cole, Bruch & Vogel, 2012). Contrary, if team members are not energized, the team will not be energized either. Presumably, non-energized teams are less successful, are less likely to reach the required performance, and are therefore less useful for the organization.

Furthermore, as previous research indicated, there is a relationship between interpersonal interactions – which refers to social networks – and human energy (Fritz, Lam

& Spreitzer, 2011). Especially it was pointed out that when these interactions are negative human energy is draining. Gossip may also be a negative interaction as it is more likely to be negative (Wert & Salovey, 2004), hence it may decrease human energy. Consequently, I hypothesize that gossip network strength is negatively related to team energy. Fritz, Lam and Spreitzer (2011) also identified strategies which employees can use in order to raise their energy. Among other factors, they explored that relational strategy, which covers positive interactions and connections (e.g.: friendships) with people, can help to sustain a sufficient human energy (Fritz, Lam & Spreitzer, 2011). Therefore, if team members are part of a friendship network, their energy will increase individually which results in an energized team.

As a consequence, I assume, that friendship network strength is positively related to team energy.

One could argue that energized teams are more likely to share information and elaborate on information (Dutton & Ragins, 2007) because their energy is sustained on an adequate level, and they want to reach their set goals. Contrary, low energy teams might do the opposite. They might not share information among team members because they are not energized and not motivated to reach their goals. Also, information elaboration might be low

(14)

14

for the same reason. In a non-scientific article the positive effect of information sharing was revealed on team energy (Sebuliba, 2017). Therefore, the relationship between the two variables is proven, although in the opposite direction. I assume, that in energized teams, information sharing and information elaboration occurs with a greater degree, thus team energy is positively related to them.

Moreover, according to Dutton and Ragins (2007), energized teams are encouraged to work more cooperatively, therefore to share information with each other, and to elaborate on information together. In this paper the authors also revealed when colleagues enjoy the positive interactions (friendship connections) and they share strong high-quality connections (friendship) with each other, they feel energized. Thus, friendship connections may increase the team energy, which makes a team more cooperative, hence more information sharing and elaboration action takes place. Furthermore, based on the above reasoning and on the paper of Fritz, Lam and Spreitzer (2011), I predict the opposite effects with gossip network strength.

When members in a team experience negative interactions (gossiping) they feel de-energized and therefore they might act less cooperatively. As a consequence they do not, or they share and elaborate less information. Hence, based on the above reasoning, I assume that team energy mediates the relationship between friendship- and gossip network strength and information sharing and information elaboration. Consequently, hypotheses 3a, 3b, 3c, 3d, 4 and 5 are the following:

H3a: Friendship network strength leads to increased team energy, which in turn leads to increased information sharing.

H3b: Friendship network strength leads to increased team energy, which in turn leads to increased information elaboration.

(15)

15

H3c: Gossip network strength leads to decreased team energy, which in turn leads to decreased information sharing.

H3d: Gossip network strength leads to decreased team energy, which in turn leads to decreased information elaboration.

H4: There is a positive linear relationship between team energy and information sharing.

H5: There is a positive linear relationship between team energy and information elaboration.

The conceptual model can be seen below, on Figure 1.

FIGURE 1 Conceptual Model

(16)

16 METHOD Data and sample

Altogether I sent my survey to 38 teams, from which I could use 30 for my analysis. I had to exclude teams due to different reasons. One reason was the insufficient filling rate of the members within a team. My threshold for the filling rate was 80%, which means that 80% of the members within one team had to fill in my survey, otherwise the network data could not be analyzed. Another reason for the exclusion was the small size of a working team. I also had a threshold for the group size which was 4. It means that there had to be at least 4 members in one team in order to be able to analyze the social network data. A third reason for not including a team was the characteristics of the questions. One of the teams considered the questions to be rather too personal and therefore the members refused to proceed with the survey. After I had the 30 teams which I could use, I had 2 datasets per each team; one was filled only by the manger and the other was filled by all the other team members. Hence, I merged the 60 different datasets into one dataset.

The characteristics of the 30 working teams which I could include in the analysis were the following: they were from 23 different companies, and consisted of 139 individuals. My research incorporated 20 Hungarian and 10 Dutch teams. The team size varied from 4 members to 10 members. Most of the teams, 17 out of 30 consisted of 4 members, there were 6 teams which consisted of 5 members, there were 3-3 teams with 6 and 7 members and there was one group with 10 people. The average age of the participants was 32.42 (SD = 6.78), and the range was between 21 – 60 years. There were more women (57,6%) among my respondents than men (42,4%). The average tenure of the employees was 4.15 (SD = 4.05) years, and the average group tenure was 2.31 (SD = 2.37) years. The range varied between 0- 39 years for the company tenure, and 0-17 years regarding the team tenure. 87,8% of the respondents were working as full time employees and 12,2% of them were trainees.

(17)

17 Survey design

My study was a correlational study, since I only observed, collected and measured data on an existing, every day situation (Gellert, 2017). I gathered the data with the help of an online survey program; Qualtrics. It incorporated 18 questions, regarding all of my main variables (friendship network strength, gossip network strength, team energy, information sharing, and information elaboration) and some additional control variables. The last question was measuring the team performance, which was only necessary to be filled by the managers of each team, the other 17 questions appeared for everybody.

Measures

Friendship Network Strength

In order to measure friendship network strength, I used the method of Kilduff (1992). I provided a list with the names of each team member, and asked them to mark who they consider as personal friends (Kilduff, 1992). Providing the names in advance, and not ask them to give the names by themselves was a conscious decision, since in this way I could control for the memory and selectivity bias (Ellwardt, Labianca & Wittek, 2012). I gave a definition for personal friends to make it clear for the respondents, and I used the definition of Reagans and McEvily (2003); “These are people with whom you like to spend your free time, people with whom you get together for informal social activities such as going out to lunch, dinner, drinks, films, visiting one another's homes, and so on” (p250). For calculating the value of the friendship network strength, I summed the indicated friendship connections in a team and divided it by the size of the particular group. As an example, in my first team there were 4 members and they indicated 4 actual connections, therefore the strength of this network was 4/4 = 1. I did the same with the other groups as well. For the testing of H1a and H1b, an additional calculation was needed. I computed the squared values of friendship

(18)

18

network strength with the ‘Compute Variable’ option of SPSS, in which I multiplied this variable by itself. With the ‘Squared Friendship Network Strength’ variable the curvilinear relationship could be tested (Burrill, 1997).

Gossip Network Strength

I used the method of Ellwardt, Labianca and Wittek (2012) for measuring gossip network strength. Regarding this variable, I also provided the list of the members and asked the respondents to indicate how often and what type of gossip do they receive from the others.

The frequency was measured on a 5 point Likert scale (1 = Never, 2 = Once in a year, 3 = Once in a month, 4 = Weekly basis, 5 = Daily basis), and they had three options for the type of gossip; positive, negative and mix. (Ellwardt, Labianca & Wittek, 2012). For calculating gossip network strength I did the same as for the friendship network strength; summed the actual gossip connections in a team and divided it by the number of the team members. To give an example, I explain this process through my first team’s answers again. The value of the actual gossip connections was 35, therefore, the strength of this network was 35/4 = 8.75.

I did the same calculation with all the other teams as well. It was also needed to calculate the

‘Squared Gossip Network Strength’ variable, in order to test the curvilinear relationship (Burrill, 1997) predicted by H2a and H2b. It was done in the same way as before, with friendship network strength.

Information Sharing

In order to measure one of my dependent variables; information sharing, I merged the items from two studies; Kearney, Gebert and Voelpel (2009) and Homan, Hollenbeck, Humphrey, Van Knippenberg, Ilgen and Van Kleef (2008). Both papers were published in the Academy of Management Journal, therefore they are valid, and can be used for further researches. I used a 4-item instrument combining these studies and measured them on a 5-point Likert scale,

(19)

19

where 1 was ’Strongly disagree’ and 5 was ’Strongly agree’. I asked the participants to what extent they agree with the statements listed in the survey regarding their work team. The items contained statements like „Team members freely exchange different ideas for solutions to team problems or decisions” or „We exchanged uncommon opinions and viewpoints about the task”. In order to test the reliability of this instrument, I used Cronbach’s Alpha, and the value was sufficient: .83.

Information Elaboration

For my other dependent variable, information elaboration, I also used several papers to be able to measure it in the way I needed. Therefore, I combined the items from three different studies; Todorova (2011) and Kearney, Gebert and Voelpel (2009) and Devine (1999), and created a 5-item measurement scale. To some extent I had to rephrase the items, in order to adapt them to my study. I measured them on a 5-point Likert scale (1 = strongly disagree; 5 = strongly agree). The items consisted of sentences such as “Team members make task-related decisions or actions with the information shared in conversations” or “Team members tend to dismiss what the others say in conversations”. The second item was a reversed item, hence before I calculated the Cronbach’s Alpha to examine the reliability of them I had to recode it with the reversed values. After this, the reliability was .72, which is an adequate value.

Team Energy

To measure my mediator variable, I used the study of Cole, Bruch and Vogel (2012). This paper was published in the Journal of Organizational Behavior, which is one of the SOM Journals therefore it is a valid and acceptable source of my items. They used an instrument which consists of 14 items, categorized in 3 dimensions; namely affective, cognitive and behavioral dimension. All the items were measured on a 5 point Likert scale, the first dimension was measured on a frequency scale (1 = never; 5 = frequently, if not always) and

(20)

20

the other two dimensions were measured on an agreement continuum (1 = strongly disagree; 5

= strongly agree). This instrument included items such as “People in my work group feel excited in their job” or “In my work group, there is a collective desire to make something happen”. The reliability of each dimensions were the following; .86 for the affective, .67 for the cognitive and .64 for the behavioral dimension. Not all of these Cronbach Alphas were acceptable separately, but the reliability of the 14 items together was .83 which means they are reliable and can be used in the analysis. In order to elevate this variable to a team level, I calculated one value for each team. Contrary to the previous calculations, which I did completely manually, I calculated this value with the help of SPSS. This variable was measured with a 14-item scale. First, with the option of ‘Compute variable’ in SPSS, I calculated the mean of the 14 answers of each respondent. After this, I aggregated the values of each participant according to the ‘TeamID’ variable, also by taking the mean of the values.

In this way, I had one value per team for the variable ‘Team Energy’.

All the items which I used in my survey are from previous validated studies. The overview of the used items can be found in Appendix A.

(21)

21 RESULTS

Data preparation

During the data cleaning process in SPSS I erased the partly-filled responses and the unnecessary variables which were being created by Qualtrics automatically such as

“Duration”, “IP address”, “Start date” or “Location latitude”. Also, I renamed the variables from ‘Q2, Q3, Q4’ etc. formulations into their actual name, like Age, Gender, Employee status and so on, in order to get a more transparent dataset. Moreover, I added two variables, namely ‘ID’ and ‘TeamID’ in order to be able to prepare the final dataset. As a next step I computed the friendship network- and gossip network strength of each team, in the way I mentioned in the ‘Method’ section. After I had one value per team for friendship network strength and one value per team for gossip network strength I calculated the value of the team energy, since it is a team-level variable, and not an individual-level variable. The calculation is already described in the ‘Method’ section. As a next step to have a complete dataset, I aggregated all the variables according to the ‘TeamID’ variable, so I had one value per team for each variable. The final step was to calculate the squared variables from both of the independent variables. The calculation is described in the ‘Method’ section. Consequently, my dataset was ready to analyze, which in the end consisted of 30 responses.

Hypotheses testing

As one can see from my models, they consist of four separate mediation analyses, moreover other relationships have been hypothesized. Therefore, after I conducted a bivariate correlation, I examined those 4 hypotheses, which assumed a curvilinear relationship between my dependent and independent variables by conducting four linear regressions with the squared independent variable values. After this, regarding my main hypotheses, I used Model

(22)

22

4 of the PROCESS module of SPSS in order to test them. The correlations between my main variables can be seen in Table 1.

TABLE 1

Results correlation analysis

Hypothesis 1a assumed a negatively skewed curvilinear relationship between friendship network strength and information sharing. To test this hypothesis a new variable needed to be computed, the square of the independent, namely the friendship network strength variable (as it is described in the ‘Method’ section). The regression analysis revealed the linear and also the non-linear regression between them. Neither of them is significant; neither the estimated linear main effect (B = .1, t = 1.15, p = .26, LLCI = -.08, ULCI = .28, R² = .05), nor the estimated non-linear main effect with the squared values (B = -.04, t = -.49, p = .63, LLCI = -.22, ULCI = .13, R² = .05). Table 2 demonstrates these results.

Hypothesis 1b predicted that there is a negatively skewed curvilinear relationship between friendship network strength and information elaboration. After conducting the same regression analysis as for the previous hypothesis testing, the results showed no significant relationships. Again, I did a linear regression analysis by including the normal independent variable and the squared independent variable, which revealed the linear and the non-linear relationship between friendship network strength and information elaboration. The results showed that neither of them is significiant; neither the estimated effect of the linear regression (B = .08, t = .92, p = .36, LLCI = -.1, ULCI = .27, R² = .03), nor the estimated effect of the

M SD α 1 2 3 4 5

1 Friendship network strength 1,69 0,89 -

2 Gossip network strength 13,60 3,88 - ,400*

3 Team energy level 3,72 0,61 0,83 -0,28 -0,10

4 Information sharing 4,11 0,42 0,83 0,21 0,24 -0,15

5 Information elaboration 4,08 0,44 0,72 0,17 0,02 0,03 ,812**

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

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

(23)

23

non-linear regression with the squared values (B = -.002, t = -.03, p = .98, LLCI = -.19, ULCI

= .18, R² = .03). The results are shown in Table 2.

TABLE 2

Results regression analyses I.

Predictor B (SE) LLCI ULCI

Hypothesis 1a

Dependent variable model 1: Information sharing

Constant 3,94 0,17 3,59 4,27

Friendship network strength 0,10 0,09 -0,08 0,03

Model R² 0,05

Constant 3,84 0,27 3,29 4,38

Squared friendship network strength -0,04 0,09 -0,22 0,13

Model R² 0,05

Hypothesis 1b

Dependent variable model 1: Information elaboration

Constant 3,94 0,17 3,58 4,29

Model R² 0,03

Constant 3,93 0,28 3,36 4,50

Squared friendship network strength 0,00 0,09 -0,19 0,18

Model R² 0,03

Notes. N=30 †p<.01,*p<.05, **p<.01, ***p<.001

I also predicted the relationships between gossip network strength and information sharing and information elaboration. Hypothesis 2a assumes that there is an inversed curvilinear relationship between gossip network strength and information sharing. To be able to test this non-linear relation, I calculated the squared variable of gossip network strength as well as I did with friendship network strength. After I had this new variable, I was able to run the linear regression analysis with the normal independent variable and the squared independent variable. Unfortunately, the results of this hypothesis testing did not show any significance either. The estimated linear regression between gossip network strength and

(24)

24

information sharing is not significant (B = .03, t = 1.3, p = .2, LLCI = -.02, ULCI = .07, R² = .06) and the estimated non-linear effect between them is also not significant (B = .00, t = -.04, p = .97, LLCI = -.01, ULCI = .01, R² = .06). The results are shown in Table 3.

Hypothesis 2b presumed an inversed curvilinear relationship between gossip network strength and information elaboration. In order to test it, a linear regression was conducted with both the normal and the squared gossip network strength variables. The results of this regression are similar to the previous hypotheses testing results; there is no signifcant linear or curvilinear relationship. The result of the linear regression, calculated with the normal independent variable is not significant (B = .003, t = .12, p = .9, LLCI = -.04, ULCI = .04, R²

= .001). The result of the estimated non-linear effect is also not significant (B = -.001, t = -.17, p = .87, LLCI = -.01, ULCI = .01, R² = .002), which suggests that there is no curvilinear relationship between these variables. Table 3 shows the above mentioned results.

(25)

25

TABLE 3

Results regression analyses II.

Hypothesis 2a

Constant 3,76 0,28 3,18 4,33

Gossip network strength 0,03 0,02 -0,02 0,07

Model R² 0,06

Constant 3,73 0,76 2,16 5,30

Squared gossip network strength 0,00 0,00 -0,01 0,01

Model R² 0,06

Hypothesis 2b

Constant 4,04 0,30 3,43 4,66

Model R² 0,00

Constant 3,92 0,82 2,23 5,60

Squared gossip network strength 0,00 0,00 -0,01 0,01

Model R² 0,00

Notes. N=30 †p<.01,*p<.05, **p<.01, ***p<.001

The first mediation hypothesis (H3a) assumes that friendship network strength leads to increased team energy, which in turn leads to increased information sharing within the team.

It can be seen from the correlation table, that neither of the relationships are significant. This suggests that friendship network strength has no influence on team energy nor on information sharing, in addition team energy has no influence on information sharing either. To test this hypothesis a mediation analysis is examined. Although, since my independent variable does not correlate to my dependent variable, it suggests, that there is no relationship and no mediation at all. The mediation analysis showed that this suggestion was right, which I elaborate in details below. Path A in the mediation analysis, which examines the regression

(26)

26

between friendship network strength and team energy is not significant (B = -.18, t = -1.52, p

= .14, LLCI = -.44, ULCI = .06, R² = .07), path B which shows the regression between team energy and information sharing is also not significant (B = -.06, t = -.49, p = .62, LLCI = -.34, ULCI = .21, R² = .05), therefore there is no mediation effect. Also, the regression between friendship network strength and information sharing is not significant (B = .09, t = 1.15, p = .26, LLCI = -.08, ULCI = .27, R² = .05), which suggests that there is no partial mediation either. The results can be seen in Figure 2 and Table 4.

FIGURE 2

Results hypothesis testing (H3a)

The next main hypothesis (H3b) assumes that friendship network strength leads to increased team energy, which in turn leads to increased information elaboration within the team. There are also no significant correlations between the variables, which again suggests that there is no relation between information elaboration and both friendship network strength and team energy, nor between friendship network strength and team energy. To test this hypothesis a mediation analysis was conducted, included the 3 main variables. This model showed the same results as the previous hypothesis. Since the independent and dependent variable is not correlated to each other (Table 1), I assumed that there is no mediation effect again. The results supported this assumption, which I report below. Path A, which shows the regression between friendship network strength and team energy is not significant and it is the same as before (B = -.18, t = -1.52, p = .14, LLCI = -.44, ULCI = .06, R² = .07).

Unfortunately, path B (regression between team energy and information elaboration) is not .09 (.08)

-.06 -.18

Friendship network strength Information sharing

Team energy

(27)

27

significant either (B = .06, t = .45, p = .66, LLCI = -.22, ULCI = .35, R² = .04), which shows that there is no mediation effect in this model. Moreover, path C, the regression between the independent and dependent variable is not significant either (B = .084, t = .92, p = .36, LLCI

= -.1, ULCI = .27, R² = .03), therefore there is no partial mediation. Altogether, there is no mediation in this model at all. The results are shown in Figure 3 and in Table 4.

FIGURE 3

Results hypothesis testing (H3b)

.084 (.09) -.18 .06

Friendship network strength Information elaboration

Team energy

(28)

28 TABLE 4

Results mediation analyses I.

Hypothesis 2a

Mediator variable model: Team energy level

Constant 4,03 0,23 3,55 4,51

Friendship network strength -0,18 0,12 -0,44 0,06

Model R² 0,07

Dependent variable model: Information sharing

Constant 4,2 0,56 3,03 5,37

Team energy level -0,06 0,13 -0,34 0,21

Model R² 0,05

Hypothesis 5a

Constant 4,03 0,23 3,55 4,51

Friendship network strength -0,18 0,12 -0,44 0,06

Model R² 0,07

Dependent variable model: Information elaboration

Constant 3,67 0,59 2,44 4,9

Team energy level 0,06 0,14 -0,22 0,35

Model R² 0,036

Notes. N=30 †p<.01,*p<.05, **p<.01, ***p<.001

In this study 15 control variables were added (age, gender, tenure, group tenure, salary, employee status, task complexity, task interdependence, individual energy, Big5 personality, team performance), from which only task complexity, task interdependence and agreeableness from Big5 personalities were significantly correlated to my dependent variables. However, I did not include them in the analyses in order not to restrict the models too much. Therefore, I only report the results with my main variables. The full correlation table including all the variables used in the study is showed in Appendix B.

Regarding the next main hypothesis (H3c), the results are also not significant. The third mediation hypothesis predicts that gossip network strength leads to decreased team energy, which in turn leads to decreased information sharing. None of the correlations in this model are significant (Table 1), which shows that there is no relation between the variables. I also conducted the mediation analysis in order to test this hypothesis, although since there are no correlations between the variables included in this model, I did not expect any significant results. The mediation analysis reassured this expectation. The regression between the

(29)

29

independent variable and the mediator was not significant (B = -.015, t = -.53, p = .59, LLCI = -.07, ULCI = .04, R² = .01), nor the regression between the mediator and the dependent variable (B = -.08, t = -.66, p = .51, LLCI = -.35, ULCI = .17, R² = .07), which proves that there is no mediation. In addition path C shows that the regression between the independent and the dependent variable is also not significant (B = .03, t = 1.3, p = .2, LLCI = -.01, ULCI

= .06, R² = .06), hence there is no mediation at all. This is also shown in Figure 4 and in Table 5.

FIGURE 4

Results hypothesis testing (H3c)

The last main hypothesis (H3d) presumes that gossip network strength leads to decreased team energy, which in turn leads to decreased information elaboration. This model’s results are very much alike as the previous models; no significant correlations (Table 1). While testing the hypothesis by the mediation analysis, it can be seen that the regressions are also not significant. This suggests that there is no mediation here either. Path A shows the regression between gossip network strength and team energy (B = -.016, t = -.53, p = .59, LLCI = -.07, ULCI = .04, R² = .01), path B shows the regression between team energy and information elaboration (B = .026, t = .19, p = .85, LLCI = -.26, ULCI = .31, R² = .002), while path C demonstrates the regression between gossip network strength and information elaboration (B = .0026, t = .12, p = .9, LLCI = -.04, ULCI = .05, R² = .0005). One can see from these values that no mediation occurred in this model. The results of the mediation analysis are shown in Figure 5 and in Table 5.

.03 (.02)

-.08 -.015

Gossip network strength Information sharing

Team energy

(30)

30 FIGURE 5

Results hypothesis testing (H3d)

The testing of hypothesis 4 and hypothesis 5 are already included in the above mentioned results (“Path B” of each mediation analyses). Hypothesis 4 assumes that there is a positive linear relationship between team energy and information sharing, while hypothesis 5 predicts that there is a positive linear relationship between team energy and information elaboration. It can be seen from the mediation analyses before, that these relations are not significant either.

.026 -.016

Gossip network strength Information elaboration

Team energy

-.0026 (-.003)

(31)

31 TABLE 5

Results mediation analyses II.

Hypothesis 2b

Constant 3,92 0,41 3,08 4,77

Gossip network strength -0,01 0,03 -0,07 0,04

Model R² 0,01

Dependent variable model: Information sharing

Constant 4,09 0,57 2,9 5,28

Team energy level -0,08 0,13 -0,35 0,17

Model R² 0,07

Hypothesis 5b

Constant 3,93 0,41 3,08 4,77

Gossip network strength -0,015 0,03 -0,07 0,04

Model R² 0,01

Dependent variable model: Information elaboration

Constant 3,93 0,63 2,65 5,22

Team energy level 0,02 0,14 -0,26 0,31

Model R² 0,002

Notes. N=30 †p<.01,*p<.05, **p<.01, ***p<.001

(32)

32

DISCUSSION AND CONCLUSION

The three main contributions of this paper were to reveal the effects of friendship/gossip network strength on information sharing and on information elaboration, to examine whether team energy mediates these relationships (between friendship network strength and information sharing, friendship network strength and information elaboration, gossip network strength and information sharing and gossip network strength and information elaboration), and to give practical hints for executives about how to form teams in which information sharing and elaboration is ideal. The results of this paper showed, that none of the predicted relationships are significant, therefore all my hypotheses were rejected. Mediation and partial mediation did not occur, neither did single linear regressions nor curvilinear regressions. It means, that this paper does not support previous studies which examined some of the relationships from my models with slightly different variables.

It is now clear, that examining the strength of a network does not give the same results as when one considers the density of a network. In other words, the results of the study by Ingram and Roberts (2000) which showed a significant positive relationship between friendship network density and information sharing could not be repeated by examining social network from a different point of view. This study showed no significant effect of friendship network strength on information sharing. It may be due to the difference regarding the network properties. Network strength covers also the one-way connections in a network, while density only covers the two-way connections; therefore, network strength is a broader aspect. As network strength incorporates the connections which are only indicated by one of the people and not both of them, it may mean that for example the friendship connection is only present from one side, and not interpreted by both people in a dyadic relationship.

Consequently, network strength is not as accurate as network density and may give biased results.

(33)

33

In the theory section, I assumed from the study of Knippenberg et al. (2004) and Van Duijn et al. (2003) that in a friendship network information elaboration supposed to occur with a higher chance. However, my study showed no significant relationship between friendship network strength and information elaboration. This result might be explained by the fact that a vague speculation was made based on these two studies. As Van Duijn et al.

(2003) reveal in their results, visible similarity only affect friendship formation in the initial stage, while invisible similarity has a very weak influence on it. Therefore, it is not a strong argument for the statement that people rather become friends with people who are similar to them. Additionally, Knippenberg et al. (2004) only proved that people are less willing to engage in information elaboration activities with diverse others, Therefore, based on this, I assumed that people are more willing to elaborate on information with peers who are similar to them. Hence, I hypothesized the opposite case of the proven results. It is clear now, that an opposite case is not necessarily supported, even if the original relationship was supported.

Taking together the above mentioned reasoning, this assumption was built on an instable base.

Other results disclosed in this study also opposed a previous study, written by Fritz, Lam and Spreitzer (2011). They revealed four themes of tactics with which human energy can be sustained; among others the relational tactic. This strategy refers to the positive connections and interactions between team members. Since friendship is a positive interaction, it can be concluded that with friendship connections human energy can be maintained high. Therefore, in a friendship network, team energy is supposed to be sustained on a high level. In contrast to these findings, the present study showed no significant relationship between friendship network strength and team energy. Fritz, Lam and Spreitzer (2011) also emphasized that negative interactions (gossiping) can drain human energy.

Therefore, there was supposed to be a negative significant relationship between gossip network strength and team energy. However, my social network data revealed no significant

(34)

34

relationship between gossip network strength and team energy. These findings may be a consequence of the incomplete or inadequate aspects of positive and negative interactions.

Friendship networks in itself do not cover the whole concept of positive interactions, since the positivity of an interaction covers the extent to which it is friendly, pleasant and agreeable (Brondolo, Rieppi, Erickson, Bagiella, Shapiro, McKinley, & Sloan, 2003). Gossip networks in turn are not considered as part of negative interactions at all, since negative interaction means that there is anger, discomfort, upset or tension in an interaction (Brondolo et al., 2003). As a consequence, friendship network strength and gossip network strength alone does not have a significant effect on team energy, as there are other factors under the notion of positive/negative interactions.

Lastly, it was expected that team energy has a positive effect on information sharing, however, my dataset showed no significant relationship between them. This might be due to the fact, that the non-scientific source which I used (Sebuliba, 2017) to theorize this relationship, revealed this connection in the opposite direction. Hence, it was stated that information sharing can enhance team energy (Sebuliba, 2017). The results of this study suggest that the direction of this relation is irreversible. This result supports the findings of Fritz, Lam and Spreizer (2011), who also examined the relationship between energy and information sharing in the opposite direction. They identified several work-related strategies with which employees can manage their energy at work. One of them was ‘talking to a co- worker or to the supervisor’, which can be interpreted as sharing information with others.

However, the results of their study showed no significant relationship between this strategy and vitality.

Since no linear regression occurred, it is not possible to reveal the reasons for the non- significant results of the mediation models. There are two reasons for this, which is written below. First, there cannot be a significant indirect effect between the independent and the