1
Isomorphic Pressures affecting Knowledge Sharing Inside Virtual
Communities.
Master Thesis by:
Koen Rijper
Student Number:
10188924
Supervisor :
Lori Divito
Word Count:
13269
Abstract:
This research studied whether isomorphic pressures affects knowledge sharing inside virtual communities. To test this, this study created a new survey, and gathered data from 285 respondents. After deletion of data for various reasons a sample of 89 was used, and after controlling for Education, Work Experience, Gender, Age, Self-Efficacy, Enjoyment of Helping, Exchange Ideology, Country, Forum type, and Nature of Presence, this study found that coercive isomorphic pressures have a positive relation with knowledge sharing inside virtual communities. Furthermore, enjoyment of helping others, self-efficacy, and education were also shown to have a positive relation with knowledge sharing inside these communities. By conducting this research, I did not only contribute to the virtual community research field, but also to the knowledge sharing, isomorphic pressures, and globalization research fields. Furthermore, practical implications were found for marketers, businesses, the government, and controllers of these virtual communities.
2
Statement of Originality
This document is written by Student Koen Rijper 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.
1. Table of Content
Introduction ... 3 Literature Review ... 4 3.1 Knowledge Sharing ... 5 3.2 Motivations to share knowledge ... 6 3.3 Virtual Communities ... 8 3.4 Isomorphic Pressures ... 13 Method ... 16 4.1 Sample and Data Collection ... 16 4.2 Measures ... 18 4.2.1 Measures Dependent Variables ... 18 4.2.2 Measures Independent Variables ... 18 Results ... 22 5.1 Recoded Variables ... 22 5.2 Reliability and Validity ... 22 5.3 Testing the Hypothesis ... 25 Discussion ... 30 6.1 Findings and Limitations ... 30 6.1 Implictions ... 33 Conclusion ... 37 Appendices ... 38 Bibliography ... 473
2. Introduction
Grant (1996) developed in his research the knowledge based theory of the firm, stressing the importance of knowledge creation inside firms to gain a competitive advantage. This was only a start of recognizing the importance of knowledge, and considering the new technological advances in the 21st century, the accessibility to, and appetite for knowledge can only be seen to have increased even further (Howells & Bessant, 2012). Barnatt (1998) already realized this in 1998 where he stated that “As the user-base of the Internet expands, online virtual communities may have the potential to become the key customer-infomediaries, social forums, and trading arenas of the early twenty-first century” (Barnatt, 1998, pp. 161). This has actually happened through, for example, the introduction of the famous Bloomberg trading terminals (Okada & Azuma, 2013). This insight evolved, and the field of research saw an increasing usage of the virtual communities becoming more and more important for sharing and attaining knowledge fast (Maclaran & Caterall, 2002). This can be seen back in sites or programs like blogs, Facebook, Secondlife, Twitter, forums, among others, whom can be considered big contributors to the increasing access and appetite for knowledge (Kodama, 2005).
Since the emergence of virtual communities, they have been on the research agenda. Many studies have gone out to the social impact of virtual communities (Hercheui, 2011), control (Jones, 1995; Steinmuller, 2002), purchase behaviour (Leal, et al., 2014), and marketing options (Hercheui, 2011). More importantly is that a significant part of the research on virtual communities addresses the contribution of virtual communities in knowledge sharing. These studies mainly try to discover why and how the virtual community is such an effective tool to share knowledge and how best to design a virtual community to support knowledge sharing at its best. All of these studies, however, fail to address one very important aspect of these virtual communities. The institutional aspect, which because of the globalization has become an increasingly important subject, has been mostly ignored (Hercheui, 2011). This is odd, since Virtual Communities have their own share in the constantly evolving globalization, and are arguably the most globalized communities out there (Kodama, 2005).
The few that did discuss the institutional aspect inside virtual communities, first state that institutions influence online interactions, without any empirical support (Baym, 1995;
4
Kelemen & Smith, 2001; Bakardjieva, 2003; Fernback, 2007). Furthermore, the aspect of institutions is often researched as a secondary theme (Venkatesh 2003; De Cindio et al., 2003; Hauben & Hauben, 1997; Holtgrewe, 2004; Gattiker, 2001; Ester & Vinken, 2003; Nguyen, 2006; Willson, 2006; Ross, 2007). Institutional influence as primary influence has been only limitedly researched (Matzat, 2004; Hercheui, 2011), however these studies are mostly concerned with government dependant virtual communities. Because virtual communities can be considered a source of knowledge sharing and learning, and is becoming increasingly important in the current globalizing world (Kodama, 2005), this research is going to address the until now neglected institutional, or more specifically, isomorphic pressures, having an effect on knowledge sharing within virtual communities. This leaves this research with the following research question: Do isomorphic pressures affect knowledge sharing inside virtual communities?
The following chapters will address this research gap by first conducting an extensive literature review. In this first the literature surrounding knowledge sharing and the motivation to share knowledge will be outlined. Secondly it will be discussed how we can define a virtual community, and discuss the research that has addressed this subject to date. Thirdly, isomorphic pressures will be discussed in which the possible effect on knowledge sharing and virtual communities will come forward. After the Literature review the method, including the survey and sample will be discussed. Finally, the results will be presented, after which the results will be discussed in the discussion part of the thesis, ending with the conclusion.
3. Literature Review
The literature review will surround three main topics. First of all, the literature on knowledge and knowledge sharing will be reviewed, in which a definition will be found about what “knowledge sharing” actually is, and how knowledge sharing is influenced by certain factors. The second topic that will be reviewed is going to be about the virtual communities. This is done to get a clear view on the literature surrounding virtual communities, including research concerned with virtual communities in relation to knowledge sharing. Finally, the research gap will then be further outlined by reviewing the literature about isomorphism. This will be done in threefold, hence mimetic, normative, and coercive isomorphism. All the literature will be searched through electronic databases, which where filtered through the years 1992 – 2016. Furthermore, the ISIwebofknowledge score will be checked in which the 5-year impact factor of the journal which published the article should notgo below 1.000.There were 36 journals
5
that were taken into account in this research, of which a small selection is: MIS quarterly, Journal of Knowledge Management, Media Culture and Society, Journal of Leadership & Organizational Studies, Technovation, Journal of Business Research, Journal of Management Accounting Systems, among others. Some of the journals provided more articles than others, simply because they had more articles. The key words that were used in the literature search are: Virtual Communities, Online Communities, Forums, Blogs, virtual Networks, Online Networks, Online Professional Networks, Learning, Knowledge Sharing, Information Sharing, Knowledge sharing inside virtual communities, Institutional forces affecting knowledge sharing, Isomorphism, Isomorphic pressures, Isomorphic pressures affecting knowledge sharing.
§ 3.1 Knowledge sharing.
Knowledge sharing has been researched extensively, and gained rapid ground inside of the research agenda after Grant (1996) introduced the knowledge based theory of the firm. It stressed the importance of knowledge within firms, and ever since researchers are searching for motives and other factors affecting the knowledge sharing between firms, within firms, within networks, between network, and between people in general. This section is going to address what knowledge sharing actually is, what motivates people to share knowledge, and other factors affecting knowledge sharing in general.
What can be considered knowledge, and when are we actually sharing knowledge with others? The term knowledge is often confused with information (Wang & Noe, 2010). Wang and Noe (2010) argue that knowledge is information processed by individuals, which can be anything ranging from facts to judgements (Wang & Noe, 2010; Alavi & Leidner, 2001; Bartol & Srivastava, 2002). Knowledge sharing therefore “Refers to the provision of task information and know-how to help others and to collaborate with others to solve problems, develop new ideas, or implement policies or procedures” (Wang & Noe, 2010, pp. 117). Concurrently, knowledge can be considered to be shared when people truly learn from each other, instead of merely stating facts. This needs the realization of knowledge to be of importance to the receiver of this knowledge (Hendriks, 1999; Cummings, 2004).
It can be argued that knowledge sharing is unnatural, since one could imagine people believe their information to be valuable and want to keep it for themselves. This leads some researchers to be hesitant towards new phenomenon’s such as virtual communities as becoming the next knowledge sharing networks (Davenport & Prusak, 1998; Dixon, 2000).
6
To get to the heart of whether and why knowledge will or will not be shared, and whether or not knowledge will be shared in virtual communities, one first needs to create a good understanding about the motivation and factors affecting knowledge sharing.
§ 3.2 Motivations to share knowledge
There are a lot of articles surrounding the motivation to share knowledge (Wang and Noe, 2010). Wang and Noe (2010) propose a framework to summarize this (Appendix 1) and identified three large subdivisions within the research on knowledge sharing, namely Individual, Motivational, and Environmental factors. First of all, the individual characteristics are addressed as a subdivision. Wang and Noe (2010) argue that this subdivision has been limitedly researched. They found that exchange ideology (whether you believe knowledge sharing to be beneficial) (Lin, 2007), Confidence (Lin, 2007), Openness to Experience (Cabrera, et al., 2006), Work Experience and Higher Education (Constat et al., 1994) are all positively related to knowledge sharing. This research would like to add that knowledge is easier shared when superior skills and willingness to both absorb and share knowledge can be found with both the sender and receiver of knowledge (Minbaeva, 2007). Furthermore, Hsu et al. (2007) found that self-efficacy has a positive effect on outcome expectations, which concurrently affected the tendency to share knowledge.
Secondly, motivational characteristics are considered important in the knowledge sharing literature. Wang and Noe (2010) argue that for an individual to share knowledge, the individual needs to believe it has valuable knowledge (Constant et al., 1994). Furthermore, the Social Exchange Theory suggests that in a social exchange, like sharing knowledge, the cost of interaction will always be weighed against the benefits (Blau, 1964). The knowledge sharing literature agrees with this statement in which possible costs of knowledge sharing (time, among others) will negatively influence the tendency to share knowledge, and that possible benefits (such as reputation, respect, among others) will positively influence knowledge sharing behaviour. Furthermore, interpersonal trust and justice is seen as a significant predictor of knowledge sharing behaviour. This trust and feeling of justice can be between two individuals, or between team members (Organ, 1990; Robinson, 1996; Wu & Lee, 2007). However, whether trust is a fair predictor is not as straight forward as Wang and Noe (2010) argue. Chiu et al. (2006) found no relation between trust of two individuals and the quantity of knowledge sharing. Finally, as can be seen in the figure in appendix 1
7
motivational characteristics can also be seen as a mediator and moderator for environmental and individual factors, making them central to the knowledge sharing literature.
Finally, Wang and Noe (2010) propose three forms of environmental factors that affect knowledge sharing behaviour. First of all, Wang and Noe (2010) stress the importance inside the literature on the effect of organizational context on knowledge sharing. This has already been argued by Magnier-Watanabe & Senoo (2008) who state that knowledge management is strongly affected by the characteristics of a firm, and more specifically the structure, internal relationships, job specifications, and the strategy of the firm. Wang and Noe (2010) add to this point that the support of management is another important factor. Ardichvilli et al. (2003) furthermore state that when employees view knowledge to be belonging to the entire organization, knowledge will flow more easily. At the same time, when there is a lot of critique from managers, employees tend to be hesitant to share knowledge. This suggests that the environment in which knowledge is shared affects the quantity and quality of knowledge shared.
Secondly, Wang and Noe (2010) argue that the interpersonal and team characteristics affect knowledge sharing. Whether knowledge is shared is not a wholly affected by the type of knowledge. Hence, Minbaeva (2007) argues that whether knowledge is effectively shared depends mainly on the senders and receivers of this knowledge, and more importantly their relationship with each other. Cummings (2004) argued that external knowledge sharing is more effective when work groups are structurally more diverse. This diversity can be found inside differences in organization affiliations, roles, or positions. Hsu et al. (2007) furthermore proposed a view from the social cognitive theory of knowledge sharing. From his theory he argues around a relation between knowledge sharing, outcome expectations for personal influences, and multi-dimensional trust for environmental influences. Bartol and Srivastava (2002) agree with this statement arguing that to effectively share knowledge, mutual trust needs to exists between individuals. Looking back at the argument from Chiu et al. (2007) whom stated that trust does not affect the quantity of knowledge shared, arguing from the research of Bartol and Srivastava (2002) one could argue that not quantity, but quality of the knowledge shared will be affected by trust. Bartol and Srivastava (2002) concurrently propose rewards based on collective performance, team-based rewards, and companywide incentives to organize the organisational environment to best support knowledge sharing behaviour.
8
Finally, the cultural characteristics have an effect on knowledge sharing. Ford & Chan (2003) and Minbaeva (2007) argue that language differences incur difficulties with knowledge sharing. Language differences can, for example, create knowledge blocks (Ford & Chan, 2003). Furthermore, it has been seen that Chinese and American employees differ significantly in terms of knowledge sharing due to the collectivistic nature of the Chinese economy (Hwang & Kim, 2007; Chow et al., 2000). This is also something that Wang and Noe (2010) acknowledge. Ardichivili, et al. (2006) added to this view by arguing that differences in modesty, competition amongst employees were seen to be significant barriers to information sharing. Furthermore, he found that power distance in a country does not significantly affect knowledge sharing behaviour.
To determine what of the above mentioned literature can be deemed to be applicable to the virtual communities, and to determine what else has been researched in the virtual community literature to clearly outline the research gap that this research will be addressing, the following chapter will address the phenomenon virtual community in detail after which a clear framework will be proposed of the currently existing literature to help future researchers in their endeavour of studying virtual communities.
§ 3.3 Virtual Communities.
Knowledge sharing can go through a number of channels, of which the virtual channel can be considered an effective tool (Cummings, 2004). The realization that virtual communities are effective platforms for knowledge sharing increases in popularity (Yao et al., 2015; Maclaran & Caterall, 2002; Kodama, 2005; Howells & Bessant, 2012; Barnatt, 1998). This is argued to be the case since everything is documented, and the effect of rewards and tracking is very significant (Bartol & Srivastava, 2002). Virtual communities furthermore create huge opportunities to support a better and more effective knowledge management. Kodama (2005), for example, argues that virtual networks are becoming more and more important for knowledge networks. Hendriks (1999) argues that the ICT developments have a positive influence on knowledge sharing motives within these virtual networks. One can argue that Virtual communities are increasingly used to attain knowledge and to solve problems in the acquirer’s personal life. This creates possibilities for employees to use these networks to improve their productivity. Knowledge sharing can thus be viewed as a very important aspect of virtual communities, and thus to create an effective online virtual community one needs to foster and motivate the users to share knowledge (Chiu et al., 2006; Hsu, et al., 2007). To
9
fully address all the literature surrounding virtual communities, this research will first discuss the definition of a virtual community, the past literature addressing virtual communities will be discussed, concurrently the motivations concerning knowledge sharing inside virtual communities will be addressed, after which the effect of the design of the virtual community will be reviewed.
The first question that comes to mind is: what exactly is a virtual community? The definition virtual community is used in many different ways. Virtual communities can, for example, be seen as geographical dispersed people with a same interest, community networks that are bound to neighbourhoods, networks of friends, or social interchange through computer mediated communication (Hercheui, 2011; Komito, 1998; Graham, 1999; Dimaggio et al, 2001; Preece, 2001; Rheingold, 2000; Mansell & Steinmueller, 2000). However, it can also be argued that a virtual community implies having a boundary where a difference can be spotted between members and non-member (Hercheui, 2011; Kling & Courtright, 2003; Fernback, 2007). Others also may refer to virtual communities as social networks, which would then represent online spaces of interaction (Mitra, 1997; Slevin, 2000). One could argue that there is some overlap. Hercheui (2011) goes into detail and argues that in communities, a group of individuals come together voluntarily because of shared interests, and states that “If an online group presents the characteristics of common interests, rules, and voluntary membership, it has boundaries and thus may be called a virtual community” (Hercheui, 2011, pp. 5).
Virtual communities are researched in relation to many different aspects. This research identified 3 recurring subjects that are being studied in retrospect of virtual communities. First there is the research that analyses why people use virtual communities, which Ridings and Gefen (2006) have extensively analysed. This pool of researchers believe that people are attracted to virtual communities because of the fast access to information (Furlong, 1989; Jones, 1995; Wellman, et al., 1996). The virtual community in this case would differ from the normal website information, because of its interactivity and member-generated content which is arguably more interesting (Hagel & Armstrong, 1997). Another reason for people to join a virtual community is because of social support exchange. Hercheui (2011) found that virtual communities can be considered a space to build social capital, and thus attain information and emotional support. This could be compelling for the act of social needs, which can be attained through these virtual communities (Thoits, 1982). Finally, some researchers believe the virtual communities also are used as a way to attain friendships through, again, the basic need
10
for social interaction, and purely for recreation purposes (Igbaria, 1999; Wellman & Gulia, 1999).
The second big topic that is being addressed extensively in the virtual community research are the possible benefits, or influences that the phenomenon virtual communities may have, and how businesses might be able to use or exploit this new phenomenon. Hercheui (2011) found that virtual communities can be considered a space to build social capital, and thus attain information and emotional support. This creates the possibility of democratizing the public debate (Graham, 1999; Rheingold, 2000). Virtual communities are also useful for surveillance and control (Jones, 1995; Steinmueller, 2002), which creates possibilities for better tracking of contributions inside a company, or even as far as surveilling for terrorist attacks. Researchers have furthermore identified the effect of virtual communities on purchasing intentions (Leal, et al., 2014), which is very useful for marketing efforts of firms. Another way virtual communities are beneficial is that they can be seen as an easy way to disperse information more quickly, creating a bigger reach for media firms to share information, and an increasingly informed public. A firm can also use a virtual community to increase customer loyalty, which concurrently affects the amount of purchases an individual makes with that specific company (Kim, et al., 2004). However, above all, virtual communities are a very effective tool in sharing knowledge fast and easy (Yao et al., 2015; Maclaran & Caterall, 2002; Kodama, 2005; Howells & Bessant, 2012; Barnatt, 1998), hence, “The web has become an irreplaceable source for knowledge creation and consumption and online communities have turned out to be the new form of socialization platforms for fulfilling certain needs such as providing or acquiring information, sharing experience” (Seraj, 2012, pp. 209).
The original research concerning knowledge sharing in general discussed above has been partly replicated inside the virtual communities. Whelan (2007) researched the motives of people sharing knowledge in electronic networks of practice. They found that people share their knowledge online when it enhances their professional reputation, when they are embedded in the network, and when they have the experience to share. Chiu et al. (2006) argued that trust, norm of reciprocity, identification, shared vision and shared language will influence knowledge sharing online. Hsu & Lin (2007) argued that the ease of use and enjoyment of sharing knowledge were positively related towards blogging. Big contributors to this line of work are Chen and Hung (2010) whom study knowledge sharing inside virtual communities. They collected data from 323 participants through a web-based survey, in
11
which they tried to find what contextual factors, and individual factors affected the knowledge contribution behaviour and knowledge collection behaviour. Their results suggest that the norm of reciprocity, self-efficacy, and perceived relative advantage where significant predictors of knowledge sharing behaviour. Although he tested a large number of the factors affecting knowledge sharing in general discussed above, they failed to study a very important aspect, hence the institutional environment (Hercheui, 2011).
Chandra and Leenders (2012) argue that the virtual world and the real world are looking alike in terms of communication. This is also something that Evans (2012) argues, but refers more to the experience in virtual communities to be not that different from experiences in real life communities. Considering this, it could be argued that the factors affecting knowledge sharing in general identified by Wang and Noe (2010) could also be applicable to knowledge sharing in the real world. This would suggest that older research of knowledge sharing in general would also be applicable to knowledge sharing in the virtual community (Evans, 2012; Chandra & Leenders, 2012). Although some researchers have tried to provide evidence for different aspects of knowledge sharing in virtual communities, the most part of the physical virtual community studies remain empirically untested, which makes this research hesitant towards taking this perspective.
The third big topic that is analysed in the virtual community research is how to best design a virtual community adjusted for specific purposes, like participation in the general conversation, or enhancing service quality, among others. Aside from the realization that virtual communities are effective platforms for knowledge sharing (Yao et al., 2015; Maclaran & Caterall, 2002; Kodama, 2005; Howells & Bessant, 2012; Barnatt, 1998), creating a community online is difficult to achieve (Kling & Courtright, 2003). The current literature identified a number of factors and models which affect this knowledge sharing inside virtual communities. According to Seraj (2012) there are three characteristics that can create value for members, which makes them more prone to stay inside the virtual community. These are that members need to be goal driven, that the content on the site needs to be of high quality, and that an interactive environment needs to be made so that relationships between members can be build. This can be put together with the research of Tsai and Bagozzi (2014) which argues that “we intentions” are necessary to create an interactive environment, and are a result of cognitive emotional and social drivers.
Furthermore, social identity within the group will affect the we-intentions, or the tendency to see the community as a collective, which is, in its turn, affected by the interactive
12
environment that needs to be put in place. This social identity can also be found inside the research of Yao et al. (2015), which argues that the social capital is positively related to team learning, which concurrently affects the knowledge sharing and knowledge creation inside the virtual community. This mediating role of team learning thus also needs to be taken into account when looking at knowledge sharing inside virtual communities. Finally, other studies contribute to the conversation by arguing that long term sustainability can be achieved by additionally creating rules of governance and behaviour, by using avatars and reputational rankings inside the system to create a better sense of efficacy and trust. This in turn influence knowledge sharing and other contributions to the community, and meta-media, which have been proven to be a factor of long-term stabilization of social structures in the virtual community, meaning that this positively affects the longevity of the community (Wang & Noe, 2010).
Figure 1: Virtual Community Literary Framework
Looking at the literary framework proposed above, one can see the topics addressed surrounding virtual communities provided by the research to date. Dimaggio, et al. (2001) states that the ultimate social implications of this new technology, e.g. knowledge sharing, depends on economic, legal, and policy decisions that are shaping the Internet as it becomes institutionalized (Dimaggio, et al., 2001). However, Dimaggio et al. (2001) do not provide further research in this matter, Hercheui (2011) furthermore argues that there is one big item which is neglected in current research. The institutional effect inside the virtual community research has been only taken into account as a secondary variable, and thus been mostly
13
neglected. Dimaggio and Powell (1991), and Scott (2001) agree with this point by arguing that because institutional environments and forces should affect online behaviours, research will benefit from understanding more about the mechanisms that are behind these behaviours, and understanding more about sanction mechanisms and legitimacy. It is however not clear what kind of institutional perspective needs to be introduced, since institutions may change or disappear overtime, which is maybe even more likely to happen online since a lot of different institutional forces are brought together by the different users from all over the world (Jeperson, 1991; Dimaggio, et al., 2001).
Considering the above, this research will aid to the call of Dimaggio et al., (2001) whom states that “Sociologists need to study the Internet more actively and, particularly, to synthesize research findings on individual user behaviour with macroscopic analyses of institutional and political-economic factors that constrain that behaviour” (Hercheui, 2011, pp 5). There is however no consensus on how to look at this new media in terms of institutions (Dimaggio et al., 2001). This research will therefore contribute to this line of work by researching isomorphic pressures, and its effect on knowledge sharing in virtual communities. Looking at virtual communities from this angle creates an interesting contribution, since this research can create new insights in the increasing debate on globalization to which virtual communities are argued to play a role in. Before any hypothesis can be made, this research will first discuss isomorphic pressures in the coming part.
§3.3 Isomorphic Pressures.
For the analysis to make sense, first isomorphic pressures need to be explained. Dimaggio and Powell (1983) came with these isomorphic changing the research agenda from original
diversity and variation among organizations, towards a process of institutionalization and homogenization (Lewin & Massini, 2002). The public sector is affected by isomorphic pressures (Frumkin and Galaskiewicz, 2004), including national level sports associations (Slack & Hinnings, 1994), and even ranges into the marriage Literature (Weir, 2009). Currie and Suhomlinova (2006) argues that there is an ongoing convergence because of isomorphic pressures within groups and/or communities, but a divergence between these groups and communities. A recent study argues that institutional pressures are argued to have an effect on the knowledge sharing in general (Wang et al., 2014). One could thus propose that the
knowledge sharing is affected by institutional pressures also in the virtual world. However, this research, as stated before, is hesitant to assume this research to also be applicable to the
14
online context, because there exists a lot of empirical invalidity and untested factors inside this literature.
Isomorphic pressures can be subdivided into three different mechanisms. First, there is the coercive isomorphic pressure. Coercive Isomorphism can be seen as the pressures of organizations that a firm is dependent on, including the influence of cultural expectations (Dimaggio & Powell, 1983). Coercive pressures can then, for example, be seen as a result from political pressures for legitimacy. This legitimacy is then exercised by the external environment and large firms with significant market shares and/or resources (Lewin & Massini, 2002). Coercive isomorphism also explains the homogenous reporting in the
accounting industry (Verbruggen, et al., 2011). Furthermore, Coercive isomorphism is argued to have an influence on the homogenization of the big banks (Deephouse, 1996), the
adoptions of CSR initiatives (Othman et al., 2005), firms also adopt CRM systems mostly to gain legitimacy while supply chain management is influenced by both gaining legitimacy and seeking knowledge (Cheng, 2010; Lai et al., 2006). Most importantly however is that
knowledge management of firms is affected by coercive isomorphism (Magnier-Watanabe & Senoo, 2009). The latter is argued to be mainly a result of coercive isomorphic pressures in terms of a term called technical isomorphism, which argues that, for example, firms adopting the same IT systems will give a technical isomorphic pressure on its users (Bender et al., 2006). Furthermore, Hercheui (2011) researched the effect of coercive institutional pressures on the interaction in virtual communities and found a significant relationship between the two. To test coercive isomorphism, this study will mainly focus on the social pressures from within the community, and including the legitimization internal to the virtual community itself on its own user. Since it is argued that the real life is not that different from the virtual world
(Chandra & Leenders, 2012), and the coercive isomorphism has been found in the virtual interaction inside virtual communities this study hypothesises the following:
H1: Coercive Isomorphic Pressures will have a positive relation with knowledge sharing behaviour inside virtual communities.
Second, mimetic pressures can be seen as a result of the increasing technological and strategic uncertainties. Firms are prone to mimic the strategies and dominant designs of the leaders inside their industry. This means that mimetic pressures lead firms to imitate others (Lewin & Massini, 2002). The above explained isomorphic pressures have been researched quite extensively. The mimetic isomorphic pressures are argued to explain the relative concentrated market share allocation inside the accounting industry, where the medium sized
15
firms are following the leaders (Tuttle & Dilliard, 2007). Mimetic Isomorphism is also shown to influence IT managers choosing different IT systems (Tingling & Parent, 2002). Mimetic isomorphism has not been researched before in relation to knowledge sharing in communities, however one could argue that inside virtual communities this relation could have a much larger effect compared to physical communities. This mainly since contributory tracking systems and rankings are in place in a lot of these virtual communities, which creates the possibility to clearly identify knowledge sharing leaders. Because of this, and since this study acknowledges the possibility that the institutional pressures do have an effect on knowledge sharing inside virtual communities the following hypothesis comes forward:
H2: Mimetic Isomorphic Pressures will have a positive relation with knowledge sharing behaviour inside virtual communities.
For the above effect to make sense, the user should be able to identify knowledge sharing leaders. Hence, this study argues that additionally, the influence of mimetic isomorphic pressures on knowledge sharing is moderated by the fact that the participant perceives there to be leaders. Thus this research hypothesises:
H3: Leadership perception can be seen as a moderator for the relation between Mimetic
Isomorphic Pressures and knowledge sharing behaviour inside virtual communities.
Finally, the normative isomorphic pressures are the result from a need to meet professional and educational standards. This can be seen in the education and industry standards. “Thus once an innovation begins to mature in the form of standards, normative isomorphic pressures explains spread of the standard throughout the population (Lewin & Massini, 2002, pp. 15). One could argue that to share knowledge in certain virtual
environments that certain education or work experience is necessary. Furthermore, one could imagine that, considering the international nature of the virtual community, educational and work experiences have an effect on the way people communicate with each other, and thus share knowledge. Because of this the following hypothesis comes forward:
H4: Normative Isomorphic Pressures will have a positive relation with knowledge sharing behaviour inside virtual communities.
Considering factors that could possibly distort the research results, e.g. our control variables, identified in the literary framework on knowledge sharing and virtual communities in appendix 3 leaves us, including the hypothesis, with the following conceptual framework.
16
Figure 2: Conceptual Framework
4. Method
§4.1 Sample and Data Collection.
To get the data specific for this research I will be conducting a survey study (Appendix 2). Since there are no prior survey studies on isomorphism, especially related to knowledge sharing, this survey will be created. To test the hypothesis, the sample will contain participants who are active inside virtual communities. Furthermore, to be able to test the level of knowledge sharing, only virtual communities will be used, which have contributory tracking devices in the form of rankings. These rankings are especially useful for this research since this ranking consists of the quantity of knowledge shared, but also quality, since the amount of up-votes (stating it is the best answer) increases the ranking by the sharing user. In the research on virtual communities, studies also found that an interactive environment, avatars, and high quality content positively affect knowledge sharing behaviour. These are also factors that need to be found inside these virtual communities to control for these factors. This leaves three very large virtual communities to focus on, namely, StackExchange (or StackOverflow), which is a forum with 5 million users in which mainly programmers come together to exchange knowledge about different type of coding languages, Quora, which is a forum with 100 million users where one can find and share information with each other, and Reddit which is similar to Quora, but has 3.4 million users.
17
The data was collected through solely web based posting and email. The first batch of answers was obtained through the forums themselves, but very little response was given. This lead this study to raffle 10 euro vouchers for amazon through a number of survey exchange sites and Facebook groups (“Respondenten Gezocht” Facebook group, Surveystudent.nl, among others). Finally, the survey has been distributed through a class of Communication and Multimedia design students. The amount of filled in surveys was 285, of which 158 were deleted from the data set, because the participant was not actively involved in the right virtual communities discussed above. Of the 127 filled in surveys left, 38 were furthermore deleted because the ranking they filled in did not match the amount of questions answered/asked. These have to be in line to a certain extent or else it would suggest false information has been entered. This has been done in a simple way. First the questions asked, answered, and the reputation were pooled in four different groups. Concurrently the average is taken of the recoded questions asked and questions answered data. This average will be weighed against the recoded reputation that the participant has given. If this deviates more than 50%, the participant will not be taken into account, since it will be very unlikely these scores match. Through this methodology clear outliers where deleted from the usable data. This left us with a total pool of 89 surveys that are applicable to this survey.
Of the usable data 52 % was male, which suggests that there is more or less an even distribution of male population inside these virtual communities compared to female. Furthermore, the age was predominantly between 18 and 30 (60 %). The participants came predominantly out of the the US (62%), and about 25 % came from northern Europe like Germany, The Netherlands, France, and Scandinavia, among others. This study believes that the pool of participants rightly reflects the population, although one could argue that the average age could be slightly lower inside this data set, since a lot of student participation was present. However, this study believes that the age of the users inside virtual communities is already low, and the prior data gathering, which was only present inside the community itself (so no student survey exchange site) was also mainly between 18 and 30. Furthermore, the 10-euro vouchers that were used to incentivise people to fill in the survey could leave the
possibility that participants acted as if they were active on this forum. In my opinion this has been tackled by using the control mechanism of matching rankings and questions
asked/answered, since only real users of these forums would be able to give the right matching information.
18
§4.2 Measures
§4.2.1 Measure Dependant Variable
Our dependant variable is knowledge sharing in virtual communities. To measure knowledge sharing in a survey (Appendix 2), this research will ask questions surrounding the ranking system that is in place within the virtual communities this research will be focussing on. The ranking systems in place, gives everybody a rank based on the amount, and quality of
questions answered (of which the latter will be stated by the one who asked the question by up-voting the answer as best). This ranking systems gives us the possibility to link different phenomena towards knowledge sharing. First of all, we will ask the participant about their reputation (basically their rank), number of answered questions, and number of questions asked. This will be done to determine how much knowledge the participant is sharing inside the virtual community. By asking it in threefold we also can test whether the participant is giving false information, since the reputation, and the number of answered/asked questions need to match to a certain extend (already explained in the method section). This provides the survey with some control to relieve the survey for some possible bias that could arise.
§4.2.2. Measures Independent Variables §4.2.2.1 Coercive Isomorphic Pressures
To measure Coercive Isomorphism inside the virtual community and relate that to knowledge sharing of the individuals we can look at former studies that researched this
phenomenon to get a better understanding of possible measures. Deephouse (1996) researched whether banks are affected by coercive isomorphic pressures. They checked this by looking at the news/media attention to different problems/issues inside the banks, and concurrently the changes that the banks have made accordingly. Othman et al. (2005) researched whether regulatory authorities, and environmental reputation affect CSR practices. They did this by surveying 117 companies and asking about whether they would conduct a CSR approach when regulative authorities would tell them to, and whether they would implement CSR practices looking at the environmental perspective, like social pressures.
In this study we will, just like the latter, ask questions through a survey to identify
whether coercive isomorphism affects knowledge sharing inside communities (Appendix 2). I believe, however, that it becomes too complex to test for political pressures on the
19
of social legitimacy from within the community. This is similar to the focus of Deephouse (1996) whom mainly looked into social pressures within the perspective of banks, and its concurrent actions to attain legitimacy form this social platform. To be able to do this, first the interdependency between individuals within the community needs to analysed. One could argue that individuals inside these virtual communities are dependent upon each other’s knowledge to grow their own knowledge base. Furthermore, one could argue that in order to attain legitimacy, users need to share knowledge to look good compared to others. One could also argue that the social context inside the virtual community, pressures people to share knowledge, and finally it could be argued that the community does not accept, or shares knowledge back to users who do not share knowledge themselves. To measure this, first asking the participants whether they believe to be in need of others for new knowledge. Then the participant is asked whether the participant feels the need to share knowledge in order to be taken serious/accepted/or legitimized by other participants within the community. We can ask whether the user has ever been instructed by the site, or community itself to actually share knowledge, and whether this affected their knowledge sharing behaviour. Additionally, we can ask the participant whether they feel the need to share knowledge in order to receive answers to their questions. This study believes, by considering all of the above, that the above explained questions will be sufficient to measure the variable Coercive Isomorphism.
§4.2.2.2 Mimetic Isomorphic Pressures
In previous research mimetic isomorphism mainly links certain strategies and decisions to industry leaders and others. This has, for example, been done by Han (1994) which looked which clients used which accounting firms. He identified the clients in leaders, followers, and small firms. By seeing which audit firms the leaders chose, and by comparing this with what the followers and small firms chose, he saw that the followers were tend to use the same auditors as their industry leaders. This research has also been redone in a similar way by Tuttel and Dilliard (2007). Other researchers also linked followers and leaders within a firm to geographic expansion and entry mode decisions by again identifying leaders, followers, and small firms, and looking whether the followers actually did follow the leader into a new region in the same way (Davis & Desai, 2000; Haveman, 1993)
This research will be measuring mimetic pressures affecting knowledge sharing by relating the answers to our questions in this survey (Appendix 2) towards the ranking system that is in place within the virtual communities. First, four groups where identified, namely beginners, intermediates, advanced, and leaders in knowledge sharing. This can be simply
20
done by looking at the reputation, or rank, of the different people within the community. Within these communities a rank of above 10.000 of the community is considered a leader, since their reputation lies in the top 5%. Advanced knowledge sharers are identified by having a reputation of 1000 to 10.000, which could then be argued to be between the top 5% and top 30% of the total active members. The intermediate users, are considered to have a reputation of 100-1000, which corresponds to top 30 % to 60 %. The last group, beginners, are
considered when a user has between 0 and 100 reputation, hence considered to be at the lowest 40 % inside the community. After this it has to be identified whether the knowledge sharing behaviour of the/other leaders affects the knowledge sharing behaviour of the participant. This study, for example, ask whether they believe the rank inside the virtual community to be an explanatory variable for the success of the top ranked community
members, whether they want to attain a higher ranking like the leaders, or would change from ranking with the leaders, after this, this study asks whether the participant perceives the need to share more knowledge now they know how much knowledge the leader shares, and even ask them whether they would decrease their knowledge sharing behaviour when the leaders were not sharing any knowledge. This study believes, by considering all of the above, that the above explained questions will be sufficient to measure the variable Mimetic Isomorphism. §4.2.2.3 Normative Isomorphic Pressures
Normative isomorphic pressures are the result from a need to meet professional standards. This can be seen in the education and industry standards. “Thus once an innovation begins to mature in the form of standards, normative isomorphic pressures explains spread of the standard throughout the population” (Lewin & Massini, 2002, pp. 15). To measure Normative Isomorphism inside the virtual community and relate that to knowledge sharing of the
individuals we can look at former studies that researched this phenomenon to get a better understanding of possible measures. These have mainly measured normative isomorphism by linking the educational system in a country to, for example, the decision to enter new markets (Eden & Miller, 2004). Furthermore, Tuttle & Dillard (2007) argued that normative
isomorphism affects the accounting profession, because of the accounting standards obliged by professional accounting organization (like IFRS e.d.).
This research at it twofold, from the educational standards perspective, and the work experience standards perspective. Firstly, the educational standards can be looked at in comparison to knowledge sharing, to see whether certain level of education affects virtual community users to share knowledge. This will also be done with work experience, to which
21
the amount of experience will be weighed against the amount of knowledge shared which concurrently gives information whether the amount of work experience affects the amount of knowledge shared. This study believes, by considering all of the above, that the above
explained questions will be sufficient to measure the variable Normative Isomorphism. §4.2.3 Control Variables
This study has to control for a number of variables. First, there is the design of the virtual community itself. This is tackled easily by only accepting answers from three forums with identical characteristics that are shown to have an effect on knowledge sharing inside virtual communities, namely, a Contributory Tracking System, High Quality Content, Avatars, and an Interactive Environment (Seraj, 2012; Bagozzi, 2014).
Furthermore, there are some factors from the virtual community, and knowledge sharing research that need to be controlled for. As can be seen in Appendix 3 there are, aside from design of the virtual communities, ten factors that are taken into account, of which four of these are proven to have an effect inside the virtual community. Forum type, nature of presence, gender, and age, are all taken into account to make sure no biased results will be presented. No effects on knowledge sharing were identified by prior research on these factors, but the fact that this study created a new survey, made this study hesitant to not adding these to the control variables. Furthermore, this study controlled for country, since Minbaeva (2007) found a significant relation to language and knowledge sharing. Education and work experience are also taken into account as control variable. Aside from the fact that these are also used to test normative isomorphic pressures, they are also used as control variables since Constat et al. (1994) found a positive relation between these variables and knowledge sharing. Finally, Self-efficacy, Enjoyment of helping others, and Exchange Ideology are taken into account. Exchange ideology was found to have a positive relation with knowledge sharing by Lin (2007). Self-efficacy and enjoyment of helping others were found to have a positive relation with knowledge sharing inside virtual communities (Hercheui, 2011; Hsu et al., 2007). It finally has to be noted that trust has not been taken into account as a control variable. This has mainly been done because of the insignificant results found by Chiu et al. (2007), lack of clear empirical results by Bartol and Srivastava (2002), and the increased size of the survey when added, which might discourage participants in finishing the survey.
22
5. Result
§5.1 Recoded Variables.
This research studied whether isomorphic pressures affect knowledge sharing inside professional virtual communities. To be able to test this first eight factors from the survey data that was collected needs to be recoded. These recoded factors can be found inside the following table.
Justification needs to be provided for the recodings of the Questions asked, Questions
answered, and Reputation. These are set up in this way because these 4 subgroups are related, hence, when someone has asked 0-10 questions and answered 0-10 question his or her
reputation can only reside between 0-100. These three subgroups furthermore represent a ranking inside the community hence, beginners, intermediates, advanced, and leaders in knowledge sharing on these forums. Furthermore, the reason why other has been set at empty, was aside from the fact that there were very little responses stating “other”, most of these where left blanc, or did not fit in one of the coded factors used in this research.
§5.2 Reliability and validity.
As stated before, the survey was not from an already proven data-base that has already been proven to be reliable and valid. Thus to be sure that the analysis procedures yield consistent findings, the questions will need to be tested in terms of reliability and validity. First, the reliability testing is done by looking at them per variable. There were four tests done. The first one concerns Q 4-6, which are all about knowledge sharing. The Cronbach’s Alpha of these questions together is 0.857, which cannot be increased by discarding one of the questions out of the questionnaire, thus one can state that this research uses reliable measures to test
knowledge sharing, since it is above 0.7. Then to determine whether the participant perceives there to be a leader inside the community were reliably measured Q7-8 were tested. This resulted in a Cronbach’s Alpha of 0.757, no deletion of questions could increase this, and
Table 1: Dummy Variables
Dummy variable:
Factor 1 2 3 4 5 6 7 8 9 Empty
Nature of pressence Professional Hobby Fun Other Gender Male Female
Age <14 14-17 18-24 25-34 35-44 45-54 >54 Questions Asked 0-10 11-20 21-50 >50
Questions Answered0-10 11-20 21-50 >50
Education Primary School High School Bachelors Masters/PHD Other Work Experience Intern Junior Senior Partner Other Reputation 0-100 101-1000 1001-10000 >10000
Country PT US GER CA NL BE IT FR DE
23
since the Cronbach’s Alpha is above 0.7 this variable is reliably measured. The first excluded variable due to reliability issues can be found inside the Questions surrounding mimetic isomorphism concerning Q9-12. Here, the initial reliability was 0.797 and can be increased by deleting question 11 to 0.813. Since the Cronbach’s Alpha is above 0.7 question 9, 10, and 12 reliably measure mimetic isomorphism. Furthermore, the reliability test of the questions (Q15,16,18,19) surrounding normative isomorphism results in a Cronbach’s Alpha of 0.838, no further improvement between these questions. Finally, the reliability test of coercive isomorphism (Q20-25) resulted in a Cronbach’s Alpha of 0.707. This can be improvement by deleting Question 20, increasing the Cronbach’s Alpha to 0,743, which is above 0.7, making Q21-25 reliable.
Then to check the validity of the created survey, hence whether the questions measure what we want them to measure, this study will conduct a factor analysis. A principal factoring analysis (PAF) was conducted on the scales, which was done by analysing Q 4-6, Q9-10, Q12, Q15-16, Q18-19, and Q21-25 of which the pattern matrix, including the explained variance can be seen in appendix 4. The Kaiser-Meyer-Olkin Measure verified the sampling adequacy for the analysis, KMO = 0.806, with a Barlett’s test significance of 0.000.
Furthermore, looking at the correlation matrix, no values above 0.8 where identified, so no variables were deleted due to singularity. Additionally, the fact that the determinant of the values was above 0.0001 makes the PAF analysis appropriate in this case. This lead to the following table:
Table 2: Validity Test
Coercive Normative Mimetic
Mim1/Q9 0.254 0.116 0.637 Mim2/Q10 -0.053 -0.031 0.931 Mim4Q12 0.216 -0.122 0.680 Norm1/Q15 0.048 0.862 0.022 Norm2/Q16 -0.210 0.837 0.094 Norm3/Q18 0.186 0.850 -0.289 Norm4/Q19 -0.024 0.741 0.158 Coerc2/Q21 0.570 0.030 0.205 Coerc3/Q22 0.078 0.466 0.266 Coerc4/Q23 0.646 0.034 0.068 Coerc5/Q24 0.900 -0.159 -0.012 Coerc6/Q25 0.813 0.133 -0.059 EigenValue 4.713 1.914 1.135 Variance 39.27% 15.95% 9.46%
24
As can be seen the Eigenvalues of these three variables are all above 1, meaning three
variables are significantly identified within the underlying factors. As furthermore can be seen is that the individual factor scores are, almost, all above 0.4 for the right constructs, meaning the right questions measure the right variables. There is however one error in these factors, Coerc3, or Question 22, can be seen to be pooled with the wrong variable. The statement in this matter “I expect others to share, if I do so”, is therefore argued to explain the variance for normative isomorphism. Looking at the question this study does not believe the question to be representable for normative isomorphism, resulting in deletion of Question 22 from the
dataset. An additional reliability analysis was done to check whether deleting this construct would cause reliability issues. The additional test pointed out that the Cronbach’s Alpha is 0.761, and impossible to further improve by deleting more constructs.Although the explained variance is not very high for these constructs, which is mainly the case for normative and mimetic, the eigenvalues and underlying constructs prove to be valid, which is why this study argues the variables in this research are reliably and validly measured with the right
underlying factors.
In table 3 one can see the Cronbach’s Alphas per construct on the diagonal. The table furthermore present correlations between the different constructs. This shows that knowledge sharing only has the tendency of a positive relation with two of the proposed control
variables, namely enjoyment of helping, and education. It is also noteworthy to state that the three isomorphic pressures all have the tendency for, or a high, positive relation towards each other. Furthermore, it seems that only work experience, gender, and age, do not have a tendency to, or high positive relation to the isomorphic pressures, meaning for the sake of the analysis. Table 3: Means, Standard Deviations, Correlations. M SD 1 2 3 4 5 6 7 8 9 10 11 12 1. Knowledge Sharing 1.66 0.74 (0.86) 2. Mimetic Iso 2.91 1.00 0.05 (0.81) 3. Normative Iso 3.37 0.87 0.12 .41** (0.84) 4. Coercive Iso 2.83 0.84 0.08 .54** .30** (0.74) 5. Education 2.60 0.79 .35** .29** 0.18 -0.02 -6. Work Experience 2.26 0.91 0.18 0.04 0.12 -0.05 0.21 -7. Gender 1.49 0.50 -0.16 -0.13 -0.04 -0.15 -0.12 -0.04 -8. Age 4.17 1.34 -0.02 -0.17 -.22* -.23* 0.02 0.15 0.11 -9. PerOfLeader 3.41 1.00 0.19 .42** .39** 0.09 .27* 0.12 -0.17 -0.18 (0.76) 10. Self efficacy 3.54 1.14 -0.01 .35** 57** 0.10 0.13 0.05 -0.03 -0.15 .37** -11. Enjoyment of helping 3.70 1.23 .248* .26* .34** 0.11 0.17 -0.01 -0.04 -.22* 38** .49** -12. Exchange Ideology 3.56 1.11 0.18 .48** .55** 0.20 .26* 0.05 -0.01 -0.13 .34** .60** .50**
-25
§5.2 Testing the Hypothesis.
To test hypothesis 1, hence whether Coercive Isomorphic Pressures will have a positive relation to knowledge sharing behaviour inside virtual communities, this study performed a hierarchical multiple regression analysis, after controlling for Education, Work Experience, Gender, Age, Self-Efficacy, Enjoyment of Helping, Exchange Ideology, Country, Forum type, and Nature of Presence. In the first step of the hierarchical regression model, presented in table 4 below, only the control variables were tested. This model was statistically significant F (10, 46) = 2.23; p < .05 and explained 18 % of the variance. After the entry of the variable coercive Isomorphism at step two the total variance explained by the model was 23.2 % F (11,45) = 2.542; p < 0.05. The introduction of coercive isomorphism significantly explained an additional 5.2% of the variance in knowledge sharing, after controlling for the ten before mentioned variables (R2 Change = 0.052; F (1, 45) = 4.11; p<0.05). in the first model two of ten variables were statistically significant, in the second model of the eleven predictors, three were statistically significant, with Enjoyment of Helping having the highest Beta value (β = .48, p < 0.01) then Education (β = .40, p < 0.01) and finally Coercive Isomorphism (β = .28, p < 0.05). Since Coercive Isomorphism significantly explained part of the variance of
knowledge sharing, it is argued that coercive isomorphism is positively related to knowledge sharing, hence the first hypothesis is supported.
Table 4: Hierarchical regression model Coercive Isomorphism R Adjusted R Square R Square Change B SE β t Step 1: 0.572 0.180* Forum -0.05 0.10 -0.08 -0.51 Nature of Pressence -0.08 0.15 -0.08 -0.52 Education 0.29 0.13 0.34 2.33 WorkExp 0.13 0.11 0.16 1.14 Gender 0.08 0.21 0.06 0.39 Age -0.09 0.08 -0.16 -1.12 Self efficacy -0.18 0.11 -0.28 -1.67 Enjoyment 0.26 0.10 0.43 2.60 Exchange Ideology -0.06 0.11 -0.08 -0.49 Country -0.03 0.07 -0.07 -0.39 Step 2: 0.619 0.232* 0.052 Forum -0.01 0.10 -0.02 -0.14 Nature of Pressence -0.02 0.15 -0.02 -0.16 Education 0.34 0.12 0.40 2.77 WorkExp 0.12 0.11 0.16 1.15 Gender 0.14 0.21 0.09 0.67 Age -0.09 0.08 -0.15 -1.09 Self efficacy -0.20 0.11 -0.30 -1.85 Enjoyment 0.29 0.10 0.48 2.95 Exchange Ideology -0.11 0.11 -0.16 -0.98 Country -0.04 0.07 -0.09 -0.53 Coercive Isomorphism -0.25 0.12 0.28 2.03
26
To test hypothesis 2 and 3, three different analysis need to be done. First, to test the whether Mimetic Isomorphic Pressures will have a positive relation with knowledge sharing behaviour inside virtual communities. This study performed a hierarchical multiple regression analysis, after controlling for Education, Work Experience, Gender, Age, Self-Efficacy, Enjoyment of Helping, Exchange Ideology, Country, Forum type, Nature of Presence, and Leader Perception. In the first step of the hierarchical regression model, presented in table 5 below, only the control variables were tested. This model was statistically significant F (11, 45) = 2.02; p < .05 and explained 17 % of the variance. After the entry of the variable Mimetic Isomorphism at step two the total variance explained by the model was 15 % F (12,44) = 1.811; p > 0.05. The introduction of Mimetic Isomorphism lowered the explained variance in knowledge sharing by 2%, after controlling for the eleven before mentioned variables (R2 Change = -0.019; F (1, 44) = 1.811; p>0.05). In the first model two of eleven variables were statistically significant, in the second model of the eleven predictors, again two were statistically significant, with enjoyment of helping having the highest Beta value (β = .43, p < 0.01), right before education (β = .33, p < 0.01). Since mimetic isomorphism did not seem to significantly explain the variance of knowledge sharing, it is argued that mimetic isomorphism is not related to knowledge sharing, hence the second hypothesis is unsupported.
Hypothesis 3 was tested by using the statistical software “Process”. The regression coefficient for XM is b3=-0.11 and is not statistically different from zero, t(83)=-1.61,
Table 5: Hierarchical regression model Mimetic Isomorphism R Adjusted R Square R Square Change B SE β t Step 1: 0.575 0.167* Forum -0.06 0.10 -0.10 -0.61 Nature of Pressence -0.08 0.15 -0.08 -0.54 LeaderToT 0.06 0.12 0.08 0.51 Education 0.28 0.13 0.33 2.13 WorkExp 0.13 0.11 0.16 1.13 Gender 0.12 0.22 0.08 0.52 Age -0.09 0.08 -0.15 -1.02 Self efficacy -0.20 0.12 -0.31 -1.73 Enjoyment 0.26 0.10 0.43 2.60 Exchange Ideology -0.05 0.11 -0.07 -0.45 Country -0.02 0.07 -0.06 -0.34 Step 2: 0.575 0.148 -0.019 Forum -0.06 0.10 -0.10 -0.59 Nature of Pressence -0.08 0.16 -0.08 -0.52 LeaderToT 0.06 0.14 0.08 0.41 Education 0.28 0.13 0.33 2.11 WorkExp 0.13 0.11 0.16 1.12 Gender 0.13 0.23 0.08 0.51 Age -0.09 0.09 -0.15 -1.01 Self efficacy -0.20 0.12 -0.31 -1.70 Enjoyment 0.27 0.10 0.43 2.53 Exchange Ideology -0.05 0.13 -0.07 -0.41 Country -0.02 0.07 -0.06 -0.34 Mimetic Isomorphism 0.00 0.13 0.01 0.03
27
p>0.05. Thus, the effect of mimetic isomorphism on knowledge sharing is not present, including the moderation effect of leadership perception on this relation. Since thus mimetic isomorphic pressures does not significantly explain knowledge sharing, and there was no interaction effect on the fact given that a leader should be perceived, hypothesis 3 is unsupported.
To test hypothesis 4, hence whether Normative Isomorphic Pressures will have an effect on knowledge sharing behaviour inside virtual communities, this study performed a hierarchical multiple regression analysis, after controlling for Education, Work Experience, Gender, Age, Self-Efficacy, Enjoyment of Helping, Exchange Ideology, Country, Forum type, and Nature of Presence. In the first step of the hierarchical regression model, presented in table 4 below, only the control variables were tested. This model was statistically significant F (10, 46) = 2.23; p < .05 and explained 18 % of the variance. After the entry of the variable Normative Isomorphism at step two the total variance explained by the model was 16 % F (11,45) = 1.991; p > 0.05. The introduction of Normative Isomorphism insignificantly lowered the explained variance in knowledge sharing with 1.7%, after controlling for the ten before mentioned variables (R2 Change = -0.017; F (1, 45) = 1.911; p>0.05. In the first model two of ten variables were statistically significant, in the second model of the eleven
predictors, still only two were statistically significant, with enjoyment of helping having the highest Beta value (β = .43, p < 0.01) right before education (β = .35, p < 0.01). Since Normative Isomorphism did not seem to significantly explain the variance of knowledge sharing, it is argued that Normative Isomorphism is not related to knowledge sharing, hence the second hypothesis is unsupported.
Table 7: Moderation effect of Leader Perception
Coefficient SE t p
Leadership Perception Unstandardized Boots effect
Boot SE Boot LLCI Boot UCLI
Intercept i1 0.33 0.64 0.5 0.61 conditional effect on all levels of leadership perception Knowledge Sharing (X) b1 0.35 0.18 2.23 0.02 Low 0.0989 0.1165 -0.1329 0.3307 Leader Perception (M) b2 0.41 0.25 1.41 0.16 Medium -0.0061 0.088 -0.1812 0.169 Knowledge Sharing x Leader Perception b3 -0.11 0.065 -1.61 0.11 High -0.111 0.1018 -0.3136 0.0915 R Square 0.065 F (3, 83) = 1.93 p>0.05
28
Finally, there were two additional tests performed. First, an additional One-Way Anova analysis is done to determine whether mimetic isomorphic pressures differ across the four different levels of knowledge sharing, hence beginners, intermediate, advanced and leaders in knowledge sharing. In this analysis the means of the scores on mimetic isomorphism were tested across these four groups. The results suggest there to be no
significant differences in the means of the four different groups: F (3, 83) = 1.512, p > 0.05. Furthermore, the Tukey post hoc tests revealed that no significant difference in perceived mimetic isomorphic pressures was found across beginners, and leaders (p = 0.136).
The second additional analysis done by this research is a multiple regression test to see whether work experience and/or education are positively related to knowledge sharing. To test this a hierarchical multiple regression analysis was performed, after controlling for Education,
Table 6: Hierarchical regression model Normative Isomorphism R Adjusted R Square R Square Change B SE β t Step 1: 0.572 0.180* Forum -0.05 0.10 -0.08 -0.51 Nature of Pressence -0.08 0.15 -0.08 -0.52 Education 0.29 0.13 0.34 2.33 WorkExp 0.13 0.11 0.16 1.14 Gender 0.08 0.21 0.06 0.39 Age -0.09 0.08 -0.16 -1.12 Self efficacy -0.18 0.11 -0.28 -1.67 Enjoyment 0.26 0.10 0.43 2.60 Exchange Ideology -0.06 0.11 -0.08 -0.49 Country -0.03 0.07 -0.07 -0.39 Step 2: 0.572 0.163* -0.017 Forum -0.05 0.10 -0.08 -0.52 Nature of Pressence -0.09 0.16 -0.08 -0.54 Education 0.29 0.13 0.35 2.31 WorkExp 0.13 0.11 0.16 1.14 Gender 0.09 0.21 0.06 0.41 Age -0.10 0.09 -0.17 -1.09 Self efficacy -0.18 0.12 -0.26 -1.51 Enjoyment 0.26 0.10 0.43 2.56 Exchange Ideology -0.05 0.12 -0.08 -0.46 Country -0.03 0.07 -0.07 -0.40 Normative Isomorphism -0.03 0.14 -0.31 -0.19
