!
!
Universiteit van Amsterdam
!
Master Thesis
The effect of customer-company identification on online sharing
Daan Hermans
10317422
August 3th, 2014
Supervisor: Dr. K.A. Venetis
!
!
!
!
!
!
!
!
Abstract
Earlier research showed that high identified customers promote an organisation more active offline than low identified customers. However little was known concerning the online sharing behaviour of these high identified customers. This research takes a quantitative approach to answer the question what the relationship is between customer-company identification and online sharing. The results from this study indicate that people share information online for personal gain, what is called self-enhancement. In contrast with the hypothesis, customer-company identification does not have a significant effect on sharing organisational information online because of self-enhancement. This means that if a person wants to share an interesting text for self-enhancement they share it regardless if they are high identified with the organisation or not. Furthermore the outcome of this study shows that people don't share organisational information online because of helping an organisation. This will only happen when a person is extremely high identified with the organisation. By proving this, the research shows for the first time that customer-company identification has a moderation effect on online information sharing for altruistic reason (towards the organisation). The present research contributes to the academic debate concerning the effects of customer-company identification on online information sharing. Furthermore, for marketing professionals this research will shed light on the online sharing behaviour of their most identified customers and helps them by designing and implementing successful viral marketing campaigns.
!
!
!
!
!
!
!
Table of content
!
1.0. Introduction 5 2.0. Research Question 7 3.0. Literature Review 7 3.1. Customer-company identification 7 3.1.1. Effects of identification 93.1.2. Reasons for identification 10
3.2. Sharing information 11
3.2.1. Differences in online and offline sharing 12
3.2.2. Reasons for sharing information online 13
3.2.2.1. Self-Enhancement 13
3.2.2.2. Helping an organisation or cause 14
3.2.2.3. Helping other people 15
3.2.2.4. Self-therapy 15 4.0. Conceptual model 15 4.1. Self-Enhancement 16 4.2. Helping an organisation 16 5.0. Methods 18 5.1. Sample 19 5.2. Pre-tests 19 5.2.1. Pre-test 1 19 5.2.2. Pre-test 2 20 5.2.3. Pre-test 3 20 5.3. Variables 20 6.0. Results 22
6.1. Reliability of the scales 23
6.2. Manipulation check 24
6.2.1. Interestingness texts - subject of text 24 6.2.2. Interestingness texts - adding supplement 25
6.2.3. Customer-company identification 25
6.2.4. Usefulness for others 26
6.2.5. Control variables - Altruism and Awareness of digital brand 26
6.3. Bivariate Pearson correlation matrix 27
6.4. Hypotheses testing 27
6.4.1. Interestingness (H1) 27
6.4.2. Helping the organisation (H3) 28
6.5. Moderation analysis 29
6.5.1. Interestingness (H2) 29
6.5.2. Helping the organisation (H4 and H5) 30
7.0. Conclusion and Discussion 32
8.0. Limitations and further research 34
9.0. Bibliography 35
10.0. Attachement 40
10.1. Questions survey 40
10.2. Texts 42
10.2.1. Text 1: Without supplement 42
10.2.2. Text 2: With supplement 42
10.2.3 Text 3: Without supplement 43
10.2.4. Text 4: With supplement 43
10.3. Moderation effect - Helping the organisation 44
!
!
!
!
!
!
!
1.0. Introduction
In 2006 George Wright became the marketing director of Blentec. With a small budget of a couple hundred dollars he introduced the
“
Will it Blend?”
campaign. This campaign existed of short films on YouTube, whereby the CEO of Blentec tried to blend the most extreme products, like IPhones or glass marbles. The short films became a hit and gained more than 300 million views. The success of the campaign resulted in an increase of 700 percent in sales in the two years following (Berger, 2013). In 2013 Ditzo, a Dutch health insurance firm, activated a film online which pledged to give one euro for every view to a hospital which is specialised in the treatment of cancer, and two euros for every time the film got shared. They set a maximum of one million euro, hoping to reach this amount within 30 days. During the first three days more than 600.000 people saw and shared the video and the maximum amount of one million euro was reached. Although Ditzo gave one million euro to the hospital, this was far less than the budget which would have been used for a conventional marketing campaign on television. Both marketing techniques described in the examples above can be called viral marketing.Viral marketing can be defined as a marketing message which is created by an organisation and encourages people to forward the message online to their contacts (Phelps, 2004). So instead of having information flowing from the organisation towards the consumer, the information is spread by the consumers themselves. The distinctive advantage of viral marketing is that it can reach a large quantity of customers in a short period of time with low costs (Lans e.a., 2009). Viral marketing is further enabled by the emergence of new connected technologies like social media and smartphones which make information-sharing more easily accessible for users (Belk, 2013; Sun e.a., 2006). Recently, several well-known firms including Microsoft, Sony, Phillips, Ford and Procter and Gamble designed and implemented viral marketing campaigns (De Bruyn and Lilien, 2005).
The key success factor of a viral marketing campaign is the effectiveness of the unsolicited, electronic referrals. These referrals should create awareness, trigger interest, and generate sales or product adoption (De Bruyn and Lilien, 2005). These referrals can be send intentionally and unintentionally. An example of unintentional referrals are the ones
created automatically by the email-service Hotmail: each message sent using the service automatically includes the sentence “Get Your Private, Free E-mail at http:// www.hotmail.com” in their signatures. Intentional referrals are send by the consumers themselves for a variety of personal reasons (De Bruyn and Lilien, 2005). The four main motivational factors behind information-sharing over the internet are of self-enhancement (Labrecque e.a., 2010; McLure-Wasko and Faraj, 2000; Zhao, 2005; Schau and Gilly, 2003), of self-therapy (Buechel and Berger, 2012; Belk, 2013; Forest and Wood, 2012) and of altruistic reasons, whereby information is shared for helping other people (e.g. Belk, 2013; Ho and Dempsey, 2009) or an organisation or a cause they support (Brett, 2013; Hennig-Thurau e.a., 2004).
From the company perspective, marketing tools, including viral campaigns are crucial in connecting with their customers. The consequences of a strong relationship between a customer and a company have also been wildly studied (e.g. Bhattacharya and Sen, 2003; Homburg, 2009). Bhattacharya and Sen (2003) showed that customers can really identify themselves with an organisation. This is confirmed by later studies (e.g. Brown e.a., 2005; Einwiller, 2006; Ahearne e.a., 2005) which found that customers with a high customer-company identification purchase more products and promote the brand more to their peers than customers with lower levels of identification.
Consequentially, within this research there is assumed that involving high-identified customers in spreading the message can be vital to the success of viral campaigns. This assumption is further supported by several studies which found that high-identified customers share information in offline environments (e.g. Bhattacharya and Sen, 2003; Brown e.a., 2005; Einwiller, 2006; Ahearne e.a., 2005). At the same time, important differences between online and offline information sharing processes have been emphasised by the literature (e.g. Belk, 2013; Sun e.a. 2006; Zhao, 2005) and cast doubts over the transferability of insights into offline impact to predominantly online viral campaigns. In conclusion, evidence is lacking on the online information-sharing activities of high-identified customers.
2.0. Research Question
The present study will focus on the effects of customer-company identification on sharing information of an organisation online. The research question of the present thesis can be summarised as follows: “What is the relationship between customer-company identification and sharing organisational information online?”. This research will be extended by analysing the effect of customer-company identification on specific reasons for sharing information online.
The present research will contribute to the academic debate concerning the effects of customer-company identification on online information sharing. Furthermore, for marketing professionals this research will shed light on the online sharing behaviour of their most identified customers. Finally, the thesis will give marketing professionals additional insights in the determinants of the success of viral marketing campaigns, and how the high identified customer-base can be involved and leveraged with these campaigns.
3.0. Literature Review
3.1. Customer-company identification
Customer-company identification can be defined as the amount of overlap between a customer’s own social identity and the identity of an organisation (Bargozzi and Bergami, 2000). Previous research has showed that this link exists in formal relationships such as employee-employer relationships or alumni-university relationships (Dutton e.a., 1994; Mael and Ashforth, 1992; Ashforth and Mael, 1989). Bhattacharya and Sen (2003) later proved that this link can also exist with non-formal relationships, for instance the relationship between the customer and the company.
Customer-company identification has been studied within the framework of social identity approach comprising both social identity theory and self-categorisation theory (Bhattacharya and Sen, 2003; Homburg, 2009). The self-categorisation theory is the more outward-oriented approach of the two. According to this framework, individuals categorise others through the use of social categories (Homburg, 2009; Bergami and Bargozzi, 2000; Turner, 1985; Mael and Asforth, 1989). Examples of these practices are seen when
subjects categorise other people by for example their nationality, occupation, age or other social categories. According to several researchers (Homburg, 2009; Mael and Asforth, 1989) self-categorisation is a tool mainly used for simplification of mental processes.
Social identity theory is the inward-looking equivalent of self-categorisation. However, instead of defining others along social categories, in this case the subject categorises him-or herself. Similarly to self-categorisation, characteristics such as religion, age or social status are often at the basis of the construction of the identity (Homburg, 2009; Mael and Ashforth, 1989; Tajfel and Turner, 1985; Brewer, 1991; Turner e.a., 1994). Ultimately, the subject’s social identities are defined by this process of categorising oneself. Social identities can be therefore seen as the categorisations of the self in more inclusive social units. This cumulative social identity defines the individual by the shared similarities with members of certain social categories in contrast with members of other groups (Turner, 1994; Brewer, 1991). Therefore, ultimately, the individual actually constructs a “definition of themselves” (Brewer, 1991) and answers the question “Who am I?” (Mael and Ashforth, 1989). A social identity is not limited to inclusion into one group at a time, but can simultaneously consist of multiple different categorisations. A person can for example at the same time think “I am a man” as “I am Dutch” (Homburg, 2009; Mael and Ashforth, 1989).
People classify themselves in order to increase their own self-esteem (Homburg, 2009; Mael and Ashforth, 1989; Bergami and Bargozzi, 2000; Hogg and Turner, 1985). Conversely, Mael and Ashforth (1989) theorise that people continuously compare their own social identity with others, and therefore continuously participate in the failures and successes of their group. An everyday example of this behaviour is the way in which fans identify themselves with their favourite football clubs, the successes and failures of this football club reflect on themselves and are perceived as their own.
As discussed above, social identities are made on the basis of different social characteristics. Mael and Ashforth (1989) extended these characteristics by making a link between the social identity theory and organisational socialisation. Therefore, their theory posits that the organisation where a person works will become part of their social identity and therefore become a characteristic by which a person measures his or her own social identity.
This is confirmed by Dutton e.a. (1994) who shows that the image of an organisation shapes the social identity of its members and employees. They also show the importance of identification by demonstrating that the members or employees determine the attractiveness of the organisation by their identification. Bhattacharya and Sen (2003) showed later the possibility of non-formal links, such as the identification between a consumer and an organisation. They were the first to demonstrate that the social identity of a person can also be fulfilled by an organisation when there is no formal relationship. This connection is at the basis of customer-company identification.
Individual’s reasons for identifying themselves with a company is within those outlined above in the general framework of social identity theory. Simplification is one of these motivations. By identifying themselves and others with an organisation, customers clarify their own social identity and the social identity of their peers. Customer-company identification also happens for other reasons than simplification. For instance, a subject’s personal image has been found to increase when they link themselves to a company with a positive appraisal (Homburg, 2009). This is in line with the findings of Scott and Lane (2000) which state that the high level of attractiveness of an organisation with which a customer identifies him- or herself with, is beneficial for the customers self-esteem. This view is shared by Bhattacharya and Sen (2003) who found that people try to satisfy their self-enhancement need, and thereby their own self-esteem, by linking their identity with organisations which have a prestigious identity. As a result of this link, individuals then reflect themselves with the “glory” of the organisation. Thereby it is important for the individuals that peers also find the organisation very well regarded (Bergami and Bargozzi, 2000).
3.1.1. Effects of identification
The more people identify with a group, the more motivated they are reach the group goals (Scott and Lane, 2000; Brewer and Gardner, 1986; Dutton e.a., 1994). They also become more helpful and supportive towards the group and its members (Bhattacharya e.a., 1995; Dutton e.a., 1994; Mael, 1988; Turner, 1982). Besides the fact that someone becomes identified with a group, individuals tend to accentuate the similarities among members of the
group, and start to emphasise the differences between people outside of the group (Hogg and Abrams, 1988).
The effects described above occurs similarly in consumer groups.. According to Ahearne e.a. (2005), “customers who identify more strongly with a company tend to purchase more and recommend both the company and its products more often”. This is because the moment a customer perceives that the characteristics that define the company can be used for their own social identity, they start to feel more attracted towards the company. In a certain way, their newly formed link to the company becomes a tool for expressing their social identity. This is in line with the studies of Brown e.a. (2005) and Homburg (2009) who both state that customers who are more identified are more committed, and are therefore more willing to purchase the products of the organisations. Similarly, Bhattacharya e.a. (1995) ‘s case study on museum membership showed that that members of a museum with a higher identification are more active in their membership and give higher donations, than members who are less identified with the museum.
Customer identification can be leveraged to increase the reach of a company’s message due to the fact that customers with higher identification perform more positive word-of-mouth (e.g. Brown e.a, 2005; Bhattacharya and Sen, 2003; Ahearne ea, 2005). According to Brown e.a. (2005), customer-company identification indirectly positively influences positive word-of-mouth, while Bhattacharya and Sen (2003) claim that high identified customers ‘enthusiastically promote the company and its products towards others’. Einwiller (2006) and Ahluwalia e.a. (2000) also show that customers with a high identification or high commitment will stay more positive about an organisation and will promote this organisation more to peers, when shown moderately negative content than people with low commitment or identification. 3.1.2. Reasons for identification
The amount of identification people have with an organisation depends on three different factors (Bhattacharya and Sen, 2003). The first factor is the similarity between the organisation
’
s identity and the identity of the consumer. Consumers find it easier to focus on, to process and retrieve an image which is similar to their own image (Markus and Wurf, 1987). Therefore it is easierfor customers
toidentify
with a company that is similar totheir
ownsocial identity. Furthermore, individuals have been found to have a stronger feeling of authenticity when they identify themselves with an image which is already similar to their own image (Pratt, 1998).
The second factor is the matter of distinctiveness the organisation possesses (Bhattacharya and Sen, 2003; Brewer, 1991). This seems contradicting to the first factor, however Brewer (1991) shows that an organisation should be somewhat distinctive to respond to people’s need to be unique and individual. However, when an organisation becomes too distinct customers will stop identifying with it because it ceases to be sufficiently similar to their own self-image. Brewer (1991) has defined the optimal distinctiveness level that an organisation can possess for optimising the attractiveness for identification.
The third factor which influence the extent to which customers will identify with an organisation, is the prestige of the organisation (Bhattacharya and Sen, 2003). Like mentioned before, customers identify themselves with an organisation for increasing their own appraisal and self-esteem (Homburg, 2009; Bhattacharya and Sen, 2003; Scott and Lane, 2000). Customers find it easier to increase their own appraisal and self-esteem when the organisation has a more prestigious identity (Ashfort and Mael, 1989; Dutton e.a., 1994). Therefore the level of prestige determines the attractiveness of the organisation for identification.
Perceptions of the similarity, distinctiveness and prestige of a company are affected by a series of other factors. For instance companies which have high employee identification have been found to have a higher percentage of high-identified customers than companies with low employee identification (Homburg, 2009). Another important factor is that an organisation clearly expresses their identification in an attractive manner (Bhattacharya and Sen, 2003). 3.2. Sharing information
Sharing information is intrinsic to human nature (Belk, 2010, 2013; Masum and Tovey, 2011). People instinctively tell their acquaintances about their experiences, be it holidays or a dining out. According to Benkler (2004) sharing can be seen as a “nonreciprocal pro-social behaviour”. The definition of Belk (2007) is more extended and states sharing as “the act and process of distributing what is ours to others for their use and/or the act and process of receiving or taking something from others for our use”. The internet and other digital devices
enable people to share more easily information to a broader audience (Belk, 2013; Born, 2011). The increasing adoption of connected technologies has resulted in higher than ever amounts of information being shared on the internet. Sharing and co-creating user-generated content are at the origin of social networks, blogs and services based on customer reviews such as yelp.com to Wikipedia (Belk, 2013; Labrecque e.a., 2012; Schau and Gilly, 2003). Open innovation and open source software, where companies reveal all building templates from their latest products (e.g. Ultimaker, 2014), are another aspect of this trend towards increased sharing and cooperation enabled by connectivity. Based on the recent trends towards growth in these services, there is assumed that sharing in online networks will be a persistent trend in the near future (Burnham, 2013).
3.2.1. Differences in online and offline sharing
Like discussed earlier (3.1.1. Effects of identification) several researchers (e.g. Brown e.a., 2005; Bhattacharya and Sen, 2003; Ahearne e.a., 2005) claim that customers who are high-identified share more positive information concerning the organisation than customers who are low-identified. However, the cited studies all tested information-sharing behaviours in offline environments, which are fundamentally different from online sharing behaviour patterns. Offline, face-to-face communication creates a barrier for easily sharing personal information (Ridley, 2012). The facelessness on the internet creates a “disinhibition effect” (Belk, 2013; Sun e.a., 2006; Zhao, 2005; Ridley, 2012). Due to this effect, people free-up towards self-disclosure and share more information of themselves in an online environment (Belk; 2013). Ridley (2012) claims that this is happening due to a natural reason and states that other primates and monkeys also avoid eye-contact due to the fact that they see this as a threat, similarly to the “threat” perceived by humans in face-to-face communication.
A further distinction between online and face-to-face information-sharing is in the extension of the disinhibition effect. As people online are faceless, inequality boundaries like race, income, gender are faded, which in turn increases equality in social interaction (Labrecque e.a., 2011; Eleanor, 1997). However due to the fact that more and more internet services are linked to real-life identities and include visual representations of the user (Belk, 2013; Bolter, 1996), this effect is lately less present than in the early days of the internet.
3.2.2. Reasons for sharing information online
When looked at different studies (e.g. Brett, 2011; Labrecque, 2007; Schau and Gilly, 2003; Ho and Dempsey, 2009) it is clear that people share information online for four different reasons. The reasons are for improving their own image, this is called self-enhancement, for helping an organisation or cause, for helping other individuals or for self-therapy.
3.2.2.1. Self-Enhancement
Self-enhancement can be seen as the fact that someone seeks experiences for improving their self-concept, by for example drawing attention to one’s skills or talents (Baumeister e.a., 1998). Sharing information online is one highly used channel for such self-exposure. An example of this is the use of professional information or educational background in online profiles or interaction in dedicated social networks (Labrecque e.a., 2010).
In line with sharing information for drawing attention towards the positive skills or talents, several researchers (e.g. Labrecque e.a., 2012; Zhao, 2005; Schau and Gilly, 2003) claim that people share information online for managing their personal brand online. This is done in the same way normal marketeers manage a corporate brand. Similarly to the marketing of products and services, the goal of this personal brand is to project the desired identity towards others. Controlling the type and amount of information shared online about the user is a key tool towards creating and managing this identity. An example of this control can be found in the over-representation of photos showing the social life of social network users (Labrecque e.a., 2012). This is in line with the study of Madden and Smith (2010) who show that 57% of adults are curious about their digital reputation, and search for their names online for determining what kind of information is available about them on the internet. They hereby state that “reputation management has now become a defining feature for online life for many internet users”. This desired identity created online is not totally different from the actual identity in real life: people’s online identity remains anchored to their offline identity (Labrecque e.a., 2011; Schau and Gilly, 2003; Bryant and Akerman 2009; Meadows 2008; Zhao e.a., 2008). So do people try to stay authentic and true to their real self. This element is also very visible in the creation of avatars (online personalities). When examining the relationship between internet users and their avatars, empirical studies have found that despite the possibility to endow an
avatar with any desired attribute, users tended to remain anchored to their real-life identity and create an avatar which is a mix between their desired and true selves (Bryant and Akerman 2009; Kozinets and Kedzior, 2009; Taylor and Harper, 2002). Like mentioned earlier, people try to project their desired personality and therefore filter and adjust the information that is spread online, which can result in further differences between online and offline information-sharing patterns (Labrecque e.a., 2011; Robinson, 2007; Taylor and Harper, 2002).
Information which is shared because of self-enhancement must reflect the desired identity. Extensive research concerning the content of this information is missing, however multiple researchers (e.g. Sernovitz, 2006; Berger and Schwartz, 2011; Wojnicki and Godes, 2008) state that this information at least should be interesting, for making the person more interesting and knowledgable.
3.2.2.2. Helping an organisation or cause
Hennig-Thurau e.a. (2004) claims that people also share information because they want to help an organisation. The driver behind this motivation is the satisfaction of the consumer in the products or services which leads to the desire to help the company (Sundaram e.a., 1998). Therefore, according to Hennig-Thureau e.a. (2004) “The customer is motivated to engage in sharing information online to give the company “something in return” for a good experience”. This view is shared by Sundaram (1998) who states that nearly 18% of all positive word-of-mouth is done for helping an organisation. This finding is in line with Brett (2013) who states that people share information online for getting the word out about causes or brands they support and care about. In the study, Brettt (2013) cites the example of a research subject that mentioned that forwarding an article about Proposition B to Everyone in his union with the reason that he “wanted them to learn about the issue and rally against it”. This urge was also seen in the Arab Spring in Egypt, where people shared information concerning demonstrations in order to make sure that as many people as possible would attend them (Tufekci and Wilson, 2012).
!
!
3.2.2.3. Helping other people
A respondent of the study of Brett (2013) answered the question why he shared something online with the answer “Because it’s so helpful. It makes me feel valuable”. The reported feeling resonates with the research done by Ho and Demspey (2009), which found that people who are more altruistic share more online content with the reason to help other people. The correlation between offline altruism and the use of online communication to help others has also been confirmed by further research (Phelps, 2004; Hennig-Thurau e.a., 2004) which all found that people spread information concerning a company online with the motivation to help their peers. Similarly, Berger and Milkman (2012) claim that there is a strong possibility that people share coupons or references of good restaurants for helping other people to save money or to eat better.
3.2.2.4 Self-therapy
Self-therapy is a further reason behind online information sharing explored by the literature (Buechel and Berger, 2012; Belk, 2013; Forest and Wood, 2012). The information shared in these cases concerns the personal situation of the individuals. In general, emotionally unstable people have been found to tend to share more information online than their emotionally stable peers (Buechel and Berger, 2012; Forest and Wood, 2012). The studies suppose that the reason behind this tendency is that expressing themselves online and sharing information helps unstable people to cope with negative events when there is a possibility that others will respond to their messages. Belk (2013) describes this kind of self-disclosure as “confessions”, for which the anonymity of the internet helps people to be more open.
4.0. Conceptual model
This paper will research which effect customer-company identification has on the online sharing of information. There will be focused on sharing information for the reasons of “self-enhancement” and “helping an organisation”. The following sections develop the conceptual model. To this end, the present section aims at summarising the underlying hypotheses, including the rationale.
4.1. Self-enhancement
Like mentioned before (3.2.2.1. Self-Enhancement) people share information motivated by a desire of self-enhancement (e.g. Labrecque e.a., 2012; Zhao, 2005; Schau and Gilly, 2003). When sharing information for self-enhancement people try to reflect their desired identity online with sharing interesting information, for seeming more interesting and knowledgeable (Sernovitz, 2006; Berger and Schwartz, 2011; Wojnicki and Godes, 2008).
The present work suggests that interesting organisational information, which is not helpful for an organisation or another person, is shared because of self-enhancement. Therefore there is expected that this information is shared more than not interesting organisational information, which is not helpful for an organisation or another person. This assumption will be tested in the first hypothesis.
-
H1: Organisational content which contains interesting information is shared online more than organisational content which is less interesting.As stated above (3.1.2. Reasons for identification) people feel identified with an organisation because its prestige contributes to increase of their self-esteem (Bhattacharya and Sen, 2003; Homburg, 2009). Furthermore,people have been found to share information online for creating their own desired identity, what can be called self-enhancement (Kozinets and Kedzior, 2009; Robinson, 2007; Taylor and Harper, 2002). Therefore it is plausible that people use a company with which they identify for showing their desired identity. This will mean that customer-company identification could possible moderates the effect of sharing information because of self-enhancement reasons. This assumption will be tested in the second hypothesis.
-
H2: Customer-company identification moderates the degree in which organisational content, which contains interesting information, is shared online.4.2. Helping an organisation
As discussed above (3.2.2.2. Helping an organisation or cause), multiple researchers (Hennig-Thureau e.a., 2004; Sundaram, 1998; Brett, 2013) have theorised that people share information of an organisation online because they want to help the organisation. Therefore in
!
the present research there is proposed that information which makes it more clear that the organisation is helped when the content is shared, is shared online more than information which does not make a clear connection between sharing and helping the organisation.
-
H3: Organisational content which contains information that helps the organisation is shared online more than organisational content that does not contain this information.Like stated earlier, people who identify with a group, are more motivated to help the group with its success (Bhattacharya e.a., 1995; Dutton e.a., 1994; Turner, 1982). There is assumed that this tendency means that possibly high identified people share information that helps the organisation online more than low identified people. However, this aspect of identification have not been researched in earlier studies. Therefore the present study proposes to verify whether customer-company identification moderates the effect of sharing information online to help this organisation. Therefore the fourth hypothesis will be the following:
-
H4: Customer-company identification effects the degree in which organisational content, which contains information that helps the organisation, is shared online.The final hypothesis proposes that customer-company identification has more effect on sharing information because of self-enhancement reasons, than because of helping the organisation. This is assumed as self-enhancement is a fundamental human motivation (together with belonging, understanding, controlling and trusting) (Baumeister e.a., 1998; Fiske, 2001) while no research suggest that helping an organisation or altruism is a fundamental human motivation. This leads us to assume that people lay more weight on self-enhancement than on helping an organisation. Therefore the fifth hypothesis will be the following.
-
H5: Customer-company identification has more effect on the degree in which organisational content which contains interesting information is shared, than organisational content which contains information that helps the organisation.!
!
!
5.0. Methods
In this section the proposed methodology and experimental design is described. Furthermore the sample, population, the used variables and the pre-test is specified.
For testing the earlier mentioned hypotheses a deductive research approach is applied. Due to the fact that this research will explain different relationships between existing variables, the research is explanatory (Saunders e.a., 2009). The hypotheses were tested in a quantitative manner with the use of questionnaires.
A survey approach was chosen as questionnaires can be distributed through a large population in an efficient and economical manner. The questionnaires were internet based self-administrative questionnaires. Online questionnaires have been chosen because of their efficient and non-expensive characteristics. The questionnaires were distributed by email. The cover email explained the purpose of the research and gave an indication of the time it would take to finish the questionnaire, what was 10 minutes. In order to gain a population large enough for a reliable research, data has been purchased from the commercial organisation MarketingData, a commercial supplier of business email addresses. The online questionnaire was designed with the use of Qualtrix software.
!
Contains(
informa,on(
that(helps(the(
organiza,on(
Interes,ngness(
of(informa,on(
Willingness(
to(share(the(
informa,on(
online(
C5C(iden,fica,on(
H1(
H2(
H3(
H4(
H5(
5.1. Sample
The sample exists of 25.856 randomly selected people. Minimum age to be part of the sample was 18 years old. No selection was made based on the profession or place of employment of the respondents. In order to decrease language or cultural biases the sample only existed of Dutch speaking males and females with a Dutch or Belgian nationality.
5.2. Pre-tests
In order to be able to test the hypotheses, two different elements in the main test needed to fluctuate. These elements were the level of customer-company identification and the interestingness of the texts. Both fluctuations needed to be present for being able to test the hypotheses. Three pre-tests have been performed for gaining a proper level of customer-company identification and interestingness.
5.2.1. Pre-test 1
The goal of this pre-test was to determine which organisation could be used within the main test. Within this pre-test 10 individual respondents rated 5 organisations concerning their customer-company identification. In the test, four seven-point Likert scale questions of Curras-Perex e.a. (2007) were used. The organisation in association with which the respondents showed the largest standard deviation in customer-company identification, and therefore had the highest chance of both having identified and non-identified respondents, was used within the main test. This was found to be object 3, which represented the organisation Ikea.
!
!
Mean Std. Devia-on
Object 1
3,63
0,784
Object 2
4,65
0,609
Object 3
4,10
1,028
Object 4
2,94
0,929
Object 5
4,06
0,702
5.2.2. Pre-test 2
The second variable which needed to fluctuate was the level of interestingness of the information. Therefore within the second pre-test ten individuals rated individually four different texts of Ikea. Texts were rated on interestingness by using the scale of Silvia (2005). The most interesting and uninteresting texts were selected for further research. These were found to be text 2 and text 3. The texts already existed and where published by the media online. The subject of text 2 was CSR activities performed by Ikea, and the subject of text 3 concerned the popularity of the meals of Ikea.
!
!
!
5.2.3. Pre-test 3The last pre-test was performed by distributing each text combined with the original questions among 2000 individuals. The response rate of 7.65% resulted in 306 filled out questionnaires. The results showed that the differentiation in interestingness between text 2 and text 3 was little. However, in the main test texts of substantial difference in interestingness were supposed to be used. It appeared that, when using these texts for the main tests, its output would not be interpretable. In order to decrease the chance that the output of the main test would not be meaningful, text 3 had been swapped for a new text (text 5). The subject of text 5 was ICT. According to multiple articles (e.g. Murray, 2012) ICT is generally perceived as an uninteresting topic. Choosing a topic that has been found by the previous literature to be of low level of interest, increased the chance that differentiation in interestingness between the CSR text (text 2) and the ICT text (text 5) would be larger in the final main test.
5.3. Variables
The first independent variable is the interestingness of the text. For determining if respondents found the text interesting the questions developed by Silvia (2005) have been
Mean Std. Devia-on
Text 1
4,643
1,825
Text 2
4,830
1,596
Text 3
4,226
1,675
Text 4
4,436
1,710
used. These questions have proven their internal consistency in earlier research, and were therefore deemed suitable for usage. The second dependent variable is the helpfulness of the text for the organisation, when the text is shared. There are no existing questions for determining this variable therefore an ad-hoc measurement approach was developed. In order to enable the measurement of this variable, a short additional supplement text was added to both texts, stating that sharing the text would help the company. Ultimately, this approach resulted in four different texts. Each text was later distributed towards a separate group of the sample, so a respondent would only read one kind of text.
The dependent variable was defined as the degree to which a person is willing to share the text online. This variable had earlier been researched by Berger (2011). Berger determined this variable with one question, namely “how willing would you be to share this information online with friends, family members and coworkers?”. In order to determine the reliability of this question, five questions using a seven point Likert-scale were added. The internal reliability of these questions was later determined by analysing the Crombach’s alpha.
The moderator variable is the degree of customer-company identification a person has. This variable was determined by four different seven point Likert-scale questions earlier used by Curras-Perex e.a. (2007). These four questions were also used by multiple other researchers (e.g. Ahearne e.a., 2005; Bhattacharya and Sen, 2003; Dholakia e.a., 2004; Lichtenstein e.a., 2004; Bergami and Bargozzi, 2000), whereby they were used in different compounds.
Two different control variables were also measured. These were the degree of altruism and the awareness of a person concerning his or her digital brand. The degree of altruism was determined through the deployment of scales used by Price e.a. (1995). The cited study found
Text%1% Text%2% Text%3% Text%4%
Text%
Su
pp
le
m
en
t%
W ith % W ith ou t% CSR% ICT%their internal consistency with an alpha of .90. Ho and Dempsey (2009) confirm this internal consistency whereby they show an alpha of .84.
The second control variable was defined as the degree to which people are aware of their digital reputation. This aspect had only been studied by qualitative research, therefore no scales existed to measure this variable. However Madden and Smith (2010) determined a comparable variable with the use of ten different “yes or no” questions. Examples of the questions include: “Have you ever deleted your name on a photo in which you were tagged?” or “Have you ever worried about the information that is flowing around you on the internet?”. Through the combination of these questions a new variable, “awareness of digital brand” has been created.
6.0. Results
The main test was filled out by 794 respondents (response rate 3.07%). Barely any data was missing due to the fact that all questions were obligatory. 218 respondents filled out the questionnaire for group 1, 186 for group 2, 170 for group 3 and 220 for group 4.
72% of the respondents was male and 28% was female. This was normally distributed between the four different groups. Considering the 25.856 people who were approached for the test, it merges that the percentage of women who have filled out the questionnaire was higher than the percentage of man. The fact that women are more cooperative to fill out a questionnaire has been empirically proven (Sax e.a., 2003) and therefore does not implicate a bias within the approach of the respondents.
The average age of the respondents was 49.72 years. The age of the respondents varied between 18 and 88 years. The means between the different groups in age do not vary much, what implicates that the chance on a bias is small. An independent sample T-test also showed that the difference in age between group 1 and group 4 is not significant (p=.523). The
Group 1 Group 2 Group 3 Group 4 Total Respondents Total Sample Male Female Male Female Male Female Male Female Male Female Male Female Unknown
Gende
r 152 66 138 48 125 45 157 63 572 222 22951 2372 4533 Percen
average age of the total sample is not known and therefore cannot been compared with the means of the respondents.
All levels of education were present in the sample, with HBO (University of Applied Sciences) as the mode with 42.7%. The different levels of education where normally distributed between the different groups. The total sample had a higher level of education in comparison with the whole Dutch population (CBS, 2014), which decreases the generalisability in extrapolating the results of the study to the whole Dutch population.
6.1. Reliability of the scales
The reliability of the different scales were measured by using the Cronbach’s Alpha (Field, 2009). All alpha’s where above .8 and can therefore be perceived as highly reliable. This directly indicates that there is a high internal consistency.
All variables are on a scale of 1 to 7. This means that all variables, except altruism, are below the median of 3.5. This effects reflects within the skewness of the different scores, whereby all variables are moderately skewed because they stay between -1 and 1 (Brown, 2012). When looking at the standard deviation it appears that the standard deviation of the
Age
Mean age
Group 1
49,04
Group 2
49,55
Group 3
50,40
Group 4
50,40
Total
49,72
0%# 10%# 20%# 30%# 40%# 50%# 60%# Basisschool# LBO,#VMBO#(MAVO)# Havo/VWO# MBO# HBO# WO# Anders# 1# 2# 3# 4# total# Dutch#PopulaGon#
variable “altruism” is very small. This is also shown within the Kurtosis, which is exceptionally high. All other variables have a Kurtosis between -2 and 2, what means that they stay within acceptable limits.
6.2. Manipulation check
6.2.1. Interestingness texts - subject of text
For manipulating the interestingness of the text two different kind of texts have been used. The goal of these texts was that one text would have a high amount of interestingness and that the other text would have a low amount of interestingness. Like predicted in the methods (5.2.3. Pre-test 3) one of the texts (CSR) is considered more interesting (M = 3.97; SD = 1.71) than the other text (ICT) (M = 2.38; SD = 1.34), this difference in interestingness is significant (P < 0.01).
Both texts combined contain all different degrees of interestingness, whereby there is an overrepresentation of the lower level of interestingness.
Crombach’s Alpha Mean Standard
Deviation Skewness Kurtosis Interestingness
.896
3.19
1.73
0,338
-0,890
Willingness to share.843
3.14
1.24
0,458
0,33
Customer-company identification.856
3.29
1.29
0,67
-0,550
Altruism.852
5.14
0.85
-0,917
3.413
Awareness of digital brand.841
2.47
1.76
0,424
-0,764
Interestingness text
Mean
N
Std. Devia-on
CSR
3.97
404
1.71
ICT
2.38
388
1.34
N
199
155
94
192
76
50
26
Degree of
interes-n
gness
1<2
2<3
3<4
4<5
5<6
6<7
7
6.2.2. Interestingness texts - adding supplement
For determining if people share the text because of altruism, at both texts a supplement was added. It is important that the text with a supplement is considered of similar interest as the text without the supplement. However an independent sample t-test shows that texts with a supplement are considered less interesting (M = 3; SD = 1.63) than texts without a supplement (M = 3.4; SD = 1.81) (p < .001). This is a very interesting factor, because this means that an organisation can “rally” their customers to help them. However this will be at the costs of the interestingness of the text.
The fact that a text with a supplement is considered less interesting than a text without a supplement could create a bias. This bias is created due to the fact that the decrease in interestingness could abrogate the increase in willingness to share which could be created by the extra supplement.
6.2.3. Customer-company identification
There is a low difference of customer-company identification between the different groups. This means that every group has relatively the same level of customer-company identification. The chance on a bias because of a difference in customer-company identification is therefore small. With an independent sample t-test there is also shown that the largest difference in mean between the groups (group 3 and group 4) is not significant (p = .11).
!
Interestingness text
Mean
N
Std. Devia-on
With supplement
3
406
1.63
Without Supplement
3.4
386
1.81
Customer-company identification
Mean
N
Std. Devia-on
Group 1
3,33
218
1,29
Group 2
3,28
186
1,32
Group 3
3,34
170
1,36
Group 4
3,23
220
1,23
6.2.4. Usefulness for others
Texts where graded on the degree of usefulness for other people (Likert-scale 1-5) by using the question “In which degree was this text useful for people in your surrounding?”. Hereby there is shown that the texts where seen as “not at all” or “barely” useful for people in their surrounding. It is clear that people have perceived the CSR text (M = 1.98) as more useful for others than the ICT text (M = 1.54). As people also share texts in order to help other people it is possible that this factor influences the outcome.
6.2.5. Control variables - Altruism and Awareness of digital brand
For the control variables “altruism” and “awareness of digital brand” there have been determined if there are large differences between the four groups. Outcomes show that the means and standard deviations of the different groups is comparable. For altruism the largest difference in means have been analysed by using an independent sample t-test. Hereby there is shown that the difference between 5.28 and 5.07 is not significant (p = .88). For the variable “awareness of digital brand” there has been determined that the largest differences (2.67 and 2.28) is significantly different (p = .041). This difference can have little influence on the outcomes and should therefore be considered when making conclusions.
Usefulness of others
Mean
N
Std. Devia-on
Group 1
1.98
218
756
Group 2
1.98
186
774
Group 3
1.54
170
587
Group 4
1.56
220
649
Total
1.77
794
730
Altruism
Mean
N
Std. Devia-on
Group 1
5.28
218
.89
Group 2
5.07
186
.87
Group 3
5.1
170
.84
Group 4
5.13
220
.81
6.3. Bivariate Pearson correlation matrix
A bivariate Pearson correlation matrix has been created for all variables. Multiple correlations have been found. No unusual correlations have been found, except the correlation between “interestingness” and “altruism”. However, the reason for this correlation can not be deduced within this analysis.
6.4. Hypotheses testing 6.4.1. Interestingness (H1)
For determining what effect the interestingness of a text has on the willingness of a person to share the text, a linear regression analyses has been performed. This analysis shows that the interestingness of a text significantly predicts (B = .323, t(790) = 14.207, p < .001) the willingness of a person to share the text. Interestingness also explained a significant proportion of the variances of the variable “willingness to share” (R2 = .451, F(1, 790) = 248.73, p < .001).
Awareness of digital brand
Mean
N
Std. Devia-on
Group 1
2.46
218
1.74
Group 2
2.67
186
1.87
Group 3
2.28
170
1.7
Group 4
2.44
220
1.72
Total
2.47
794
1.76
Means, Standard Devia-on, Correla-on
M
SD
1
2
3
4
5
1. Interestingness 3.19 1.73(.9)
2. Willingness to share 3.14 1.24.451** (.85)
3. Customer-company identification 3.29 1.29.216** .191** (.86)
4. Altruism 5.14 .85.122** .100** .62
(.85)
5. Awareness of digital brand 2.47 1.76
.50
.101** .075* .65
(8.4)
** Correla-on is significant at the 0.01 level (2-‐
tailed).
* Correla-on is significant at the 0.05 level (2-‐
tailed).
This indicates that when a person finds a text more interesting he or she is more willingly to share the text. This was already claimed in multiple researches (Sernovitz, 2006; Berger and Schwartz, 2011; Wojnicki and Godes, 2008) however was not statistically proven. This supports the claim that people share a text not only because they want to help an organisation or another person, but also share a text because they only find the text interesting. This is in line with the theorem that people share information for self-enhancement reasons, whereby they use the information to promote their digital self online.
This linear regression analysis shows that hypothesis 1: “Organisational content which contains interesting information is shared online more than organisational content which is less interesting” can be accepted.
6.4.2. Helping the organisation (H3)
An independent sample t-test was used to determine if people share a text online for helping an organisation. Like stated earlier (5.3 Variables) this was tested by adding a supplement to both texts. This supplement stated that the organisation would be helped when the text would be shared. An independent sample t-test was used for determining if the willingness to share the text was significantly higher at a text with a supplement versus a text without a supplement.
This test shows that a text with a supplement is shared more (M = 3.23, SE = 1.26) than a text without a supplement (M = 3.06, SE = 1.22). However this effect is not significant (t (794) = 1.913, p = 0.056). This means that people do not share a text with a supplement significantly more than a text without a supplement. It is clear that the p-value is very close to .05, what indicates that the difference is almost significant. However due to the fact that it is officially not significant the third hypothesis: “Organisational content which contains information that helps the organisation is shared online more than organisational content that does not contain this information”, is rejected.
!
!
!
6.5. Moderation analysis 6.5.1. Interestingness (H2)
The moderator effect of customer-company identification on the relationship between interestingness of the text and the degree in which people are willing to share the text online is measured by standardising the variable values. This is done by subtracting the mean from the original factor for both the degree of customer-company identification and the interestingness of the content. Both new z-variables were multiplied by each other what created a new variable, called “moderation interestingness”.
With a linear regression analysis there is shown that customer-company identity has barely any effect, which is not significant (B = .05, t( 790) = 1.135, p = .257), on the relationship between the interestingness of the text and the willingness of the person to share the text. The new variable also does not significantly predict a proportion of the variances of the willingness to share the text (R2 = .04, F(1, 790) = 1.289, p = .257). This indicates that customer-company
identification has no moderation effect on people when they share a text because of interestingness. So if a person wants to share an interesting text for self-enhancement they share it regardless if they are high identified with the organisation or not. The level of multicollinearity was at this regression analysis according to Menard (2002) and Myers (1990) within acceptable limits (Tolerance = .978; VIF = 1.023).
When there is zoomed in on the the 10% lowest and highest identified respondents, it is clear that there is a difference in willingness to share between the 10% lowest (M = 2.71, SE = 1.27) and the 10% highest (M = 3.38, SE = 1.29) identified respondents. However a univariate two-way anova shows that the fact if someone belongs to the highest or lowest group does not significantly predicts (F(1) = .538, p = .465) the willingness to share the text because of interestingness. This outcome in combination with the outcome of the linear regression analysis supports the claim that H2 (Customer-company identification moderates the degree in which organisational content, which contains interesting information, is shared online), is rejected.
!
6.5.2. Helping the organisation (H4 and H5)
The moderator effect of customer-company identification on the relationship between content that helps the organisation and the willingness to share the content online was measured with the method created by Andrew F. Hayes (Hayes, 2008). Within this test there is determined if the fact that a person is high or low identified with an organisation influences the willingness of this person to share a text because helping the organisation. This analysis shows that customer-company identification does not have a significant effect (p = .66) on the willingness of a person to share a text because of helping the organisation. What indicates that it does not matter what kind of degree of identification a person has. When a person wants to help the organisation by sharing its content, it does it regardless of the degree of identification.
This analysis have been retested with standardising the variable means. First two z-values where created which were multiplied by each other. In the second step, a linear regression analysis was performed with this new variable on the variable “willingness to share”. This analysis confirmed that customer-company identification does not have a significant moderation effect (B = -.01, t(792) = -.229, p = .82), on the relationship between adding a supplement at a text and the willingness of a person to share the text. The new variable does also not predicts a significant proportion of the variances of the variable “willingness to share“ (R2 = .008, F(1, 792) = .053, p < .819).
Both tests indicate that identification does not effect the willingness of a person to share a text online because of helping the organisation. However when looked at the 10% lowest and highest identified respondents it indicates that identification does have a certain impact. So is the difference between the means of the variable “willingness to share” between the text with a supplement at the 10% lowest identified respondents much smaller than at the 10% highest identified respondents (see graph attachment 10.3). This indicates that the supplement has more effect on the 10% highest identified respondents in comparison with the 10% lowest
Customer -‐ Company Iden-fica-on
10% Lowest
10% Highest
Mean
SD
N
Mean
SD
N
identified respondents. A univariate two-way anova hereby shows that there is a significant (F(1) = 11.449, p = .001) moderation effect of customer-company identification on adding a supplement to the text. This means that the moderation effect does not effect the whole range of identification, but only the highest identified customers. This indicates that only the highest identified customers are willing to help the organisation by sharing it’s content. When analysed if these high identified customers are also more altruistic than the low identified customers there is seen no significant difference between both groups (High identified; M = 5.19, SD = .91; Low identified; M = 5.16, SD = 1.12; t (157) = -.206, p =0.837), what indicates that the helpfulness for the organisation is really caused by the customer-company identification.
So although the method of Andrew F. Hayes (Hayes, 2008) and the linear regression model showed that customer-company identification does not have a moderation effect on the whole range of identification. The univariate two-way anova shows that customer-company identification does have a significant effect on the altruistic reasons for sharing organisational content online. However this occurs only when the identification is strong enough. Therefore the fourth hypothesis (Customer-company identification effects the degree in which organisational content, which contains information that helps the organisation, is shared online) is accepted.
Hereby there is shown that customer-company identification has a moderation effect on sharing information because of helping the organisation. However customer-company identification does not have a significant effect on sharing information because of self-enhancement. Therefore the fifth hypothesis (Customer-company identification has more effect on the degree in which organisational content which contains interesting information is shared, than organisational content which contains information that helps the organisation) is rejected.
!
Customer -‐ Company Iden-fica-on
10% Lowest
10% Highest
Mean
SD
N
Mean
SD
N
Willingness
to share
With supp
2.71
1.43
39
3.55
1.21
35
Without supp
2.70
1.11
40
3.26
1.35
45
7.0. Conclusion and Discussion
This paper has done a research in the field of online information sharing and the effect of customer-company identification on this practice. Hereby there was hypothesised that people share information because of self-enhancement and for helping an organisation (altruistic reasons). The study found that the more interesting a text is, the more willingly people are to share this text online. Although this was claimed by multiple researchers in previous literature (Sernovitz, 2006; Berger and Schwartz, 2011; Wojnicki and Godes, 2008), none have statistically investigated it before. The analysis showed that people not only share information because they want to help other people, but also because they just find the text interesting. This supports the claim that people share information online because of personal reasons; for creating an interesting appealing personal brand, so for self-enhancement. Although it was known that people share information for self-enhancement, like for example sharing information about the professional skills, it was not known that organisational information could be used for this issue. This means that when an organisation publishes information which is interesting, people will share it for showing their own (desired) personal self online. For marketing managers does this implicate that the information used for a viral campaign should be useable for people to show their personal desired identity online. Then people will use it for their personal gain and directly benefit the organisations viral marketing campaign.
Another outcome of this research is the finding that customer-company identification does not moderates the effect if people share information because of self-enhancement. This is interesting due to the fact that multiple researchers claim that people share information online for personal gain (Labrecque e.a., 2012; Zhao, 2005; Schau and Gilly, 2003), and that people identify with an organisation for increasing self-esteem (Homburg, 2009). However this research shows that people don't directly use the company with which they are high identified for personal gain. This means that the interestingness of the information is more important than the identification with that organisation. For managers this is interesting due to the fact that it supports the thought that, although the company does not have a large high-identified customer base, it still can start a viral marketing campaign with interesting content. The fact that the interestingness of the information is more important than the customer-company identification could possibly have happened because self-enhancement is a stronger human motivation than
the identification with an organisation (Baumeister e.a., 1998; Fiske, 2001). Although belonging is, besides self-enhancement, a fundamental human motivation (Baumeister e.a., 1998; Fiske, 2001), this can also refer to more formal relationships like the belongings to a family instead of an organisation. Therefore it is very plausible that people find it much more important to share information for personal gain, than showing others that they feel high identified with an organisation.
Furthermore, the study shows that people don't directly share information more because of helping an organisation. An explanation for this phenomena is the reason that people find information whereby it is clear that they help the organisation less interesting. And because interestingness has a direct relationship with the willingness to share a text, this moderates the relationship. A plausible explanation why customers find a text less interesting when it is made clear that they help the organisation is because customers are always questioning themselves “why is the organisation doing this” (Bhattacharya and Sen, 2004). Hereby customers can get suspicious about the fact why the organisation is publishing the relevant information. They can wonder that the organisation does this for personal gain, and therefore find the information less interesting. This is very interesting because when a viral marketing campaign is started, the organisation cannot rally their customers in a too obvious way, because this will backfire and makes the content less interesting. What leads towards a lower willingness to share the content.
Within this research there is also shown that the level of customer-company identification does have a significantly moderation effect on the fact that people share organisational information for helping the concerned organisation. Hereby it is very interesting to see that this effect does not occur in the whole range of identification, but only occurs with the highest identified customers. These customers are the so-called promoters (Bhattacharya and Sen, 2003) and are less suspicious about the text and are more willingly to help the organisation. This can mean for marketing managers that, when they want to start a viral marketing campaign, they should not rally their total customer base but only the most identified customers. These customers will be more willing to help the organisation and therefore share the content.
