Determinants of Passing on Viral Messages Empirical Analysis of a Viral Marketing Campaign on Facebook

Internet has evolved into a dynamic network where people can easily and constantly interconnect. Social-media-platforms allow consumers to interact and inform each other, likewise on products and services. Under pressure of increasing customer-resistance toward traditional advertising-forms marketers have to find new, effective forms. Viral-marketing on social-media therefore plays an important role, evoking the question which factors really have an impact on the success of viral-marketing-strategies. Therefore, we undertake an empirical study to analyze and compare factors influencing the behavior of forwarders and non-forwarders of viral messages, examining the effect of brand-attitude, brand-experience and self-congruity on recipients’ forwarding behavior on the basis of a real viral-marketingcampaign on Facebook about Tiger Balm. Generally, the paper provides evidence that there are significant differences between people who pass-on a viral message and those who don’t. Accordingly, it is vitally important to select adequate consumers who should be first recipients of a viral marketing message.


Introduction
For a long time traditional Word-of-Mouth-Communication (WOM) has been an important research topic in academic literature (Engel et al., 1969;Czepiel, 1974;Haywood, 1989; companies as well as the increasing credibility of a message that comes from the customers' own social network, we define viral marketing as follows: "Viral marketing is an advertisement that is in some way tied to an electronic message. It can be attached to an e-mail, video spot or posted in any form of social media. It is a form of advocacy or word-of-mouth endorsement marketing, whereas one customer passes along the message to other potential buyers, creating the potential for exponential growth in the message's exposure and influence." (Clow, and Baack, 2005, p. 177; see also Wilson, 2000) The main element for a viral marketing campaign to work is not the advertised product or brand, but the message itself (Langner, 2007). The creation of a unique message content and container is of crucial importance. A very often used container for viral messages is a video clip (Kirby, and Marsden, 2006), putting emphasis on special emotions going along with with the advertised brand/product. To sum up: viral marketing is about a special packaging of advertising messages that will be forwarded exponentially, using existing communication networks of the costumers, because it is of outstanding interest for the target group.
To initiate the viral effect it is important for companies to select adequate consumers who should be the first recipients and first to pass on the viral marketing message, as the success of a viral marketing campaign depends on their forwarding behavior (Helm, 2000;Bannan, on the principle of cause and effect, whereas inner processes in the human mind are neglected and take place in a kind of black box. To focus on the black box, we need to interconnect stimulus and response with an organism that accommodates the involved individuals of the communication act, which leads to the further developed stimulus-organism-response (SOR) model. Within this model, we can give a better description of the complexity of human behavior than before, however it only captures a single sequence (Nießing, 2007).
To elaborate on this, we come back to the already mentioned "two-step-flow-model of communication" (Lazarsfeld et al., 1948). In their study, Lazarsfeld et al. concluded that ultimately, interpersonal communication is most influential on people's behavior. Putting these findings together in a single communication model, we can draw a cycle of consequent SOR-sequences influencing each other (Luthans, 1981). Thus, the following model is used to describe the dispersion of viral messages:  Nießing, 2007, p. 112) In its centre is the traditional model of interpersonal communication, which is extended concerning input (first stimulus) and output (last response) (Nießing, 2007). The first stimulus is the beginning of a viral marketing campaign, namely the already mentioned process of seeding. The aim of the marketer should be the detection of adequate first recipients like e-fluentials (Burson-Marsteller, 2007) or e-mavens (Walsh et al., 2005). In our model organism I stands for a recipient, who becomes a prospect communicator with the receipt of the viral message (stimulus I). Then, the first SOR-model has two possible responses: if the overall evaluation of the viral message is negative, there will be no forwarding and the model is at its end. But if the overall evaluation is positive, the message will be forwarded and response I (the viral message) becomes the stimulus for organism II.
Again an individual rating of the message takes place and leads to the same options as in the first SOR-model. If these positive sequences are repeated again and again we will find a viral, exponential growth of information dispersion through the existing communication networks of the costumers.
Referring to our model, forwarding behavior is substantially influenced by inner processes taking place in the organism (recipient). These processes are individually differentiated.
Furthermore, the given overview of prior research indicates a strong relationship between the recipients' beliefs and attitudes and the actual forwarding process of viral messages. In the next step we therefore discuss constructs, which influence the overall evaluation of a viral message and form beliefs and attitudes. We explain the underlying theories and build hypotheses which are tested empirically.
quite young (56.4% less than 30 years) and very well educated with 74.3% having at least a high school degree which allows application at a university.
The viral message was forwarded by 55.8% of the sample. This group can be identified as forwarders or advocates. 44.2% did not forward the message.

Measures
Variables were measured through 7-point Likert scales, ranging from strongly agree (1) to strongly disagree (7). In all cases, items were extracted from previous research. The survey used a multi-item approach with each construct being measured by a few items for construct validity and reliability.

Differences between the mean values of forwarders and non-forwarders of viral messages
were tested for significance using the conventional t-test. As it is vitally important to differentiate between testing with equal or unequal variance, we deployed a Levene-test in preparation. Table 2 shows the results.

Constructs Source Items
Brand Attitude (BrandAtt) Batra/Stayman (1990) 1 pleasant -unpleasant 2 good -bad 3 positive -negative 4 favorable -unfavorable 5 like -dislike 6 useful -useless 7 high quality -low quality 8 beneficial -not beneficial 9 valuable -worthless 10 agreeable -disagreeable Self-Congruity (SelfCon) Sirgy et al. (1997) 1 The typical user of Tiger Balm reflects the type of person who I am 2 The typical user of Tiger Balm is similar to me 3 The typical user of Tiger Balm is very much like me We can rely on getting the truth in most advertising 2 Advertising's aim is to inform the consumer 3 I believe advertising is informative 4 Advertising is generally truthful 5 Advertising is a reliable source of information about the quality and performance of products 6 Advertising is truth well told 7 In general, advertising presents a true picture of the product being advertised 8 I feel I've been accurately informed after viewing most advertisements 9 Most advertising provides consumers with essential information Online Journal of Communication and Media Technologies Volume: 4 -Issue: 4 -October -2014 The results show that the t-test for involvement (0.494), self-congruity (0.086) and brand attitude base on equal variance, for ad skepticism and brand experience on unequal variance.
Consequently, Table 3 reports the results of the hypotheses tested: t-tests with equal variance for brand attitude (0.000), self-congruity (0.007) and involvement (0.007) show that the mean deviation between the two groups is highly significant.
Furthermore, t-tests with unequal variance show that the mean deviation of brand experience (0.000) is highly significant, only ad skepticism (0.270) is not. These findings provide evidence that the two investigated groups differ significantly from each other.
Amongst forwarders, the mean value for brand attitude is 5.2 when non-forwarders rate brand attitude with a mean value of only 4.5. Thus, forwarders rate brand attitude 0.7 points higher than non-forwarders (p < 0.01). Thus, H 1 "The value of brand attitude is significantly higher in the group of forwarders." is supported. The values for self-congruity show similar characteristics. Forwarders rate self-congruity 0.6 points higher than non-forwarders. This highly significant (p < 0.01) mean difference supports H 2 "The value of self-congruity is significantly higher in the group of forwarders." The overall low rating of self-congruity (overall mean value of 2.9) may be related to translation issues. With a highly significant (p < 0.01) mean difference of 0.9 brand experience differs most between the two groups, which