See discussions, stats, and author profiles for this publication at: [Link]

net/publication/295252407

Discriminant model for online viral marketing influencing consumers

behavioural intention

Article in Pacific Science Review B Humanities and Social Sciences · January 2016
DOI: 10.1016/[Link].2015.12.002

CITATIONS READS

54 755

2 authors:

Sharda jai Haryani Bharti Motwani

Prestige Institute of Management and Research University of Maryland, College Park
16 PUBLICATIONS 112 CITATIONS 14 PUBLICATIONS 141 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Bharti Motwani on 24 October 2023.

The user has requested enhancement of the downloaded file.

 Pacific Science Review B: Humanities and Social Sciences 1 (2015) 49e56

H O S T E D BY Contents lists available at ScienceDirect

Pacific Science Review B: Humanities and Social Sciences

j o u r n a l h o m e p a g e : w w w . j o u r n a l s . e l s e v i e r . c o m / p a c i fi c - s c i e n c e -
review-b-humanities-and-social-sciences/

Discriminant model for online viral marketing influencing consumers

behavioural intention
Sharda Haryani*, 1, Bharti Motwani 2
Prestige Institute of Management and Research, Scheme 74-C, Vijay Nagar, Opp. Life Care Hospital, Indore, M.P., India

a r t i c l e i n f o a b s t r a c t

Article history: Advancements in communication technology have given rise to the evolution of a new electronic form of
Available online 19 January 2016 word-of-mouth publicity i.e., viral marketing communication. Internet induced viral marketing is a
highly effective tool available to marketers today. Online marketing has been used by various organi-
Keywords: zations via platforms, such as blogs, company websites, and social networking sites. Online marketing is
Consumer behaviour helping organizations to grow their business by adopting strategies to reach out to their customers. Such
Internet marketing
marketing can be used to generate brand awareness among large populations in a short span of time; in
Viral Marketing
addition, consumers can refer brands to others. Word of mouth publicity can be a powerful factor to
influence the purchasing decision of customers. This paper is an attempt to predict the buying intention
of consumers as a result of internet induced viral marketing techniques. The study was performed on
primary data collected through self-structured questionnaire based on a five-point Likert Scale. Non-
probability judgemental sampling was used. Discriminate analysis was performed using statistical
software SPSS 16.0. The study determines the important factors that predict customer buying behaviour.
Marketing managers should understand these factors and appropriately target the customer. Appropriate
use of these factors can help the managers' businesses flourish. In this study, we found that the factors
immense efficacy, professed security and escalating brand were significant predictors of the consumer's
buying intention.
Copyright © 2016, Far Eastern Federal University, Kangnam University, Dalian University of Technology,
Kokushikan University. Production and hosting by Elsevier B.V. This is an open access article under the CC
BY-NC-ND license ([Link]

1. Introduction

According to Ridings et al. (2002) and Bruyn and Lilien

* Corresponding author. Tel.: þ91 9755124949 (mobile).
E-mail addresses: sharda_haryani@[Link] (S. Haryani), bharti_ (2008), in the small world concept, the demand for informa-
motwani@[Link] (B. Motwani). tion, the good intention to share experience or ideas regarding
Peer review under responsibility of Far Eastern Federal University, Kangnam products and services, the capability of the customer-to-
University, Dalian University of Technology, Kokushikan University. customer (C2C) communications as electronic peer-to-peer re-
1
She has done [Link] (Mathematics) from Holkar Science College, D.A.V.V., Indore.
ferrals on the Internet and the potential for obtaining recom-
She has seven years of teaching experience in academics. She has been distinction
holder throughout the academics. She has various publications to her credit and has mendations from mutual acquaintances have considerably
attended a number of conferences, seminars and workshops organized by various increased the potential for Viral Marketing (VM), which is the
institutes and organizations. Her areas of interest are Differential Equations, Cal- effective mixture of the traditional idea called Word of Mouth
culus, Discrete Mathematics and Statistics. (WOM). The advent of social networks, online communities,
2
She has completed Masters in Business Administration (Information Systems)
from Institute of Management Studies, DAVV, Indore in 1997 and [Link](Computers)
online videos and emails, blogs, and forums provide the ability to
from Holkar Science College, Indore in 1995. She has 7.5 years of experience in distribute information faster than ever before. Viral Marketing
software industry. She is associated with teaching industry since last 8 years. She can spread any message to millions of people in a short time
has been a distinction holder throughout her academics. Her technical expertise span in an effective manner. This technique has been used to take
include languages like Java, J2EE, [Link] etc, RDBMS like Oracle, SQL SERVER etc.,
advantage of pre-existing social networks to produce exponen-
Frameworks like JSF and IDE like Net Beans etc. She has developed and trained
many software projects on different platforms. She has also conducted training tial increases in brand awareness; this is similar to the spread of
program for students of different institutes. Currently she is pursuing Ph.d on En- an epidemic.
terprise Resource Planning. Tel.: þ91 9755958877 (mobile).

[Link]
2405-8831/Copyright © 2016, Far Eastern Federal University, Kangnam University, Dalian University of Technology, Kokushikan University. Production and hosting by Elsevier
B.V. This is an open access article under the CC BY-NC-ND license ([Link]
50 S. Haryani, B. Motwani / Pacific Science Review B: Humanities and Social Sciences 1 (2015) 49e56

Viral Marketing is the current word that is used to describe the informal. Messages are spread through different channels including
revolutionary manner by which any information proliferates across e-mail, chat rooms and discussion forums. The messages may
a million people rapidly within a short period of time. Marketers encompass various types of content that ranges from text and
are now increasingly looking forward to tap this opportunity to images, to Microsoft PowerPoint files, Adobe's Flash animations,
deliver any type of a marketing message because of its ability to and YouTube's video clips (Woerndl et al., 2008). Viral marketing is
make the message contagious. Viral Marketing is not easy or un- viewed as a far more effective, penetrating and faster medium
complicated; it has been observed in the past that all such mar- compared with traditional WoM communication (Helm, 2000).
keting campaigns have not been always successful, with certain Spero and Stone (2004) state that it is estimated that in the UK,
exceptions. The complexity of Viral Marketing campaigns arises twelve- to sixteen-year-olds spend, on average, £3 Billion per year
from understanding how consumers react or respond to online on products and services. More traditional advertising such as
marketing messages. It is extremely important to tap the appro- television ads no longer satisfies the youth sectors, who increas-
priate type of consumer behaviour and attitude to leverage the ingly desire more information regarding products and brands
opportunities available with the marketers. The effectiveness of before they decide to make a purchase. However, more importantly,
viral marketing as a promotion relies on the types of social in- it appears the “consumer generated media”, as described by Frost
teractions and on identifying the most active media in viral mar- (2007), is being used by over 50% of the population to make their
keting campaigns. Hence, in this study, we will examine certain purchasing decisions. It has been suggested that 82% of the fastest
factors that directly impact the implementation of Viral Marketing growing private companies use word-of-mouth techniques, which
campaigns and how they play a role in creating positive sales for is why viral marketing is becoming more popular (Frost, 2007).
any product or service. Jurvetson and Draper (1997) also argued that viral marketing
For the purposes of this study, we will consider viral marketing utilises the recommendations by friends and has a snowball effect.
as a technique that utilises the Internet to transmit and spread Viral marketing is more powerful than other means of advertising
messages among individuals who will filter and forward the mes- because it conveys an implied endorsement from a known person.
sages to their peers, who may be interested in the message's con- Therefore, companies should personalize their referral e-mail so
tent. Due to the extensive use of the internet in sharing information that it clearly shows that it is originating from a friend. Currently,
among young adults, marketers should focus on viral marketing as Viral Marketing Communication is still considered to be at a pre-
one of the new marketing sources. The growth in the number of mature phase of development; therefore, information regarding its
internet users has been dramatic since it entered into existence. behaviour and magnitude has yet to be clearly established and
The marketing managers want to observe the factors that determined. The Internet allows significantly more interaction,
discriminate between those who have the intention of buying from targeted communication, increased reach and better results eval-
those with the intention of not buying the products as a result of uation, all at a low cost (Derbaix and Vanhamme, 2003; Kozinets
online viral marketing techniques, because once this can be iden- et al., 2010). Although WOM occurs naturally between con-
tified, certain intervention measures can be put in place to ensure sumers, researchers note that there are marketing activities
the businesses flourish. In our pilot study, we identified seven implemented to intentionally influence the per-to-peer commu-
factors that affect the consumer attitude towards the internet nications (Kozinets et al., 2010). At the same time, as we will
induced by the viral marketing technique. In this study, these fac- observe in the following, word-of-mouth is an important aspect in
tors were considered as predictors to discriminate between the the modern communication media, such as e-mail, blogs and social
customer who wants to buy the product and the customer who media (Steyer et al., 2007; Smith et al., 2007). Consumer behaviour
does not buy the product as a result of an online viral marketing has been identified as a risk taking behaviour since Raymond Bauer
technique. proposed his risk theory in 1970 (Taylor, 1974). Since then, risk
taking studies have examined consumer responses to reduce the
2. Literature review risks of purchasing certain products and brands. Specifically, Sheth
and Venkatesan (1968) observed that consumers rely on seeking
According to Vilpponen et al. (2006), certain of the terms used to information, such as a brand image or purchasing experience, to
describe electronic WoM include “interactive marketing” (Blattberg overcome a perceived risk. In another investigation, Roselius (1971)
and Deighton, 1991), “Internet word-of-mouth” and “word-of- proposed word of mouth seeking as a searching method that
mouse” (Goldenburg et al., 2001), “stealth marketing” (Kaikati and consumers use to manage risk and uncertainty. Word of mouth
Kaikati, 2004) and “referral marketing” (De Bruyn and Lilien, 2004). (WOM) is a primary information channel that influences consumer
Thomas (2004) attempted to unify these ideas in the term called decision-making (Arndt, 1967; Engel et al., 1969; Katz and
“buzz marketing”. Knight (1999) states that viral marketing is Lazarfeld, 1955; Richins, 1983; Richins and Root-Shaffer, 1988).
similar to a “digitalized sneeze”, one characterized by the release of Accompanied by the prevalence of the online marketplace, eWOM
“millions of tiny particles that can infect others who come into is rapidly flourishing. McGuigan (2008) stated that 77% of online
contact with them”. Welker (2002) emphasizes the contagious consumers check online reviews and the ratings of products before
power of a virus and suggests that a “virus replicates with deciding to make a purchase, and 92% of consumers reported that
geometrically increasing power, doubling with each interaction”. online reviews are extremely helpful (Kee, 2008).
Viral marketing communication is viewed as a vital electronic
extension of word-of-mouth (WoM) communication, which in- 3. Research methodology
volves the principle of passing on or referring news, information or
entertainment to another person. Where WoM has occurred, it The research design of this study is exploratory in nature. The
simply implies that informal, ad hoc communication between in- objective of exploratory research is to explore or search through a
dividuals concerning products and services has occurred (Bayus, problem or situation to provide insight and understanding (Kothari,
1985). WoM communication is widely perceived as a dominant 2004). The analysis of collected data was performed using Statis-
force in the marketplace where information is filtered out and tical Package for Social Science (SPSS 16.0) and MS Excel 2007. In a
passed on by friends or family who are viewed as free, unbiased pilot study on a sample of 110, seven factors were extracted using
sources of advice (Cruz and Fill, 2008; Carl, 2008). In viral mar- principal component factor analysis, which are Immense Efficacy (%
keting, the communication style used for transmission is usually of Var. ¼ 19.693), Supportive Access (% of Var. ¼ 13.17), Professed
 S. Haryani, B. Motwani / Pacific Science Review B: Humanities and Social Sciences 1 (2015) 49e56 51

Security (% of Var. ¼ 10.289), Message Material (% of Var ¼ 10.18), Many of the users show a lack of trust in terms of the credibility
Irrepressible (% of Var. ¼ 6.503), Consumer dependency (% of of the medium, thereby considering the material as spam and are
Var. ¼ 6.029) and Escalating brand (% of Var. ¼ 5.142). The total reluctant to provide their personal information to companies
percentage of variance for factors was 71.004%, and the Eigen values because they often think of advertisers as dishonest (Kelly et al.,
for each dimension was more than one. The details of these factors 2010). Higher source credibility results in a more favourable con-
are provided in Annexure 1. sumer response towards the SMS advertisements (Muzaffar and
Discriminant analysis was applied using these seven factors Kamran (2011)). The credibility and trustworthiness of the viral
as dependent variables and buying intention as the independent message source is also closely linked with the perceived risk. If the
variable; buying intention is a nominal variable with two levels viral marketing message originates from a trusted source, the
of categories i.e., 1 ¼ buy the product and 2 ¼ not buy the perceived risk associated with the message is low. The message
product. recipients express no security and privacy concerns when they
Discriminant or discriminant function analysis is a parametric receive viral message from their social network (Palka et al., 2009).
technique to determine which weightings of quantitative variables According to Chu and Kamal (2008), if the blogger's trustworthi-
or predictors best discriminate between 2 or more than 2 groups of ness is high, the blog readers will be willing to trust the information
cases and do so better than chance (Cramer, 2003). The analysis provided on the blog and would read the arguments made and vice
creates a discriminant function, which is a linear combination of versa. Hence, it is proposed that
the weightings and scores of these variables. The maximum num-
H3. Professed Security is a good predictor of the intention to buy
ber of functions is either the number of predictors or the number of
as a result of viral marketing.
groups minus one, whichever of these two values is smaller.
Reyck and Degraeve (2003) maintain that advertisements that
Zjk ¼ a þ W1 X1k þ W2 X2k þ ………: þ Wn Xnk contain interesting and customized information that matches
customer preferences will result in their positive attitude toward
Where: mobile advertisements. Furthermore, message variety and appro-
priate message delivery timings with an appropriate message fre-
Zjk ¼ Discriminant Z score of discriminant function j for object k. quency could also result in a positive customers' attitude toward
a ¼ Intercept. mobile marketing messages. It is necessary that viral marketing
Wi ¼ Discriminant coefficient for the Independent variable i. messages contents are concise and are able to capture consumers'
Xik ¼ Independent variable i for object k. attention. Moreover, if the message content becomes excessively
manipulative, it also irritates message recipients and results in a
negative consumer attitude (Palka et al., 2009; Haghirian et al.,
2005). It has been found that message clarity was considered as a
4. Hypothesis
significant factor to the success of advertising, particularly in the
context of digital advertising (Taylor et al., 2006); this influences
The heightened focus on viral marketing in the computer and
the consumer's attitudes in the area of viral advertising. Therefore,
management literature is a sign that there can be significant ben-
the hypothesis formulated is:
efits to be gained from viral marketing.
Informativeness is an important factor in determining the H4. Message material is a good predictor of the intention to buy as
effectiveness of advertisements (Saadeghvaziri and Hosseini, 2011). a result of viral marketing.
Informativeness has an impact on consumers' attitude toward on- Minimal control occurs over information because social media
line advertising because it helps them in making purchase decisions sites are represented as “anything goes” communication channels
(Tsang et al., 2004). The information provided to the consumers by because anyone can post anything. Hence, people do not trust these
means of mobile phones should consist of features such as accuracy, posts or the people posting those messages (Kelly et al., 2010).
timeliness, and usefulness for generating positive consumers' atti- However, in addition to these significant benefits, there are risks
tude. Consumers are interested in receiving messages that are and challenges that marketers must confront when engaging in
relevant to them (Haghirian et al., 2005). Entertainment is a sig- viral marketing campaigns. Probably the largest risk is the lack of
nificant predictor of the value of web-based advertisements and control associated with viral marketing campaigns: organisations
hence, determines their efficiency. Interesting and pleasing ads have no means of controlling the spread of the message and the
have a positive impact on consumers' attitude towards brands. content of the transmission (Dobele et al., 2005; Helm, 2000;
Moreover, entertainment is the most important feature that affects Kaikati and Kaikati, 2004; Welker, 2002) that the receivers of a
consumers' attitude towards mobile advertising (Tsang et al., 2004; message may consider. Based on this argument, it is proposed that:
Bauer et al., 2005). Therefore, we hypothesized that
H5. Irrepressible is a suitable predictor of the intention to buy as a
H1. Immense Efficacy is a good predictor of intention to buy as a result of viral marketing.
result of viral marketing. It is generally agreed that viral marketing involves the spreading
One important benefit is that viral marketing is relatively of a marketing message via ‘word of mouse,’ which ensures that the
inexpensive in comparison to many other forms of advertising and receivers have the interest to pass along the message to their ac-
marketing campaigns (Dobele et al., 2005; Kaikati and Kaikati, quaintances (Kaikati and Kaikati, 2004). The dependency on the
2004; Welker, 2002). The other major benefits relate to the posi- consumer for message transmission is a further risk because con-
tive diffusion characteristics; viral marketing can, for example, sumers, for example, may want a return from the organisation for
reach audiences within a short period of time (Kaikati and Kaikati, passing on a viral message (Helm, 2000). Thus, it is predicted that:
2004) as messages spread exponentially at a rapid speed (Helm,
H6. Consumer enslavement is a good predictor of the intention to
2000; Welker, 2002). This rapid diffusion can significantly boost
buy as a result of viral marketing.
the speed of the adoption of the marketed product or service
Brand familiarity has been identified, by Alba and Hutchinson
(Dobele et al., 2005). Hence, it is hypothesized that
(1987), Hoch and Deighton (1989), and Kent and Allen (1994), as
H2. Supportive access is a good predictor of the intention to buy the consumers level of direct and indirect experience with products
as a result of viral marketing. or brands. It is clear that consumers are likely to respond to ads if
52 S. Haryani, B. Motwani / Pacific Science Review B: Humanities and Social Sciences 1 (2015) 49e56

they are sent from the company that they know or are familiar with. and consumer dependency in the groups are significantly different,
Another area of opportunity for social marketing is “brand build- and the means of the variables, irrepressible and escalating brand,
ing”, which connects enthusiastic online brand advocates with the have nearly the same values. Otherwise, the same relation has been
company's product development cycle (Ferguson, 2008, p. 181). In established in the tests of equality of the Group Means Table, where
this area, research becomes marketing; product developers are in excluding irrepressible and escalating brand, all of the other five
now using social forums to spot reactions after they modify an offer, predictors have a statistically significant difference. The pooled
a price, or a feature in a product or service. Such brand managed within group matrix (refer to Table 2) also establishes the fact that
communities can have real success. Thus, this statement leads to the intercorrelations among all the variables are very low and
the seventh hypothesis: statistically insignificant.
The maximum number of discriminant functions is equal to the
H7. Escalating brand is a good predictor of the intention to buy as
number of groups in the dependent variable minus one, or the
a result of viral marketing.
number of variables in the analysis, whichever is smaller. In our
study, the maximum number of discriminant functions is one. The
5. Analysis canonical correlation is the multiple correlations between the
predictors and the discriminant function. With only one function,
Before proceeding with the analysis, we split the sample into the correlation provides an index of the overall model fit, which is
two portions. Of the sample, 65% was used for analysis, and the interpreted as being the proportion of variance explained (R2). In
remaining 35% was used as a control sample. The splitting follows this analysis, a canonical correlation of .684 (refer Table 3) suggests
the analysis sample and the control sample testing, which typically that the model explains 46.7% (square of canonical correlation) of
needs another data set to be collected for prediction purposes. This the variation in the grouping variable A high canonical value de-
data set is achieved by splitting the sample for which we develop a scribes a good overall fit of the analysis, which, in our findings, is
function using the analysis and then use that function to predict the relatively good (0.684).
control sample to gauge the predictive accuracy of the model Wilks' lambda indicates the significance of the discriminant
developed (Ramayah, 2006; Ramayah et al., 2004). To split the function. Table 4 below indicates a highly significant function
sample, we compute a variable using the function, as follows: (p ¼ 0.000, which is less than .05, and we accept the Null Hy-
RANDZ ¼ UNIFORM (1) > 0.65. In our study, the analysis sample pothesis). The lower the Wilks' lambda, the better the model is.
is composed of 134 cases, whereas the control sample has 67 cases.

Table 3
6. Results and discussion
Eigenvalues and correlation.

The data were analysed using the enter method of discriminant Function Eigenvalue % of Variance Cumulative % Canonical correlation
analysis in the statistical package SPSS Statistics 16.0 An approxi- 1 .877a 100.0 100.0 .684
mate idea of variables that may be important can be obtained by
inspecting the group means and standard deviations. If there are no
significant group differences, it is not worthwhile to proceed any Table 4
further with the analysis. In Table 1, the group statistics mean Wilks' lambda.
values clearly indicate that there is a large separation among all the Test of Function(s) Wilks' Lambda Chi-square df Sig.
predictors within each other. The means of the variables, immense
1 .533 80.922 7 .000
efficacy, supportive access, professed security, message material

Table 1
Mean values, test of equality of group means.

Independent variable Mean values customer intention

Buy Not buy Wilks' Lambda F df1 df2 Sig.

Immense efficacy 3.69 2.69 .698 57.231 1 132 .000

Supportive access 3.60 2.78 .789 35.329 1 132 .000
Proffessed security 3.77 2.86 .822 28.663 1 132 .000
Message material 3.94 3.03 .746 44.884 1 132 .000
Irrepressible 3.91 3.88 .999 .085 1 132 .772
Consumer dependency 3.74 3.08 .894 15.728 1 132 .000
Escalating brand 3.86 3.86 1.000 .000 1 132 .989

Table 2
Pooled within-groups matrices.

Immense Supportive Professed Message Irrepressible Consumer Escalating

efficacy ACCESS security material dependency brand

Correlation Immense efficacy 1.000 .222 .375 .409 .211 .081 .117
Supportive access .222 1.000 .161 .445 .083 .040 .203
Professed security .375 .161 1.000 .238 .015 .021 .168
Message material .409 .445 .238 1.000 .171 .171 .253
Irrepressible .211 .083 .015 .171 1.000 .091 .155
Consumer dependency .081 .040 .021 .171 .091 1.000 .051
Escalating brand .117 .203 .168 .253 .155 .051 1.000
 S. Haryani, B. Motwani / Pacific Science Review B: Humanities and Social Sciences 1 (2015) 49e56 53

We examine the probability of the Box's M statistic to determine the holdout sample exceeded both the maximum and proportional
whether we meet the assumption of equal dispersion of the chance values. The prediction accuracy is also judged by press Q
dispersion or covariance matrices (multivariate measure of vari- statistics; the formula for press Q is
ance). At the alpha level 0.01, we reject the null hypothesis for this
analysis, which means the covariance matrices differ between
groups formed by the dependent. Thus, we re-run the discriminant ðN  n  kÞ2
Press Q ¼
analysis requesting the use of separate covariance matrices in N*ðk  1Þ
classification. Consequently, we found that p > 0.01; thus, we fail to
reject the null hypothesis that the covariance matrices do not differ Where:
between groups formed by the dependent (refer Box M results
below). N ¼ Total sample size.
n ¼ Number of observations correctly classified, k ¼ number of
groups.
Test results
Substituting the values for this problem (total unselected
Box's M 5.939
F Approx. 5.895
67cases, 56 correct classifications, and 2 groups), we obtain press
df1 1 Q ¼ [67  (56  2)] ^ 2/67 * (2  1) ¼ 30.22. This value exceeds the
df2 5.198E4 critical chi-square value of 6.63 with degree of freedom one.
Sig. .015 Therefore, we conclude that the prediction accuracy is greater than
that expected by chance. By all three criteria, we would interpret
our model as having an accuracy above that expected by chance.
Furthermore, we compared the accuracy rate for the holdout
Thus, this is a valuable or useful model that supports predictions of
sample, which is 83.6% in our study, by chance accuracy rates. From
the dependent variable.
prior probabilities for groups, we observe that the two groups
To remove the outliers from the data, Mahalanobis distance
contained .522, and .478 proportion of the sample of 134 cases used
scores for each case in the table of casewise statistics were used.
to derive the discriminant model.
Cases with large values of the Mahalanobis Distance from their
The proportional chance criteria to assess model fit is calculated
group mean can be identified as outliers. For large samples from a
by summing the squared proportion of the sample that each group
multivariate normal distribution, the square of the Mahalanobis
represent, we obtain 0.522*0.522 þ 0.478*0.478 ¼ 0.501. Based on
distance from a case to its group mean is approximately distributed
the requirement that model accuracy be 25% better than the chance
as a chi-square statistic with degrees of freedom equal to the
criteria, the standard to use for comparing the model's accuracy is
number of variables in the analysis. The critical value of chi square
1.25  0.501 ¼ 0.626, which is 62.6%. Our model accuracy rate of
with 7 degrees of freedom and an alpha of 0.01 is 18.48. We scan the
83.6% exceeds this standard.
table of casewise statistics to identify any cases that have a squared
The maximum chance criteria uses the proportion of cases in the
Mahalanobis distance greater than 18.48 for the group to which the
largest group, 0.522 in this problem. Based on the requirement that
case is most likely to belong, i.e., under the column labelled ‘Buy’. In
model accuracy be 25% better than the chance criteria, the standard
this particular analysis, we find no cases with a sufficiently large
to use to compare the model's accuracy is 1.25  52.2% ¼ 65.25%.
Mahalanobis distance to indicate that they are outliers.
Our model accuracy rate of 83.6% exceeds this standard.
An additional means of interpreting discriminant analysis re-
As shown above in classification Table 5, the predictive accuracy
sults is to describe each group in terms of its profile using the group
of the model for the analysis sample was 9.9%, and the holdout
means of the predictor variables. These group means are called
sample was 83.6%. The values indicate that the hit ratio of 83.6% for
centroids; these are displayed in the Group Centroids table. In our

Table 5
Classification results.

Intention Predicted group membership Total

1 2

Cases selected Original Count Buy 60 10 70

Not Buy 17 47 64
% Buy 85.7 14.3 100.0
Not Buy 26.6 73.4 100.0
Cases not selected Original Count Buy 33 4 37
Not Buy 7 23 30
% Buy 89.2 10.8 100.0
Not buy 23.3 76.7 100.0

Comparison of goodness of results

Measure Value Hit ratio for the holdout sample

Maximum chance 0.522 83.6

Proportional chance 0.501 83.6
Comparison with Hair et al. (2010) 1.25 times higher than chance 1.25* 0.501 ¼ .626,
1.25*0.522 ¼ .6525
Press Q table value 6.63
Press Q calculated value 30.22**

a. Of the selected original grouped cases, 79.9% are correctly classified.

b. Of the unselected original grouped cases, 83.6% are correctly classified.
**p < 0.01.
54 S. Haryani, B. Motwani / Pacific Science Review B: Humanities and Social Sciences 1 (2015) 49e56

study, the intention to buy has a mean of 0.889, whereas the The unstandardized canonical discriminant function coefficients
intention to not buy produces a mean of 0.972. Cases with scores are used to create the discriminant function (equation). This
close to Centroids are predicted as belonging to that group. equation operates similar to a regression equation. In this study, we
have:
D ¼ (0.52  Immense Efficacy) þ (0.55  Message
Functions at Group Centroids Material) þ (0.461  Consumer Dependency) þ (.399  Supportive
Access) þ (0.235  Professed Security) þ (0.207 
Intention Function
Irrepressible þ (0.275  Escalating Brand) þ (5.374).
1
The discriminant function coefficients b or standardized form
1 .889 beta both indicate the partial contribution of each variable to the
2 .972 discriminate function, controlling for all other variables in the
equation. The coefficients can be used to assess each predictor's
Calculation of the cutting score: unique contribution to the discriminate function and therefore,
For unequal groups, the formula for finding cutting scores is provide information on the relative importance of each variable.
Suggested Research Frame Work

NA ZB þ NB ZA
Zcs ¼
NA þ NB

Where:

Zcs ¼ Optimal cutting score between group A and B.

NA ¼ Number of observations in group A ¼ 70
NB ¼ Number of observations in group B ¼ 64
ZA ¼ Centroid for Group A ¼ .889
ZB ¼ Centroid for Group B ¼ 0.972

Thus, the cutting score in our study is Zcs ¼ 0.083.

A respondent scoring Z score of more than 0.083 will belong to
group one, which means that the consumer intends to buy the
product, whereas those who score less than 0.083 intend not to buy,
as influenced by viral marketing over the internet.
The Standardized Canonical Discriminant Function Coefficients 7. Hypothesis testing
provide an index of the importance of each predictor such as the
standardized regression coefficient. According to this index, the On the basis of this study, we can observe that the F value for
variable consumer dependency has the highest positive score of Immense efficacy was found to be 57.23 with p < 0.01(refer
0.0446. Thus, consumer dependency is the strongest predictor. Table 6). Thus, we accept H1, which indicates that Immense Effi-
However, the variables irrepressible and escalating brand have low cacy is a good predictor of the intention to buy as a result of viral
negative scores and thus, are less successful predictors. Many re- marketing. The study shows that if the online viral marketing
searchers use the structure matrix correlations because they are message is entertaining, informative, cost effective, and time
considered more accurate than the Standardized Canonical saving, and if it contains a wide range of products, it attracts
Discriminant Function Coefficients. The structure matrix table customers to buy the product. The results of this research are
shows the correlations of each variable with each discriminate consistent with the results of the previous research studies con-
function. These Pearson coefficients are structure coefficients or ducted by Tsang et al. (2004), Haghirian et al. (2005),
discriminant loadings; they serve as factor loadings in factor anal- Saadeghvaziri and Hosseini (2011) and Muzaffar and Kamran
ysis. By identifying the largest loadings for each discriminate (2011). According to all these research studies, informativeness
function, the researcher gains insight to develop the function. In is an important predictor of the value of viral marketing and plays
our study, efficacy is ranked first with a highest loading of 0.703. In a significant role in determining consumers' attitude. Entertain-
addition, escalating brand and irrepressible have very low loadings; ment is a significant predictor of the value of web-based adver-
therefore, these variables are not associated with the intention of tisements and hence, determines their efficiency. Interesting and
the customer. The arguments above are supported by the results pleasing ads have a positive impact on consumers' attitude to-
given in Table 6. wards brands. Moreover, entertainment is the most important

Table 6
Discriminant function coefficients.

Independent variable Unstandardized canonical Standardized canonical Discriminant loading (rank) Univariate
discriminant function coefficients discriminant function coefficients structure matrix F ratio

Immense efficacy .520 .397 .703(1) 57.231**

Supportive access .399 .318 .552(3) 35.329**
Proffessed security .235 .231 .498(4) 28.663**
Message material .551 .433 .623(2) 44.884**
Irrepressible .207 .162 .027(6) .085
Consumer dependency .461 .446 .369(5) 15.728**
Escalating brand .275 .258 .001(7) .000
(Constant) 5.374

**p < 0.01.

 S. Haryani, B. Motwani / Pacific Science Review B: Humanities and Social Sciences 1 (2015) 49e56 55

feature that affects consumers' attitude towards mobile adver- 8. Conclusions

tising (Tsang et al., 2004; Bauer et al., 2005).
Second, H2 is also accepted, which indicates that supportive Immense efficacy, message material, supportive access, con-
access is a good predictor of the intention to buy as a result of viral sumer dependency and professed security were found to be sig-
marketing. Easy navigation, rapid circulation, wide reachability, nificant factors that influence the consumer's purchasing
anytime anywhere access, and no geographical boundaries for cir- decisions. As internet and online businesses have become preva-
culation are the important characteristics of viral marketing mes- lent, online viral marketing is rapidly becoming an important and
sage that impact the buying decision of the customer. This study is popular communication tool. Researchers have perceived the sig-
supported by previous research (Zamri and Idris, 2013), which in- nificance of online viral marketing and studied the impact on
vestigates the consumer's purchasing intention of buying the consumer purchasing decisions. Researchers examined the effec-
product using six different independent variables of the perceived tiveness of online viral marketing as a decisive factor for con-
ease of use, perceived usefulness, information privacy and security, sumers' purchasing behaviour (Gretzel and Yoo, 2008; Pan et al.,
product and service quality, social influences and the role of 2007). However, the diverse findings of the online viral market-
experiential online shopping motives. However, the result solely ing research have yet to be synthesized to improve the in-depth
supports the perceived ease of use as being significant to the con- understanding of the online viral marketing mechanism and to
sumer's intention to purchase. integrate the influential factors that affect online viral marketing
Hypothesis H3 is also accepted, which indicates that professed performance. Thus, this study intended to review and analyse the
security is a good predictor of the intention to buy as a result of viral factors and hypothesized relationships in previous studies that
marketing. This finding supports the assumption that these sites relate to online viral marketing and develop the conceptual
are trustworthy and provides information from credible sources; in framework to represent the online viral marketing mechanism.
addition, people believe that these sources do not lead to unau- Marketers can also find the market potentiality of their brand in a
thorized association. In our study, H4 is also accepted, which in- new market through this research in addition to determining
dicates that message material is a good predictor of the intention to problems in the existing market in terms of the predictors so that
buy as a result of viral marketing. Sufficient information and clarity appropriate marketing policies can be implemented to tap the
in the message attracts the customers. We reject H5, which in- market. It has been established that the five abovementioned
dicates that Irrepressible is a good predictor of the intention to buy variables, which were included in the model, make it possible to
as a result of viral marketing. This finding shows that people do not discriminate between the two groups of customers. Effective
believe that online marketing circulates beyond limits, and they do execution of viral marketing campaigns can create an instanta-
not believe that the reviews and comments by the consumer can neous buzz and help boost the promotion of brands, products and
create disaster. Hypothesis H6 is also accepted, which indicates that services.
consumer dependency is a good predictor of the intention to buy as
a result of viral marketing. The study supports the argument that
viral marketing is dependent on the consumer because it is suc- 9. Limitation
cessful solely when someone forwards or shares the information
with his or her peer group. In addition, people are willing to for- The sample used for this study is relatively small and was
ward or share the information if they are provided with incentives. limited to a specific geographic region. The study was conducted by
The information age has enhanced and accelerated the ability for taking the views of the 201 respondents. A self-structured ques-
everyday Internet users to communicate and spread the word tionnaire was designed for the purpose. Respondents for the survey
regarding brands, products, experiences and events. Although were taken from Indore city. This study could have been more
digital media makes sharing information and content more acces- reliable and would have yielded different results if the target de-
sible, the motivation to share is essential for the virality of a mes- mographic was broader. Similar to every study that involves human
sage to occur. Thus, viral success is dependent on a consumer's feedback, there is always large space for bias. Respondents could
willingness to share a message to others within his or her social have provided false information due to the belief that it may reflect
circle (Tang, 2006). Finally, H7 is rejected, which indicates that their personality. However, increasing the sample size and
escalating brand is a good predictor of the intention to buy as a including respondents from varying demographics will remove this
result of viral marketing. limitation.

Annexure 1. Factor analysis details

Factor Items Factor load Initial Eigen values % of variance

Immense efficacy Detailed information regarding the product can be assessed anytime .783 7.294 19.693
It saves time .746
It is cost effective .717
Wide range of products can be displayed .616
It is entertaining .609
Relevant information is provided .607
Supportive access Wide reach ability .735 13.170
Rapid diffusion of information .694
No geographical boundaries for promotion .683
Provides easy navigation on web .653 2.554
Enlarged images of products can be assessed. .582
Professed security VM can lead association to unauthorized companies .703 10.289
The site that provides the information is virus free .619
Information is provided by credible sources .611 2.358
VM is trustworthy .610
The information that is downloaded is free of viruses .500
(continued on next page)
56 S. Haryani, B. Motwani / Pacific Science Review B: Humanities and Social Sciences 1 (2015) 49e56

(continued )

Factor Items Factor load Initial Eigen values % of variance

Message material Sufficient information is provided .778 10.180

The message is clearly understood .743 1.509
Irrepressible The communication is uncontrolled .741 1.204 6.503
Frequent messages can lead to mail box traffic .659
Consumer dependency People share the information when incentives are offered .805 6.029
VM is effective solely when consumers share it with others .518 1.090
Escalating brand It is a useful marketing tool for creating brand awareness .821 1.032 5.142

References Knight, C.M., 1999. Viral marketing e defy traditional methods for hyper growth.
Broadwatch Mag. 13 (11), 50e53.
Kothari, C.R., 2004. Research Methodology-Methods & Techniques, second ed. New
Alba, Joseph W., Hutchinson, J. Wesley, 1987 March. Dimensions of consumer
Age International.
expertise. J. Consum. Res. 13, 411e454.
Kozinets, Robert V., Valck, Kristine de, [Link], Andrea, Wilner, Sarah J.S., 2010.
Arndt, J., 1967. Word-of-mouth advertising and informal communication. In:
Networked narratives: understanding word-of-mouth marketing in online
Cox, D.F. (Ed.), Risk-Taking and Information-Handling in Consumer Behavior.
communities. J. Mark. 74, 71e89.
Harvard University Press, Boston, pp. 188e239.
McGuigan, D., 2008. 70% of Online Shoppers Read Multiple Product Reviews. http://
Bauer, H.H., Reichardt, T., Barnes, S.J., Neumann, M.M., 2005. Driving consumer
[Link]/2008/02/25/70-of-online-shoppers-read-multiple-
acceptance of mobile marketing: a theoretical framework and empirical study.
product-reviews/.
J. Electron. Commer. Res. 6 (3), 181e192.
Muzaffar, F., Kamran, S., 2011. SMS advertising: youth attitude towards perceived
Bayus, B.L., 1985. Word-of-mouth: the indirect effects of marketing efforts. J. Advert.
informativeness, irritation and credibility. Interdiscip. J. Contemp. Res. Bus. 3
Res. 25, 31e39.
(1), 230e245.
Blattberg, R.C., Deighton, J., 1991. Interactive marketing: exploiting the age of
Palka, W., Pousttchi, K., Wiedemann, D.G., 2009. Mobile word-of-mouth e a
addressability. Sloan Manag. Rev. 33 (1), 5e14.
grounded theory of Mobile viral marketing. J. Inf. Technol. 24 (2), 172e185.
Bruyn, A.D., Lilien, G.L., 2008. A multi-stage model of word-of-mouth influence
Pan, B., MacLauri, T., Crotts, J., 2007. Travel blogs and the implications for destina-
through viral marketing. Int. J. Res. Mark. 25, 151e163.
tion marketing. J. Travel Res. 46, 35e45.
Carl, W.J., 2008. The role of disclosure in organized word-of-mouth marketing
Ramayah, T., 2006. Doing e-research with e-library: determinants of perceived ease
programs. J. Mark. Commun. 14 (3), 225e241.
of use of e-library. Int. J. Technol. Knowl. Soc. 1 (4), 71e82.
Cramer, D., 2003. Advanced Quantitative Data Analysis. Open University Press,
Ramayah, T., Aafaqi, B., Ignatius, J., 2004. E-library usage among students of a public
Buckingham.
University in Malaysia. In: The Proceedings of the International Conference on
Chu, S.C., Kamal, S., 2008. The effect of perceived blogger credibility and argument
Management Education, 28the29th June, Kuala Lumpur, Malaysia.
quality on message elaboration and brand attitudes: an exploratory study.
Reyck, B.D., Degraeve, Z., 2003. Broadcast scheduling for mobile advertising. Oper.
J. Interact. Advert. 8 (2), 26e37.
Res. 51 (4), 509e517.
Cruz, D., Fill, C., 2008. Evaluating viral marketing: isolating the key criteria. Mark.
Richins, M.L., 1983. Negative word-of-mouth by dissatisfied consumers: a pilot
Intell. Plan. 26 (7), 743e758.
study. J. Mark. 47 (Winter), 68e78.
De Bruyn, A., Lilien, G.L., 2004. A Multi-stage Model of Word-of-mouth through
Richins, M.L., Root-Shaffer, T., 1987. The role of involvement and opinion leadership
Electronic Referrals. eBusiness Research Centre Working Paper, The Pennsyl-
in consumer word-of-mouth: an implicit model made explicit. In: Houston, M.J.
vania State University, State College, PA.
(Ed.), Advances in Consumer Research, vol. 15, pp. 32e36.
Derbaix, Christian, Vanhamme, Joelle, 2003. Inducing word-of-mouth by eliciting
Ridings, C.M., Gefe, D., Arinze, B., 2002. Some antecedents and effects of trust in
surprise epilot investigation. J. Econ. Psychol. 24, 99e116.
virtual communities. J. Strat. Inf. Syst. 11 (3e4), 271e295.
Dobele, A., Toleman, D., Beverland, M., 2005. Controlled infection! Spreading the
Roselius, T., 1971. Consumer rankings of risk reduction methods. J. Mark. 35, 56e61.
brand message through viral marketing. Bus. Horizons 48 (2), 143e149.
Saadeghvaziri, F., Hosseini, H.K., 2011. Mobile advertising: an investigation of factors
Engel, J.E., Blackwell, R.D., Kegerreis, R.J., 1969. How information is used to adopt an
creating positive attitude in Iranian customers. Afr. J. Bus. Manag. 5 (2),
innovation. J. Mark. Res. 9, 3e8.
394e404.
Ferguson, R., 2008. Word of mouth and viral marketing: taking the temperature of
Sheth, J.N., Venkatesan, M., 1968. Risk-reduction processes in repetitive consumer
the hottest trends in marketing. J. Consum. Mark. 25 (No. 3), 179e182.
behavior. J. Mark. Res. 5, 307e310.
Frost, D., 2007. Viral Campaign Spreading [online], SFGate. Available from: http://
Smith, T., Coyle, J.R., Lightfoot, E., Scott, A., 2007. Reconsidering models of influence:
[Link]/[Link]/[Link]?file¼/c/a/2007/06/09BUG9VQC1DT
the relationship between consumer social networks and word-of-mouth
(accessed 24.02.08.).
effectiveness. J. Advert. Res. 47 (4), 387e397.
Goldenburg, J., Libai, B., Muller, E., 2001. Talk the network: a complex systems
Spero, I., Stone, M., 2004. Agents of change: how young consumers are changing the
look at the underlying process of word-of-mouth. Mark. Lett. 12 (3),
world of marketing. Qual. Mark. Res. Int. J. 7, 153e159. Source: Emerald [online].
211e223.
Steyer, A., Garcia Bardidia, R., Quester, P., 2007. Mode lisation De La Structure Sociale
Gretzel, U., Yoo, K., 2008. Use and impact of online travel reviews. In: O’Connor, P.,
Des Groupes De Discussion Sur Internet: Implications Pour Le Contro ^le Du
Hopken, W., Gretzel, U. (Eds.), Information and Communication Technologies in
Marketing Viral. Rech. Appl. Mark. 2 (3), 29e44.
Tourism. Springer-Verlag, Wien/NewYork, pp. 35e46.
Tang, Yu Chieh, 2006. The Effects of Commercial Intention, Humour, and Media
Haghirian, P., Madlberger, M., Tanuskova, A., 2005. Increasing advertising value of
Richness on Receivers' Forward Intention in Internet Viral Marketing Cam-
mobile marketing- an empirical study of antecedents. In: Proceedings of 38th
paigns. National Central University Library Press, Jhongli City.
Hawaii International Conference on System Sciences, Big Island, HI, USA.
Taylor, J.W., 1974. The role of risk in consumer behavior. J. Mark. 38, 54e60.
January 3e6, 2005.
Taylor, C.R., Franke, G.R., Bang, H., 2006. Use and effectiveness of billboards per-
Hair, J., Black, W.C., Babin, B., Anderson, R., 2010. Multivariate Data Analysis.
spectives from selective-perception theory and retail-gravity models. J. Advert.
Prentice Hall, Upper Saddle River.
35 (4), 21e34.
Helm, S., 2000. Viral marketing e establishing customer relationship by, word-of-
Thomas, G.M., 2004. Building the buzz in the hive mind. J. Consum. Behav. 4 (1),
mouse. Electron. Mark. 10 (3), 158e161.
64e72.
Hoch, Stephen J., Deighton, John, 1989 April. Managing what consumers learn from
Tsang, M.M., Ho, S.C., Liang, T.P., 2004. Consumer attitudes toward Mobile adver-
experience. J. Mark. 53, 1e20.
tising: an empirical study. Int. J. Electron. Commer. 8 (3), 65e78.
Jurvetson, Steve, Draper, Tim, 1997. Viral Marketing Found at. [Link]
Vilpponen, A., Winter, S., Sundqvist, S., 2006. Electronic word-of-mouth in online
files/[Link].
environments: exploring referral network structure and adoption behavior.
Kaikati, A.M., Kaikati, J.G., 2004. Stealth marketing: how to reach consumers sur-
J. Interact. Advert. 6 (2), 71e86.
reptitiously. Calif. Manag. Rev. 46 (4), 6e22.
Welker, C.B., 2002. The paradigm of viral communication. Inf. Serv. Use 22 (1), 3e8.
Katz, E., Lazarsfeld, P., 1955. Personal Influence. Free Press, New York.
Woerndl, M., Papagiannidis, S., Bourlakis, M., Li, F., 2008. Internet-induced mar-
Kee, T., 2008. Majority of Online Shoppers Check at Least Four Reviews Before
keting techniques: critical factors in viral marketing campaigns. Int. J. Bus. Sci.
Buying. [Link]
Appl. Manag. 3 (1), 33e45.
san&s¼76727&Ni¼39482&p¼368788.
Zamri, N.B., Idris, I., 2013. The effects of attitude, social influences and perceived
Kelly, L., Kerr, G., Drennan, J., 2010. Avoidance of advertising in social networking
behavioural. In: 3rd International Conference on Management Proceedings,
sites: the teenage perspective. J. Interact. Advert. 10 (2), 16e27.
pp. 124e144. Penang, Malaysia. ISBN: 978-967-5705-11-3. Website: www.
Kent, R.J., Allen, C.T., 1994 July. Competitive interference effects in consumer
[Link].
memory for advertising: the role of brand familiarity. J. Mark. 58, 97e105.

View publication stats