i
The Interaction between Electronic Word of Mouth
Communications on Willingness, Trust, and Tie
Strength
研 究 生︰楊孝康
Student︰Conna Yang
指導教授︰張力元
博士
Advisor︰Dr. Charles V. Trappey
國立交通大學
經營管理研究所
碩士論文
A Thesis
Submitted to Institute of Business and Management
College of Management
National Chiao Tung University
in Partial Fulfillment of the Requirements
for the Degree of
Master of Business Administration
June 2010
Taipei, Taiwan, Republic of China
中華民國 九十九 年 六 月
ii
探討網路口碑、信任、溝通意願、與關係強度之間的
關係研究
學生：楊孝康 指導教授：張力元 博士 國立交通大學經營管理研究所碩士班 摘 要 口碑一向是企業在打響一個產品的知名度最省錢，也最能快速達到成效的一個手 段。如今在網路發達的社會，除了傳統的面對面的口碑傳播，我們更想要了解的是網路 口碑的影響跟傳遞。本研究對於人與人之間分享網路口碑的原因作了一個探討，去測量 個人的信任程度，是否會影響他願意去溝通以及對話的結果。另外，本研究也將關係強 度當作一個重要的變數納入研究範圍。我們透過問卷調查，針對在台北縣市內的大學生 作了一項調查，有效問卷樣本為 171 份。本研究透過 SPSS 17 之回歸分析、信度分析， 與 Pearson 相關係數分析進行資料的分析與假說檢定。 研究結果顯示，信任與溝通意願有顯著的關係。而溝通意願與網路口碑活動及正面 口碑亦有顯住的關係。這代表我們假設的溝通意願的確是扮演中介變數的功能。同時， 關係強度與正面口碑和負面口碑皆有顯著的關係。這結果顯示個體之間的關係程度會影 響是否會主動去分享與傳播對於產品的心得。 關鍵字：網路口碑、信任、溝通意願、關係強度iii
The Interaction between Electronic Word of Mouth
Communications on Willingness, Trust, and Tie Strength
Student：Conna Yang Advisor：Dr. Charles V. Trappey
Institute of Business and Management
National Chiao Tung University
Abstract
Word of mouth is one of the quickest and most efficient ways to help increase a product’s awareness among consumers. Nowadays, people all have access to the internet and this has become another cable for information and experience sharing. In addition to traditional word of mouth, electronic word of mouth has been an extremely important issue due to the heavy use of internet in modern society. Therefore, this is the research topic that we aim to study and discuss. Our research aims to study if people share electronic word of mouth due to his degree of willingness to communicate by measuring an individual’s trust. Furthermore, we also include tie strength as one of our variables to study its influence on electronic word of mouth. Using paper based questionnaires, we chose college students in Taipei as our participants. We use the statistical analysis software SPSS 17 to conduct Regression Analysis, Reliability Test, and Pearson Correlation Coefficient with a valid sample size of 171.
Our research results support our hypothesis of trust and willingness to communicate having a significant relationship. In addition, willingness to communicate and electronic word of mouth activity and positive electronic word of mouth also have significant relationships. This means that the most important part of our framework, willingness to communicate as a mediator between trust and electronic word of mouth is accepted.
iv ACKNOWLEDGMENTS
First of all, I want to thank my advisor, Dr.Charles Trappey for being such an intelligent and kind teacher. It was truly a pleasure and an honor to be your advisee. Next, I want to thank my parents, Chyan and Anna Yang for their moral support and confidence in me. And of course, many thanks to all my friends who have always been there for me and believed I could indeed achieve my goals. Some have said that the process of research can be lonesome and quite dull sometimes, but I am very glad that this has never been the case for me. For me, “research is the process of satisfying one’s curiosity.” I have greatly enjoyed my masters’ program which has given me so much in terms of happiness and academic knowledge.
I realized that the most important thing is to believe in yourself and never lose faith of what you can achieve. “Hope does not disappoint us, because God has poured out his love into our hearts (Romans 5:3-5).” Dear God, thank you.
Conna Yang National Chiao Tung University Institute of Business and Management
v TABLE OF CONTENTS
ABSTRACT ... ii
TABLE OF CONTENTS ... …v
LIST OF TABLES ... vi
LIST OF FIGURES ... vii
CHAPTER 1 INTRODUCTION ... 1
1.1 Research Motivation ... 1
1.2 Research Objectives ... 2
1.3 Research Process ... 3
CHAPTER 2 LITERATURE REVIEW ... 4
2.1 Definition of Word of Mouth (WOM) ... 4
2.2 Definition of Electronic Word of Mouth (eWOM) ... 9
2.3 Definition of Willingness to Communicate (WTC) ... 11
2.4 Trust ... 20 2.5 Tie Strength…… ... 25 CHAPTER 3 METHODOLOGY ... 27 3.1 Research Framework………...…………...27 3.2 Research Hypotheses……….29 3.3 Sampling Design………30 3.4 Questionnaire Design……….…31 3.5 Pretest……….40
CHAPTER 4 RESEARCH RESULTS AND ANALYSIS ... 45
4.1 Reliability Analysis………45
4.2 Regression Analysis……….………..46
4.3 Pearson Coefficient Analysis……….47
CHAPTER 5 CONCLUSION AND DISCUSSION ... 50
5.1 Conclusion……….50
5.2 Suggestions and Managerial Implications……….52
5.3 Suggestions for Future Research………53
5.4 Research Limitations………..53
REFERENCES ... 54
APPENDIX A………...60
vi LIST OF TABLES
Table 2-1. Definitions of Word of Mouth………P.8 Table 2-2. A Comparison of traditional word of mouth and electronic word of mouth…... P.10 Table 2-3. Comparison of different types of electronic word of mouth………... P.11 Table 2-4. Definition of willingness to communicate………...P.20 Table 2-5. Definition of Trust………P.22 Table 3-1. Demographic variables……… P.32 Table 3-2. Measuring an individual’s electronic word of mouth items……….P.33 Table 3-3. Measuring an individual’s interpersonal trust items………....P.34 Table 3-4. Measuring an individual’s willingness to communicate items………P.38 Table 3-5. Measuring an individual’s tie strength with another person in eWOM items…..P.40 Table 3-6. Reliability statistics of first pretest………...P.42 Table 3-7. Reliability statistics of second pretest………..P.43 Table 4-1. Cronbach Alpha Analysis Results………...……….P.45 Table 4-2. Regression Analysis Results……….P.46 Table 4-3. Pearson Coefficient Analysis of Variables……… P.47
vii LIST OF FIGURES
Figure 1-1. Research Process………..……….P.3 Figure 2-1. The Two Step flow of Word of Mouth……….P.6 Figure 2-2. Model of variables influencing WTC……….P.14 Figure 2-3. Relationship of trust and willingness to communicate...………P.21 Figure 2-4. The Interdisciplinary Model of High-Level Trust Constructs………P.24 Figure 3-1. Proposed research framework……….P.27 Figure 5.1 Correlations and Significance Relationship proven to exist………P.50
CHAPTER 1
INTRODUCTION
In Chapter 1, we will introduce the research background and give a brief introduction of
the constructs that are studied in our study. First, the research motivation will be defined;
explaining what drives us to conduct this study. Then, we state our research objectives which
describe the main aim of our research. And finally, the process of our research is divided into
8 different steps and depicted as a graph for readers to understand.
1.1 Research Motivation
The Internet provides a means of has become a new way of communication that
substitutes of traditional ways of promoting products, brands, and services. Through Word of
Mouth (WOM) communication, passing information and shared experience is well known as
an efficient way of spreading commercial messages. Since, “pre-usage attitudes about a
product can be influenced by WOM communications.” (Bone, 1995), there is research interest
in measuring and influencing WOM effects. WOM effects are even more important with the
emergence of Internet. This research evaluates Internet based sources of WOM
communications and proposes a study to measure and influence WOM commercial
communications. Varying among different industries, the strength of WOM might have
diverse effects yet the outcome is undoubtedly able to influence the purchase intentions of
consumers. Due to the advanced technology nowadays, the internet has become a new
platform where electronic word of mouth takes place. Different from traditional word of
mouth where the information is passed on from person to person, electronic word of mouth is
eliminated.
1.2 Research Objectives
This study intends to analyze interpersonal trust, willingness to communicate, and the tie
strength among individuals. Also, we aim to examine whether or not willingness to continue
is the mediator between trust and electronic word of mouth. If this is proven to be true, then in
the future corporations will have a more concrete idea of how to control electronic word of
1.3 Research Process
Figure 1-1. Research Process Develop Research Objective
Literature Review
Create Research Framework
Research Methodology
Questionnaire Design
Data Gathering & Statistical Analysis
Research Results
CHAPTER 2
LITERATURE REVIEW
Chapter 2 is the foundation of our thesis. It includes a very thorough review of all the
past studies on the constructs concerning our study. Through this review of literature, we will
have a clear insight of how these variables are established and defined. By understanding each
construct and its influence it has with each other, we can then set our research framework.
2.1 Definition of Word of mouth
Word of mouth can be traced back to ancient times of many different cultures and
countries when there was not yet an effective way for both government and social
communities to enable information or messages between people. Individuals relied on this as
one of the main sources to gain access to information and news. Up to this day, word of
mouth is still one of most familiar function anyone can use to spread and gain information on
certain issues and topics. It is defined as “informal communication directed at other
consumers about the characteristics, ownership, or their sellers (Westbrook, 1987).” From the
marketing and advertising perspective, word of mouth is no doubt one of the most important
mechanisms that have an impact over the purchase intentions of consumers. The production
of output WOM is thought to be an outcome of customer experiences with a product or
services (Buttle, 1998). Both positive and negative Word of mouth has influences on whether
or not a potential consumer would choose to buy the product. The most obvious example is
when we ask suggestions from friends or family and the information given could easily
influence our purchase decision. A disappointed customer’s reactions can be categorized into
known to the suppliers or company (Hirschman, 1970). Out of these three, complaining to
others, also known in our literature as “negative WOM,” is the most destroyable for a
company’s reputation and sales: Consumer purchase intentions lead to the search of shared
experience which greatly influences the outcomes of their behavior, that is, to buy the product
or service or not. Customers often depend on word of mouth to decide whether or not they
should patronize. One of the most direct ways for consumers to find such shared information
is through the search of word of mouth. It is mostly free and easy to access. Furthermore,
consumers seem to find it hard to keep such information to themselves and they tend to like to
share their experiences with others. According to statistical data, individuals usually only
keep 10% of our emotional experiences, whether good or bad, to ourselves; the majority will
be shared through the sharing of our conversation (Flynn, Goldsmith, and Eastman, 2006).
Surprisingly, after acquiring word of mouth, sometimes consumers even exclude their own
opinions and private information they already have and prefer the information given by others
(Godes and Mayzlin, 2004). In other words, the effect and influence of word of mouth is
stronger when the consumer is faced with an ambiguous experience which they have to make
an immediate or delayed decision (Bone, 1995). This is one of the main reasons why word of
mouth has become one of the major concerns of corporate organizations in hope to gain more
customers. The advice from friends, family, or other individuals whom have persuasive power
serves as an important factor for them. The significance and influence of word of mouth has
long been documented. “Word of mouth communications (WOM) is an interpersonal
traditional depiction of word of mouth is divided into a model of 2 parts (Figure 1.),
information given by the corporate and content based on their experience will be passed on
from opinion leaders on to other potential consumers (Haywood, 1989). Yet word of mouth
does not necessarily rely on opinion leaders to be passed on, ordinary consumers can also
generate word of mouth. This research will focus on step 2, the spread of information and
subjective point of view by consumers.
Figure 2-1. The Two Step flow of Word of Mouth
Reference: Haywood, K. Michael (1989), “Managing Word of mouth Communications” The
Journal of Services Marketing, Vol.3, No.2, 55-67
According to Godes and Mayzlin (2004), the study of WOM can be categorized into
three different streams: (1) a driver of consumer purchase behavior, (2) a result of consumer
behavior, (3) the social structure in the flow of WOM. This study will focus on how word of
mouth works as the driver of purchase behavior.
Word of Mouth: A two step flow hypothesis
Step 1
Step 2
Marketing Activities Target Market
Initial consumers, adaptors, evaluators
A decision maker looks at the former decision maker that has already chosen as a
foundation and base for his own choice. This is a rational choice for him since the former
decision maker must have some sort of information that the latter one lacks. Therefore, when
a person does not have sufficient information to make a perfect choice, this is the most often
chosen way to make a decision. Benerjee proposed a model on the ‘herding’ phenomenon
which quite depicts the situation of reliance and trust of consumers on word of mouth, “if an
agent has a signal, then he follows that signal, unless someone before him has already
followed someone else. In that case, he follows suit” (Banerjee, 1992). For example, if we had
to choose between two different restaurants and had neither information nor any sort of clue
which was better, then normally we would choose to observe the choices of past customers.
Assuming they made their choice according to some source of information, we then would
also make the same decision. In contrast, if one person already had preferred to choose
restaurant A over restaurant B, and yet knew that the customer in front of her chose B, then
the struggle between the two restaurants would be ruled out, whereas he would then pick B,
trusting and following the former consumer’s choice.
In our study, we believe that word of mouth consists of three different dimensions that
make the construct complete which are word of mouth activity, positive word of mouth, and
negative word of mouth. These three constructs can be seen as different aspects to a single
and more generalized word of mouth construct that should be separately studied due to a
certain level of independency each construct possesses (Harrison-Walker, 2001). Word of
between consumers on products. Each individual will choose according to his own preference,
character, and other antecedents the content and frequency of sharing electronic word of
mouth. Moreover, Richins (1984) argued that people are more likely to spread negative
attitudes to others compared to positive attitudes. When unsatisfied or disappointed with a
certain product, customers are likely to spread negative word of mouth by actions such as
giving bad reviews or furthermore, even advising other customers not to make the same
purchase.
Table 2-1. Definitions of Word of Mouth
Scholar Definition of word of mouth
Westbrook ,1987 Informal communication directed at other
consumers about the characteristics,
ownership, or their sellers
Haywood, 1989 Information given by the corporate and
content based on their experience passed on
from opinion leaders on to other potential
consumers
Bone, 1995 An interpersonal communication in which
none of the participants are marketing
sources
Buttle, 1998 An outcome of customer experiences with a
product or services
Stern, 1994 The exchange of ephemeral oral or spoken
Stern, 1994 recipient who communicate directly in real
life…(also)consumers are not assumed to
create, revise, and record pre-written
conversational exchanges about products or
services
Armdt, 1967 Oral, person to person communication
between a receiver and whom the receiver
perceives as non-commercial regarding a
brand, product, or service.
2.2 Electronic Word of Mouth (eWOM)
New media offers different ways of the spread of word of mouth. The Internet in
particular has been the main portal of electronic word of mouth. It has greatly changed the
marketing communications. Similar to traditional word of mouth, electronic word of mouth
has even higher reliability and level of credit than other forms of marketing information and
strategies on the web (Bickhart and Schindler, 2001). The rise of the Internet makes the access
of information available more easily with different choices of platforms that consumers can
choose from. Blogs, forums, and websites are some examples of where electronic word of
mouth can be spread. This is different from the traditional spread of word of mouth since the
barriers such as distance, time, and cost of expense on gaining information is decreased or
eliminated. The Internet enables consumers to share their opinions on, and experiences with,
goods and services with a multitude of other consumers; that is, to engage in electronic word
gather in virtual forms of social communities and exchange opinions and share advice where
the flow of content is free and wide. Consumers gather shared experiences posted on websites
to learn more about a product before making a purchase (Doh and Hwang, 2009). Electronic
word of mouth not only benefits consumers, but also is a new tool for players in the business
and advertising industry. Media players can freely interact with consumers and advertisers
through the new media. Electronic word of mouth shortens the distance between consumers
and corporations and can reach out to a larger group of audience (Hung and Li, 2007).
Table 2-2. A Comparison of traditional word of mouth and electronic word of mouth
Different types of Word of Mouth Communication platforms
Traditional word of mouth P2P
Electronic word of mouth
Bulletin Board System, Blogging, Micro-blogging,
Instant Messaging
In traditional word of mouth, the source of information is more visible and people
communicate in person whereas in electronic word of mouth, there are many more
possibilities for information to be passed on. One of the most common used electronic
platforms in Taiwan is the Bulletin Board System, or more commonly known as “BBS” for its
initial abbreviations. Textual messages are the only kind of communication available in this
platform. The main viewers are students and often gather together on different boards to share
their purchase experience on certain products ranging from digital cameras to makeup and
Internet sites (Nardi et al. 2004). New forms of micro-blogging that simplify the functions and
make it easier for users to share their status and information is a new form of communication
in which people can post short messages or upload pictures through mobile phones or the
internet. The main difference between traditional blogging and micro-blogging is the later
offers an even faster way of sharing information by encouraging users to keep their posts
short. Not only does this lower the required time to share information, it also increases the
frequency of posting. A traditional blog might have 1 or more articles per day, whereas in
micro-blogging, users tend to renew their status several times a day. In addition, instant
messaging is a type of technology on the Web that allows users to send and receive mainly
short text based messages and check to see who of their friends are also online and available
(Cameron et al. 2004).
Table 2-3. Comparison of different types of electronic word of mouth
Different types of eWOM Example(s)
Bulletin Board System (BBS) PPT
Blogs Wretch, Yam, Blogger
Micro-blogs Twitter, Facebook, Plurk
Instant Messaging MSN, Yahoo Messenger, AIM
2.3 Definition of Willingness to Communicate (WTC)
Each person has different levels of tendency whether or not they like to speak, write, or
in any other form communicate with other people. The willingness to communicate is a
tendency to approach or avoid communication with others. This construct most originally
originated from Burgoon’s 1976 research on the unwillingness to communicate which was
based on factors such as introversion, lack of communication competence, alienation, anomie,
and communication apprehension (MacIntyre, 1994). Mortenson, Arntson, and Lustig (1977)
used this construct to farther develop it into the construct “willingness to communicate” to
measure predisposition towards verbal communication. McCroskey and Baer gave a more
detailed description of this construct as “the intention to initiate communication when given
the opportunity.” The word “intention” here should be highlighted and emphasized since this
slight change in definition brings this construct makes it more complete. Past research has
shown that the willingness to communicate is also positively related to how much a person
chooses to listen and comprehend with others. The willingness to communicate is a
personality variable that decides how much or how little a person likes to talk. The
willingness to communicate will result in good personal images in the work place. Yet, it
varies in different situations, from the mood of the speaker to whom he is speaking to. More
certainty leads to a higher level of willingness to communicate and develop interpersonal
relationships (McCroskey, 1985).
The two different layers of willingness include L1 and L2. L1 is our native language,
whereas L2 the second or foreign language one learns (MacIntyre et al. 1998). ManIntyre and
other scholars focus more on L2 as the area of research for willingness to communicate, yet
there has been a gap of research between the willingness to communicate and spread of word
communicate was first conceptualized based on L1 by McCroskey and Baer in 1985.
Although this construct has more commonly been studied as personality trait that might vary
across different situations rather than a situational variable, we do not have to limit the
construct and its character.
Willingness to communicate is a study that has been researched thoroughly under
different constructs. According to McCroskey (1992) these different constructs, the
willingness to communicate can be divided into three different groups that focus on various
aspects. First group of constructs focuses on the anxiety or apprehension about
communication. Second group of constructs focuses on the frequency of talking. The third
L 2 Us e W illin g n es s to C o mmu n ic ate D es ir e to C o mmu n ic ate w it h a s p eci fi c p er so n S ta te C o mmu n ic ativ e S el f C onf ide nc e L2 S el f C onf ide nc e Int er gr oup A ttitu d es C o mmu n ic ativ e C o m p et en ce L ay er 1 L ay er 2 L ay er 3 L ay er 4 L ay er 5 L ay er 6 C om m uni ca ti on B eha vi or B eha vi or al I nt ent ion S itu ate d A n teced en ts M o tiv atio n al P rope ns it ie s A ffe ct iv e-C ogni ti ve C ont ext P er so n ality Int er gr oup M o tiv atio n S o cia l S itu atio n Int er gr oup C lima te In ter p er so n al M o tiv atio n S o cia l & Indi vi dua l C ont ext F igur e 2-2. M ode l of va ri abl es i nf lue nc ing WT C R ef er en ce : M ac Int yr e P .D . a nd Z . D or ny ei a nd R . C le m ent a nd K .A .N oe ls ( 1998) “ C onc ept ua li zi ng W illin g n es s to C o mmu n icat e i n a L 2: A S it ua ti ona l M ode l of L 2 C onf ide nc e and A ff il ia ti on ” T he M ode rn L anguage J our nal 82 545 -5 62
This model with 6 layers represents the two basic structures that underlie willingness to
communicate. Layers 1 to 3 describe how individuals will react under certain circumstances
or situations. On the other hand, layers 4 to 6 represent the stable influencing factors that
influence our willingness to communicate.
Layer I
In the top layer, layer 1, we have the use of a second language, L2. Communication behavior
using the second language can be observed by the frequency of how an individual chooses to
engage in behaviors that use L2. Reading magazines and articles in L2 language, or choosing
to speak up in a L2 class are signals that the L2 education has succeeded in this level of
willingness to communicate.
Layer II
An interesting example can be used to explain the second layer, willingness to communicate.
If we have a group of students taking a foreign language class, those who choose to answer or
respond to the teacher’s question, regardless whether or not they are chosen to answer, but as
long as they raise their hands, this is a demonstration of their willingness to communicate.
Layer 2 tends to explore the urge of raising one’s hands to express oneself in L2. A
combination of different reasons behind this motivation has been given: self-confidence,
affiliation and control motives.
Layer III
Layer 3 proposes 2 different precursors of willingness to communicate, desire to
and control motives can influence one’s desire to communicate. Yet, it is not always definite
that these two factors will be potent every time. According to research in the psychology field,
affiliative motives usually exist when those we communicate with are close to us, have more
common attributes, or are more physically attractive to us (Lippa, 1994). In other words,
when we encounter a person who attracts us, we will be more willing to communicate with
him. According to Clement, the two key constructs that are the foundation of state
communicative self-confidence are “perceived competence” and “lack of anxiety”. Perceived
competence means that “the feeling of competence one has to communicate efficiently at a
particular moment” (MacIntyre et al. 1998). When one has a confident feeling that he will be
able to talk and communicate in a certain situation, that signifies he is in a state of perceived
competence. The state of anxiety is a mixture of tension and apprehension that also arouses
the autonomic nervous system (Spielberger, 1983). Any cause that triggers the nervous
system and raises the sense of anxiety will cause the confidence in communication to decrease,
which will lead to the weakening of one’s willingness to communicate.
Layer IV
Layer 4 includes three different variables which are interpersonal motivation, intergroup
motivation, and L2 confidence and are all motivational propensities. Motivational
propensities are defined as “stable individual differences that can be applied in different
situations” (MacIntyre et al. 1998). Based on cognitive and affective contexts of one’s
interpersonal interaction with others, motivational propensities decide the degree of one’s
be categorized under two different motives, control and affiliation. As a motivational aim,
control restricts the cognitive, affective and behavioral freedom of the other person. Often this
type of communication occurs more in hierarchal relationships such as workplaces or
situations where one has more authority over the other. For example, during the beginning of
exams, teachers communicate in order to let students understand the rules and restrictions on
exams. Different from control’s task oriented situations, affiliation originates from the desire
to establish a relationship with another. Although the motive to control cannot be entirely
eliminated, yet the degree is limited. The degree of affiliation is influenced by different
personal characteristics such as attractiveness, similarity, yet it is most often relevant with
one’s personality. Past research on the differences of personality points out that “the
personality trait of an individual leads to the different preference of the need for affiliation (or
control).” In addition, intergroup relations can also be divided into control and affiliation. The
orientation and definition is the same except for the interlocutors in this dimension are groups
of people instead of individuals. In Layer 4, interpersonal and intergroup relations are more
concerned with the affective and social aspects of motivational propensities, whereas L2
self-confidence studies the relationship between an individual with L2. The two elements that
make up L2 self-confidence is one’s perception of L2 efficiency and anxiety when
communicating in L2. These two components also are classified as cognitive and affective.
The concept of one’s evaluation and anxiety of L2 usage also is linked to the willingness to
communicate in L1. The core idea of Layer 4 is that control and affiliation are important
Layer V
Layer 5 includes variables that diverse according to each individual and his attitudes and
motives. The three components, intergroup attitudes, social situation, and communicative
competence are less influenced by situational differences; instead, they are individually based.
The first component, intergroup attitudes consists of three different constructs: integrativeness,
fear of assimilation, and motivation to learn L2. Integrativeness is the desire to mix with
different ethnic groups and learn their language in order to be identified as “one of the group.”
It is a desire to be part of the L2 group which will lead to positive results such as increased
involvement with the group. However, when gaining the recognition of the L2 group, the
individual also has a fear of assimilation, of turning into a member of the L2 group and losing
identification or companionship in his L1 group. The two construct mentioned are reverse
elements that an individual has to face. The third construct concerning intergroup attitudes is
the motivation to learn L2 which can be either a positive or negative attitude, depending on
the individual. Yet a positive attitude sometimes has to do with the type of language you are
learning.
The second variable in Layer 5 is the social situation one faces. When the identity of the
interlocutor or the content of our conversation is different, one might behave differently in
attitude and confidence when using L2. For example, when college students in Taiwan use
English to chat with each other in the classroom, it appears to be casual and fun. Yet when the
occasion is in a foreign land or at the Airport, it is often that we shy away from speaking in
variable in layer 5. Since an individual’s L2 proficiency will have a very influential effect on
his willingness to communicate, this component is part of the cognitive-affective context.
Layer VI
Under Layer 6, communication is divided into two broad dimensions: society and
individual. According to Gardner and Clement (1990), intergroup climate is made up of “two
complementary elements which are the structural characteristics and their perceptual and
affective correlates. The structural characteristics of a group are made up of the
enthnolinguistic vitality and personal communication networks. Enthnolinguistic vitality,
defined in Giles et al. in 1977 is the “relative demographic representation of both L1 and L2
groups and their socioeconomic power in the society.” Generally, a language which has
higher ethnolinguistic vitality would be more commonly learned and used. On the other hand,
a personal communication network is the group of people whom we communicate with most
often in daily life. This construct is important since the people with whom we communicate
more often can be influenced by our promotion of L2. The second element of intergroup
climate is the perceptual and affection correlates which are the attitude and value towards a
L2 group and the motivation to adapt to their community. A positive attitude towards the L2
community will have a positive effect on an individual’s interaction with people in a L2 group.
Furthermore, positive learning motives such as hoping to gain knowledge on another culture
or making friends will also lead to positive effects.
Although personality does not have a direct influence on language learning
members of L2 group. Altemeyer’s study in 1981 and 1988 used two different personality
patterns as examples to describe the difference of language learning. An Authoritarian type of
person considers is more conventional and would not likely engage in language learning if he
thought the other group to be inferior. This type of personality can be explained by the
construct “ethnocentrism,” meaning having strong pride in one’s ethnic group and looking
down on others.
Table 2-4. Definition of willingness to communicate
Scholar Definition
MacIntyre, 2001 Intention to initiate a communication when
one has choice
McCroskey, 1992 A tendency to approach or avoid
communication with others
McCroskey & Baer, 1985 The intention to initiate communication
when given the choice
2.4 Trust
Trust is a very abstract construct that has been the interest of many fields of scholars. It
is because of this construct that one chooses to believe another person’s words or advice.
Therefore, it is also a crucial element whether word of mouth is accepted by consumers or not.
One of the classical definitions of trust is that it is “a generalized expectancy held by an
can rely on the information given to us by others our trust towards them will gradually build
up. Our establishment of willingness to share our information and expectancy towards that
certain person can be seen as the foundation that trust builds upon. The role of trust is an
important factor to understand the consumer behavior. For example, since it is the degree of
willingness we have to rely on someone we have confidence in (Moorman et al., 1993); it is
unmistaken that we must have a certain amount of confidence in the people we trust. In
addition, not only do we choose to put our confidence in them, but often some sort of positive
outcome is also expected although negative outcomes cannot be ruled out. Pruitt (1981) bests
describe this attribute of the construct as, “the belief that one party’s word is reliable and that
it will fulfill its obligation in an exchange.” This paradoxical attribute is one of the main
reasons why it is not easy to earn a person’s trust. The first party will still have a feeling of
fear for negative outcomes or loss if they do not have a certain amount of trust in the one they
choose to rely on.
Figure 2.3 Relationship of trust and willingness to communicate Trust
Willingness To Communicate
`Our study focuses on Interpersonal Trust, which is “the expectancy held by an individual or a
party that the word, promise, verbal, promise, or written statement of another individual or
party can be relied on (Rotter, 1967.)” According to Rotter’s widely used definition of
Interpersonal Trust, it is a construct that can be used across different situations and is part of
our personality traits.
Table 2-5. Definition of Trust
Scholar Definition
Rotter (1967)
A generalized expectancy held by an individual that the word of
another…can be relied on.
Pruitt (1981)
Trust is the belief that one party’s word is reliable and that it will
fulfill its obligation in an exchange
Moorman (1993)
Trust is defined as a willingness to rely on an exchange partner in
whom one has confidence.
Morgan (1994)
We conceptualize trust as existing when one party has confidence in
an exchange partner’s reliability and integrity.
McKnight &
Chervany ( 2002)
A relative feeling of security in a situation of risk
Trust is a very complex and multi-dimensional construct that has been studied from
different views including psychology and sociology. As long as risk and information sharing
exists, the issue of trust will be studied because of its importance in both interpersonal and
business relationships. Although scholars have very similar definitions of trust, yet there has
in 2002 which included different areas of definitions (Figure 3). This model was revised and
modified from McKnight, Chervany, and Cummings research on trust in organizational
relationships in 1998. Three different constructs are dispositional, institutional, and
interpersonal trust which are alike yet have very distinct differences in their definitions and
usage. These three constructs of trust differs from one another for the following reasons
(McKnight & Chervany, 2002). Psychologists believe that disposition to trust comes mainly
from past experience that shapes our character. For example, our actions origin from
childhood derived attributes. Institutional trust is molded from different situations which
allow one to act according to different situations. The factor that decides our action is not our
personality or inner traits, but the environmental factors. As for interpersonal trust, it can be
divided into two parts: trusting beliefs and trusting intention which defines interpersonal trust
as the relationship that is constructed through the interpersonal communication and
interactions which arouse our cognitive-emotional reactions. An unique trait of this model that
should be noticed is that “trust related behaviors” is not inside of the model since it is already
described in other labels that often are under the category of “cooperation, information
sharing, risk taking, etc.”
According to research done by Rempel and Holmes (1985), there are four key
conclusions that can sum up the features of trust. First, it is built upon past experience and a
series of events that strengthen the feeling of trust. Second, dispositional characteristics are
made towards the trustor, including attributes such as reliability, dependability, and
Disposition to Trust Institution -Based Trust Trusting Beliefs Trusting Intentions Trust-Related Behaviors Dispositional Trust Institutional Trust Interpersonal Trust
the risk of being hurt, or loss of something either abstract or concrete material. The last
conclusion raised from past researches is that trust can be defined by the sense of security and
confidence in the responses given and the intimacy and strength of the relationship.
Figure 2-4. The Interdisciplinary Model of High-Level Trust Constructs
(Reference: McKnight, D.H. & Chervany, N.L. (2002) What Trust means in e-commerce
customer relationships: An Interdisciplinary Conceptual Typology International Journal of
Electronic Commerce 6(2)35-59 )
There have always been three basic questions of trust development which are often asked.
These are the basic foundations of this construct and must be understood in order as the
Interpersonal trust, we will mainly discuss the latter two questions.
In early studies, most results pointed out that trust begins at zero, that is, when we are
unfamiliar or strangers with a person then there is a lack of trust between us. Since there is no
past experience that we can use as indicators or reference, individuals have to rely on their
own analysis and objective view of the situation whether to rely or not. Different experiments
mainly labeled “trust game” were conducted in the research of trust to explain and explore
how initial trust worked in relationships. When exploring how the level of trust changes over
time, most studies show that trust will build gradually as time goes by. This means that we
will undergo a process in which we evaluate the circumstances and individual carefully to be
sure that we are placing out trust in the right position. If not, then we could withdraw our trust
as soon as possible.
2.5 Tie Strength
The social relationship of individuals can be categorized according to the frequency of
contact, and how well you know each other. This is known as the “potency of a bond between
members of a network (Granovetter, 1973),” or “tie strength.” Strong tie sources are friends
and family, whereas weak ties are mainly acquaintances and strangers (Duhan, 1997). The
influence of tie strength can affect the outcomes in consumer decision such as the spreading
of word of mouth, intergroup interactions, and other relationship relevant issues (Mittal,
2008). Word of mouth communication is a social behavior that includes mainly the exchange
of information between individuals. Past research has proven that tie strength has an influence
relationship, or more intimate relationship have more opportunities to interact and share
information. On the contrary, in a weak tie relationship, there is a very limited amount of
chances for individuals to spread information to each other. Due to this reason, it is reasonable
to believe that strong ties will lead a positive relation with the spread of word of mouth and
vice versa. There is a higher amount of word of mouth generated in a network or group which
has strong ties (Bone, 1995). There is a complex relationship between tie strength and word of
mouth due to the consideration of one’s personal image. Since complaints and negative
reviews might have a negative effect on one’s personal image, an individual might not want to
express his unsatisfactory to an acquaintance or stranger. Instead, those that have a stronger
tie with the individual have a more likely possibility of accessing both positive and negative
word of mouth about a product, service, or purchase made. Yet in cyberspace where internet
is the medium for communication, the network that people communicate consists of a greater
population and has weaker ties. Thus it is proposed that perhaps unlike in the physical world
where the strength of ties has a positive relationship on the spread of word of mouth; weak
CHAPTER 3
METHODOLOGY
In Chapter 3, we establish our research framework based upon the literature review and
the hypotheses in our study. Scales used and generated from past researchers will be used in
our research as tools to evaluate the constructs we want to measure. Organized into a
paper-based questionnaire, 2 pretests are conducted before sending out the official
questionnaire to our participants.
3.1 Research Framework
Figure 3-1. Proposed research framework
Our research aims to explore the interaction between electronic word of mouth,
communication on willingness, trust, and tie strength. A thorough introduction and
description on the constructs has already been given in the previous chapter which is the
foundation of this study. According to past research, there is a strong reason to believe that a
definite relationship between these constructs exists. This belief will be examined and tested
in order to bring the issue of electronic word of mouth into a more complete view. eWOM Activity Positive eWOM Negative eWOM Tie Strength Trust WTC + + + + + + + - - -
In Ben-Ner and Putterman’s 2009 study on trust and communication’s interaction in
economic behaviors, they believed that if there is a contract relationship that one desires to
establish, then individuals will take the communication process more seriously. This is
because they believe in depth communication will promote the trustworthy behaviors. And
indeed, their study proves their hypothesis to be correct; communication does promote the
level of trust and trustworthiness. In addition, communication is seen as an antecedent factor
in trust in Morgan and Hunt’s. Various other researchers have also proven the interaction
between trust and communication in different fields of study. “Communication is critical to
build a trusting relationship which will create stability (Anderson et al., 1989).” However, few
studies have explored the relationship of how the level of interpersonal trust affects
willingness to communicate. Although the content of communication is not completely
associated with personal privacy or other matters that one might be cautious about, yet it is
still reasonable to believe that the more one is willing to be exposed, the more he will be
willing to share information with other. Therefore our study proposes that interpersonal trust
will have a positive effect on willingness to communicate.
Word of mouth is an initiative action where an individual chooses to express his opinions
or share past experiences with others. One must be willing to communicate in either reality or
in cyberspace to perform the act of word of mouth. Since this is an action that depends on the
individual’s intention of whether he would like to spread the word, we believe that an
individual’s willingness to communicate with another person should have a positive effect on
“willingness to communicate” as the third construct in our framework.
The strength of social relationships between individuals, known as “tie strength,” is an
interesting variable since modern technology and use of internet has already expanded our
social networks and those that we communicate with on the internet does not necessarily have
strong ties with us. Already in Granovetter’s 1983 research, he proved that weak ties actually
have a contribution that strong ties do not; (in a social system) those that lack weak ties will
have difficulty integrating with people that are less similar than themselves. When exploring
the interaction between tie strength and electronic word of mouth, Granovetter’s research can
be used to imply that there should be a negative relationship between tie strength and
electronic word of mouth.
3.2 Research Hypotheses
Hypothesis 1: Trust has a significant and positive relationship with an individual’s
willingness to communicate.
Hypothesis 2a: Trust has a significant and positive relationship with an individual’s
electronic word of mouth activity.
Hypothesis 2b: Trust has a positive relationship with an individual’s positive electronic
word of mouth.
Hypothesis 2c: Trust has a significant and positive relationship with an individual’s
negative electronic word of mouth.
Hypothesis 3: Willingness to communicate will mediate the relationship between trust
Hypothesis 4a: An individual’s willingness to communicate has a significant and
positive relationship with an individual’s electronic word of mouth activity.
Hypothesis 4b: An individual’s willingness to communicate has a significant and
positive relationship with an individual’s positive electronic word of mouth.
Hypothesis 4c: An individual’s willingness to communicate has a significant and
positive relationship with an individual’s negative electronic word of mouth.
Hypothesis 5a: An individual’s tie strength with another individual on the internet has a
significant and negative relationship with electronic word of mouth activity.
Hypothesis 5b: An individual’s tie strength with another individual on the internet has a
significant and negative relationship with positive electronic word of mouth.
Hypothesis 5c: An individual’s tie strength with another individual on the internet has a
significant and positive relationship with negative electronic word of mouth.
3.3 Sampling Design
Our study’s main primary targets is the mass population, since internet is already a
common tool for people living in the metropolitan area, the population of this study will be
defined as the college students in Taipei. The sample size of our study is decided by Roscoe’s
4 principles in sampling design as listed below. Due to the population size, and limited time
and resources, our study uses convenience sampling, or also known as subjective sampling.
Convenience sampling is often taken place near school campus, parks, or other places where
thumb, there are several principles to be followed. A sample of more than 30 and less than
500 is appropriate for most studies, and the size must be 10 times or more of the researched
variables for multiple regression analysis to be applied (Chew et al., 2009).
3.4 Questionnaire Design
All our data will be collected by paper-based questionnaires. Most items will be
answered on a five-point Likert scale ranging from 1 to 5, with 1 being “strongly disagree”
and 5 being “strongly agree.” Willingness to communicate to others is a self-report scale
which is scored by the participants themselves. All the original items were in English and then
translated into Traditional Chinese and then back into English by another native speaker who
is also fluent in Chinese to ensure the content and meaning remained the same during the
translation process. There are two different parts of this questionnaire design. According to
Churchill, “once a construct is clearly defined, the next step is to generate a set of items that
clearly capture the domain of the construct (Churchill, 1979).” A pre-test has to be conducted
in order to be sure the wording and expressions are clear to our participants. The first step of
our questionnaire is to have 30 college students fill out the questionnaire and give opinions
and suggestions on the wording and content. After making appropriate adjustments, a revised
version will be handed out to college students in Taiwan.
Our official questionnaire is divided into 5 sections, electronic word of mouth, trust,
willingness to communicate, tie strength, and demographic variables. Electronic word of
were 13 items to measure the construct, yet Harrison-Walker (2001) eliminated 7 items after
performing a scale purification process which refined the scale into a “more reliable and
meaningful scale in assessing word of mouth (Harrison-Walker, 2001).” Positive electronic
word of mouth was measured by 3 items from Liljander and Strandvik (1997)’s loyalty scale
which demonstrated high reliability statistics. On the other hand, the scale used to measure
electronic negative word of mouth was originally used in Liu and McClure’s research on
cross-cultural customer complaint behavior study. The scale was adapted to suit the
contextual characteristics of this particular study. In order to help our participants be clear of
the difference between electronic and traditional word of mouth, examples of electronic
platforms and the definition of traditional word of mouth was given at the beginning of
section 1 and section 2. The original set of items was measured on a 7 point Likert type scale.
Yet to help improve the efficiency and increase the unity of our questionnaire, scales were
modified from 7 point to 5 point.
Section 1: Demographic Variables
Table 3-1. Demographic variables
Construct Scales of measurement Item
Gender Nominal scale 1. Male
2. Female
Age Ordinal scale 1. Under 18
2. 19~23
3. 24~28
Construct Scales of measurement Item
5. 34~38
6. Above 39
University
Section 2: Electronic Word of Mouth
Table 3-2. Measuring an individual’s electronic word of mouth items
Construct Dimension Scales of Measurement Items
Electronic word of mouth Electronic word of mouth activity
5 point Likert type scale 1. I mention the product which I
use to others quite frequently.
2. I’ve told more people about
certain products than I’ve told
about most other products.
3. I seldom miss an opportunity
to tell others about certain
products.
4. When I tell others about a
certain product/service, I tend
to talk about it in great detail.
Positive
Electronic
word of
mouth
1. I only have good things to
say about certain
products/services.
2. I am proud to tell others that I
Section 3: Trust
Rotter’s Interpersonal trust scale is an extremely often used tool to measure the level of
trust one has towards others. Developed in 1967 with 25 items measuring trust, and 15 filler
items, this scale is “designed to measure a person’s generalized expectancy that the promises
of another individual which can be relied on (Rotter et al., 1971)”
Table 3-3. Measuring an individual’s interpersonal trust items
Construct Dimension Scales of Measurement Items
Trust Interpersonal
trust
5 point Likert type
scale
1. Hypocrisy is on the increase in
our society.
2. This country has a dark future
unless we can attract better
people into politics.
3. Using the honor system of not
having a teacher present during
exams would probably result in
increased cheating.
4. Taiwan will never be an
effective force in keeping world
peace.
5. Most people would be horrified
if they knew how much news
the public hears and sees is
distorted.
Construct Dimension Scales of Measurement Items
6. Even though we have reports
in newspapers, radio, and
T.V., it is hard to get objective
accounts of public events.
7. If we really knew what was
going on in international
politics, the public would have
reason to be more frightened
than they now seem to be.
8. Many major national sports
contests are fixed in one way or
another.
9. Most people can be counted on to do what they say they will
do.
10. In dealing with strangers one is
better off to be cautious until
they have provided evidence
that they are trustworthy.
11. Fear of social disgrace or
punishment rather than
conscience prevents most
Construct Dimension Scales of Measurement Items
12. Parents usually can be relied
upon to keep their promises.
13. The judiciary is a place where
we can all get unbiased
treatment.
14. It is safe to believe that in spite
of what people say most people
are primarily interested in their
own welfare.
15. The future seems very
promising.
16. Most elected public officials are
really sincere in their campaign
promises.
17. Most experts can be relied upon
to tell the truth about the limits
of their knowledge.
18. Most parents can be relied upon
to carry out their threats of
punishment.
19. In these competitive times one
has to be alert or someone is
likely to take advantage of you.
Construct Dimension Scales of Measurement Items
20. Most idealists are sincere and
usually practice what they
preach.
21. Most salesmen are honest in
describing their products.
22. Most students in school would
not cheat even if they were sure of getting away with it.
23. Most repairmen will not
overcharge even if they think
you are ignorant of their
specialty.
24. A large share of accident claims
filed against insurance
companies are phony.
25. Most people answer public
opinion polls honestly.
Section 4: Willingness to Communicate
In 1985, McCroskey proposed a scale to measure willingness to communicate which is a
personality-based trait like scale that is consistent across between different receivers and
different communication contexts. McCroskey has proved that the level of willingness of a
particular context (small group) is correlated with this individual’s willingness in a different
with a certain type of audience (for example, friends) is also correlated with the willingness to
talk with other different types of people (strangers and acquaintances). Yet this does not mean
that a person will be equally willing to communicate in all different contexts and receivers.
What this means is that they will be correlated. This scale has 20 items, with 8 items being
filler items that will not count nor influence the final results of our scale. The willingness to
communicate scale includes four communication contexts and three different receivers. The
scale can measure the overall willingness to communicate represents an individual’s general
personality orientation to communicate with others as well as 7 different sub-scores.
Participants will fill out the percentage of how much they are willing to communicate in each
different scenario.
Table 3-4. Measuring an individual’s willingness to communicate items
Construct Dimension Scales of
Measurement Items Willingness to communicate Talking with strangers
1. Present a talk with a group of
strangers
2. Talk in a small group of
strangers
3. Talk with a stranger while
standing in line
4. Talk in a large meeting of
strangers
Talking with
acquaintances
1. Talk with an acquaintance
Construct Dimension Scales of Measurement Items Willingness to communicate Talking with acquaintances
2. Talk in a large meeting of
acquaintances
3. Talk in a small group of
acquaintances
4. Present a small talk to a group
of acquaintances
Talking with
friends
1. Talk in a large meeting of
friends
2. Talk with a friend while
standing in line
3. Present a talk to a group of
friends
4. Talk in a small group of friends
Filler items 1. Talk with a service station
attendant
2. Talk with a physician
3. Talk with a salesperson in a
store
4. Talk with a
policeman/policewoman
5. Talk with a waiter/waitress in a
restaurant
Construct Dimension Scales of Measurement Items Willingness to communicate
Filler items 7. Talk with a garbage collector
8. Talk with a spouse
(girlfriend/boyfriend)
Section 5: tie strength between individual and electronic word of mouth receiver
In section 4, we study the tie strength of each participants and the electronic word of
mouth receiver whom he has interacted with most recently. Hansen’s two item scale
measuring the closeness of a working relationship and frequency of contact developed on
1999 is used in our study to measure tie strength. According to Levin’s pre-test, instructions
were given to “choose 7 for these two questions if you have never had prior contact with this
person (Levin, 2002).” Again, to help simplify the procedure of filling out this questionnaire,
this set of items were modified from a 7 point Likert scale to a 5 point scale.
Table 3-5. Measuring an individual’s tie strength with another person in eWOM items
Construct Dimension Scales of measurement Items
Tie
Strength
5 point Likert scale 1. How close was your relationship
with him/her?
2. How often did you communicate
with him/her?
3.5 Questionnaire Pretest
Cronbach Alpha of 0.7. The reliability of a scale means the consistency and stability of a scale.
In 1951, Cronbach proposed that a method that is now known as “Cronbach Alpha” that could
measure the reliability. The standard of how much the Alpha has to be for the instrument to be
accepted is different for each area of research. In most studies, 0.6 is enough for the
instrument to be declared acceptable. In the first pretest executed, many participants
responded that the items of the trust scale were hard to understand and confusing. Therefore
there was no surprise when the internal consistency of this scale received a fairly low score of
0.593. Item 8 had a negative item to total correlation, and was removed since the scale could
achieve a 0.657 Cronbach Alpha if this was done. The original electronic word of mouth scale
consisted of 2 dimensions which are word of mouth activity and positive word of mouth.
Electronic word of mouth activity reached a Cronbach Alpha of 0.622, with the last item
having a negative correlation with the entire scale. If the item were to be deleted, the
Cronbach Alpha would rise to 0.886. Therefore, the fourth item was deleted in our second
pretest. Along with the replacement of a new positive word of mouth scale that consisted of 3
items, 2 more items measuring negative word of mouth were added to strengthen the
contribution of our study. As mentioned, the positive word of mouth scale was replaced by a
new one due to the extremely low Alpha of 0.197 which could not be fixed by removing items
or revision of translation and wording. Although electronic word of mouth activity also
achieved a low internal consistency score, yet participants also responded that the wording
was difficult and hard to understand. For the same reason, this scale was given another chance
to communicate achieved a high Alpha of 0.803 as well as tie strength’s 0.910.
Table 3-6. Reliablity statistics of first pretest
Construct Dimension Cronbach α Item(s) removed Cronbach α
after item
removed
eWOM eWOM activity 0.622 Item 4 0.886
Positive eWOM 0.197 Scale Removed --
Tie Strength -- 0.910 -- --
Trust -- 0.593 Item 8 0.657
Willingness to
Communicate
-- 0.803 -- --
In our second pretest, the questionnaire made several revisions. First, demographic
variables were moved from Part 1 to Part 5. This was because most scholars believe that
demographic questions might be more sensitive and lead to a low return rate of questionnaires
if they were placed at the beginning of the survey. Second, electronic word of mouth was
rated with a total of 8 items including newly added 2 item scale on negative electronic word
of mouth. Third, 1 item was removed from our trust scale in hope of improving the internal
consistency. Furthermore, more directions were added to each section to help our participants
understand each part of the questionnaire and how to respond to the questions better. A total
Table 3-7. Reliability statistics of second pretest
Construct Dimension Cronbach α Item removed Cronbach α
after item
removed
eWOM eWOM activity 0.792 -- --
Positive eWOM 0.806 -- -- Negative eWOM 0.957 -- -- Tie Strength 0.958 -- -- Trust 0.553 Items 4, 10. 17 19 0.713 Willingness to communicate 0.712 -- --
The results for our second pretest all passed the bar of 0.7 internal consistencies except
for “trust.” Four items that were negatively correlated with the scale were removed, which
resulted in a Cronbach’s α 0.713 if so. The final official questionnaire had a total of 53
questions, including demographic variables.
In our study, we will use several types of statistical analysis method to explain the
collected data and help understand our research framework. These methods include:
1. Regression Analysis: Regression analysis is used to explore the relationship
between dependent and independent variables. In our study, we will use regression analysis
to analyze the relationships hypothesized in our study.
we have retrieved. It will test whether or not the data is internally consistent. Therefore, the
higher the constructs achieve on this score, the higher the constructs are internally related
and reliable. According to Roberts and Wortzel, the alpha coefficient between 0.7 and 0.98
reflects high reliability. Therefore 0.7 will be the standard to examine whether out constructs
are internally consistent.
3. Pearson Coefficient: Pearson Coefficient Correlation is used to examine the
relationship between the variables in our study. Whether they are positively or negatively
correlated, and if there is a significant relationship and the strength of that relationship is
CHAPTER 4
RESEARCH RESULTS AND DATA ANALYSIS
With the help of acquaintances who are students in different colleges in Taipei,questionnaires were distributed in classrooms and a total of 171 valid questionnaires were
retrieved. Using SPSS version 17 software, the procedure of statistical analysis will be
conducted to examine our study and test the hypotheses.
4.1 Reliability Analysis
Our study conducted a reliability analysis for the 4 different variables included in the
study. The results were fairly good, with each constructs’ Cronbach’s α above 0.7. This
means that the items in the questionnaire are highly correlated and also consistent. Electronic
word of mouth activity reached a Cronbach α of 0.783, positive electronic word of mouth
0.775, negative electronic word of mouth 0.811. Our trust scale has a Cronbach αof 0.756,
scoring much higher than the 2 pretests conducted. Furthermore, the 12 items of willingness
to communicate had an internal consistency of 0.810, while tie strength had the highest
internal consistency of all, 0.921.
Table 4-1 Cronbach Alpha Analysis Results
Construct Cronbach α Number of Items
Trust 0.765 20 WTC 0.810 12 Tie Strength 0.921 2 eWOM Activity 0.783 3 Positive eWOM 0.775 3 Negative eWOM 0.811 2
4.2 Regression Analysis
Table 4-2 Regression Analysis Results
Independent Variable Dependent Variable P-Value R-Square
Trust WTC 0.048 0.023
WTC eWOM Activity 0.001 0.059
WTC Positive eWOM 0.003 0.051
WTC Negative eWOM 0.271 0.007
Trust eWOM Activity 0.547 0.002
Trust Positive eWOM 0.677 0.001
Trust Negative eWOM 0.461 0.003
Through table 4-2, we can see that the R-square is relatively low. This means indicates
that the research framework can only explain a very low percentage of the dependent
variables. However, these results do not mean that the model should be rejected since there
might be other possibilities that influence the result. The regression analysis also points out
that four of our hypotheses are accepted with a P-Value that is significant. Trust has a
significant effect on willingness to communicate, whereas willingness to communicate has a
significant relationship with electronic word of mouth activity and positive electronic word of
