Luận văn thạc sĩ Vận dụng thẻ điểm cân bằng (Balanced Scorecard) tại công ty CP vận tải dầu khí Cửu Long

TR NG TH TUY T DUNG INFLUENTIAL FACTORS ON WORD OF MOUTH IN SERVICE INDUSTRIES THE CASE OF HIGHER EDUCATION IN HOCHIMINH CITY, VIETNAM Subject: Master of Business Administrator... COM

TR NG TH TUY T DUNG

INFLUENTIAL FACTORS ON WORD

OF MOUTH IN SERVICE INDUSTRIES

IN HOCHIMINH CITY, VIETNAM)

MASTER THESIS IN BUSINESS ADMINISTRATION

TR NG TH TUY T DUNG

INFLUENTIAL FACTORS ON WORD

OF MOUTH IN SERVICE INDUSTRIES (THE CASE OF HIGHER EDUCATION

IN HOCHIMINH CITY, VIETNAM)

Subject: Master of Business Administrator

ACKNOWLEDGEMENT

I would like to express my great gratitude to my supervisor, Dr Tr n Hà Minh Quân, for his wonderful knowledge, patient and kind guidance throughout this research

I would like to thank all of my respected lecturers of the eMBA19 course for their efforts and dedication in training us

I am appreciative my classmates for their kind support and encouragement Finally, I would like to dedicate this study to my mom, my husband, and my sister, who always love and encourage me to pursue with my study

Ho Chi Minh City, January, 2014,

Tr ng Th Tuy t Dung

COMMITMENT

I would like to commit that this thesis, “Influential factors on word of mouth

in service industries (the case of higher education in Hochiminh city, Vietnam)”, was accomplished based on my independent and serious study and scientific research The data was collected in reality and it has clear origins In addition to that, the data would be trust-worthily handled and it has never been released in any menu

Tr ng Th Tuy t Dung

TABLE OF CONTENT

ABSTRACT 7

CHAPTER 1 INTRODUCTION 8

1.1 Introduction 8

1.2 Research Objectives 10

1.3 Research methodology and research scope 11

1.4 Significances of the research 11

1.5 Structure of the research 12

CHAPTER 2 LITERATURE REVIEW 12

2.1 Introduction 12

2.2 Higher education as a service 13

2.3 Word-of-Mouth (WOM) 13

2.3.1 Definition & characteristics 13

2.3.2 Role of WOM in decision of consumers 15

2.3.3 Effectiveness of WOM 17

2.4 Development of research model and hypothesis 18

2.4.1 Satisfaction 18

2.4.2 Loyalty 19

2.4.3 Service quality 20

2.4.4 The perceived value 21

2.4.5 Trust 22

2.4.6 Reference intention to enroll 23

2.4.7 Proposed conceptual model and hypotheses 23

2.5 Summary 24

CHAPTER 3 RESEARCH METHODOLOGY 25

3.1 Introduction 25

3.2 Research design 25

3.2.1 Research methodology 25

3.2.2 Research process 26

3.3 Measurement scales and questionnaire development 27

3.3.1 Measurement scales 27

3.3 2 Questionnaire development 29

3.3.3 Questionnaire design 29

3.4 Pilot study 30

3.4.1 Qualitative pilot study 30

3.4.2 Quantitative pilot study 30

3.5 Quantitative main study 31

3.5.1 Research sampling 31

3.5.2 Collecting data 31

3.5.3 Methods of data analysis 31

3.5.4 Data cleaning 32

3.5.5 Reliability assessment of measurement scales (Cronbach's alpha) 32

3.5.6 Exploratory factor analysis (EFA) 32

3.5.7 Confirmatory factor analysis (CFA) and SEM: 34

3.5.8 Bootstrap Method 36

3.6 Summary 36

CHAPTER 4 DATA ANALYSIS AND FINDINGS 37

4.1 Introduction 37

4.2 Descriptions of sample 37

4.3 Reliability and validity of the measurement scale 38

4 3.1 Reliability (Cronbach’s alpha) 39

4.3.2 Exploratory factor analysis (EFA) 41

4.3.2.1 EFA implementation for independent variables 41

4.3.2.2 EFA implementation for dependent variable 45

4.3.2.2.1 EFA for Word of Mouth (WOM): 45

4.3.2.2.2 EFA for Reference Intention (INT): 46

4.3.3 Confirmatory Factor Analysis: 47

4.4 Adjust the research model and the hypotheses 50

4.4.1 The adjusted research model 50

4.4.2 The adjusted research hypothesis 51

4.5 Testing the adjusted research model and the hypotheses 51

4.5.1 Testing the adjusted research model 51

4.5.2 Results of testing hypotheses 54

4.5.3 Boostrap test 56

4.6 Summary 57

CHAPTER 5 CONCLUSION AND IMPLICATION 58

5.1 Introduction 58

5.2 Findings 58

5.3 Recommendations 58

5.4 Limitation and future research 59

LIST OF REFERENCES 60

APPENDIX 1 QUESTIONNAIRE 71

APPENDIX 2 CRONBACH’S ALPHA ANALYSIS 79

APPENDIX 3 EXPLORATORY FACTOR ANALYSIS (EFA) 82

APPENDIX 4 CFA , SEM RESULTS 87

LIST OF TABLE

Table 1.1: Definitions of Word-of-Mouth 13

Table 3.1 Measurement scales of variables 28

Table 4.1 Demographic characteristics of the respondents 38

Table 4.2 Cronbach‟s alpha of variables 39

Table 4.3 KMO and Bartlett's Test 41

Table 4.4 Total Variance Explained 42

Table 4.5 Rotated Component Matrix 43

Table 4.6 EFA results of independent variables 44

Table 4.7 EFA results of WOM variable 46

Table 4.8 EFA results of Reference Intention Variable 47

Table 4.9 Decriptive statistics, factor loads and Cronbach‟s alpha coefficients 48

Table 4.10 Pearson Correlation 49

Table 4.11 Results of testing hypothesis 55

Table 4.12 Square multiple Correlations (R2) 56

Table 4.13 Boostrap results with N = 500 56

LIST OF FIGURE

Figure 2.1 Proposed conceptual model of the research 24

Figure 3.1 Research process 26

Figure 4.1 Standardized measurement model of variables 50

Figure 4.2 The adjusted research model 51

Figure 4.3 Initial Structure Model 52

Figure 4.4 Nonstandard regression of the modified structural model 53

Figure 4.5 Standard regression of modified structural model 54

ABSTRACT

This study aims to exam the factors effective in the formation of mouth about the services of higher education and their results among students in Ho Chi Minh City

word-of-A structural model with dimensions of factors affecting word-of-mouth, word-of-mouth and their results was tested with a sample of 223 students of higher educational institutions in Ho Chi Minh City, Vietnam

The results of the study indicate that to Vietnamese consumers, perceived value, satisfaction and loyalty are factors effective in the formation of Word of mouth about higher educational institutions and Word of Mouth can affect on students‟ intention to refer to enroll at the institutions These findings suggest that educational managers should make efforts to improve their mental image in students and to make students more satisfied in order to encourage them producing positive word-of-mouth about the organization

KEYWORDS: word of mouth, Hochiminh City, higher education, consumer‟s behavior

CHAPTER 1 INTRODUCTION

1.1 Introduction

From marketer perspective, Word-of-mouth (WOM) is an important area and

a key player for marketing research as this form of communication is regarded to have greater influences on other people‟s choices compared to other forms of communication (Murray, 1991; Day, 1971) and thus, more effective than other marketing tools and conventional advertising media (Engel et al., 1969; Katz and Lazarfeld, 1955) Although a WOM communication can be very influential in any purchase decision, previous researches suggest that its influence seems to be greater

in services contexts (Ettenson and Turner, 1997; Murray, 1991) Even within the service sector, WOM seems to have different influences in different service contexts because risk is regarded higher in some service purchase situations than it

is in others (Bansal and Voyer, 2000; Still et al., 1984) Education is a special service situation in which WOM seems to be significant (Cook and Zallocco, 1983; Mazzarol and Soutar, 2002)

In Vietnam, education always plays a central role in culture and society With a densely-populated country with a population of 86 million people and with more than 60% under the age of 35 (Source: General Department of Statistics of Vietnam, 2009), Vietnam‟s education emerges huge needs for this young population Moreover, as a developing country with strong industrial growth, expanding foreign investment and a booming private sector, Vietnam need a variety

of workplaces skills, including technology and management knowledge; therefore, high demand for good quality training is generated for education in Vietnam (Maine International Trade Center, 2010 )

The Higher Education Reform Agenda 2006-2020 has built to aim on changing education environment, resulting in a growth in the number of public,

semi-public and private higher education institutions However, despite the number

of higher institutions in Vietnam has doubled for the last 10 years (from 153 in 2000

to 369 in 2008, and even more today) (cited in Runckel, 2012) and the government has issued $456.5 million in loans (Source: World Bank, 2010) aimed at improving Vietnam‟s higher education system, only about 1.6 million (about 2% of Vietnam population) are at higher education institutions (Runckel, 2012); i.e., Vietnam‟s proportion of college students is still half that of Thailand and a third that of South Korea (Maine International Trade Center, 2010) Moreover, according to data from MOET in 2007, higher education in Vietnam faces many challenges such as ratio of students to faculty is low at 1/30; quality of faculty is not as high as needed (for example, only 14% of faculty have a doctoral degree and 33% have a masters degree); quality assurance in education remains weak; enrollment is concentrated in

a few academic disciplines (nearly 50% of all students major in economics/ business) Therefore, it is a requirement for higher education institutions to make efforts to meet demand of modern business society

In addition, as higher education provision is a service and students are expected to pay fees for their educational trainings, it seems appropriate that higher education institutions make a shift from being product-led ( i.e., relying on the product to sell) towards a more “customer-led” approach (Angell et al., 2008) When consumers (in this context, students) makes an uncertain and high-risk decision in service industries (like choosing a university), they may rely heavily on WOM communications such as the advice and suggestions from others who have experienced the service (Kinard and Capella, 2006) Moreover, WOM‟s importance

is highlighted as consumers often trust each other more than they trust communication from organizations (Zeithaml and Bitner, 1996) As a result, educational managers need to understand the formation of WOM and its role in their organizations‟ success

Studies on WOM have exposed that its effectiveness is relied on the overwhelming influence which it makes on consumer behavior Reicheld (1996)

points out that these effects are enlarged by a higher degree of customer loyalty and profitability Researchers have found that WOM is strongly and positively connected with consumers‟ levels of trust (Bergeron at al, 2003), service quality (Parasuraman et al, 1988), satisfaction (Anderson, 1998), perceived value (Hartline

& Jones, 1996), relationship quality (Boles et al., 1997), and with consumers‟ intention to purchase (Crocker, 1986) Furthermore, within the higher education context, Athiyaman (1997) found that student attitude to the university is positively related to positive WOM subsequent to their enrolment (post-enrolment communication behavior) More recently, Jayawardhena & Wright (2009) add further distribution when they find excitement positively influences on positive WOM Especially, Yasvari et al (2012) propose and empirically test a comprehensive model to examine factors effective on WOM in a service context (Iran Airline Company) and their results Their study indicates that five factors - (1) satisfaction, (2) trust, (3) service quality, (4) the perceived value and, (5) loyalty are found to be effective on WOM and can affect on consumers‟ decision about using the service In this study, the author uses the model of Yasvari et al to test factors effective in the formation of WOM and their outcomes in context of higher education in Hochiminh City

1.2 Research Objectives

As mentioned above, studies into word-of-mouth (WOM) communication have recently gained increasing attention in marketing research From a practical standpoint, managers are more interested in generating positive WOM as positive WOM is recognized as a powerful tool for promoting products and services Moreover, the formation of positive WOM about a service organization can be depended on different factors and hence, will have noticeable outcomes However, little research has been devoted to examining these factors and outcomes in the service industries such as higher education in developing countries like Vietnam Therefore, the present study has these objectives :

 Identifying important factors effective in the formation of positive word-of-mouth in higher educational sector in Hochiminh City

 Adjusting measurement scale and testing theoretical framework on the factors effective on the formation of positive word-of-mouth about higher education service

1.3 Research methodology and research scope

This study was conducted by 2 steps: pilot study and main study

In the pilot study, an approach was conducted by interviewing the students who have experiences in studying in higher education institutions in Ho Chi Minh city to explore whether the scales for measuring the construct sections are suitable

or not After that, some necessary adjustments have been made

In the main study, a quantitative approach was used Data analysis is conducted to test reliability of the measurement scales relied on Cronbach‟s alpha coefficient and Exploratory Factor Analysis (EFA) Then the Confirmatory Factor Analysis was used to test the validity of the measurement scales After that, structural equation modeling (SEM) was employed for testing the proposed model and hypotheses SPSS software version 16 and Amos 21 was used for data analysis

in this study

1.4 Significances of the research

This study is practically meaningful in WOM literatures in service industry

in Vietnam with the result from examining factors effective on WOM and the effect

of WOM, as the outcomes of these factors, on the intention of consumers to refer to higher education institutions

The results of this study are expected to enhance educational managers‟ understanding of WOM formation and its roles in their organizations Such an understanding will help higher education institutions to build suitable programs to meet demands of students and father, to provide qualified workers, managers for the

country as well as to promote their reputations in a burgeoning industry like Hochiminh city

The result of the study is also distributed as a source of reference for marketing researchers in Vietnam market

1.5 Structure of the research

The present study is structured in five chapters Chapter 1 covers research background, research objectives, research methodology, and necessary of the study Chapter 2 comprises the current knowledge of marketing scholars regarding WOM, some important factors effective on it and their results as well as proposes a conceptual model for the research Chapter 3 assigns the research methodology used

to modify and evaluate the measurement scales, the hypotheses and the research model of the study Chapter 4 presents discussions about the analysis of data related

to the evaluation of measurement scales, research model and hypotheses Lastly, Chapter 5 gives some main conclusions and recommendations based on the results

of the study in chapter 4 as well as points out some limitations of this study

CHAPTER 2 LITERATURE REVIEW

2.1 Introduction

This chapter reviews the theoretical basis of the study about influential factors on words of mouth in service industries and also presents a conceptual model including hypotheses of the study Chapter 2 is composed of four main parts: (1) literature review of higher education as a service, (2) literature of word-of-mouth, (3) literature review of factors affecting on word of mouth in services industries and reference intention as well as proposed conceptual model and hypothesis of the study

2.2 Higher education as a service

Higher education can be regarded as a service industry (Oldfield and Baron,

2000) Students are regarded as consumers of service, being “partners” in the learning process as they consciously pick, choose and buy the service (Yorke, 1999) Moreover, researchers such as Barnett (2011) also consider students as key stakeholders in the process of higher education while Dill and Soo (2005) value the importance of their opinion for improvement in quality of higher education Hennig-Thurau et al (2001) point out that educational services “fall into the field

of services marketing” and differ from other professional services in the manner that educational services play a central role in the students‟ lives and students require huge amounts of motivation and intellectual skills to attain their goals Furthermore, educational services have several service characteristics such as intangible, complex, perishable, heterogeneous (Swanson and Davis, 2000; Voss et al., 2010) In addition, the professor‟s teaching efforts are simultaneously

“produced” and “consumed” with both professor and student being part of the teaching experience (Shank et al., 1995) Therefore, as a service, it is believed that higher educational institutions‟ success primarily depends on the efforts of both students as consumers and professors as service providers (Cooper, 2007)

2.3 Word-of-Mouth (WOM)

2.3.1 Definition & characteristics

There are many definitions about WOM from past to present Table 1 below

is derived from a chronological list about WOM definitions made by Goyette et al (2010) We implement a definition of WOM by Silverman (2011) to the last row of the table

Table 1.1: Definitions of Word-of-Mouth

Author/ Researcher Definition

Arndt (1967, p.3) WOM is defined as oral, person to person communication

between a receiver and a communicator whom the receiver perceives as non-commercial concerning a brand, a product, or

a service Richins (1983, p.17) the WOM communication was defined as the act of telling at

least one friend or acquaintance about the dissatisfaction Westbrook (1987, p

261)

In a post purchase context, consumer WOM transmissions consist of informal communications directed at other consumers about the ownership, usage, or characteristics of particular goods and services and/or their sellers

Singh (1990, p 1) telling others about the unsatisfactory experience (that is,

negative WOM) Bone (1992, p 579) WOM communication is conceptualized herein as a group

phenomenon – an exchange of comments, thoughts, and ideas among two or more individuals in which none of the

individuals represent a marketing source Anderson (1998, p.6) WOM refers to information communications between private

parties concerning evaluations of goods and services WOMMA (2006) WOM is “an act by consumers providing information to other

consumers.”

Silverman (2011, p 51) WOM is “the exchange of information about a product or

service among people who are independent of the producer”

As it is indicated from table 1 that WOM is often defined as being informal and non-commercial communication and as an exchange of information between two or more individuals regarding a product or a service Moreover, Soderlund & Rosengren (2007) expose that WOM is the informal transfer of positive or negative purchase and consumption-related behavior between consumers Brown et al (2005) extend the definition when they propose that in its broadest sense, a WOM communication includes any information about a target object (e.g company, brand) transferred from one individual to another These communications can be mutual conversations or one-way suggestions and recommendations ; live or recorded; in person, by telephone, by email, or by any other means of communication; one-to-one, one-to-many, or group discussion as long as they are from or among people perceived as non-commercial interest in encouraging others to a product or a service (Silverman, 2011) These people can be friends, family, acquaintances or even strangers (Duhan et al, 1997) In addition, WOM sources can be classified in to personal and impersonal sources based on the means the consumers use Friends, family, and acquaintances are regarded as personal sources of recommendations (Duhan, et al., 1997) whereas columns, articles, and commentary by journalists,

columnists, consumers, and experts appeared in newspapers, magazines, specialized publications, online discussion forums, and expert systems are regarded as impersonal sources of WOM recommendations (Senecal & Nantel, 2004)

It is suggested consumers are stimulated to share their experiences with others by cognitive elements such as satisfaction, acquiescence and irritation (Neelamegham & Jain, 1999) People generating WOM communications are likely motivated by personality, sociability, and a desire to help others (Lau & Ng, 2001)

In addition, researchers find that WOM‟s ability to occur between consumers seems

to be influenced by the tie strength (i.e the intensity of the social relationship between consumers) (Bansal and Voyer, 2000), and by homophile (i.e how similar)

or heterophony (how dissimilar) such consumers are in terms of their backgrounds, opinions, likes and dislikes (Steward & Conway, 1996; Gilly et al., 1998)

2.3.2 Role of WOM in decision of consumers

Evidence indicates consumers‟ buying decision process of goods and services is strongly influenced by WOM communications (Richins, 1983) Reichheld (1996) reveals that WOM impacts significantly on consumer behavior and consequently on sales Researches emphasize that when consumers who involve deeply in decision making and purchase having little or no expertise in a product or service, they tend to refer to others‟ opinion more frequently and rely on WOM communications more than on other sources of information (Bansal, 2000) According to Price et al (1995), information exchange about a product through WOM makes a consumer more powerful and decreases the asymmetry in the information exchanged between the consumer the producer and accordingly, increases (in case of positive WOM) or decreases (in case of negative WOM) the acceptance speed of the product Especially, WOM is more important in the final stages of purchase process because it reassure consumers as well as reduces post-purchase uncertainty (Martilla, 1971)

WOM is likely to be more important in service contexts as service is intangible, non-standard, non-guaranteed, and thus, difficult to evaluate prior to purchase, higher perceived risk than goods, and impossible for consumers to undertake a pre-purchase trial to reduce perceived risk (Zeithaml et al., 1985) Consumers tend to engage in WOM conversation with the aim to gain more information that will help them eventually reduces the risk and understand the service prior to consumption (Bristor, 1990) Service consumers usually think of asking for opinions first before making purchase decision (Murray, 1991); therefore, they rely on the opinions of other consumers to form expectations and evaluate the service (Bansal, 2000) For instance, a research by Walker (1995) reveals that over

40 percent of Americans actively seek the advice of family and friends when they shop for services

In service sector, researchers find that WOM‟s influence is likely different from one service to another Because of the characteristics of services, consumers make evaluations by three criteria : 1) search qualities of the service which encompass the attributes of the service that consumers can determinate prior to purchase; 2) experience qualities of the service which are described as the attributes

of the service that consumers can ascertain during the consumption after the acquisition (Nelson, 1970); and 3) credence qualities of the service which are very difficult to evaluate (Darby and Karni, 1973) Even if it is hard for consumers to evaluate the credence qualities of a service, they can rely on the reputation criterion, stimulated through positive WOM, to make more detailed evaluations (Bharadwaj, 1993) According to Harrison-Walker (2001), services with high in experience and credence qualities can be tagged as “natural candidates for WOM communication among consumers” As a consequence, WOM which is seen as a highly credible source of information seems to be the most significant in high risk, high credence qualities, and high involvement contexts such as professional and financial services (Hogan et al, 2004)

2.3.3 Effectiveness of WOM

In an environment in which consumers‟ trust of both organizations and advertising has been reduced, WOM gives a way to gain a significant competitive advantage (International Word of Mouth Marketing Conference, 2005) Both of WOM and traditional advertising can be seen as forms of advocacy; however, WOM is perceived free of vested interest while advertising and commercial communication is information from a source having vested interest in presenting the information in a particular way (Silverman, 2011) It is evident that consumers generally view WOM as more credible and trustworthy than marketing communications (Lau & Ng, 2001) Podoshen (2008) attributes to researches in this area by showing that opinions of consumers‟ friends, family and acquaintances are considered credible because they are sincere and without any prejudice In addition, Smith and Swinyard (1982) investigates that a WOM given through direct product experience is more credible, and thus, more effective than advertising The main point to the concept of WOM is the belief that the producer of the product or the service does not control the information; therefore, it is expected to be freer of bias, more relevant, more complete, more trustworthy, and thus more accurate than commercial information (Silverman, 2011)

Research studies generally support the fact that WOM is widely recognized

as one of the most effective communication tools and has a greater influence on consumers‟ decisions compared to other forms of communication (Murray, 1991; Day, 1971) In their study as early as 1955, Katz and Lazarsfeld find out that WOM

is seven times more effective than newspaper ads, four times more effective than direct sales, and twice as effective as radio advertising Day (1971) adds further evidence to the view by stating that WOM is nine times more effective than advertising in changing consumer attitudes from negative or neutral attitudes into positive ones, while Morin (1983) shows that “other people‟s recommendations” are three times more effective than advertising in terms of stimulating purchases of over

60 different products Rusticus (2006) proves the WOM effectiveness when he

investigates that only 14 per cent of people believe what they see, read or hear in advertising while, surprisingly, 90 per cent of people will believe their family, friends, or colleagues who have experienced a service or product because they know they do not have a vested interest in it More recently, Hogan et al (2004) note that WOM can be three times effective than advertising Marsden at al (2005) concludes that WOM is at least twice as powerful as traditional marketing communications in term of influencing sales and is about 50% more effective than it was in 30 years‟ time due to the growth of electronic WOM via mobile and internet

In addition, Trusov et al (2009) go on to conclude that WOM communication plays a significant role when firms want to acquire new customers and that WOM is likely to have larger and longer lasting effects than other marketer-controlled sources Furthermore, nowadays, WOM is considered as powerful communication tool because of its two distinctive functions – persuasion and the ripple effect caused by WOM diffusion (Hogan et al., 2004) WOM‟s diffusion is regarded as a help to generate a ripple effect for marketing activities, which is also called BUZZ marketing or viral marketing (Verlegh and Moldovan, 2008)

2.4 Development of research model and hypothesis

2.4.1 Satisfaction

Within the context of higher education, student satisfaction is regarded as “a short-term attitude resulting from an evaluation of a student‟s educational experience” (Elliott and Healy, 2001) With respect to the factors that affect WOM,

it is believed that there is relationship between satisfaction and the desire for customers to make recommendations for the service provider (e.g Parasuraman et al., 1988) Satisfaction in this view is regarded as a post-purchase evaluation of a product or service quality given pre-purchase expectations (Kotler et al., 2003) Organizations tend to believe that satisfaction is a sufficient marketing objective (Anderson, 1998) and that satisfied customers will automatically spread WOM (Gremler and Brown, 1999) WOM seems to occur when people are either satisfied

or dissatisfied experiencing a product or service (Anderson, 1998) The satisfied mode is based on the level the product or service performance exceeding from customers‟ expectation and is probably resulted in positive WOM , referring to pleasant experiences (Maxham and Netemeyer, 2002;) , while dissatisfied emotion depends on the level customer‟s expectations are not met and may lead to negative WOM, including product denigration, unpleasant experiences, negative feelings, rumor and private complaining to (Richins, 1983) These results illustrate that it is crucial for companies to be able to maximize WOM from highly satisfied customers and to minimize WOM from customers with low levels of satisfaction Findings of Ranaweera & Prabhu (2003) state two types of purchase behavior, including repurchase intention and WOM, are affected by the level of customer‟s satisfaction Furthermore, most authors agree that satisfaction is a key antecedent of WOM (Anderson, 1998; Babin et al., 2005; de Matos & Rossi, 2008) and that in cases of extreme customer satisfaction WOM occurs more frequently (Anderson, 1998) More recently, researchers such as Brown et al (2005), Heitmann et al (2007) and Wangenheim and Bayon (2007) demonstrate that customer‟s satisfaction regarding

a service or product significant affects on the formation of WOM about a company Therefore, within higher education context, we put forward the first hypothesis as follows:

The first hypothesis (H1): student's satisfaction has a positive effect on the formation of positive WOM about higher education institutions

2.4.2 Loyalty

Loyalty, a key factor to achieve company success and sustainability, may be defined as intention or predisposition of a customer to repurchase from the same organization (Edvardsson et al., 2000), which results from the belief that the value received from one provider is greater than the value available from other alternatives (Hallowell, 1996) Helgesen and Nesset (2007) suggest that student loyalty can be related both to the period when a student is enrolled at the university

as well as after the student finishes his or her studies Loyalty is considered to be depend on a favorable attitude that is based on cognitive, affective and conative antecedents and on positive repurchase intentions (Dick and Basu, 1994; Gremler and Brown, 1999) It is indicated that loyal customers tends to make positive recommendation to friends and relatives, have more incentives to get new information as well as resist more negative information about the organization (Dick and Basu, 1994) Conversely, if customers have no loyalty to the firm, they tend to switch to another alternative and probably publish negative words of mouth about the firm to reduce their cognitive dissonances (Wangenheim, 2005) As a consequence, loyalty can be seen as one factor effective on WOM (Bowman and Narayandas, 2001) Based on the above discussion, within higher education context, the following hypothesis is proposed:

The second hypothesis (H2): loyalty has a positive effect on the formation of

positive word of mouth about higher education institutions

2.4.3 Service quality

Service quality is defined as the level of service offered to meet customer expectations and hence, emerged as the result of the interaction between consumer and service provider (Leisen and Vance, 2001) So far, views remain divided when

it comes to the best measure of the service quality construct Parasuraman et al (1988) suggest service quality is a five-dimensional construct, with tangible (facilities, equipment and appearance of staff), reliability (ability to perform the promised service dependably and accurately), responsibility (willingness to help customers and provide prompt service), assurance (knowledge and courtesy of staff and their ability to convey trust and confidence) and empathy (caring, individualized attention the organization provides to its customers) dimensions, and develop the famous SERVQUAL scale to measure these dimensions While SERVQUAL is popular, some researchers have questioned its generalizability (Carman, 1990), as the number of dimensions obtained seems to vary across

industries Studies conducted by Boulding et al (1993) indicates that customers‟ perceptions of service quality are importantly connected with their behavioral reactions beneficial to the service provider (e.g generate positive WOM, recommend the service, etc.)” Empirical studies propose that service quality is a factor determining WOM (Bloemer et al., 1999) because in the context of the service quality field, it can be stated that the consumer with high service quality perception will have tendencies to recommend the service provider, and thus, a positive WOM will occur in this manner In addition, recently researches consistently suggest that service quality leads to customer satisfaction, positive WOM, attraction of new customers, enhanced corporate image, and so forth (Zabkar et al., 2009; Zeithaml et al., 2006) Following the above discussion, we suggest the third hypothesis as below:

The third hypothesis (H3): service quality has positive effect on the

formation of positive WOM about higher education institutions

2.4.4 The perceived value

The perceived value is overall evaluation of customers in concern of the desirability of a product based on perceptions from they receives and what they gives for it (Zeithaml, 1988) In other words, perceived value is the equity concept which refers to customers‟ evaluation of what is fair, right, or deserved for the perceived cost (money, time, energy consumption, etc.) of the offering (Bolton & Lemon, 1999) According to Sirdeshmukh et al (2002), the perceived value of customers makes equable on behavioral intentions of loyalty toward the service provider In addition, Hartline and Jones (1996)‟s research indicates that the perceived value affects on the behavioral intentions of customers, especially on WOM The reason for these intentions is that the more satisfied customers are with their current service providers, the more likely that they believe they are receiving a valuable service from their supplier, and thus, have more commitments to the service provider and try to recommend the members of the reference group to be

loyal to that organization (McKee et al., 2006) Researches in behavioral intentions propose that perceived value is positively correlated to WOM as one of the factors predicting the positive WOM (Hartline and Jones, 1996; Keiningham et al 2007) Hence, the fourth hypothesis is posited as follows:

The fourth hypothesis (H4): the perceived value has positive effect on the

formation of positive WOM about higher education institutions

2.4.5 Trust

Trust, an important factor that affects human relationships at all levels, is generally defined as “perceived credibility (i.e., the trustor believes that the trustee has required ability to take his role effectively and reliably) and benevolence (i.e., the intention of the trustee to do his best for the trustor) of a target of trust (i.e the other party)” (Ganesan, 1994) Tax et al (1998) finds that trust (together with commitment) is a consequence of satisfaction in concern of handling complaint Consistently, Kau and Loh (2006) indicate that customers‟ trust, whether in the whole organization or in the employees, is significantly affected by their satisfaction Moreover, Ranaweera and Prabhu (2003) implement empirical studies

to show that when customers have high level of trust in an organization, they have high tendencies to publish WOM about it Findings of Hennig-Thurau at el (2002) reveal trust may affect on WOM because of its indirect effect through satisfaction

In this case, trust benefits customers in some ways such as descending the level of anxiety, non-assurance and vulnerability and in turn, these benefits affect on customers satisfaction which have influences on the WOM, especially in the field of service which is more complex than the field of product Furthermore, in the service quality literature, trust may also be regarded as „„trust in the service itself‟‟ (Parasuraman et al., 1988) Studies have shown that trust, as one element of relationship quality (along with satisfaction and commitment), has positive influences on WOM (Dwyer et al., 1987; Hennig-Thurau et al., 2002) According

to the literature stated, the fifth hypothesis can be suggested as follows:

The fifth hypothesis (H5): trust has a positive effect on the formation of positive WOM about higher education institutions

2.4.6 Reference intention to enroll

Purchase intention involves subjective judgment for future behavior (Engel

et al., 2001); in other words, it stands for what we would like to buy in the future According to Lin and Lu (2010), purchase intention may cover some meanings, such as referring to the possibility for consumers to be „„willing‟‟ to consider buying; representing what a person „„wants‟‟ to buy in the future; revealing the decision of a consumer to „„buy‟‟ a company‟s product „„again.” Researches state that repurchase and positive WOM would be positively correlated to one another (Bloemer et al., 1999; Ewing, 2000) Moreover, studies of different service sectors have proved effects of words of mouth on purchasing the services provided by service companies in different companies For example, Smith and Vogt (1995) finds in their study of WOM and advertising and the hotel industry that negative WOM mitigates consumers „trust in advertisements, brand preference, and purchase intention Oneil et al (2002) do a research in tourism in Australia and find that verbal recommendations of communicators lead to the increase in the sale of local drinks Consistently, Yasvari et al (2012) make a research in Iran airline companies and their results prove that WOM can affect on customers‟ intention of referring to service companies Based on the discussions above, we prose the following hypothesis in context of higher education:

The sixth hypothesis (H6): Positive WOM has a positive significant effect

on the intention of students to refer to enroll higher education institutions

2.4.7 Proposed conceptual model and hypotheses

With a review of the literature in the field of word of mouth above, six proposed hypotheses can be displayed as the conceptual model in figure 2

H4

H2

H3

Figure 2.1 Proposed conceptual model of the research

H 1 : Satisfaction has a positive influence on positive word-of-mouth

- H 2 : Loyalty has a positive influence on positive word-of-mouth

- H 3 : Service quality has a positive influence on positive word-of-mouth

- H 4 : Perceived value has a positive influence on positive word-of-mouth

- H 5 : Trust has a positive influence on positive word-of-mouth

- H 6: Positive word-of-mouth has a positive influence on reference intention

CHAPTER 3 RESEARCH METHODOLOGY

3.1 Introduction

This chapter presents the research methodology used to modify and evaluate the measurement scales, the research model, and the hypotheses This chapter includes 4 main parts: (1) research design, (2) measurement scales of observation variables, (3) pilot test of measurement scales, and (4) analysis method of main study

in the measurement scales Based on the results, necessary adjustments were made and the final draft questionnaire in Vietnamese could be completed for the main study

In the second step, the quantitative pilot study was conducted with an initial collection of 102 samples to test the reliability of the variables in order to refine the measures Cronbach‟s alpha reliability and exploratory factor analysis (EFA) were employed to assess the scales

After the pilot study, the final measure scales were tested in terms of validity and reliability using Cronbach‟s alpha test and EFA, and confirmatory factor analysis (CFA) Then, the structural model of the research was statistically measured to test the hypotheses and the fitting model

3.2.2 Research process

This study has been conducted in a process showed in Figure 3.1

Figure 3.1 Research process

SEM

Report Writing

Qualitative (n=5) In-depth discussion

Cronbach‟s Alpha + EFA

3.3 Measurement scales and questionnaire development

3.3.1 Measurement scales

Data collection was done through a questionnaire having seven sections including satisfaction, loyalty, service quality, perceived value, trust, word-of-mouth and intention of referring to the higher education institutions Each of these sections was measured by three items These items were based on Yasvari (2012) All items were measured using 5-point Likert scales anchored by 1 = strongly disagree and 5 = strongly agree

Table 3.1 Measurement scales of variables

SAT1 I‟m satisfied of my decision to enroll at X

SAT2 I‟m satisfied of the services provided by X

SAT3 Services of X have met my needs and demands

LOY1 I would enroll at X if I wanted to study

LOY2 My future study will be by X

LOY3 I‟m loyal to X

SER1 The quality of the services of X are at high level

SER2 The service quality of X is better than other universities

SER3 The services of X have a high standard

PER1 Compared to other universities, X charges better prices for the quality of services

TRU1 X always meets its commitments

TRU2 I trust to the professors of X for service provision

TRU3 X tries to protect from information of students

WOM1 I‟ll say positive things about X to the others

WOM2 I‟ll recommend X to people asking my opinion

WOM3 I encourage my friends and my relatives to enroll at X

INT1 I‟ll refer and enroll at X in the future

INT2 I prefer the services of X to other universities

INT3 Compared to other universities, X will be my first choice

3.3 2 Questionnaire development

According to Malhotra (1996), a questionnaire (or called measuring instrument) has three specific objectives: (1) transforming necessary information related to the study into a list of questions that the respondents could answer, (2) having good dynamic to encourage respondents to participate in the interviews and complete the questionnaire; and (3) ensuring minimum response errors

In this study, the questions to measure seven sections of variable were translated into Vietnamese and back-translated (i.e translating version back to the source language) in line with cross cultural translation procedures as instructed by Brislin (1980) (cited in Harkness & Schoua-Glusberg, 1998) Craig and Douglas (2000) pointed out that the original and back-translation should be compared with each other in order to ensure the meaningful of the questionnaires Parallel translation into Vietnamese was done by the participation of two MBA friends of the author as translators Then the author asked for the help of two English teachers

to back-translate in to English version Based on the original questionnaire, the author compared two simultaneous and independent translations and chose the final one for draft questionnaire in Vietnamese

3.3.3 Questionnaire design

In term of questionnaire structure, questionnaire was divided into 3 main sections In the first section, a screening question was asked to find whether the

respondent have experiences in enrolling at a high educational institution in Hochiminh city If yes, the interviewee went on to the second section; if no, they stopped In the second section, the interviewees were asked to state the extent they agree/disagree about 21 statements related to their behaviors to the higher education institution in Hochiminh city which they have experiences in enrolling These 21 statements were used to measure 7 observation variables (three statements for each variable) in the research model: satisfaction, loyalty, service quality, perceived value, trust, word of mouth, and reference intention Each item was measured on five-point Likert-type scales on which 1 represents “strongly disagree” and 5 represents “strongly agree” Lastly, in the third section, the interviewees were asked

to provide some their personal information: gender, age, education, university Detailed content of the questionnaire was presented in Appendix 1

3.4 Pilot study

3.4.1 Qualitative pilot study

When the draft questionnaire was completed, the author conducted the pretest (or called alpha test) with the participation of a group of 5 people studying in MBA course at Hochiminh City University of Economics Participants of this study would not be interviewed in the main study in next step Comments on the translation of the questionnaire were mainly discussed Some revised translations about structures and terms were contributed to make the content of questionnaire more understandable After the revision, the author got the final draft questionnaire

3.4.2 Quantitative pilot study

In this quantitative pilot study (or called beta test), the final draft questionnaire was used to interview Based on the 62 initial samples collected, the evaluation of the questionnaire was examined (e.g., did the respondents of the study understand the questions correctly? Did they have information to answer? Were they willing to provide information?) Moreover, measurement scales were preliminarily tested in term of reliability and validity by using Cronbach‟s alpha test

and exploratory factor analysis Consequently, the author had the final questionnaire ready for the main study

3.5 Quantitative main study

3.5.1 Research sampling

Sample size: Sample size is often identified based on (1) minimum sample size and (2) number of measurement variables (Nguyen Dinh Tho, 2011) According to Hair et al (2006) (cited in Nguyen Dinh Tho, 2011), the sample size should be at least 50, at best 100 and the observation/items ratio should be 5:1, so that the exploratory factor analysis could be applied It means that each item would require at least 5 observations In this study, because there were a total of 21 measurement items, the required sample size of this study was 105 In addition, Ding at all (1995) also suggests that minimum sample size should be at least 100 to

150 to apply Maximum Likelihood method in SEM Therefore, the total of 223 valid responses for sample size received in this study was satisfied

Sample strategy: People who have experienced in enrolling higher educational institutions in Hochiminh city is regarded as population in this research Samples were selected based on non-probability convenience sampling method

3.5.2 Collecting data

In the quantitative main study, the author collected data by handing out 260 paper datasheet of questionnaires The author received back 244 sheets After cleaning, the author had a data of 223 qualified respondents to use in data analysis

3.5.3 Methods of data analysis

Data collected from questionnaires was analyzed using SPSS version 16.0 and Amos version 21.0 For data processing, some statistical techniques were used for different purposes, including frequency, descriptive statistics, validity and reliability test, correlation analysis and structural equation model

3.5.4 Data cleaning

Data would be cleaned before analyzing Data cleaning was made to prevent any possible mistakes which are probably missing data or unreasonable answers (Nguyen Dinh Tho, 2011, p.27) The author reviewed all the questionnaire sheets and eliminated 21 disqualified sheets for missing data

3.5.5 Reliability assessment of measurement scales (Cronbach's alpha)

Cronbach‟s alpha is a tool for assessing the reliability of scales It was provided by Lee Cronbach in 1951 and is widely used the social science, business, and other fields The formulation of Cronbach‟s alpha is:

Where K is the number of components or variables,2X is the variance of the

observed total test scores, and

2 i Y

the variance of component i for the current samples of interviewees

A measurement scales would be considered reliable if Cronbach‟s alpha coefficient varies from 70 to 80 If the Cronbach‟s alpha is bigger or equivalent

to 60, the scale is acceptably reliable (Nunnally & Bernstein, 1994) For variables that the item-total correlations were higher or equivalent to 30, they would be qualified (Nunnally & Bermstein, 1994)

3.5.6 Exploratory factor analysis (EFA)

After the reliability of the scale was evaluated by using the Cronbach‟s alpha analysis, the author implemented the exploratory factor analysis (EFA) to assess the preliminary validity of the scales The EFA method would help to identify the

collection of variables necessary for the study matters and find out the relations between variables The author follows some criteria as follows in EFA stage :

(1) Sample size: Hair et al (2006) indicate that using the EFA method, the sample size should be at least 50, better 100 and the observation/items rate should

be 5:1, meaning that 1 measurement variable would require at least 5 observations,

at best 10:1 In this study, the sample size was 223 and more than 10 times bigger than the observed variables (21 items) So the sample size was eligible for the use of EFA method

(2) Kaiser–Meyer–Olkin (KMO) and Bartlett‟s Test: If the KMO indicator runs from 5 to 1, EFA would be applicable (Kaiser, 1974) The Bartlett‟s Test shows that p < 05, it means that the variables have correlation together and the application of EFA is suitable

(3) Factor Loading: According to Hair et al (1998), factor loading is an indicator ensuring practical significance of EFA method If factor loading ≥ 30, it‟s at minimum level; if factor loading ≥ 40, it is necessary and if it is ≥ 50, it is considered practical significance Hair et al (1998) advised that if factor loading ≥ 30, the sample size should be at least 350 If the sample size is about 100, we should choose the factor loading ≥ 55 If the sample size is about 50, the factor loading should be ≥ 75

(4) Only factors with Eigenvalues value ≥ 1 would be kept in the model (Anderson & Gerbing, 1988)

(5) Variance explained criteria: The total variance explained should not be lower than 50%

(6) To satisfy discriminant validity of the scale, the differences in factor loading values of variables in different factors should be bigger or equivalent to 30 (Jabnoun et al., 2003)

3.5.7 Confirmatory factor analysis (CFA) and SEM:

In this step, the author relied on some statistical tests determine how well the model fits to the data such as Chi-square test, the ratio of Chi-square and its degree

of freedom (CMIN/df); GFI, NFI, CFI, TLI, AGFI, RMSEA

* Chi-squared test: The chi-squared test indicates the difference between observed and expected covariance matrices For models with about 75 to 200 cases, the chi square test is a reasonable measure of fit But for models with more cases (400 or more), the chi square is almost always statistically significant Chi square is also affected by the size of the correlations in the model: the larger the correlations, the poorer the fit For these reasons alternative measures of fit have been developed (Kenny, 2013)

* Root mean square error of approximation (RMSEA): The RMSEA avoids issues of sample size by analyzing the discrepancy between the hypothesized model, with optimally chosen parameter estimates, and the population covariance matrix (Hooper et al., 2008) The RMSEA ranges from 0 to 1, with smaller values indicating better model fit According to Arbuckle (1999), if RMSEA=0 the model

is exact/good fit; if RMSEA<0.05: close fit, RMSEA<0.08: mediocre fit If 0.08<RMSEA<0.01 , the model can be regarded as reasonable error of approximation (Arbuckle, 1999) or mediocre fit (Byrne, 2001)

* Goodness of fit index and adjusted goodness of fit index: The goodness of fit index (GFI) is a measure of fit between the hypothesized model and the observed covariance matrix The adjusted goodness of fit index (AGFI) corrects the GFI, which is affected by the number of indicators of each latent variable The GFI and AGFI range between 0 and 1, with a cutoff value of 9 generally indicating acceptable model fit (Baumgartner & Hombur, 1996)

* The normed fit index (NFI) and non-normed fit index (NNFI): The NFI analyzes the discrepancy between the chi-squared value of the hypothesized model and the chi-squared value of the null model However, this NFI was found to be

very susceptible to sample size (Bearden et al., 1982).The non-normed fit index (NNFI) also known as the Tucker-Lewis index (TLI) resolves some of the issues of sample size Note that the TLI (and the CFI which follows) depends on the average size of the correlations in the data If the average correlation between variables is not high, then the TLI will not be very high (Kenny, 2013)

* Comparative fit index: The comparative fit index (CFI) analyzes the model fit by examining the discrepancy between the data and the hypothesized model, while adjusting for the issues of sample size inherent in the chi-squared test of model fit, and the normed fit index CFI values range from 0 to 1, with larger values indicating better fit; a CFI value of 90 or larger is generally considered indicating acceptable model fit (Hu & Bentler, 1999)

Fit refers to the ability of a model to reproduce the data (i.e., usually the variance-covariance matrix) A good-fitting model is one that is reasonably consistent with the data and so does not necessarily require respecification Also a good-fitting measurement model is required before interpreting the causal paths of the structural model (Kenny, 2013) If a model has GFI, TLI, CFI (Bentler & Bonett, 1980); and CMIN/df <2 or some case CMIN/df <3 (Carmines & McIver, 1981); RMSEA ≤ 0.08, it fits the data (Cited in Nguyen Khanh Duy, 2009) This rule is also used in SEM stage

Cronbach's alpha coefficient was used again to determine the reliability of the questionnaire in this step CFA was used to test the validity of the questionnaire The validity test includes two convergent and discriminate validities The convergent validity shows that whether the questions can state the factor corresponding to them The discriminate validity measures the significance of the difference between two factors Factor loads are used to examine the convergent validity If the values of the factor loads are more than 0.5 at significant P <0.05, the convergent validity is desirable (Gerbring & Anderson, 1988) If the correlation

coefficients among the factors are lower than 0.85, it can be argued that the measurement tool has a suitable discriminate validity (Kline, 2005)

Structural Equation Modeling (SEM) using the AMOS Software version 21 was employed to test the hypotheses and the fitting of the model The parameters and the total fit index of the measurement model were estimated based on the Maximum Likelihood (ML) method The fit indexes are same as CFA above

3.5.8 Bootstrap Method

In this study, the author used bootstrapping as a method for assigning measures of accuracy to sample estimates This method was applied to test the final model with N=500 bootstrap samples (more than 2 times compared to 223 samples of the research) If absolute Critical ratio (CR) values of 1.96 (t value in Student's t-distribution at significant level 0.05) or less it is meant there is non-significant kurtosis and non-statistical meaning and hence, estimations in the model is reliable (Nguyen Khanh Duy, 2009)

21

4.2 Descriptions of sample

This research was undertaken in Ho Chi Minh City The population of this study was Vietnamese people who had experiences in enrolling in higher educational institutions in Hochiminh city The qualified responses were 223 samples Results of the descriptive statistics show that 58.3% (130 respondents) were female and 41.7% (93 respondents) were male 20.2 % (45 respondents) were under 22 years old, 21.5% of them were 22 to 25 years old, 30.9% of them were 26

to 30 years old, 24.7% of them were 31 to 35 years old and 2.7% (6 respondents) were over 35 years old In concerning of

higher educational institutions, 30.9% respondents were enrolling at Hochiminh

City University of Economics ( i h c Kinh t ), 19.3% of them were at Hochiminh City University of Technology ( i h c Bách Khoa); 6.3% of them were at Hochiminh City University of Natural Sciences ( i h c Khoa h c T Nhiên); 7.2% of them were at Hochiminh City Open University; 4.5% of them were at Ton Duc Thang University ( h c Tôn c Th ng), 4.9% of them were at Banking University of Hochiminh City; and 26.9% (60 respondents) were from other 26 higher educational institutions in Hochiminh City In addition, 32.7% of them had master degree, whereas bachelor‟s degree, second bachelor‟s degree certificate, and

junior college degree distributed 52.5%, 4.9%, and 9.9% respectively

Table 4.1 Demographic characteristics of the respondents

4.3 Reliability and validity of the measurement scale

In this study, the reliability of the measurement scale was evaluated by Cronbach‟s alpha coefficient analysis and the validity of the measurement scales were evaluated by Exploratory Factor Analysis (EFA), and Confirmatory Factor Analysis (CFA)