VIETNAM NATIONAL UNIVERSITY – HO CHI MINH CITY

INTERNATIONAL UNIVERSITY
SCHOOL OF BUSINESS

IDENTIFYING FACTORS AFFECTING THE LEVEL

OF IMPULSE BUYING IN FASHION MALLS
IN HO CHI MINH CITY

In Partial Fulfillment of the Requirements of the Degree of

BACHELOR OF ARTS in BUSINESS ADMINISTRATION

Student’s name: TA BAO TRAN (BABAIU11081)

Advisor: BUI QUANG THONG, Ph.D.

Ho Chi Minh City, Vietnam

May, 2015

i
IDENTIFYING FACTORS AFFECTING THE LEVEL
OF IMPULSE BUYING IN FASHION MALLS
IN HO CHI MINH CITY

APPROVED BY: Advisor APPROVED BY: Committee

_____________________ _______________________
Bui Quang Thong, Ph.D Nguyen Minh Tuan, MBA

________________________

Le Nhat Hanh, Ph.D

________________________
Nguyen Ngoc Duy Phuong, Ph.D

THESIS COMMITTEE

ii
 ACKNOWLEDGEMENT
F irst and foremost, it is a pleasure to express my sincere gratitude to many people
who made this thesis possible. I wish to give my deepest gratitude and appreciation
to my profound advisor, Dr. Bui Quang Thong, for his continuous support and
encouragement. This thesis would not have been completed without his help. Throughout
my thesis-writing period, he encouraged me whenever I encounter problems with careful
and inspiring advice, good teaching and lots of interesting ideas. His extensive
knowledge in marketing as well as deep understanding in business research has given me
invaluable experience in conducting scientific study in marketing field. Especially in the
data analysis, thanks to his useful tutorials and recommendations, I had overcome the
difficulties of statistic process and software. In short, it is an honor for me to be instructed
by Dr. Bui Quang Thong who never accepted less than my best efforts.
Additionally, I would like to thank Dr. Le Van Chon for supporting me in raising
the thesis’s topic and widening useful knowledge through the Business Research Method
course. His instructions and advices are valuable to me.
My gratitude also goes to all professors in International University, especially
professors and staffs in School of Business Administration, who always do their best to
make all thesis procedures as convenient to students as possible.
A word of appreciation must go out to all of my close friends who aided and
supported me during the time conducting this thesis, Mr. Le Minh Khoa, Mr. Nguyen
Trung Hung, Ms. Banh Ngoc Nhu and Ms. Nguyen Phuong Hoang Cuc. Without their
cooperation, I could not have accomplished my thesis.
Last but not least, I am appreciate to those who spent time on spreading and
completing my questionnaire. Their responses are the valuable information and big
encouragement for me in private and for those who want to have further investigation in
this research topic in general.

iii
 Finally, my beloved family deserves a special mention for their unconditional
love and support throughout the whole research study.
To all of them, I dedicate this thesis.
Ho Chi Minh, January, 2015.

iv
 Table of Contents
ACKNOWLEDGEMENT ................................................................................................................. III
LIST OF TABLES .......................................................................................................................... VIII
LIST OF FIGURES ............................................................................................................................. X
ABSTRACT ........................................................................................................................................ XI
ABBREVIATIONS ......................................................................................................................... XIII
CHAPTER I......................................................................................................................................... 14
1.1. RESEARCH BACKGROUND .......................................................................................................... 14
1.2. RESEARCH OBJECTIVES .............................................................................................................. 17
1.3. RESEARCH QUESTIONS ............................................................................................................... 17
1.4. SIGNIFICANCES OF THE RESEARCH ............................................................................................. 18
1.5. RESEARCH SCOPE AND LIMITATIONS ......................................................................................... 19
1.5.1. Research scope ................................................................................................................... 19
1.5.2. Research limitations ........................................................................................................... 19
1.6. RESEARCH STRUCTURE .............................................................................................................. 19
1.7. THESIS TIMELINE ....................................................................................................................... 20
CHAPTER II ....................................................................................................................................... 22
2.1. DEFINITION OF TERMS................................................................................................................ 22
2.1.1. Optimum Stimulation Level ................................................................................................ 22
2.1.2. Hedonic Purchase .............................................................................................................. 23
2.1.3. Consumer’s Need for Uniqueness ...................................................................................... 23
2.1.4. Novelty- Fashion Consciousness........................................................................................ 26
2.1.5. Consumers Self- Spending Control .................................................................................... 26
2.1.6. Level of Impulse Buying ..................................................................................................... 27
2.2. RESEARCH HYPOTHESES ............................................................................................................ 28
2.2.1. Relationship between Optimum Stimulation Level and Level of Impulse Buying .............. 28
2.2.2. Relationship between Hedonic Purchase and Level of Impulse Buying ............................ 28
2.2.3. Relationship between Consumer’s Need for Uniqueness and Level of Impulse Buying .... 29
2.2.4. Relationship between Novelty – Fashion Consciousness and Level of Impulse Buying .... 30
2.2.5. Relationship between Consumer Self-Spending Control and Level of Impulse Buying ..... 30
2.3. RESEARCH MODEL ..................................................................................................................... 32
CHAPTER III...................................................................................................................................... 34
3.1. NATURE OF THE RESEARCH........................................................................................................ 34
3.2. RESEARCH METHOD ................................................................................................................... 34
3.3. DATA COLLECTION METHOD ..................................................................................................... 36
3.4. RESEARCH DESIGN ..................................................................................................................... 36
3.5. SAMPLING DESIGN ..................................................................................................................... 38
3.5.1. Sample Size ........................................................................................................................ 38
3.5.2 Sampling Technique ............................................................................................................ 38
3.5.3. Sampling Locations ............................................................................................................ 38

v
 3.6. QUESTIONNAIRE DESIGN ............................................................................................................ 39
3.7. PILOT STUDY .............................................................................................................................. 44
3.8. DATA COLLECTION PROCEDURE ................................................................................................ 44
3.9. DATA ANALYSIS METHOD ......................................................................................................... 46
CHAPTER IV ...................................................................................................................................... 48
4.1. SAMPLE DEMOGRAPHIC ............................................................................................................. 48
4.1.1. Response Rate .................................................................................................................... 48
4.1.2. Demographic Analysis ....................................................................................................... 48
4.1.2.1. Gender ........................................................................................................................ 49
4.1.2.2. Age ............................................................................................................................. 50
4.1.2.3. Occupation .................................................................................................................. 51
4.1.2.4. Income ........................................................................................................................ 52
4.2. DESCRIPTIVE STATISTICS ........................................................................................................... 52
4.2.1. Descriptive Statistics of Optimal Stimulation Level ........................................................... 53
4.2.2. Descriptive Statistics of Hedonic Purchase ....................................................................... 53
4.2.3. Descriptive Statistics of Consumer’s Need for Uniqueness (in sub -dimensions) .............. 54
4.2.4. Descriptive Statistics of Novelty – Fashion Consciousness ............................................... 56
4.2.5. Descriptive Statistics of Consumer Self-Spending Control ................................................ 57
4.2.6. Descriptive Statistics of Level of Impulse Buying .............................................................. 58
4.3. EXPLORATORY FACTOR ANALYSIS (EFA) ................................................................................. 59
4.3.1. Reliability Analysis............................................................................................................. 59
4.3.2. Factor Analysis .................................................................................................................. 64
4.3.2.1 EFA for Independent Variables ................................................................................... 65
4.3.2.2 EFA for Dependent Variable ....................................................................................... 68
4.4. DETECTING MULTIVARIATE OUTLIERS ...................................................................................... 70
4.5. CONFIRMATORY FACTOR ANALYSIS (CFA) ............................................................................... 73
4.5.1. Check for Model fit............................................................................................................. 76
4.5.2. Check Convergent Validity ................................................................................................ 83
4.5.3. Check Reliability ................................................................................................................ 86
4.5.4. Check Discriminant Validity .............................................................................................. 89
4.6. STRUCTURAL EQUATION MODELING (SEM).............................................................................. 90
4.6.1. Check for Model Fit ........................................................................................................... 92
4.6.2. Hypotheses Testing in SEM .............................................................................................. 107
4.6.3. Revised Research Model .................................................................................................. 108
CHAPTER V ..................................................................................................................................... 111
5.1. CONCLUSION ............................................................................................................................ 111
5.2.1. Recommendations for Retailing Strategies ...................................................................... 113
5.2.2. Recommendations for Future Research Study ................................................................. 114
5.3. LIMITATIONS ............................................................................................................................ 115
REFERENCES .................................................................................................................................. 117
APPENDIX A .................................................................................................................................... 127
QUESTIONNAIRE IN ENGLISH ................................................................................................. 127
QUESTIONNAIRE IN VIETNAMESE .......................................................................................... 130

vi
APPENDIX B .................................................................................................................................... 135
APPENDIX C .................................................................................................................................... 141
APPENDIX D .................................................................................................................................... 143
APPENDIX E .................................................................................................................................... 151

vii
 LIST OF TABLES
Table 1: Research Timeline .............................................................................................. 21
Table 2: Measurement Items for the Research Model ...................................................... 40
Table 3: Valid Response Rate ........................................................................................... 48
Table 4: Demographic Information .................................................................................. 48
Table 5: Descriptive Statistics of Optimum Stimulation Level ........................................ 53
Table 6: Descriptive Statistics of Hedonic Purchase ........................................................ 53
Table 7: Descriptive Statistics of Consumer's Need for Uniqueness's sub-dimensions ... 55
Table 8: Descriptive Statistics of Novelty-Fashion Consciousness .................................. 56
Table 9: Descriptive Statistics of Consumer Self-Spending Control................................ 57
Table 10: Descriptive Statistics of Level of Impulse Buying ........................................... 58
Table 11: Rules of Thumb for Evaluating Cronbach's Alpha ........................................... 59
Table 12: Corrected Item-to-Total Correlation and Cronbach's Alpha (First Round) ...... 60
Table 13: Rotated Component Matrix of Independent Variables ..................................... 65
Table 14: KMO and Barlett's Test of Independent Variables ........................................... 67
Table 15: Total Variances Explained of Independent Variables ...................................... 68
Table 16: Component Matrix of Dependent Variable ...................................................... 69
Table 17: KMO and Bartlett’s Test of Dependent Variable ............................................. 69
Table 18: Total Variance Explained of Dependent Variable ............................................ 69
Table 19: Observations Farthest from the Centroid (Mahalanobis Distance) .................. 70
Table 20: Criteria for Measurement Model ...................................................................... 74
Table 21: Model Fit of CFA (First Round) ....................................................................... 76
Table 22: Modification Indices Value and Questionnaire Per Items ................................ 78
Table 23: Model Fit of CFA (Final Round) ...................................................................... 83

viii
Table 24: Regression Weights: (Group number-Default model) ...................................... 84
Table 25: Standardized Regression Weights .................................................................... 85
Table 26: Composite Reliability and Extracted Variance of Each Factor ........................ 87
Table 27: Correlations (CFA) ........................................................................................... 89
Table 28: Model Fit of SEM (First Round) ...................................................................... 92
Table 29: Regression Weights of SEM Testing: (Group number 1-Default Model) (First
Round)............................................................................................................................... 92
Table 30: Covariance: (Group number 1-Default model) (First Round) .......................... 95
Table 31: Model Fit of SEM (Second Round) ................................................................ 100
Table 32: Covariances (Group number 1: Default model) (Second Round) .................. 101
Table 33: Regression Weights of SEM Testing (Group number 1-Default) (Second
Round)............................................................................................................................. 101
Table 34: Regression Weights of SEM Testing (Group number 1-Default model) (Third
Round)............................................................................................................................. 103
Table 35: Model Fit of SEM (Third Round)................................................................... 106
Table 36: Covariances: (Group number 1-Default model) (Third Round) ..................... 106
Table 37: Standardized Regression Weights .................................................................. 107
Table 38: Hypotheses Testing in SEM ........................................................................... 108

ix
 LIST OF FIGURES
Figure 1: Vietnam GDP Per Capita from 2006 to 2014.................................................... 15
Figure 2: Research Model ................................................................................................. 32
Figure 3: Main Steps of Research Design......................................................................... 37
Figure 4: Gender Percentage ............................................................................................. 49
Figure 5: Age Group Percentage....................................................................................... 50
Figure 6: Occupation Percentage ...................................................................................... 51
Figure 7: Income Percentage ............................................................................................ 52
Figure 8: First Measurement Standardized Modelling ..................................................... 76
Figure 9: Final Measurement Standardized Modelling .................................................... 83
Figure 10: First Structural Equation Model ...................................................................... 91
Figure 11: Second Structural Equation Model.................................................................. 99
Figure 12: Final Structural Equation Model ................................................................... 105
Figure 13: The Revised Research Model ........................................................................ 109

x
 ABSTRACT
Impulse buying is a major research concern among researchers due to its
pervasive aspects of consumer behavior as well as its mystery in the marketing world. As
counted, impulse buying has a long story of being taken into research for the past sixty
years since 1950 by Clover until the recent research paper conducted in by Chang et al. in
2011. Thus, factors affecting impulse buying have long been explored with diverse
aspects, which associates with shopping atmosphere, shopper’s personal characteristics,
the various demographic and socio-cultural features as well as product itself. However, in
general view, most of the study on impulse purchase concentrated on the external features
impulse buying. For instance, Abratt and Goodey (1990) indicated that in-store
environment such as Point- of- Purchase (POP) posters can stimuli the impulse purchase.
Next, the early work of Rook discovered the fundamental nature of impulse buying
(Rook, 1987) and later explored the normative influences effect on it, where extrinsic
impacts were highlighted. (Rook and Fisher, 1995). In 1998, Jones and Beatty
continuously proposed that impulse buying is probable to impact on individuals across
circumstances. After that, Zhou and Goong (2003) re-confirmed that impulse buying
could be influenced by retail store stimuli such as POP. Recently, the study of Chang et
al. (2011) contributed some direct and indirect influences of retail environmental traits on
impulse purchase. However, most of these researches has not brought a fully concept of
impulse buying. Perhaps, they did lots of investigation on external variables affecting
impulse buying behavior while fail to provide internal variables, which also caused
impact on this consumer behavior. Besides, results from past studies shows that the
extrinsic variables did not support much to the increasing of impulse buying of
consumers. As a result, it is necessary to explore more important internal variables which
are expected to cause stronger effect on impulse buying to provide a better understanding
of the impulse buying concept and assist more appropriate marketing decisions in the

xi
field of marketing activities. Thus, the objective of this research is not out of
investigating factors affecting level of impulse buying, which not only examines new
intrinsic variables influencing on impulse purchase but also check their level of effect.
In this study, to measure the level of impulse buying of factors, a research
study was formed and presented clearly in the methodology chapter. Target populations
are those living in Ho Chi Minh City and did make impulse purchases in the past or at the
time they were asked. In general speaking, data was collected from 355 respondents in
total by questionnaires at some famous fashion malls in Ho Chi Minh City as well as
other offline and online channels which will be well discussed later. The findings of the
data were assessed in terms of the level of effect on impulse buying of factors compared
to those conducted in many previous research papers. Although the research study
hypothesized five independent factors causing influences on impulse buying, which were
optimum stimulation level, hedonic purchase, consumer’s need for uniqueness,
consumer’s self-spending control as well as novelty-fashion consciousness, however, it
was found that only optimum stimulation level, hedonic purchase and consumer’s need
for uniqueness had positive impact on the level of impulse buying while consumer’s self
– spending control was pointed out having negative effect on the dependent factor.
Novelty-fashion consciousness was explored to have no impact on impulse buying as
concluded by the result from data analysis.
Thus, it will be much valuable for both overseas and local fashion brands/
retails to implement the results and turn into practical marketing activities to attract more
impulse buyers. Besides, it also useful for the researchers to develop the theoretical
frameworks and replicate in other industries.
Key words: impulse buying, impulsiveness, consumer behavior, fashion
brands, retailing, hedonic consumption, optimum stimulation, novelty-fashion
consciousness, consumer’s need for uniqueness, consumer self –spending control.
Paper type: Research paper

xii
 ABBREVIATIONS
AVE: Average Variance Extracted
CFA: Confirmatory Factor Analysis
CNFU: Consumer’s Need for Uniqueness
CSSC: Consumer Self-Spending Control
CSI: Consumer Style Inventory
EFA: Exploratory Factor Analysis
HP: Hedonic Purchase
IB: Impulse Buying
NFC: Novelty – Fashion Consciousness
OSL: Optimum Stimulation Level
POP: Point – of – Purchase

xiii
 CHAPTER I

INTRODUCTION
This chapter provides information about the rationale of this paper. From that
background, research problems are stated and research objectives are set in order to
solve such problems. Furthermore, scope and limitations as well as implications of this
study are discussed. Lastly, the overall structured of the whole research has been taken
into consideration.
1.1. Research Background
Shopping has long been explored as an endless demand for human being,
which was proved that can bring the joyfulness as well as satisfaction for the buyers.
Thus, it is commonly understand that people are easily to fall into impulse purchase when
they do not even tend to buy these products, or do not need as a must-have-item in the
shopping list. Most of them only recognized how overspending they were when they
went home, however, they may continuously made spontaneous decisions many times
later when seasons sales came or new collections arrived. It has been reported that the
number of shopaholics is increasing remarkably in developed countries, especially
women, and most of them are called as slaves to fashion and debt. (Moussa, 2007).
Nevertheless, under the point of view of fashion brands/ retailers, it is a good news due to
the growing sales when more and more people are impulse buyers.
The power of impulse buying has been accredited by marketing practitioners
and researchers through various statistical data sources from past studies. In 1978,
Bellenger et al. has reported that about 40% of annual volume from department stores is
generated by impulse purchases. Mogelonsky (1998) also gave more evidence on the
power of impulse buying by showing that around $4.2 billion of total revenue gained in
fashion stores originated from impulse buying. In particular product categories, impulse

14
purchase takes a percentage of 80% among consumers buying behavior. (Smith, 1996;
Abraham, 1997). This denotes that impulse buying has strong effects on consumer
behavior, which is worth to be conducted in research paper in every aspects.
However, most research paper conducted has origin from developed countries
such as United States, the United Kingdom (McConatha, Lightner, and Deaner, 1994;
Dittmar, Beattie, and Friese 1995; Bayley and Nancarrow, 1998) and Singapore as
Shamdasani and Rook (1989). Almost no previous research study about impulse buying
has empirically examined in developing countries before, thus, this study is conducted in
Vietnam – the promising developing countries to find out the level of impact of factors
on impulse buying, then give implications managements for both local and international
firms.
As reported in 2014, Vietnam has a population of over 90 million, which
ranked 14th in the list of countries by population. Besides, Vietnam has higher population
density in urban areas, especially in big cities such as Ho Chi Minh City, Ha Noi or Da
Nang. (Statistics reported from General Statistics Office of Vietnam). Furthermore, GDP
per capita kept increasing steadily with the latest record around US $1,030 in 2014.

Figure 1: Vietnam GDP Per Capita from 2006 to 2014

15
 (Sources: www. tradingeconomics.com)
As reported by Vietnam General Statistics Office, in 2014, GDP per capita of
Vietnam reached to US $1,030. Furthermore, total personal disposable income of
Vietnam was US$127 billion in 2013. Next, the total consuming volume of Vietnamese
consumers was about US$111 billion in the same year (Deloitte). Continuously reported
by Deloitte, Vietnam’s retail sales, which already reached to US$53.9 billion in 2011 in
total volume, is forecasted to grow up to 109 billion USD in 2017. Especially Ho Chi
Minh City – the most dynamic city as well as the economic central of Vietnam, had the
retail’s sales up to US$28.9 billion in 2013 (cited by Deloitte in Retail in Vietnam,
Emerging Market, and emerging growth).
Since Vietnam’s retailing industry was ranked at 14th out of 30 countries, it
can be said in confidence that Vietnam is worth to be invested and conducted many
researches on consumer behavior. Thus, Ho Chi Minh City, especially fashion malls in
this city, were chosen as sampling locations to conduct this research paper. Ho Chi Minh
City is not only the most dynamic city of Vietnam but also the economic central,
accordingly, impulse buying is expected to occur in these potential locations. In
particular, the underlying purpose of this research is to explore more stimuli behind the
decisions of buying spontaneous within a physical shopping circumstances like fashion
malls in Ho Chi Minh City.
Fashion industry always takes central stage all over the world compares to
other industries, however, this industry has not received equals treat for its development
among countries when in some countries, fashion industry is still in low consideration. In
Vietnam, fashion industry is moving to the development phase, which is recognized as
“slowly marches ahead” by Fibre2fashion News Desk – India. Currently, foreign and
local retailers are competitively fight for a share in this promising market. While high-
income customers are aimed to be targeted by foreign investors, local competitors
identify an increasing needs for clothing from the low and middle income customers,
thus, they invested in casual clothing products for this group at the affordable prices. In

16
2011, the largest categories in Vietnam clothing market are women outerwear and
footwear, which is highlighted for the development in this type of clothing products.
(Cited by Euromonitor, “Vietnamese consumers find their passions for fashion).
The study’s conclusion can be useful for marketing practitioners and
merchandisers in improving the sales of clothing products in fashion malls by
implementing the advantages of impulse buying. However, it should be taken into
consideration since the increasing sales from impulse purchasing will harm to consumers’
financial budgets and cause negative consequences in post-purchase feelings, social
responses as well as decrease overall self-esteem. (Rook 1987; Rook and Hoch 1985).
1.2. Research Objectives
The purpose of this study, as mentioned above, is to investigate how optimum
stimulation level, hedonic purchase, consumer’s need for uniqueness, novelty – fashion
consciousness and consumer self-spending – control affect the level of impulse buying in
fashion malls. Thus, to achieve this goal, the research objectives are stated as follows:
- Identifying factors affecting the level of impulse buying in fashion malls in
Ho Chi Minh City.
- Examining the level of influence of each factor on the level of impulse buying
in fashion malls.
- Providing practical recommendations for fashion brands and retailers in
increasing impulse purchase behavior of consumers, especially those living in Ho
Chi Minh City.
1.3. Research Questions
To reach the research objectives, the questions were formed as below:
- What are factors that affect the level of impulse buying in fashion malls in Ho
Chi Minh City?
- In what level does each factor affect the level of impulse buying in fashion
malls in Ho Chi Minh City?

17
 - What aspects of fashion brands or retailers need to be improved to increase
the level of impulse purchase of consumers?
1.4. Significances of the Research
Although there has many researchers conducted research about impulse
buying behavior and the factors affecting on this consumer behavior, in developing
countries, the research studies are hard to be found. In other words, the level of impulse
buying and the factors affecting on this level has never been conducted before in an
official way in developing countries, or Vietnam market in particular. Thus, this study
aims to provide supports for the value of measuring the level of impulse buying in
fashion malls with clothing products.
The outcome of the research would provide manager and researcher a
conceptual framework to describe the relationships between the level of impulse and
variety seeking and the factors affecting on them. Thanks to that, business owners in
fashion industry can be able to target at the right customers and link these factors with
their present business situation in order to revise their current marketing strategies and
plans for better improvement.
Besides, the scientific recommendations from this research are well discussed
and this can be used as a good solution for the business owners on the way of reaching
the business’s objectives. Last but not least, thanks to this study, the business owners can
understand the level of influence that each factor contributes to the level of impulse
buying. Therefore, they can focus on the factors which have the greatest influences to the
level of impulse purchase in order to increase sales and develop their businesses and
utilize their resources in the most effective way.
Finally, this study can serve as a reference for those who wish to study about
impulsive behavior in Vietnam. The results from this study are also significant points that
can add value to further researches about Vietnamese consumer attitude in particular and
Vietnamese consumer behavior in general afterwards.

18
1.5. Research Scope and Limitations
1.5.1. Research scope
Ho Chi Minh City was chosen to be a scope of this research paper thanks to its
convenience and higher rate of impulse buyers. Ho Chi Minh City is not only the
economic center of Vietnam but also is the most modern and dynamic city. Besides, it
has lots of shopping malls and famous fashion brands to attract a huge of impulse buyers
every day. Thus, it was believed to have a good sample size which suitable with the
research’s purpose. In short, the paper mainly focuses on consumers having impulse
purchase toward clothing products as perceived by people in Ho Chi Minh City.
1.5.2. Research limitations
The major limitation of this study was the undiversified sample – not a
representative population due to the limited geographic accessibility capacity (All
surveys were delivered and collected around Ho Chi Minh City). Therefore, the result of
the study only helped explain partly the concept of impulse buying and its factors.
1.6. Research Structure
The study was designed with five main parts placed in five chapters as
following:
Chapter I: Introduction
This chapter will explain why it is necessary to conduct this paper. Moreover,
the introduction also provides the overall background, research objectives and research
questions, research scope and limitations as well as the structure of this study.
Chapter II: Literature Review
The second chapter mainly focuses on reviewing the important concepts,
definitions used and previous researches associated with the problem addressed in this
study. Most importantly, the theoretical framework and hypotheses are proposed in this
chapter based on the research questions.
Chapter III: Research Methodology

19
 This chapter gives the detailed information about the research process,
research design and research procedure as well as data analysis. Besides, the criteria for
selecting research sample and data collecting process are provided. Additionally, chapter
3 shows the questionnaire items in details as well as describes the survey development.
The statistical SPSS programs used in the research study is also mentioned and discussed.
Chapter IV: Data Analysis and Findings
Chapter 4 will provide the research results of the data analysis obtained from
the collected data through the main surveys. Besides, this chapter aims to give clear
explanation and recommendation of the statistic numbers as well as present the results of
the proposed research model after being tested.
Chapter V: Conclusion and Recommendations
The final chapter will summarize the findings, offer the discussion for
practical implications as well as give recommendation for further research. Lastly, the
limitation of the study is also provided.
1.7. Thesis Timeline
Thesis Duration: 6 months (10/11/2014 – 18/5/2015) (including Tet Holiday)

20
 Table 1: Research Timeline

No Activities

18 May
29 Mar

30 Mar
08 – 14
10 Nov

19 Dec

26 Jan

30 Jan

01 –22

7 Apr
Mar
Feb
1 Register for thesis topic
2 Apply for approval
3 Oral defense for thesis
proposal
4 Design questionnaire
5 Pilot test for questionnaire
6 Distribute and collect
questionnaires
7 Mid way report
8 Collect data and start
analyzing
9 Write thesis
10 Submit thesis

21
 CHAPTER II
LITERATURE REVIEW
In this chapter, 5 hypotheses are proposed. The first four hypotheses concern direct
impacts of optimum stimulation level, hedonic purchase, consumer’s need for uniqueness
and novelty – fashion consciousness respectively on the level of impulse buying. The final
hypothesis proposes the negative relationship between the consumer self-spending –
control and the level of impulse buying.
2.1. Definition of Terms
2.1.1. Optimum Stimulation Level
According to Wilhelm Wundt – founder of experimental psychology,
Optimum Stimulation Level (OSL) can be understood as a level of arousal, which
produces a positive feeling or an optimal level of sensation (cited by Soares & Maria,
2004). Raju (1980), OSL was defined as a general personality trait that characterizes
individual in terms of their response to environmental stimuli. Due to dissimilar
responses to stimulation from different kinds of people, there is a diverse OSL reaction
with regard to cognitions, perceptions, attitude and behavior. Accordingly, those who
have low OSL can comfortably satisfy with commonplace situation while individuals
having high OSL can easily have negative feelings in a steady and usual environment
since they have higher propensity to discover latest inducement to fulfil their demand on
stimulations. (Kish & Donnenwerth, 1972; Raju, 1980). In addition, new experiences
tends to be accepted straightforwardly by group having high OSL (Aluja et al.,2003;
Vries et al.,2009) and thus they are more engaging and willing to check out new
products, services and brands (Raju 1980; Steenkamp & Burgessm, 2002). High-OSL
persons also prefer a wide variety of usage of different products and services or brands

22
having common classification, (Menon & Kahn, 1995) and those products and services
are put on use creatively (Mittelstaedt et al., 1976; Raju, 1980).
2.1.2. Hedonic Purchase
According to Bhatnagar (2004), hedonic purchase refers to “emotional needs
of individuals for enjoyable and interesting shopping experiences”. As discussed by
Holbrook & Hirschman (1982), hedonic buying motive relates to emotional arousal,
which happens while carrying out purchasing process. In other words, hedonic purchase
occurs in a high-involvement situation, where any single individual has a deep fall into a
consumption event (Hopkinson & Pujari, 1999). For example, shoes’ consumers with
higher hedonic purchase level will have interest in tracking shoes’ sales off event more
often than normal consumers. Therefore, the changing levels of involvement have a
positive influence on the level of hedonic purchase. The level of hedonic purchase is
estimated to be higher in high – involvement consumption situations and vice versa.
2.1.3. Consumer’s Need for Uniqueness
Consumer’s Need for Uniqueness (CNFU) was early defined as “the need to
differentiate one-self and competes with other motives in situations that threaten the self-
perception of uniqueness (i.e., situations in which individuals see themselves as highly
similar to others in their social environment)”. (Snyder & Fromkin 1970, 1972).
Accordingly, the level of sensitivity to similarity and the desire to be different to others
will tend to be higher in those people who have stronger needs for uniqueness (Snyder,
1992). In a study implemented by Tian et al. (2001), CNFU was re-defined as “the trait of
pursuing differences relative to others through the acquisition, utilization, and disposition
of consumer goods for the purpose of developing and enhancing one’s self-image and
social image”. Therefore, CNFU is claimed to be associated with products which have
some symbolic meanings to enhance individual’s self and social image as an expression
of uniqueness. (Tian et al., 2001; Tian and Mckenzie, 2001). Also, since individuals may
complete their aspiration to be unique in various ways such as possession displays (Belk,
1988), style of interaction in communication between people (Maslach, Stapp, and

23
Santee, 1985), or the domains of knowledge in which they initiate competence (Holt,
1995). They are probable to be dissimilar in their ways to fulfill their uniqueness
aspiration through consumer behaviors and possessions.
Based on the previous research on need for uniqueness and the nonconformity
in consumer behavior, the conceptualization of consumer’s need for uniqueness is
divided into three behavioral dimensions. Each will be defined in turn as follows.
Creative Choice Counter -Conformity.
Creative choice counter-conformity demonstrates consumers’ attempt to
pursue the distinction from most others, however, these selections made by one consumer
can be contemplated as a good decision by others (Tian, et al., 2001). As explained by
Knight. & Kim (2007), in creative choice counter -conformity, goods are purchased by
consumers must be unique and approved by most of people in society in order to express
consumers’ uniqueness. Accordingly, brand names that satisfy this type of consumer
must have some distinguishing characteristics such as exclusiveness, prestige and unique
features.
For example, limited edition of Gucci’s hand bags is one kind of product that
not only expresses the exclusiveness, prestige but also the uniqueness since these hand
bags of Gucci are produced with very limited items. Besides, Gucci is one of the most
famous fashion brands all over the world which was known for hundred years, thus, their
products can satisfy the need of being praised by most of people of those having creative
choice counter-conformity. In short, people who want to be unique and recognized by
society for their choices will tend to buy products with limited edition but still luxury and
good looking under the view of others.
Unpopular Choice Counter -Conformity.
Unpopular choice counter-conformity mentions about the preference or usage
of products and brands that diverge from normal standards of society, therefore, it can
cause high risk of being disapproval from most of people. However, these purchaser
cannot resist their desire to be different from others, accordingly, they still want to be

24
unique in this way. In comparison to the aspect of creative choice counter-conformity,
unpopular counter- conformity may terminate in a common image and encourage the
improvement of self-concept. In short, people following this type of uniqueness are
considered as rules breakers. However, according to Heckert (1989), through times, such
choices can gain social approval and individuals who used to be viewed as “rules
breakers” before can now become fashion leaders. In other words, it cost a lot of time for
these individuals to be viewed as “fashionista” and be recognized by society that they are
not that weird. Thus, according to Simonson & Nowlis (2000), these consumers are not
worried about others’ criticism; in fact; the most concerns them is to make purchase
decisions that might be viewed as bizarre from others.
For example, blue jeans – one kind of clothes that are too famous for not only
young people but also for the whole fashion industry and society, used to be viewed as
“abnormal clothing” in the 19 century. At that time, people who wore jeans to seek for
new image and enhance self-concept were easily considered as weird people. Until later
of 1990s, jeans were recognized by most of people, especially celebrities, jeans became
famous and those wore jeans before had become fashion leaders.
Avoidance of Similarity.
Avoidance of similarity refers to individual’s effort to avoid the similarity by
losing interest and discontinue using popular products. In other words, avoiding
similarity also involves with the devaluation and avoidance of some kinds of products or
brand that are considered to be well – recognized. However, since the consumer’s success
in trying to differentiate themselves to others by creating distinctive self-images and
social images doesn’t last long, they tend to discontinue to use or purchase products
which can cause the similarity. Therefore, these individuals tend to purchase goods that
are not widely accepted by most of people, and able to distinguish them from others.
(Knight. & Kim, 2007).
For example, these customers will find items which have origin from less
popular brands such as local brands or unknown products to avoid the similarity from

25
others. Besides, they may stop using products that were widely used by many people. In
summary, different from creative choice conformity and unpopular choice conformity,
these kinds of customers are not trying to be unique in an outstanding way that can make
people remember them longer than others or call them as fashion leaders, the only thing
they need is to be different from others by using non-popular products or stop using
products that is becoming popular.
2.1.4. Novelty- Fashion Consciousness
Consumer Style Inventory (CSI) was conceptualized by Sproles and Kendall
(1986), which is an early effort to systematically measure shopping orientations by
applying decision-making orientations. It was originally contained 50 items to measure
the general customer’s orientations towards shopping, and later, thanks to Sproles and
Kendall, the CSI was developed in a short-form scale with 40 items which validated from
a sample of US high school students. Novelty – fashion consciousness is one in eight
dimensions, together with Perfectionism (high-quality consciousness), Brand
consciousness, Recreational (hedonistic shopping consciousness), Price consciousness,
Impulsiveness and Confusion from over choice. As defined by Sproles and Kendall
(1986), Novelty -Fashion Consciousness (NFC) is a characteristic that describes
individuals who are fashionable with novelty conscious and love to experience new
things. These customers always seek for new fashions and fads with excitement and
pleasure. For them, it is necessary to be up-to-date with styles and be trendy.
2.1.5. Consumers Self- Spending Control
According to Haws and Bearden (2010), consumer self- spending control
describes individuals having ability to regulate themselves from spending, or self-
regulations in other words. Therefore, self-spending control relates to the consumers’
decision of controlling their financial budget and can be explained in terms of some
responses such as thoughts (e.g. by discarding undesirable thoughts or putting oneself
into concentration), altering emotions, control over impulsiveness and changing
performances. However, there has an example of a person which self -control is high

26
finds difficulties in regulate spending while another person with average self-control
could become an extremely well-managed financial. That can only be explained in the
way consumers’ decision on self-allocating their financial budget.
2.1.6. Level of Impulse Buying
Early marketing literature defined impulse buying in a simple way as
unplanned purchasing (Cobb and Hoyer, 1986). However, as discussed by Rook (1987),
impulse buying occurs when “a consumer experiences a sudden, often powerful and
persistent urge to buy something immediately”. In other words, impulse buying can be
understood as an immediate and sudden purchase behavior with no thoughtful or future
implications before. Also, impulse buyers are claimed to have unreflective thinking,
which is “prompted by physical proximity to a desired product, dominated by emotional
attraction to it, and absorbed by the promise of immediate gratification”. (Hoch and
Loewenstein, 1991; Thompson, Locander, and Pollio, 1990). Therefore, these kinds of
consumers are likely to act without carefully hesitation in mind and have quick response
to their impulse purchase. In negative speaking, impulse buying can be almost entirely
driven by stimulus, which is translated directly into an unconditional response.
Regarding above definition of terms, it can be seen that it was expected to
have relationships between impulse buying and the other independent factors such as
optimum stimulation level, hedonic purchase, consumer’s need for uniqueness, novelty-
fashion consciousness and consumer self-spending control. However, there was no
previous theoretical framework that transform these independent factors above into the
proposed framework, which cause direct impact on the impulse buying. In this case, only
exhibiting definition was not enough to create rational writing and propose hypotheses.
Therefore, next part will focus on discussing the relationships between independent and
dependent factors before proposing a theoretical framework for this research study.

27
2.2. Research Hypotheses
2.2.1. Relationship between Optimum Stimulation Level and Level of Impulse
Buying
As discussed by Raju, (1980), Steenkamp and Baumgartner (1992),
individuals with high Optimum Stimulation Level (OSL) have lower arousal level which
encourages them to seek for activities in order to reach their desired stimulation level.
Thus, individuals having low arousal level will tend to have indulgence in impulse
buying since the extra stimulation that high OSL individuals need to achieve their OSL
would be provided. It is consistent with the general theory of Eysenck (1993) that
inconsiderate, impatient, risk-taking, sensation-seeking and pleasure seeking are some
attributes that describes an individual with low arousal level (Dickman, 2000). Lastly,
almost all of the purchasers’ characteristics related to impulse buying or optimum
stimulation level (Baumgartner and Steenkamp, 1996) are likely to derive from isolated
personal attribute defined as impulsivity (Eysenck, 1993) or impulsive sensation seeking
(Zuckerman, 1993).
Therefore, this research expects OSL to have a great impact on impulse
buying, which means individuals with higher OSL will have higher possibility of falling
into impulse purchase. A review of relative literature led to the subsequent hypothesis:
H1: Optimum Stimulation Level has significant positive influence on
Level of Impulse Buying.
2.2.2. Relationship between Hedonic Purchase and Level of Impulse Buying
As Engel and Blackwell (1982), impulse buying is described as an action
undertaken without having consciousness or intention before entering the store. As
defined by Bayley and Nancarrow (1998), “hedonic purchase is marked with pleasure; in
contrast to the utilitarian behavior where the shoppers seek for functional benefits and
economic value in the shopping process”. In addition, a research conducted by Sharma et
al. (2010) classified impulse buying as a hedonic behavior that is incorporated with
feelings and psychosocial incitements without concentrating on functional benefits.

28
Based upon the different description, we conclude that impulse buying involves hedonic
purchase decisions which are made inside a store and excludes the reminder purchasing
activities. Besides, research by Beatty and Ferrell (1998) demonstrates that impulse
purchasing is incorporated with sensory stimulation and hedonic motivation. Recently, as
suggested by Yu and Bastin (2010), shopping value of a person result in impulse
purchase and are inseparably associate with each other. Thus, thanks to the support from
relevant literature, we suggest the hypothesis as follows:
H2: Hedonic Purchase has significant positive influence on Level of
Impulse Buying.
2.2.3. Relationship between Consumer’s Need for Uniqueness and Level of
Impulse Buying
As Tian et al. (2001), “Consumer’s need for uniqueness (CNFU) allows
individuals to enjoy improved self- and social-image”. Besides, “consumers’ image is
enhanced internally and externally through the use of products when they recognize some
symbolic meanings in these products” (Tian and Mckenzie, 2001). Moreover, as cited by
Dee& Eun Young (2007), new products or brands can be acquired more rapidly by
purchasers, who have greater demand of distinction, comparing with those who have that
demand at a lower level. This means that these kinds of customers do not hesitate to
spend money on seeking items which can differentiate them from others. Since it may
cost lots of time to find stuffs with unique features, these customers may fall into impulse
purchase whenever they find suitable accessories that best suit for their clothes. Besides,
people with high need for uniqueness may purchase more new arrivals than others. In
short, it can be inferred that consumer’s need for uniqueness has a positive relationship
with impulse buying, which individuals having higher need for uniqueness will have
higher level of impulse buying in other words. Thus, the study proposes H3 as follows:
H3: Consumer’s Need for Uniqueness has significant positive influence on
the Level of Impulse Buying.

29
 2.2.4. Relationship between Novelty – Fashion Consciousness and Level of
Impulse Buying
As simply defined by Bae (2004), Novelty – fashion consciousness describes
a shopper who is always aware of new styles while impulse buyers are labeled as
individuals’ tendency to have spontaneous, unreflective, immediate and kinetic buying
motives. As fully stated above, buyers with high level of impulse will have faster
approach to buying ideas and have “opening” shopping lists at any time. Thus, it can be
inferred that buyers with novelty-fashion consciousness are likely to experience impulse
purchase. Additionally, in 1986, Sproles and Kendall together developed a consumer
decision-making style inventory (CSI), which helps define and evaluate eight mental
traits of consumer in decision making. Among eight consumer’s characteristics, novelty –
fashion consciousness and impulsiveness are mentioned as traits which have greater
impact towards consumption. Accordingly, it could be said that there has a positive
relationship between Novelty – Fashion Consciousness and Level of Impulse Buying,
which consumers who have high novelty and fashion consciousness will easily get
involves with impulse buying and vice versa.
Therefore, this study proposes H5 as follows:
H4: Novelty – Fashion Consciousness has significant positive influence on
Level of Impulse Buying.
2.2.5. Relationship between Consumer Self-Spending Control and Level of
Impulse Buying
Previous studies suggested that the concept of self-control affects the level of
impulse buying, which its central theme underlies impulse buying on many recent
research (Baumeister 2002; Faber and Vohs, 2004). Additionally, research by Youn and
Faber (2000) also demonstrates that impulse buying has its origin from purchaser
characteristic; for instant, impulsiveness and optimum stimulation level, shopping
enjoyment, or lack of self-control.
Originated from the survey of self-control issue and failure conducted by

30
Baumeister et al. (1994), development of the characteristic measurement for self-control
has been carried out by Tangney and Baumeister (2001). For those researchers attracted
by the first phase of analysis and evaluation of how self-control forecast purchaser
reactions, it is considered as a supportive instrument for a variety of assessment from
adjusting emotions, tracking and administering performances as well as impulses,
sustaining self-control, escape from inadequate practices, and the like. As discussed by
Romal and Kaplan (1995), individuals having greater self-control administer their finance
more appropriate than the others as well as preserving more and expending less.
Regarding above differences, the ability of being able to receive stimuli to
various kind of marketing strategies can be forecasted by self-control. Individuals having
low self-control is unprotected to be attracted by the moment, and a sales pitch focusing
on instantaneous satisfaction would be engaging. On the other hand, individuals having
greater self-control persuasively consider long-term value and advantages before
purchasing. Low self-control person tends to respond in a way as “You will look and feel
great in this car”. Conversely, “This car will be reliable and durable and will have a high
resale value” tends to be the consideration of people having high self-control. Thus,
impulse buying occurs when the desire of customers for a product beats their intentions to
avoid purchasing. This indicates that the need to purchase and the capability of self-
control are two discrete procedures taken into account with impulse buying.
In summary, the purchaser having low self-control tends to have impulse
buying practice, and vice versa. Accordingly, for those who have lower purchaser self-
control, a conclusion can be withdrawn that those individual will repeat impulse buying
practice, having high optimum stimulation level and the requirement for being distinct as
well as hedonic purchase behavior.
Therefore, we can come up with the following hypothesis is proposed:
H5: Consumer’s Need for Uniqueness has significant negative influence
on Level of Impulse Buying.

31
2.3. Research Model
To sum up, all of the hypotheses above will be presented in the theoretical
framework for this study which is shown in the figure below:

Figure 2: Research Model

H1: Optimum Stimulation Level has significant positive impact on Level of Impulse
buying.
H2: Hedonic Purchase has significant positive impact on Level of Impulse buying.

32
H3: Consumer’s Need for Uniqueness has significant positive impact on Level of
Impulse buying.
H4: Novelty – Fashion Consciousness has significant positive impact on Level of
Impulse buying.
H5: Consumer Self-Spending Control has significant negative impact on Level of
Impulse buying.

33
 CHAPTER III
RESEARCH METHODOLOGY
This chapter aims to present about the research design that consist of research
approach, questionnaire design and sampling technique. Moreover, data analysis method
is also explain in details in this chapter.
3.1. Nature of the Research
According to Strauss and Corbin (1998), an academic research can be
approached by various methods. However, this section tends to introduce briefly about
some well-known methods and explain for the methods used in this study research.
There are two different kind of research, which are inductive and deductive
research. If the process of inductive research starts with the gathering data, then looking
for patterns in the data and end up with developing a theory that could give explanations
on these patterns, researchers that apply deductive research begins examining a pre-
existing theories and testing its association with data. In other words, while inductive
research goes from data to build theories, deductive research involves with investigating
data and testing hypotheses derived from existed theories. Accordingly, this research
study applies deductive research.
3.2. Research Method
As stated by Spencer, Ritchie and O’Connor (2004), there are two kinds of
methodologies used in study research to gain knowledge in social contexts which are
qualitative and quantitative research method. While quantitative research is expounded as
a set of statistical and numerical data and is explained based on research with a wide
range of variables on a large number of factors, qualitative methodology deals with non-
numerical data. These methodologies are used with purpose of achieving “a better

34
understanding of the surrounding society as well as better comprehension of how
individuals, groups and institutions act and influence one another”. (Sogunro, 2001).
Qualitative research is mainly viewed as exploratory research. It is applied to
achieve an understanding of casual reasons, ideas and stimuli. It delivers perceptions into
the problem or supports to develop hypotheses for prospective quantitative approach.
Besides, it is used to discover trends in thought as well as opinions, and provides
researchers a deeper look into problems. Thus, the main purpose of this method is to gain
“a deeper understanding of a specific phenomenon” instead of “drawing general
conclusion about the situations” (Lundahl and Skärvad, 1999; Strauss and Corbin, 1998).
Qualitative research mainly concentrates on quality and prejudice a social practices.
Some common methods of qualitative research can be listed as focus group (or group
discussion in other words), individual interviews and participant observations. Moreover,
the sample size to conduct qualitative research is usually small and the respondents are
well designated to satisfy a given quota.
Comparing to qualitative method, quantitative method is more formalized, and
is utilized to measure to measure with regard to the quantity of the issue by initiating
numerical data or data which can be transfigured into usable statistics. Besides, it also
used to measure attitudes, viewpoints, behaviors and other determined variables, then
results from a greater sample population. Facts and exposed models in research can be
formulated by measurable data utilized by quantitative research. Additionally, collection
methods of quantitative data are more organized in a close order compared with
qualitative data collection methods. On the other hand, quantitative data collection
methods incorporate with a wide variety of surveys, from online and paper surveys to
mobile and kiosk surveys, as well as direct interviews or telephone interview, long-
distance studies, website interceptors, online polls, and structured monitoring. To sum up,
in quantitative method, few variables are examined on a greater number of entities
(Neuman, 2006).

35
 In summary, for the purpose of getting solutions from enquiry hour numerical
proof, quantitative research should be applied; on the other hand, for the aim of clarifying
the occurrence of particular event, or explain the reason of any particular phenomenon,
then qualitative research may applied.
3.3. Data Collection Method
There are two ways to collect data: primary and secondary sources. While
primary data is defined as original data that collected first hand for a specific study
research; secondary data is data that already existed and accumulated for another uses
before conducting the study research. (Tull and Hawkins, 1993; Yin, 1994). Saving cost
and time are the two biggest advantage of secondary data in the process of data collection
in comparison with primary data. Nevertheless, since secondary data has been gathered
for various different purposes, its content might not well suitable with the current
requirements of the researchers (Hair et al., 2003; Yin, 1994). Thus, regarding above
reasons, this study is conducted by collecting data from primary source.
3.4. Research Design
In this part, the process chart will be used to summarize the main steps of
research design. There has 8 steps in total, which starts from Data collection and ends up
in Data analysis.

36
 Data Collection

Research Question

Research Model

Sampling Design

Questionnaire Design

Pilot Test

Data Procedure

Data Analysis

Figure 3: Main Steps of Research Design

37
3.5. Sampling Design
3.5.1. Sample Size
Exploratory Factor Analysis is applied in this research study, thus, variables
should follow the standard ratio which is 5:1 (with 5 respondents answering for 1 item) in
EFA (Gorsuch, 1983 and Hatcher, 1994). This questionnaire consists of 39 items, thus,
the sample size must reach at least 195 to be qualified. Furthermore, as proposed by
Comfrey and Lee (1992), the sample size of 50, 100, 200, 300 can generate a very poor,
poor, fair and good research outcomes correspondingly. Accordingly, the sample size of
320 is chosen to ensure the good result for the research study.
3.5.2 Sampling Technique
Due to the lack of time, budget and human resources, non – probability
sampling was chosen among the sampling methods. Specifically, there are two kinds of
sampling technique applied that are convenient sampling and snowball sampling. As
defined by Cooper and Schindler (2006), convenient sampling is a “non-probability
sampling where researchers use any readily available individuals as participants”. It is
obvious that using convenient sampling is a reasonable and practical choice of
investigators to approach the objects. Besides, snowball method is also applied in this
study thanks to its convenience in reaching a large sample of respondents regardless of
limited time, which mostly based on the researcher’s relationship or acquaintances. In
other words, this method allows researchers utilized their network relationship to reach a
wider target participants in order to gain more unbiased results.
3.5.3. Sampling Locations
Since the research topic covers the impulse buying in fashion malls in Ho Chi
Minh City, it is a must to collect data in shopping malls to ensure the quality of the data
as well as avoid bias. Besides, almost buyers couldn’t remember whether they did make
an impulse purchase or not if they weren’t at shopping malls at the time they were asked,
thus, most of the data should be collected in hand at the malls. However, due to the limit
of time and geographical barriers, four well-known shopping malls were chosen as

38
sampling locations, which are Diamond Plaza, Vincom Center, Parkson Hung Vuong and
Cresent Mall. These fashion malls not only have ideal geographic locations but also are
famous for the huge attraction of shopping buyers every day. While Diamond Plaza and
Vincom Center are placed at the most crowded streets in District 1 – the central business
district of the city; Parkson Hung Vuong is in the center of District 5, which is one of the
most districts have highest population in Ho Chi Minh city. Besides, Cresent Mall in
District 7 is considered as the most potential shopping malls among all since it was built
in the high standard of residential area (Phu My Hung metropolitan) where concentrates
most of the senior classes of the society. Most of people living in District 7 have high to
very high income, thus, they are believed to be indulged in hedonic consumption as well
as impulse buying.
In short, fashion malls are the most suitable choices to conduct research
toward impulse buying among consumers, especially those living in such a dynamic and
modern city like Ho Chi Minh City. Exact number of data collected will be well-
discussed in the Data collection procedure. The next part will mention about the
Questionnaire Design.
3.6. Questionnaire Design
According to Churchill and Iacobucci (2006); Hair et al. (1998); Hair et al.
(2006), there are 3 most widely used measurement scale in social sciences, including
nominal, ordinal, interval or ratio. Apart from the nominal scales or demographic
questions in other words, interval scales are mainly adapted in this research study since
this scale is common applied in the field of academic marketing research thanks to its
optimization in multivariate statistical techniques (Hair et al. 1998; Tull and Hawkins
1993). Besides, due to its fitness, Likert scale (Rensis Likert, 1932) is adopted for this
research.
“The Likert Scale, developed by Rensis Likert, is the most frequently used
variation of the summated rating scale”, cited by Cooper et al. (2006). In the Likert scale,
if the scale is too small or too large, the differences may not be seen clearly or

39
respondents may find it difficult to discriminate, respectively. Thus, various discussions
have been carried out to decide the optimal number to best classify for the Likert scale.
As argued by Green and Rao (1970), six or seven point scale is acceptable. Conversely,
Ghiselli (1955) stated that we should develop the scale based on the diverse sets of
circumstances. In this research study, a five-point scale (from 1 = Totally Disagree to 5 =
Totally Agree) is applied in which the participants are required to give their level of
agreement among these 5 scales.
Besides, the questionnaire consists of two main parts. The first part focuses on
the groups of factors which are 5 independent variables and 1 dependent variable. The
second part concentrates on demographic questions to gather personal information of
impulse buyers.
As presented clearly above, all items were measured by a five-point
Likert scale, anchored by 1: totally disagree and 5: totally agree.
Table 2: Measurement Items for the Research Model

Construct Sub- Scale Item Measurement Reference

OSL1 I like to experience novelty
and change in daily routine. Baumgartner
OSL2 I am continually seeking new and
OPTIMUM ideas and experiences. Steenkamp,
STIMULATION OSL3 I like continually changing (1996)
LEVEL activities.
OSL4 When things get boring, I
like to try something
different.
HP1 Shopping to me is truly a
joy.
HP2 I shop not because I have to,

40
 but because I want to.
HP3 The time spent in shopping is
truly enjoyable to me.
HP4 I enjoy being immersed in Babin et al.
exciting new products while (1994).
shopping.
HP5 I enjoy shopping for its own
sake and not because of that I
need to purchase something.
HP6 While shopping I can feel the
HEDONIC excitement of the hunt.
PURCHASE HP7 While shopping, I am able to
forget my other problems.
HP8 While shopping I feel a sense
of adventure.
HP9 Any shopping is a very nice
time out to me.
CNU1 I stop wearing fashions when
they become popular with
the general public.
Avoidance
CNU2 I dislike brands bought by
of
everyone.
similarity
CNU3 When a sport-wear brand
becomes too popular, I wear Snyder and
it less. Fromkin
CNU4 I look for one-of-a-kind (1977)
CONSUMER’S products to create my own

41
 NEED FOR style.
UNIQUENESS Unpopular CNU5 The thing that I buy shapes a
choice more unusual personal
image.
CNU6 I buy unusual brands to
create a more distinctive
personal image.
CNU7 An important goal is to find a
brand that communicates my
Creative uniqueness.
choice CNU8 The brands that I like best
are the ones that express my
individuality.
NFC1 I usually have one or more
NOVELTY – outfits of the very newest Sproles
FASHION style. (1985) &
CONSCIOUSNESS NFC2 I keep my wardrobe up –to- Kendall
date with the changing (1986)
fashions.
NFC3 Fashionable, attractive
styling is very important to
me.
NFC4 To get variety, I shop
different stores and choose
different brands.
CSC1 I am able to work effectively
toward long term financial

42
 goals.
CSC2 I carefully consider my needs
before making purchases.
CSC3 I often delay taking action Kelly L.
CONSUMER SELF - until I have carefully Haws,
SPENDING considered the consequences William O.
CONTROL of my purchase decisions. Bearden and
CSC4 When I go out with friends, I Gergana Y.
keep track of what I am Nenkov
spending. (2010)
CSC5 I am able to resist temptation
in order to achieve my
budget goals.
CSC6 I know when to say when
regarding how much I spend.
CSC7 In social situations, I am
generally aware of what I am
spending.
CSC8 Having objectives related to
spending is important to me.
IB1 When go shopping, I buy
things that I had not intended
to purchase.
IB2 I am a person who makes
IMPULSE BUYING unplanned purchases.
IB3 It is fun to buy Beatty and
spontaneously. Ferrell

43
 IB4 When I bought (the item), I (1998)
felt a spontaneous urge to
buy it.
IB5 When I saw (the item), I just
couldn't resist buying it.

3.7. Pilot Study

According to Cavana et al. (2001), a pilot test should be conducted with a
reasonable sample size to test the comprehensiveness and possibility of the survey. Thus,
90 respondents including teaching assistants and staff at ILA – an English center in Ho
Chi Minh City as well as a part of International University’s students were invited to
participate in this pilot test. In specific, 15 out of 90 surveys were given directly to the
ILA’s teaching assistants during two weeks, then 10 of them were continually delivered
to the staffs working at this center in the next weekends. At the same time, the data were
collected in the two Psychology classes of Dr. Bui Quang Thong on Tuesday and
Thursday, which the main respondents are International University’ students and the
number of surveys returned was 43. The rest questionnaires were 22, and it were given
face-to-face to the students of Organizational Behavior class on Friday of Dr. Mai Ngoc
Khuong. The process of collecting data for doing pilot test ended up at 90 respondents.
Although this sample size was not large enough to run the main data analysis,
it still brought an overview of how the questionnaires were designed and check the level
of understanding of respondents. In other words, it cannot deny that the result of the pilot
test not only pointed out misunderstandings in questionnaires but also showed partly of
the result of the upcoming test. Thanks to that, the questionnaire can be designed again to
avoid ambiguity or confusion caused from respondents in a more convenient way.
3.8. Data Collection Procedure
This study aims of targeting the potential consumers toward clothing products,
especially who did have impulse purchase in the past. Since the consumer’s buying

44
behavior is complicated, the research targets this group of respondents with purpose of
having better understanding on this segmentation as well as giving suitable implication
management for fashion industry.
As presented clearly above, in order to avoid bias, questionnaires were spread
out mostly in the fashion malls. By giving directly to respondents and asking for fulfilling
the paper questionnaire, there are 196 out of 355 surveys were delivered in these four
fashion malls in total. Specifically, 35 data were collected in Diamond Plaza, 42 in
Vincom Center, 58 in Parkson Hung Vuong and 61 data went from Cresent Mall. The
respondents from this channel varies from students to executives, but most of them were
planning to buy clothing products or related accessorizes at the time they were asked.
Thus, this method of collecting data can be said in confidence that it has brought an
objective result toward consumer buying behavior.
However, to reach the target sample size, data has to be gathered via online
channel and some other offline locations apart from shopping malls. Thus, 159 surveys
left had been collected from online channel and other offline approaches.
Firstly, for the online approach, Google Form was used to build online survey.
The researcher can distribute this online survey via social networks or emails then collect
data from the “View Response” column. In details, there were 100 responses within a
week after spreading out the online questionnaires. Obviously, the rest of surveys went
from other offline channels besides the fashion malls which will be discussed clearly as
below.
About the respondents’ occupation, in specific speaking, the total of
respondents for both online and offline channels was 355, and 210 out of them were
students from Vietnam National University such as University of Technology, University
of Social Sciences, University of Social Sciences and Humanities, University of
Economics and Laws as well as International University. Constantly, 74 data were
collected among students from other universities and colleges such as Foreign Trade
University, RMIT University, Hoa Sen University and the National College of Education

45
of Ho Chi Minh City. As can be recognized, most of respondents in this study were
college and university students, which also were the young people with high fashion
consciousness and high frequency of visiting shopping malls. Thus, students are the main
respondents of this research paper, which were approached equally by both online and
offline channels.
Besides, offline participants also reached to office workers, housewives and
executives (teachers, doctors, engineers, etc.). Apart from respondents reached through
fashion malls, the researcher also sought for more in other offline sources. Firstly, since
the researcher both have mother and father working as teachers in Nguyen Tri Phuong
secondary high school and Tran Boi Co secondary high school, 23 paper questionnaires
were delivered face-to-face to their colleagues within 4 days. Lastly, 48 surveys left went
to parents who currently have students studying in ILA English center. The very small
part of them are executives while almost they are working as office worker in companies
or housewives/ retailers. In short, it took more than half of two weeks to complete
collecting data from these participants.
Generally, the whole data procedure needs nearly a month to be finished.
Exact figures will be calculated and presented clearly in descriptive statistics part in the
following chapter, thus, the next section will introduce about how the data were analyzed.
3.9. Data Analysis Method
SPSS statistical software was used to analyze collected data in this research.
Analyzed contents were categorized into descriptive statistics, reliability, validity,
correlation, and AMOS model testing. With the usage of SPSS software, a two-step
approach in structural equation modeling (SEM) was employed to analyze the data
(Anderson and Gerbing, 1988) by using AMOS (adds-on SPSS).
Beside, to test the validity and reliability of the data, Cronbach’s Alpha was
applied also. Additionally, we ran Factor analysis (Exploratory Factor Analysis – EFA) to
attain the amount of factors that would give explanation on the maximum variance in the
data. Lastly, Confirmatory factor analysis (CFA) was performed to validate the measures

46
and SEM was used to test the theoretical model. As explained by Anderson and Gerbing
(1988), CFA should be adapted firstly to check the quality of the measurement structure.
It could be inferred that any measurement problems existing in the measurement model
would be leftover to avoid impact on the overall model fit in SEM. Then, SEM analysis
was used to estimate the fitness of the desire model. Moreover, this research utilize the
demographic analysis and descriptive statistics.

47
 CHAPTER IV
DATA ANALYSIS AND FINDINGS
4.1. Sample Demographic
4.1.1. Response Rate
In this study, there has 355 responses were approached directly through online
surveys and paper questionnaires. However, there were 320 out of 355 returned with
acceptable and valid quality after deleting 35 unreliable answers from respondents. It also
means that among 355 respondents, there were 320 of them qualified the criteria to
become right subjects for this study. Invalid responses are excluded since the respondents
had misunderstanding with reversed –scale questions or left blank the require questions.
As a result, the response rate is about 90% in which the analysis based on.
Table 3: Valid Response Rate

FREQUENCY PERCENTAGE
INVALID 35 9.86
VALID 320 90.14

4.1.2. Demographic Analysis

In this part, the data of respondent profiles were summarized into important
categories such as gender, age, occupation and monthly income with purpose of
classifying the group of customers.
Table 4: Demographic Information

DETAIL FREQUENCY PERCENTAGE

Male 111 34.7
GENDER Female 209 65.3

48
 15 – 18 7 2.2
AGE 19 – 30 302 94.4
31 – 40 9 2.8
> 40 2 0.6
Students 266 83.1
Office Executive 36 11.3
Retailer/ Housewife 2 0.6
OCCUPATION Executive (Engineer, 5 1.6
Doctor, Teacher)
Others 11 3.4
<3 million VND 194 60.6
INCOME 3 – 7 million VND 79 24.7
8 – 15 million VND 31 9.7
> 15 million VND 16 5

4.1.2.1. Gender

Figure 4: Gender Percentage

49
 The pie chart indicates about the ratio of the two gender among the
respondents in this study. It can be seen clearly that the percentage is higher in female
buyers with nearly 65.4% while it only takes approximately 34.7% in male. Obviously, it
can be concluded that female customers more potential than male since they have strong
interest in fashion industry as well as impulse buying in the field of this research study.
4.1.2.2. Age

Figure 5: Age Group Percentage

The diagram shows the percentage of respondents’ age group. According from
this, the highest percentage of the age group lies in the range from 19 – 30 years old with
the proportion up to 94%. As can be explained, 19 – 30 is the age group that buyers are
usually concerned about fashion trend and easily have impulse purchase based on their
financial budgets. Besides, 3% are buyers at the age of 31 – 40, 2% from 15 – 18 years
old and there is only 1% of buyers with age over 40.

50
 4.1.2.3. Occupation

Figure 6: Occupation Percentage

Figure 6 demonstrates the percentage of occupation of the respondents. There

are 83% are students, 11% of office executive (including ILA’s staffs and other officers),
2% are retailers and housewives, 1% are executives which are engineers, doctors,
teachers and the others job accounts for 3%.

51
 4.1.2.4. Income

Figure 7: Income Percentage

Generally, the lower – middle income accounts for 60% of the total
respondents and is the highest income among the 3 groups. It can be explained that the
students are the largest respondents, thus, the income level of these students usually
ranges from under to 3 million VND. Besides, middle income level (3 – 7 million VND)
takes the second place with 25%, upper – middle income level (8 – 15 million VND) lies
in third place with 10% and the high income level (> 15 million VND) has the lowest
proportion with only 5%.
4.2. Descriptive Statistics
To gain a preliminary understanding about the data, Descriptive statistics are
used as the first statistical step that must be conducted. In the descriptive statistics table,
min, max, mean and standard deviation are calculated to describe the characteristics of
collected data. While mean helps reveal the central tendency of distribution, standard
deviation is a measure of dispersion in a frequency distribution. Therefore, according to
the definition, a low standard deviation means that most of the numbers are very close to
the mean, and vice versa.

52
 4.2.1. Descriptive Statistics of Optimal Stimulation Level
Table 5: Descriptive Statistics of Optimum Stimulation Level

Mean Std. Average

Deviation Mean
OSL1 (I like to experience novelty and change in 3.96 0.91
daily routine).
OSL2 (I am continually seeking new ideas and 3.97 0.89
experiences). 3.83

OSL3 (I like continually changing activities). 3.41 0.96

OSL4 (When things get boring, I like to try 3.99 0.85
something different).
Valid N (listwise): 320

The table shows that the average means among these items of Optimum
Stimulation Level ranges from 3.41 to 3.99. Therefore, the means in these items are very
close to each other. The item OSL1, OSL2 and OSL4 with the very high means value
which are 3.96, 3.97 and 3.99 respectively shows that the customers’ opinions towards
these 3 items are quite similar. Although the OSL3 (I like continually changing activities)
gains the lowest mean value among the three, its mean value still is an average value
compares to the standard. Finally, the average mean of all items is 3.83, which is very
high reveals that most of respondent agree with all the items of Optimum Stimulation
Level.
4.2.2. Descriptive Statistics of Hedonic Purchase
Table 6: Descriptive Statistics of Hedonic Purchase

Mean Std. Average

Deviation Mean

53
HP1 (Shopping to me is truly a joy). 3.91 1.03
HP2 (I shop not because I have to, but because I 3.19 1.18
want to).
HP3 (The time spent in shopping is truly enjoyable 3.71 1.1
to me).
HP4 (I enjoy being immersed in exciting new 3.48 1.22
products while shopping).
HP5 (I enjoy shopping for its own sake and not 2.92 1.18 3.35

because of that I need to purchase something).

HP6 (While shopping I can feel the excitement of 3.73 1.05
the hunt).
HP7 (While shopping, I am able to forget my other 3.05 1.22
problems).
HP8 (While shopping I feel a sense of adventure). 2.98 1.17
HP9 (Any shopping is a very nice time out to me). 3.22 1.19
Valid N (listwise): 320

In general, it could be concluded that the respondents hold a neutral attitudes

toward this factor due to the average mean value of total items (3.35). In specific, all of
the items of Hedonic Purchase have mean values ranging from 2.92 to 3.91, which means
there has a huge gap among variables. While HP1 (Shopping to me is truly a joy) stands
out as a promising item with highest mean value (3.91), HP5 (I enjoy shopping for its
own sake and not because of that I need to purchase something) does not receive many
agree answers from respondents when having the lowest mean value (2.92).
4.2.3. Descriptive Statistics of Consumer’s Need for Uniqueness (in sub -
dimensions)

54
 Table 7: Descriptive Statistics of Consumer's Need for Uniqueness's sub-
dimensions

Mean Std. Average

Deviation Mean
CNU1 (I stop wearing fashions when they become 2.64 1.14
popular with the general public).
CNU2 (I dislike brands bought by everyone). 2.76 1.16
CNU3 (When a sport-wear brand becomes too 2.75 1.14
popular, I wear it less).
CNU4 (I look for one-of-a-kind products to create 3.13 1.16
my own style).
CNU5 (The thing that I buy shapes a more unusual 3.17 0.99
personal image).
CNU6 (I buy unusual brands to create a more 2.85 1.12
distinctive personal image).
CNU7 (An important goal is to find a brand that 3.34 1.09
communicates my uniqueness). 3.02

CNU8 (The brands that I like best are the ones that 3.52 1.04
express my individuality).
Valid N (listwise): 320

Table 7 continually gives us information about descriptive statistics of 8 items

in Consumer’s Need for Uniqueness. The item CNU8 (The brands that I like best are the
ones that express my individuality) achieves the highest mean value with 3.52 and is
considered as the most agreeable item by respondents. In contrast, the item CNU1 (I stop
wearing fashions when they become popular with the general public) reaches the lowest

55
value for mean (2.64). Finally, the average mean (3.02) proves that the consumers have
the neutral attitude towards this factor.
4.2.4. Descriptive Statistics of Novelty – Fashion Consciousness
Table 8: Descriptive Statistics of Novelty-Fashion Consciousness

Mean Std. Average

Deviation Mean
NFC1 (I usually have one or more outfits of the 2.87 1.12
very newest style).
NFC2 (I keep my wardrobe up –to-date with the 2.78 1.11
changing fashions). 3.05
NFC3 (Fashionable, attractive styling is very 3.05 1.16
important to me).
NFC4 (To get variety, I shop different stores and 3.49 1.11
choose different brands).
Valid N (listwise): 320

There are 4 items in the Novelty – fashion conscious (NFC) and this table
presents the result of NFC dimensions after running descriptive statistics calculation. In
general speaking, there is a huge gap between the highest and the lowest mean value.
According to the result, item NFC4 (To get variety, I shop different stores and choose
different brands) reaches the highest mean value with 3.49 among all items. On the
contrary, the lowest one is NFC2 (I keep my wardrobe up –to-date with the changing
fashions) with 2.78. The next two mean values are 2.87 and 3.05 which go to item NFC1
(I usually have one or more outfits of the very newest style) and NFC4 (To get variety, I
shop different stores and choose different brands) respectively. In short, it can be seen
clearly that consumers care most about the variety in expressing their novelty – fashion
consciousness but care less about changing their wardrobe up on fashion trend.

56
4.2.5. Descriptive Statistics of Consumer Self-Spending Control
Table 9: Descriptive Statistics of Consumer Self-Spending Control

Mean Std. Average

Deviation Mean
CSC1 (I closely monitor my spending behavior). 3.49 0.98
CSC2 (I am able to work effectively toward long 3.35 1.01
term financial goals).
CSC3 (I carefully consider my needs before 3.76 0.94
making purchases).
CSC4 (I often delay taking action until I have 3.61 0.96
carefully considered the consequences of my 3.49

purchase decisions).
CSC5 (When I go out with friends, I keep track of 3.36 1.04
what I am spending).
CSC6 (I am able to resist temptation in order to 3.32 1.11
achieve my budget goals).
CSC7 (I know when to say when regarding how 3.12 1.14
much I spend).
CSC8 (In social situations, I am generally aware of 3.74 0.86
what I am spending).
CSC9 (Having objectives related to spending is 3.69 0.88
important to me).
Valid N (listwise): 320
In general speaking, the mean values of all items of Consumer Self –
Spending Control are in the ranges from 3.12 to 3.76 which proves that the level of
agreement from respondents is quite high. In specific, the highest mean value equals
3.76, which belongs to item CSC3 (I carefully consider my needs before making

57
purchases) and the lowest one equals 3.12 which goes to item CSC7 (I know when to say
when regarding how much I spend).Therefore, it can be concluded that these items are
important to the respondent.
4.2.6. Descriptive Statistics of Level of Impulse Buying
Table 10: Descriptive Statistics of Level of Impulse Buying

Mean Std. Average

Deviation Mean
IB1 (When go shopping, I buy things that I had not 3.28 1.04
intended to purchase).
IB2 (I am a person who makes unplanned 2.85 1.15
purchases).
IB3 (It is fun to buy spontaneously). 3.39 1.1 3.2
IB4 (When I bought (the item), I felt a spontaneous 3.43 1
urge to buy it).
IB5 (When I saw (the item), I just couldn't resist 3.03 1.03
buying it).
Valid N (listwise): 320

Table 10 presents the descriptive statistics of level of impulse buying. Overall,

3.2 is the average mean value among these 5 items. The most agreeable items from
respondents goes to item IB4 (When I bought (the item), I felt a spontaneous urge to buy
it) thanks to the highest mean value (3.43). The least agreeable one is recorded for the
item IB2 (I am a person who makes unplanned purchases) with lowest mean value (2.85).
In short, the items of Level of Impulse buying receive high level of agreement from
respondents due to the high average mean value.
To sum up, all the means have the average value ranging from 2.51 to 3.83.
Thanks to the descriptive statistics, the research paper can reveal partly the level of

58
agreement among respondents in HCMC which is illustrated in the Likert scales.
Moreover, the standard deviation among these items ranges from 0.85 to 1.22, shows that
the gap between them is quite small and very close to the mean.
4.3. Exploratory Factor Analysis (EFA)
Exploratory factor analysis is a statistical method needed to determine
underlying constructs for a large set of measured variables and identify the relationships
among them. Besides, EFA can suggest some new possibilities of the output for further
analysis. (Field, 2000; Rietveld & Van Hout, 1993). In reliability test, there are two
factors needed which are: Item-to-total correlations and Cronbach’s alpha. Each will be
explained in turn below. Next, underlying pattern of factors is determined in the EFA
test. Furthermore, to check whether the correlations among the observed variables exist
or not to run factor analysis, the significant level must be lower than 0.5 according to the
Barlett's Test of Sphericity (Tabachnick & Fidell, 2007). Moreover, Kaiser-Meyer-Olkin
(KMO) is used as a criterion to weight the EFA’s appropriateness and to ensure the
satisfactory analysis, the KMO index should be achieved in the range of 0.5 and 1
(Kaiser, 1974). Finally, according to Hair et al. (1998), factor loading should be
considered as a criterion to make sure the EFA’s practical significance, and the minimum
level of factor loading is over 0.3. To be more detailed, the factor loading will be seen as
significant if it is greater than 0.4 and practical significant if is larger than 0.3.
4.3.1. Reliability Analysis
Reliability test is performed to measure the internal consistency of the
construct. In other words, testing for reliability is to discover and remove failures before
conducting factor analysis. Internal consistency is measured by Cronbach’s alpha, which
is the most widely – used method. According to George and Mallery (2003), the rule for
Cronbach’s Alpha is stated as follows:
Table 11: Rules of Thumb for Evaluating Cronbach's Alpha

Cronbach’s Alpha < 0.5 Unacceptable

59
 0.5 < Cronbach’s Alpha < 0.6 Poor
0.6 < Cronbach’s Alpha < 0.7 Acceptable
0.7 < Cronbach’s Alpha < 0.8 Good
Cronbach’s Alpha > 0.8 Excellent

However, Slater (1995) reported that Cronbach’s Alpha is acceptable when

equals to 0.6. Besides, the rules have stated that if Cronbach’s Alpha if Item deleted is
greater than the overall Cronbach’s Alpha and Corrected Item - Total Correlation is less
than 0.3, the variable should be deleted from the list. The table below shows the results
after conducting reliability test.
Table 12: Corrected Item-to-Total Correlation and Cronbach's Alpha
(First Round)

Before removing inappropriate items

No. Factor Initial Corrected Items Cronbach’s
Cronbach’s – Total Alpha If
Alpha Correlation Items Deleted
Optimum Stimulation Level
1 OSL1 (I like to experience novelty 0.712 0.716
and change in daily routine).
2 OSL2 (I am continually seeking 0.648 0.748
new ideas and experiences).
3 OSL3 (I like continually changing 0.808 0.588 0.779
activities).
4 OSL4 (When things get boring, I 0.557 0.790
like to try something different).
Hedonic Purchase

60
 5 HP1 (Shopping to me is truly a 0.730 0.913
joy).
6 HP2 (I shop not because I have to, 0.747 0.911
but because I want to).
7 HP3 (The time spent in shopping 0.768 0.910
is truly enjoyable to me).
8 HP4 (I enjoy being immersed in 0.808 0.907
exciting new products while 0.922

shopping).
9 HP5 (I enjoy shopping for its own 0.695 0.915
sake and not because of that I
need to purchase something).
10 HP6 (While shopping I can feel 0.749 0.911
the excitement of the hunt).
11 HP7 (While shopping, I am able 0.650 0.918
to forget my other problems).
12 HP8 (While shopping I feel a 0.610 0.920
sense of adventure).
13 HP9 (Any shopping is a very nice 0.743 0.911
time out to me).
Consumer’s Need for Uniqueness
14 CNU1 (I stop wearing fashions 0.612 0.880
when they become popular with
the general public).
15 CNU2 (I dislike brands bought by 0.695 0.872
everyone).
16 CNU3 (When a sport-wear brand 0.726 0.869

61
 becomes too popular, I wear it
less). 0.889
17 CNU4 (I look for one-of-a-kind 0.706 0.871
products to create my own style).
18 CNU5 (The thing that I buy 0.633 0.878
shapes a more unusual personal
image).
19 CNU6 (I buy unusual brands to 0.731 0.869
create a more distinctive personal
image).
20 CNU7 (An important goal is to 0.638 0.878
find a brand that communicates
my uniqueness).
21 CNU8 (The brands that I like best 0.561 0.885
are the ones that express my
individuality).
Novelty – Fashion Consciousness
22 NFC1 (I usually have one or more 0.674 0.787
outfits of the very newest style).
23 NFC2 (I keep my wardrobe up – 0.782 0.738
to-date with the changing 0.835
fashions).
24 NFC3 (Fashionable, attractive 0.665 0.791
styling is very important to me).
25 NFC4 (To get variety, I shop 0.548 0.841
different stores and choose
different brands).

62
Consumer Self-Spending Control
26 CSC1 (I closely monitor my 0.645 0.867
spending behavior).
27 CSC2 (I am able to work 0.629 0.869
effectively toward long term
financial goals).
28 CSC3 (I carefully consider my 0.668 0.866
needs before making purchases).
29 CSC4 (I often delay taking action 0.882 0.692 0.863
until I have carefully considered
the consequences of my purchase
decisions).
30 CSC5 (When I go out with 0.609 0.871
friends, I keep track of what I am
spending).
31 CSC6 (I am able to resist 0.663 0.866
temptation in order to achieve my
budget goals).
32 CSC7 (I know when to say when 0.573 0.875
regarding how much I spend).
33 CSC8 (In social situations, I am 0.644 0.868
generally aware of what I am
spending).
34 CSC9 (Having objectives related 0.556 0.875
to spending is important to me).
Level of Impulse Buying
35 IB1 (When go shopping, I buy 0.672 0.796

63
 things that I had not intended to
purchase).
36 IB2 (I am a person who makes 0.632 0.808
unplanned purchases). 0.838
37 IB3 (It is fun to buy 0.670 0.796
spontaneously).
38 IB4 (When I bought (the item), I 0.654 0.802
felt a spontaneous urge to buy it).
39 IB5 (When I saw (the item), I just 0.576 0.822
couldn't resist buying it).

As can be seen clearly in the table 12, within the initial reliability analysis, the
Cronbach’s alpha of all the items above are higher than 0.8 and the item-to-total of
correlation of all variables are higher than standard level (0.5), which means that all the
scales have high internal consistency reliability. To be specific, the initial cronbach’s
alpha of Optimum Stimulation Level, Hedonic Purchase, Consumer’s Need for
Uniqueness, Novelty – Fashion Consciousness, Consumer Self-Spending Control and
Level of Impulse Buying are 0.808, 0.922, 0.889, 0.835, 0.882 and 0.838 respectively.
Thus, there is not necessary to remove any items to help increase Cronbach’s alpha value
for any factors. It can be claimed that the scales are well – design and the respondents
have clear understanding about the questions.
4.3.2. Factor Analysis
After completing the reliability test, Factor analysis will be conducted to
demonstrate the scale’s validity. In this study, Principal Component Analysis is used as a
means of compressing and classifying data through Varimax rotation in the sample of
observations. This method is able to make higher factor’s loadings become much higher
and lower factor’s loadings become much lower. Besides, variables will be deleted if its
loading is under 0.3. Furthermore, in case that items are loaded in many components, the

64
difference between them must be greater than 0.3 or they will also rejected. Next, second
round of factor analysis will be run in order to define the new model structure.
Additionally, the total variance extracted has to be equal or greater than 50%. In short,
Kaiser-Meyer-Olkin and Barlett’s test with Varimax rotation and Principal Component
Analysis extraction method will be conducted in this part.
4.3.2.1 EFA for Independent Variables
Table 13: Rotated Component Matrix of Independent Variables

65
66
 Table 14: KMO and Barlett's Test of Independent Variables

As can be seen clearly from Table 13 that all the items are loaded on 5
columns with no items are on the same components. Moreover, the KMO value from
Table 14 equals 0.891 (> 0.6) and the Bartlett test is significant (sig = 0.000). Thus, it can
be concluded that the EFA is applicable.

67
 Table 15: Total Variances Explained of Independent Variables

According from the table of Total Variance Explained, the result of

Cumulative Rotation Sums of Squared Loadings equals 60.425% > 50%, which means
that 5 extracted components may explain 60.425% of total variance in the 34 variables.
4.3.2.2 EFA for Dependent Variable

68
 Table 16: Component Matrix of Dependent Variable

According from the table 16, there is only one component extracted.
Table 17: KMO and Bartlett’s Test of Dependent Variable

Table 18: Total Variance Explained of Dependent Variable

69
 As can be seen clearly from the table 17, the KMO is 0.809 which greater than
0.5. Besides, total variance explained for the dependent variable equals 60.811% in Table
18 indicates that Cumulative explain 60.811% of total variance in 5 variables. With no
doubt, EFA is appropriate.
In conclusion, after running EFA, there is no items are eliminated from the
list. Next, before coming to Confirmatory Factor Analysis (CFA) to test the model
structure, it is important to detect potential outliers since they may indicate bad data and
reduce the quality of the model testing. However, multivariate outliers need to be
detected by specifying the numbers of suspected outliers exactly. Thus, the following part
will describe carefully how multivariate outliers were detected.
4.4. Detecting Multivariate Outliers
As defined by Sharif (2013), “Multivariate outliers refer to sets of data points
that do not fit the standard sets of correlations exhibited by the other data points in the
dataset with regards to your causal model”. By removing outliers, the results of data
analysis can change in a better way. Multivarite outliers can be detected by Mahalanobis
distance, which is described as “a distance of a data point from the calculated centroid of
the other cases where the centroid is calculated as the intersection of the mean of the
variables being assessed”. (Mahalanobis, 1936).
As the rules of detecting outliers, high Mahalanobis d-squared values with
both p1 and p2 equal .000 and .000 are potential of outliers (then related cases may be
considered for deletion). Therefore, according from Table 19, cases 178, 179, 72, 182,
12, 208, 318, 169, 130, 197, 107, 275 were excluded at the same time. In total, there was
12 items that were deleted from the data file. Thus, the total responses left was 308.
Table 19: Observations Farthest from the Centroid (Mahalanobis Distance)

Observation number Mahalanobis d-squared p1 p2

178 92.963 .000 .001
179 91.538 .000 .000
72 90.567 .000 .000

70
Observation number Mahalanobis d-squared p1 p2
182 85.592 .000 .000
12 80.226 .000 .000
208 79.836 .000 .000
318 79.686 .000 .000
169 79.062 .000 .000
130 78.151 .000 .000
197 77.071 .000 .000
107 76.889 .000 .000
275 75.526 .000 .000
45 74.270 .001 .000
235 74.184 .001 .000
315 73.044 .001 .000
50 71.822 .001 .000
134 71.363 .001 .000
78 71.159 .001 .000
170 70.225 .002 .000
229 69.273 .002 .000
112 68.242 .003 .000
257 67.256 .003 .000
210 66.741 .004 .000
85 66.215 .004 .000
187 66.116 .004 .000
294 65.777 .005 .000
136 65.135 .005 .000
163 64.590 .006 .000
37 64.466 .006 .000
307 63.950 .007 .000
296 62.829 .009 .000
212 62.346 .010 .000
80 61.874 .011 .000
154 61.531 .012 .000
217 61.158 .013 .000
145 60.475 .015 .000
203 60.045 .017 .000
34 59.997 .017 .000
48 59.626 .018 .000

71
Observation number Mahalanobis d-squared p1 p2
153 59.550 .019 .000
177 59.453 .019 .000
133 59.193 .020 .000
41 58.854 .022 .000
120 58.684 .022 .000
109 58.035 .025 .000
181 57.966 .026 .000
155 57.925 .026 .000
9 57.845 .026 .000
226 57.741 .027 .000
99 57.442 .029 .000
106 57.029 .031 .000
101 56.321 .036 .000
119 55.758 .040 .000
232 55.588 .041 .000
295 55.139 .045 .000
51 54.716 .049 .000
16 54.431 .051 .000
213 54.331 .052 .000
116 53.878 .057 .000
260 53.585 .060 .000
68 53.519 .061 .000
215 53.406 .062 .000
52 53.285 .063 .000
205 53.229 .064 .000
253 53.074 .066 .000
302 53.025 .066 .000
61 52.971 .067 .000
73 52.791 .069 .000
297 52.399 .074 .000
35 52.015 .079 .000
157 51.948 .080 .000
267 51.268 .090 .000
132 51.227 .091 .000
174 50.399 .104 .000
211 50.080 .110 .000

72
 Observation number Mahalanobis d-squared p1 p2
143 50.003 .111 .000
186 49.884 .114 .000
313 49.700 .117 .000
195 49.326 .124 .000
270 49.306 .125 .000
194 49.158 .128 .000
193 49.023 .130 .000
75 48.939 .132 .000
89 48.066 .151 .000
149 47.799 .158 .000
233 47.713 .160 .000
18 47.530 .164 .000
90 47.157 .174 .000
67 47.019 .177 .000
122 46.990 .178 .000
204 46.895 .180 .000
100 46.545 .190 .000
59 46.471 .192 .000
201 46.013 .205 .000
159 45.776 .211 .000
156 45.773 .212 .000
214 45.531 .219 .000
288 45.514 .219 .000
5 45.434 .222 .000
166 45.231 .228 .000

4.5. Confirmatory Factor Analysis (CFA)

In statistics, Confirmatory Factor Analysis (CFA) is a commonly used method
in social research. The purpose of conducting CFA is testing the relationship between a
set of observed variables and their latent constructs. In CFA context, the researcher has “a
strong idea about the number of factors, the relations among the factors, and the
relationship between the factors and measured variables” (Ullman, 2006). Besides, CFA

73
does not include factor extraction and rotation. In this part, AMOS 21 software is applied
to run the CFA with the main reason is to test whether the model fits or not with the
survey’s data.
The table below shows the criteria to evaluate the measurement model fit.
Table 20: Criteria for Measurement Model

Measure Threshold
Chi-square/DF (CMIN/DF) < 3 good; < 5 sometimes permissible
p-value for the model > 0.05
CFI (Comparative Fit Index) > 0.95 great; > 0.9 traditional; > 0.8 sometimes
permissible
GFI (Goodness-of-Fit Index) > 0.95 great; > 0.9 traditional; > 0.8 sometimes
permissible
AGFI > 0.8
SRMR < 0.09
RMSEA (Root Mean Squared < 0.06: good fit
Error of Approximation) 0.06 – 0.08: acceptable fit
0.08 – 0.1: mediocre fit
≥ 0.1: poor fit
PCLOSE > 0.05
TLI (Tucker Lewis Index) ≥ 0.9
CR (Composite Reliability) > 0.7 and > AVE
AVE (Average Variance > 0.5
Extracted)
Standardized Regression Weight > 0.5

Source: Joreskog (1969), Bagozzi (1981), Brown and Cudeck (1993), Hair et al.
(2010)

74
75
 Figure 8: First Measurement Standardized Modelling
4.5.1. Check for Model fit
One of the most important steps in CFA is to check the fitness of the proposed
model as well as its data. As defined by Pire (2007), the model fit means “sample data
consistent with the implied model”. Besides, the model fit is better when the discrepancy
between the implied model and the sample data is smaller. Even though there are many
criteria to evaluate the model fit as presented in the table 20, the table below only shows
some recommended types of fit indices, which also are the most important ones to
evaluate the fitness of the model.
Table 21: Model Fit of CFA (First Round)

Fit Indices Thresholds Current Fit Indices

CMIN/DF < 3 good; < 5 sometimes permissible 2.628
CFI > 0.95 great; > 0.9 traditional; > 0.8 0.841
sometimes permissible
GFI > 0.95 great; > 0.9 traditional; > 0.8 0.743
sometimes permissible
TLI ≥ 0.9 0.828
SRMR < 0.09 0.0668
< 0.06: good fit 0.073
RMSEA 0.06 – 0.08: acceptable fit
0.08 – 0.1: mediocre fit
≥ 0.1: poor fit

According to Table 21, some criteria are chosen from the list to measure the
fitness of the model. In comparison with the thresholds in Table 21, it can be seen cleatly
that most of the current fit indices are not well-fitted with the criteria. In specific, the
CMIN/DF equals 2.628 (<3) together with SRMR = 0.0668 (<0.09) can be considered as

76
desired values. Besides, RMSEA = 0.073, which is in the acceptable fit compared to the
standard measurement criteria. However, the other fit indices such as CFI = 0.841 (> 0.8)
is only viewed as “sometimes permissible”; TLI = 0.828 (<0.9) is obviously not in
acceptable range of desired value or GFI = 0.743 (<0.8) which is very far from the level
of fitness. Thus, the question is stated as follows: “What if the model does not fit?”.
According to Pire (2007), there are 3 main ways to improve the model fit, which are: 1/
Model trimming and building (add parameters (LaGrange Multiplier test) and drop
parameters (Wald test)); 2/ Empirical vs. theoretical respecification and 3/ Consider
equivalent models. However, in CFA context, the second method was applied, which
used the respecification from the theory developed by Pire (2007). Specifically, there are
4 cases leading to the unfitted model, which include “Lower factor loadings of the
indicator (standardized<=0.2); High loading on more than one factor; High correlation of
the residuals and the last one is High factor correlation”. In this case, the main reason led
to the low level of fitness of implied model is the High correlation of the residuals. Thus,
according to Pire (2007), it is necessary to allow error measurements to be covered by
using the curved line with two path arrows.
Furthermore, support from Ullman (2006) indicates clearly that “Model
estimates for path coefficients and their standard errors are generated under the implicit
assumption that the model fit is very good. If the model fit is very close, then the
estimates and standard errors may be taken seriously, and individual significance tests on
parameters (path coefficients, variances, and covariances) may be performed”. Thus, it is
necessary to check the modification covariance of specific relationship among items to
remove some pair of items which are unsatisfied to improve the model fit indices.
To conduct the adjustments, some covariance between error terms were
created according from the modification indices.According to Hair et al., modification
indices “are estimated for all nonestimated parameters, so they are also generally
provided for dianosing error term correlations and also correlational relationships
between constructs that may not be initially specified in the CFA model”. Continously,

77
Hair et al.proposed that “Reseachers should consult other residual diagnostics for a
change suggested by a modification index and then take appropriate action”. In other
explaination, if a measurement model is not fitted, modification indices can be applied to
evaluate which relax restrictions would better the model. As stated by Hair et al., it is not
allowed to cover error terms with observed or latent variables, or with other error terms
that are from different factors. The most appropriate modification is to cover errors terms
that are in the same factor. Besides, we need to check the items where errors were freed
based on the covariance tables, which the Modification Indice (M.I) of the two errors
were high.
About the theoretical meaning, the high M.I values among items indicate that
these items have the similar meaning in the questionnaire designed. In other words, each
error value represeneted for each question in the scale measurement of one factor,
however, in order to test the reliability among respondents, many questions in the same
scale have similar meanings although they were described by different words an grammar
structures. To sum up, in both practical and theoretical meaning, there has high
correlations between pair of items having high M.I values, and it can be concluded that
the relationships among these items should be added by drawing curved lines with two
path arrows.
Table 22 below presented the result from the Covariance table combined with
the questions represented by these errors. As can be seen clearly, the M.I values were
ordered from the highest to the smallest one in order to ensure the logical progress in
drawing covariances.
Table 22: Modification Indices Value and Questionnaire Per Items

M.I Error Code Questionnaire per items

e16 CNU7 An important goal is to find a brand that

e16 <--> e17 103.655 communicates my uniqueness.

e17 CNU8 The brands that I like best are the ones that

78
 express my individuality.
e2 HP2 I shop not because I have to, but because I want
to.
e2 <--> e5 86.222
e5 HP5 I enjoy shopping for its own sake and not because
of that I need to purchase something.
e11 CNU2 I dislike brands bought by everyone.

e11 <--> e12 66.408 e12 CNU3 When a sport-wear brand becomes too popular, I
wear it less.
e7 HP7 While shopping, I am able to forget my other

e7 <--> e8 52.734 problems.

e8 HP8 While shopping I feel a sense of adventure.
e8 HP8 While shopping I feel a sense of adventure.
e8 <--> e9 42.344
e9 HP9 Any shopping is a very nice time out to me.
e10 CNU1 I stop wearing fashions when they become

e10 <--> e11 32.895 popular with the general public.

e11 CNU2 I dislike brands bought by everyone.
e31 IB1 When go shopping, I buy things that I had not

e31 <--> e32 28.842 intended to purchase.

e32 IB2 I am a person who makes unplanned purchases.
e13 CNU4 I look for one-of-a-kind products to create my
own style.
e13 <--> e14 27.545
e14 CNU5 The thing that I buy shapes a more unusual
personal image.
e1 HP1 Shopping to me is truly a joy.
e1 <--> e8 24.978
e8 HP8 While shopping I feel a sense of adventure.
e7 HP7 While shopping, I am able to forget my other
e7 <--> e9 23.889
problems.

79
 e9 HP9 Any shopping is a very nice time out to me.
e1 HP1 Shopping to me is truly a joy.

e1<--> e3 22.800 The time spent in shopping is truly enjoyable to

e3 HP3
me.
e11 CNU2 I dislike brands bought by everyone.

e11 <--> e16 21.987 e16 CNU7 An important goal is to find a brand that
communicates my uniqueness.

According from Table 22, the M.I value of the relationship between e16 and
e17 equals 103.655, which is also the highest M.I value in the covariances table. It
indicated that there has a high correlation between these two items, thus, it is necessary to
check the scale measurement of each item. As can be seen clearly, e16 (CNU7) standed
for “An important goal is to find a brand that communicates my uniqueness” and e17
(CNU8) indicated the question “The brands that I like best are the ones that express my
individuality”. Although these two questions were designed to ask in different ways, they
both had the same meaning and purpose. Therefore, it can be explained there has a high
correlation between two items e16 and e17. As a result, it is essential to draw a curved
line from e16 to e17 to show their tight relationships.
Similarly, e11 (CNU2) and e12 (CNU3) were found as the next pair which
have high M.I value (equals 66.408), thus it can be considered to check their correlations.
While e11 displayed for the question “I dislike brands bought by everyone”, e12 implied
for the question “When a sport-wear brand becomes too popular, I wear it less”. As can
be seen that e11 and e12 presented the questions with similar meanings, that means they
correlate with each other, thus, we continue to draw covariance between them.
We continue to draw covariance between e2 (HP2: I shop not because I have
to, but because I want to) and e5 (HP5: I enjoy shopping for its own sake and not because
of that I need to purchase something) with M.I value was 86, 222. Both were from
Hedonic Purchase factor, and the questionnaires were designed to ask the same thing.

80
Next, another pair of Hedonic Purchase factor had high correlation was e7 (HP7: While
shopping, I am able to forget my other problems) and e8 (HP8: While shopping I feel a
sense of adventure). Although these two questions were not familiar in the meaning,
however, it can be explained that the respondents had the same thoughts when they were
asked about these two questions. Thus, the curved line was drawn between e7 and e8
thanks for the above reason.
Next, we continue to draw covariances between these following pairs for their
high modification indices values as well as similarity in the questionnaire designed:
- e8 and e9 with M.I = 42, 344.
- e10 and e11 with M.I = 32.895.
- e31 and e32 with M.I = 28.842.
- e13 and e14 with M.I = 27.545.
- e1 and e8 with M.I = 24.978.
- e7 and e9 with M.I = 23.889.
- e1 and e3 with M.I = 22.800
- e11 and e16 with M.I = 21.987
In short, there had 12 pairs of item were drawn the covariances. As mentioned
above, though we do not suggest to add these relationships to the model, they do imply
the high degree of covariances between these items that are not captured by the construct.
Accordingly, we will draw the correlations between these items in order to increase the
Chi-square and better the model fit.
Again, we run the second round of CFA to check the result after drawing
correlations among these errors.

81
82
 Figure 9: Final Measurement Standardized Modelling
Next, we will see the result from the new model by checking the model fit
indices in the Table 23.
Table 23: Model Fit of CFA (Final Round)

Fit Indices Thresholds Current Fit Indices

CMIN/DF < 3 good; < 5 sometimes permissible 1.936: good
GFI > 0.95 great; > 0.9 traditional; > 0.8 0.813
sometimes permissible
CFI > 0.95 great; > 0.9 traditional; > 0.8 0.91: traditional
sometimes permissible
TLI ≥ 0.9 0.901
SRMR < 0.09 0.0647
< 0.06: good fit 0.055 (good fit)
RMSEA 0.06 – 0.08: acceptable fit
0.08 – 0.1: mediocre fit
≥ 0.1: poor fit

In general, the model fit was in an acceptable fit since its current fit indices
were almost in satisfactory to good range of the thresholds listed in the table. To be
specific, the CMIN/DF = 1.936 (<3), CFI = 0.91 (> 0.09), TLI = 0.901 (≥ 0.9), SRMR =
0.0647 (< 0.09) and RMSEA = 0.055 (< 0.06). Though GFI = 0.813 (>0.8) was only
recognized as “sometimes permissible”, the model was still worth to be evaluated as
satisfactory since 5 out of 6 current fit indices were well- accepted.
Thus, it can be concluded that the measurement model fitted the data good
enough.
4.5.2. Check Convergent Validity

83
Table 24: Regression Weights: (Group number-Default model)

Estimate S.E. C.R. P Label

HP1 <--- HP 1.000
HP2 <--- HP 1.023 .072 14.240 ***
HP3 <--- HP 1.095 .055 19.861 ***
HP4 <--- HP 1.283 .069 18.624 ***
HP5 <--- HP .962 .075 12.836 ***
HP6 <--- HP 1.063 .061 17.449 ***
HP7 <--- HP .889 .078 11.358 ***
HP8 <--- HP .790 .081 9.705 ***
HP9 <--- HP 1.020 .074 13.851 ***
CNU1 <--- CNU 1.000
CNU2 <--- CNU 1.079 .088 12.314 ***
CNU3 <--- CNU 1.108 .100 11.129 ***
CNU4 <--- CNU 1.162 .101 11.450 ***
CNU5 <--- CNU .841 .084 9.969 ***
CNU6 <--- CNU 1.195 .100 11.943 ***
CNU7 <--- CNU .959 .094 10.164 ***
CNU8 <--- CNU .807 .088 9.187 ***
CSC1 <--- CSC 1.000
CSC2 <--- CSC .938 .083 11.259 ***
CSC3 <--- CSC 1.054 .078 13.464 ***
CSC4 <--- CSC 1.040 .081 12.849 ***
CSC5 <--- CSC .920 .087 10.625 ***
CSC6 <--- CSC 1.079 .092 11.781 ***
CSC7 <--- CSC .950 .097 9.838 ***
CSC8 <--- CSC .865 .073 11.895 ***
CSC9 <--- CSC .756 .073 10.333 ***
OSL1 <--- OSL 1.000
OSL2 <--- OSL .923 .060 15.431 ***
OSL3 <--- OSL .760 .065 11.618 ***
OSL4 <--- OSL .664 .060 11.097 ***
IB1 <--- IB 1.000
IB2 <--- IB 1.017 .081 12.598 ***
IB3 <--- IB 1.217 .103 11.824 ***
IB4 <--- IB 1.068 .093 11.519 ***

84
 Estimate S.E. C.R. P Label
IB5 <--- IB .960 .094 10.227 ***
NFC1 <--- NFC 1.000
NFC2 <--- NFC 1.040 .058 17.799 ***
NFC3 <--- NFC .880 .064 13.733 ***
NFC4 <--- NFC .681 .064 10.566 ***

Table 25: Standardized Regression Weights

(Group number 1-Default model)

Estimate
HP1 <--- HP .824
HP2 <--- HP .728
HP3 <--- HP .835
HP4 <--- HP .880
HP5 <--- HP .673
HP6 <--- HP .841
HP7 <--- HP .610
HP8 <--- HP .560
HP9 <--- HP .713
CNU1 <--- CNU .662
CNU2 <--- CNU .719
CNU3 <--- CNU .743
CNU4 <--- CNU .772
CNU5 <--- CNU .657
CNU6 <--- CNU .813
CNU7 <--- CNU .667
CNU8 <--- CNU .594
CSC1 <--- CSC .729
CSC2 <--- CSC .668
CSC3 <--- CSC .796
CSC4 <--- CSC .760
CSC5 <--- CSC .631
CSC6 <--- CSC .698
CSC7 <--- CSC .585
CSC8 <--- CSC .705

85
 Estimate
CSC9 <--- CSC .614
OSL1 <--- OSL .879
OSL2 <--- OSL .815
OSL3 <--- OSL .636
OSL4 <--- OSL .612
IB1 <--- IB .689
IB2 <--- IB .638
IB3 <--- IB .795
IB4 <--- IB .767
IB5 <--- IB .665
NFC1 <--- NFC .835
NFC2 <--- NFC .891
NFC3 <--- NFC .718
NFC4 <--- NFC .582

As a result from Table 24, all P-value, which is denoted by ***, are smaller
than the conventional 0.05. Besides, according from Table 25, all the value of
standardized regression weights are greater than 0.5, which are acceptable. Next, we
continue to check the reliability.
4.5.3. Check Reliability
In order to conclude whether the measurement model adapt the requirement of
convergent validity or not, we need to check the Standardized Regression Weights as
well as Average Variance Extracted (AVE). In details, convergent validity exists when all
of items in scales are loaded significantly, or Standardized Regression Weights must
greater than 0.5 in other words. (Hair et al, 2006). All variables are categorized into
proper constructs, which denotes the convergent validity of the model. Besides, AVE
must highly above 0.5 to be considered as acceptable (Fornell & Larcker, 1981).
In addition, though Cronbach’s Alpha helped reveal the reliability of this
study, it is necessary to be reassessed by Composite Reliability, which is “calculated by
the formed formula and λ-coefficients gained from the AMOS estimated results” (Fornell
& Larcker, 1981). The condition that Composite Reliability must satisfy to make sure the

86
 reliability of overall study has consistency with Cronbach’s Alpha’s criteria is equal or
higher than 0.7.
Next, the formula to calculate Composite Reliability (CR) and Average
Variance Extracted (AVE) are shown as follows:
Equation 1: Composite reliability (Joreskog 1971)

λ: the standardized loading on its corresponding factor

1-λ 2 : the variance of measurement error of each indicator
p: the number of indicators
Equation 2: Extracted variance–VE (Fornell& Larcker 1981)

λi: the i standardized factor loading on its corresponding factor

n: the number of indicators
Table 26: Composite Reliability and Extracted Variance of Each Factor

Estimate Sum of (Sum  2

1-  2
Sum of AVE Composite
( )  of  ) 2 (1-  2 ) Reliability
HP1 <--- HP 0.824 0.679 0.321
HP2 <--- HP 0.728 0.53 0.47
HP3 <--- HP 0.835 0.697 0.303
HP4 <--- HP 0.880 0.774 0.226
6.664 44.409 3.966 0.559 0.918
HP5 <--- HP 0.673 0.453 0.547
HP6 <--- HP 0.841 0.707 0.293

87
HP7 <--- HP 0.610 0.372 0.628
HP8 <--- HP 0.560 0.314 0.686
HP9 <--- HP 0.713 0.508 0.492
CNU1 <--- CNU 0.662 0.438 0.562
CNU2 <--- CNU 0.719 0.517 0.483
CNU3 <--- CNU 0.743 0.552 0.448
CNU4 <--- CNU 0.772 0.596 0.404
CNU5 <--- CNU 0.657 0.432 0.568
5.627 31.663 4.006 0.5 0.888
CNU6 <--- CNU 0.813 0.661 0.339
CNU7 <--- CNU 0.667 0.445 0.555
CNU8 <--- CNU 0.594 0.353 0.647
CSC1 <--- CSC 0.729 0.531 0.469
CSC2 <--- CSC 0.668 0.446 0.554
CSC3 <--- CSC 0.796 0.634 0.366
CSC4 <--- CSC 0.760 0.578 0.422
CSC5 <--- CSC 0.631 0.398 0.602
6.186 8.267 4.71 0.477 0.89
CSC6 <--- CSC 0.698 0.487 0.513
CSC7 <--- CSC 0.585 0.342 0.658
CSC8 <--- CSC 0.705 0.497 0.503
CSC9 <--- CSC 0.614 0.377 0.623
OSL1 <--- OSL 0.879 0.773 0.227
OSL2 <--- OSL 0.815 0.664 0.336
OSL3 <--- OSL 0.636 2.942 8.655 0.404 0.596 1.784 0.554 0.829

OSL4 <--- OSL 0.612 0.375 0.625

IB1 <--- IB 0.689 0.475 0.525

88
IB2 <--- IB 0.638 0.407 0.593
IB3 <--- IB 0.795 0.632 0.368
IB4 <--- IB 0.767 3.554 12.631 0.588 0.412 2.456 0.509 0.837

IB5 <--- IB 0.665 0.442 0.558

NFC1 <--- NFC 0.835 0.697 0.303
NFC2 <--- NFC 0.891 0.794 0.206
NFC3 <--- NFC 0.718 3.026 9.157 0.516 0.484 1.654 0.586 0.847

NFC4 <--- NFC 0.582 0.339 0.661

Table 26 presents the Standardized Regression Weights together with the

Average Variance Extracted (AVE) and the Composite Reliability (CR). According to the
result, all items were loaded in appropriate constructs and the all the Standardized
Regression Weights are greater than 0.5. Besides, almost factors contained AVEs with
values higher than 0.5, except CSC. However, it was still acceptable in this case since the
value of CSC nearly approached to the criteria. In addition, all the CR tested presented
qualified results since they were above 0.7. Thus, as inferred from the result, all factors in
the table had sufficient reliability and the measurement model met the convergent
validity.
4.5.4. Check Discriminant Validity
Discriminant Validity tests the divergent degree between the concepts. In
other explanations, discriminant validity will be acceptable in case overlaps don’t exist
among constructs. To test the discriminant validity, the correlation between two
constructs must be lower than 1, or both AVEs of the two constructs must be higher than
r2 .
Table 27: Correlations (CFA)

Estimates r2 Both AVEs > r 2

89
HP <-> CNU 0.427 YES
0.182
HP <-> CSC -0.16 YES
0.026
HP <-> OSL 0.423 YES
0.179
HP <-> IB 0.605 YES
0.366
HP <-> NFC 0.586 YES
0.343
CNU <-> CSC 0.145 YES
0.021
CNU <-> OSL 0.403 YES
0.162
CNU <-> IB 0.463 YES
0.214
CNU <-> NFC 0.603 YES
0.364
CSC <-> OSL 0.082 YES
0.007
CSC <-> IB -0.278 YES
0.077
CSC <->NFC -0.079 YES
0.006
OSL <-> IB 0.379 YES
0.144
OSL <-> NFC 0.262 YES
0.069
IB <-> NFC 0.522 YES
0.272

The results implies that both AVEs of the constructs are higher than r 2 in each
correlation. Therefore, the study ensured the discriminant validity.
4.6. Structural Equation Modeling (SEM)
In the previous procedure, CFA was mainly used as a first step to evaluate the
proposed measurement model by concentrating on how well the variables characterized
the constructs. SEM was distinguished from CFA by the fact that in SEM, the
relationship between latent factors and their level of significance are focused through the
process of testing structural model.
The result of Structural Equation Modeling was shown as below.

90
Figure 10: First Structural Equation Model

91
 4.6.1. Check for Model Fit
Table 28: Model Fit of SEM (First Round)

Fit Indices Thresholds Current Fit Indices

CMIN/DF < 3 good; < 5 sometimes permissible 2.320
CFI > 0.95 great; > 0.9 traditional; > 0.8 0.872
sometimes permissible
GFI > 0.95 great; > 0.9 traditional; > 0.8 0.781
sometimes permissible
TLI ≥ 0.9 0.861
RMSEA < 0.06: good fit 0.066
0.06 – 0.08: acceptable fit
0.08 – 0.1: mediocre fit
≥ 0.1: poor fit

Similarly, the model was tested for its fitness. The recommended criteria
consists of CMIN/DF, CFI, GFI, TLI and RMSEA with the same standard for evaluating
in CFA. As a result, CMIN/DF = 2.320 (<3) was good, CFI = 0.872 (>0.8) was
considered as “sometimes permissible” and RMSEA = 0.066 was in the range of
acceptable fit. However, GFI = 0.781 and TLI = 0.861 were totally not acceptable since
the values were well below the desired range, thus, it cannot be concluded that the model
fit well with the data collected from the survey. In short, some adjustments would be
necessarily conducted to improve the model fit indices.
Next, the result of Regression Weight was shown as below:
Table 29: Regression Weights of SEM Testing: (Group number 1-Default
Model) (First Round)

Estimate S.E. C.R. P Label

IB <--- HP .309 .053 5.805 ***

92
 Estimate S.E. C.R. P Label
IB <--- OSL .154 .052 2.955 .003
IB <--- CSC -.302 .062 -4.854 ***
IB <--- NFC .122 .044 2.782 .005
IB <--- CNU .222 .057 3.877 ***
HP1 <--- HP 1.000
HP2 <--- HP 1.009 .073 13.912 ***
HP3 <--- HP 1.102 .056 19.841 ***
HP4 <--- HP 1.292 .069 18.610 ***
HP5 <--- HP .950 .076 12.554 ***
HP6 <--- HP 1.068 .061 17.415 ***
HP7 <--- HP .873 .079 11.063 ***
HP8 <--- HP .775 .082 9.478 ***
HP9 <--- HP 1.019 .074 13.752 ***
CNU1 <--- CNU 1.000
CNU2 <--- CNU 1.117 .090 12.356 ***
CNU3 <--- CNU 1.123 .102 11.046 ***
CNU4 <--- CNU 1.149 .103 11.148 ***
CNU5 <--- CNU .828 .085 9.691 ***
CNU6 <--- CNU 1.223 .103 11.881 ***
CNU7 <--- CNU .936 .095 9.809 ***
CNU8 <--- CNU .784 .089 8.837 ***
CSC1 <--- CSC 1.000
CSC2 <--- CSC .940 .083 11.276 ***
CSC3 <--- CSC 1.055 .078 13.452 ***
CSC4 <--- CSC 1.042 .081 12.863 ***
CSC5 <--- CSC .920 .087 10.618 ***
CSC6 <--- CSC 1.070 .092 11.676 ***
CSC7 <--- CSC .948 .097 9.811 ***
CSC8 <--- CSC .869 .073 11.927 ***
CSC9 <--- CSC .756 .073 10.335 ***

93
 Estimate S.E. C.R. P Label
OSL1 <--- OSL 1.000
OSL2 <--- OSL .906 .061 14.881 ***
OSL3 <--- OSL .739 .065 11.360 ***
OSL4 <--- OSL .639 .060 10.720 ***
IB1 <--- IB 1.000
IB2 <--- IB 1.025 .093 10.961 ***
IB3 <--- IB 1.155 .125 9.215 ***
IB4 <--- IB .914 .114 8.036 ***
IB5 <--- IB .862 .107 8.066 ***
NFC1 <--- NFC 1.000
NFC2 <--- NFC 1.116 .066 16.857 ***
NFC3 <--- NFC .871 .066 13.166 ***
NFC4 <--- NFC .652 .066 9.819 ***

According from the Regression Weights table, it can be seen that all the P-
values indicating the relationships between IB (Impulse Buying) and independent factors
were significant (below 0.05). Thus, we can apply some adjustments to increase the
model fit as mentioned above.
Apart from CFA, Structural Equation Modelling (SEM) was applied to
observe the relationships among the constructs in the measurement model and its validity.
In other words, SEM can be understood as a “series of statistical methods that allow
complex relationships between one or more independent variables and one or more
dependent variables”. (Ullman, 2006). Thus, SEM not only examines the relationship
between the independent and dependent variables but also checks the correlations among
independent factors. Conceptually speaking, if the two independent variables are believed
to have a causal effect on the other, it is necessary to draw covariance between them by
the two path arrows. As explained in the context of SEM, obtaining covariance estimates
between variables encourage ones to improve the estimate direct and indirect impact on

94
other variables, which mainly happened in complex models with numerous parameters to
be assessed. Therefore, it is allowed to draw covariance to better the model fit indices.
However, only high correlation between variables are permitted to make
covariance since it indicates clearly the causal effect among these variables. The table
below presented the correlations among independent variables. As can be seen
noticeably, 6 correlations between CNU and NFC; CNU and OSL; HP and NFC; HP and
OSL; HP and CNU were highlighted due to their high M.I (Modification Indices) values.
Table 30: Covariance: (Group number 1-Default model) (First Round)

M.I. Par Change

OSL <--> NFC 12.532 .162
CNU <--> NFC 78.546 .386
CNU <--> OSL 35.861 .231
CNU <--> CSC 5.072 .076
HP <--> NFC 79.639 .414
HP <--> OSL 43.446 .270
HP <--> CSC 6.361 -.091
HP <--> CNU 42.798 .256

In practical explanation, high M.I values indicated high correlations among

factors. In theoretical explanation, relationships existed among factors having high M.I
values. Specifically, the highest M.I values in the covariances table (M.I. = 79.639) of the
relationship between HP (Hedonic Purchase) and NFC (Novelty-fashion consciousness)
showed that the relationship between hedonic purchase and novelty-fashion
consciousness were strong. As defined by Bhatnagar (2004), hedonic purchase is referred
to “emotional needs of individuals for enjoyable and interesting shopping experiences”,
while novelty-fashion consciousness describes individuals who are fashionable with
novelty conscious and love to experience new things. Therefore, these customers were

95
found to have the same interest in shopping and were predicted to have effect on each
other. That means people having hedonic purchase behavior tend to have high
consciousness on novelty-fashion and vice versa.
Similarly, the next pairs having high M.I value is the pair of CNU
(Consumer’s Need for Uniqueness) and NFC (Novelty-fashion consciousness) (M.I =
78.546). As mentioned above, consumer’s need for uniqueness is claimed to be
associated with products which have some symbolic meanings to enhance individual’s
self and social image as an expression of uniqueness. Thus, it can be inferred that people
with high demand for distinction likely engaged with fashion since they always seek for
fashion products that differentiate them from others. Especially for those customers
having creative choice conformity-one trait of need for uniqueness, they must have high
novelty-fashion consciousness since they want to build their images in both unique and
attractive way. In short, there was a relationship between the need for uniqueness of
customers and their novelty-fashion consciousness.
Next, the relationship between HP (Hedonic Purchase) and OSL (Optimum
Stimulation Level) was highlighted thanks to the high M.I value (M.I=43.446). As
mentioned above, those who have low optimum stimulation level can comfortably satisfy
with commonplace situation while individuals having high optimum stimulation level can
easily have negative feelings in a steady and usual environment since they have higher
propensity to discover latest inducement to fulfil their demand on stimulations. In
addition, new experiences tends to be accepted straightforwardly by group having high
optimum stimulation level and thus they are more engaging and willing to check out new
products, services and brands. As mentioned by Holbrook & Hirschman (1982), hedonic
purchase relates to emotional arousal, which happens while carrying out purchasing
process. Accordingly, individuals with high optimum stimulation level are likely to have
hedonic purchase thanks to the same interest in shopping and seeking for new things.
Therefore, this study pointed out a positive relationship between optimum stimulation
level and hedonic purchase.

96
 HP (Hedonic Purchase) and CNU (Consumer’s Need for Uniqueness) was
explored to have strong relationship with M.I = 42.798. Regarding above explanation,
hedonic purchase describes people who love shopping experience, and new products or
brands can be acquired more rapidly by purchasers who have greater demand of
distinction than those having that demand at a lower level Knight. & Kim (2007). Thus, it
can be inferred that there has a positive relationship between the consumer’s need for
uniqueness hedonic purchase, which implies those customers have strong interested in
shopping activities.
Last but not least, the relationship between CNU (Consumer’s Need for
Uniqueness) and OSL (Optimum Stimulation Level) was pointed out in this research
study thanks to the high M.I value. A research study about product replacement of Miller
et al. (1993) mentioned the relationship between consumer’s need for uniqueness and
optimum stimulation level as follows “Individual differences in consumers’ need for
uniqueness might also influence product or style replacement behavior through its effect
on consumers’ optimum stimulation level”. Besides, McAlister and Pessemier’s (1982)
proved the mediating role of optimum stimulation in the relationship between CNU and
variety seeking behavior by the theoretical model of variety seeking behavior. In details,
the theory proposes that individual’s need for uniqueness affect variety seeking through
its relationship with different level of stimulation of consumers such as novelty,
complexity or change. In other words, “Consumers possessing a high need for uniqueness
relative to others will seek to avoid popular consumer preferences, will become more
familiar with product offerings during their search for unique goods, and, thus, will
require greater novelty or complexity to achieve optimal levels of arousal”. Regarding
above explanation, it can be concluded that there has a positive relationship between
consumer’s need for uniqueness and optimum stimulation level.
Finally, the last pair of OSL (Optimum Stimulation Level) and NFC (Novelty-
fashion consciousness) was considered to have correlation with the M.I value equals
12.532. As can be inferred, keep the styles always up-to-date is one of most important

97
thing for the consumers having novelty-fashion consciousness. Since the consumer in this
segment always excited in seeking for new styles, they usually pleased themselves and
“satisfy their need for stimulation and thereby achieving Optimum Stimulation Level”
(Chang, Burns & Francis, 2004). Thus, it can be said in confident that apart from the high
M.I value, there is a positive relationship between novelty-fashion consciousness and
optimum stimulation level.
Then, we run the second round of SEM testing after drawing covariance and
check the result in the figure below.

98
Figure 11: Second Structural Equation Model

99
 After running the second round of SEM, it is important to have a look at
model fit indices and examine the result in the table 31.
Table 31: Model Fit of SEM (Second Round)

Fit Indices Thresholds Current Fit Indices

CMIN/DF < 3 good; < 5 sometimes permissible 1.944: good
CFI > 0.95 great; > 0.9 traditional; > 0.8 0.909: traditional
sometimes permissible
GFI > 0.95 great; > 0.9 traditional; > 0.8 0.812:
sometimes permissible sometimes permissible
TLI ≥ 0.9 0.901
RMSEA < 0.06: good fit 0.055: good fit
0.06 – 0.08: acceptable fit
0.08 – 0.1: mediocre fit
≥ 0.1: poor fit

After the final SEM testing, CMIN/DF = 1.944 and it was good in comparison
with the threshold (<3). In addition, CFI= 0.909, also satisfied the criteria evaluating the
measurement model. Moreover, TLI = 0.901, which also exceeded 0.9 and was viewed as
a desired value. Last but not least, RMSEA (=0.055) supported the good fit of the model
by showing the value well within the desired range of threshold. Although GFI= 0.812,
which was only considered as sometimes permissible, it was still acceptable in the range
of threshold. Regarding above comparisons, it could be seen clearly that the model was in
an acceptable fit with the data from survey.
However, before coming to the conclusion of the fitness of model, it is
necessary to examine the result from the Covariances table again to check whether the P-
values of correlations between factors are significant or not. If not, we must remove the

100
covariances created and continue to run SEM testing until all the P-values are well-
significant.
Table 32: Covariances (Group number 1: Default model) (Second Round)

Estimate S.E. C.R. P Label

HP <--> NFC .447 .059 7.528 ***
CNU <--> NFC .423 .061 6.977 ***
OSL <--> NFC .188 .049 3.858 ***
CNU <--> OSL .236 .044 5.353 ***
HP <--> OSL .270 .045 5.964 ***
HP <--> CNU .265 .047 5.646 ***

As a result, all P-values are well significant. Therefore, it can be

confidentially concluded that the model are in the acceptable fit with the data collected.

In accordance with the mentioned result, we continue to check the Regression

Weights in Table 33.
Table 33: Regression Weights of SEM Testing (Group number 1-Default)
(Second Round)

Estimate S.E. C.R. P Label

IB <--- HP .293 .069 4.276 ***
IB <--- OSL .138 .062 2.207 .027
IB <--- CSC -.300 .061 -4.937 ***
IB <--- NFC .114 .067 1.717 .086
IB <--- CNU .209 .077 2.701 .007
HP1 <--- HP 1.000
HP2 <--- HP 1.017 .072 14.179 ***
HP3 <--- HP 1.094 .055 19.876 ***
HP4 <--- HP 1.283 .069 18.702 ***
HP5 <--- HP .957 .075 12.793 ***

101
 Estimate S.E. C.R. P Label
HP6 <--- HP 1.062 .061 17.507 ***
HP7 <--- HP .883 .078 11.297 ***
HP8 <--- HP .785 .081 9.663 ***
HP9 <--- HP 1.018 .073 13.876 ***
CNU1 <--- CNU 1.000
CNU2 <--- CNU 1.080 .088 12.327 ***
CNU3 <--- CNU 1.106 .099 11.136 ***
CNU4 <--- CNU 1.162 .101 11.469 ***
CNU5 <--- CNU .838 .084 9.949 ***
CNU6 <--- CNU 1.195 .100 11.955 ***
CNU7 <--- CNU .954 .094 10.133 ***
CNU8 <--- CNU .800 .088 9.134 ***
CSC1 <--- CSC 1.000
CSC2 <--- CSC .940 .083 11.276 ***
CSC3 <--- CSC 1.055 .078 13.452 ***
CSC4 <--- CSC 1.042 .081 12.862 ***
CSC5 <--- CSC .920 .087 10.617 ***
CSC6 <--- CSC 1.071 .092 11.675 ***
CSC7 <--- CSC .948 .097 9.810 ***
CSC8 <--- CSC .869 .073 11.925 ***
CSC9 <--- CSC .757 .073 10.335 ***
OSL1 <--- OSL 1.000
OSL2 <--- OSL .923 .060 15.423 ***
OSL3 <--- OSL .760 .065 11.633 ***
OSL4 <--- OSL .662 .060 11.056 ***
IB1 <--- IB 1.000
IB2 <--- IB 1.026 .081 12.722 ***
IB3 <--- IB 1.157 .103 11.187 ***
IB4 <--- IB .919 .094 9.742 ***
IB5 <--- IB .865 .091 9.480 ***

102
 Estimate S.E. C.R. P Label
NFC1 <--- NFC 1.000
NFC2 <--- NFC 1.041 .059 17.772 ***
NFC3 <--- NFC .880 .064 13.722 ***
NFC4 <--- NFC .682 .064 10.577 ***

As can be seen clearly, all P-values were acceptable, except the P-values of
the relationship between IB and NFC. (P-value = 0.086 > 0.05). This proved that IB only
had significant relationships with OSL (Optimum Stimulation Level), HP (Hedonic
Purchase), CSC (Consumer Self-Spending Control), and CNU (Consumer’s Need for
Uniqueness) or IB had no significant relationship with NFC (Novelty-Fashion
Consciousness) in other words. Therefore, it is essential to eliminate NFC from the SEM
model and continue to check the Regression Weights after removing.
Table 34: Regression Weights of SEM Testing (Group number 1-Default
model) (Third Round)

Estimate S.E. C.R. P Label

IB <--- HP .348 .062 5.635 ***
IB <--- OSL .125 .062 2.016 .044
IB <--- CSC -.313 .062 -5.056 ***
IB <--- CNU .281 .069 4.073 ***
HP1 <--- HP 1.000
HP2 <--- HP 1.012 .072 14.046 ***
HP3 <--- HP 1.100 .055 19.870 ***
HP4 <--- HP 1.282 .069 18.568 ***
HP5 <--- HP .952 .075 12.654 ***
HP6 <--- HP 1.067 .061 17.525 ***
HP7 <--- HP .880 .078 11.221 ***
HP8 <--- HP .788 .081 9.669 ***
HP9 <--- HP 1.019 .074 13.842 ***
CNU1 <--- CNU 1.000

103
 Estimate S.E. C.R. P Label
CNU2 <--- CNU 1.103 .091 12.138 ***
CNU3 <--- CNU 1.119 .103 10.843 ***
CNU4 <--- CNU 1.184 .106 11.209 ***
CNU5 <--- CNU .863 .087 9.868 ***
CNU6 <--- CNU 1.218 .104 11.669 ***
CNU7 <--- CNU .974 .098 9.977 ***
CNU8 <--- CNU .820 .091 9.052 ***
CSC1 <--- CSC 1.000
CSC2 <--- CSC .940 .083 11.280 ***
CSC3 <--- CSC 1.054 .078 13.464 ***
CSC4 <--- CSC 1.041 .081 12.864 ***
CSC5 <--- CSC .920 .087 10.618 ***
CSC6 <--- CSC 1.070 .092 11.686 ***
CSC7 <--- CSC .947 .097 9.803 ***
CSC8 <--- CSC .868 .073 11.935 ***
CSC9 <--- CSC .756 .073 10.343 ***
OSL1 <--- OSL 1.000
OSL2 <--- OSL .922 .060 15.378 ***
OSL3 <--- OSL .762 .065 11.646 ***
OSL4 <--- OSL .663 .060 11.069 ***
IB1 <--- IB 1.000
IB2 <--- IB 1.019 .080 12.752 ***
IB3 <--- IB 1.154 .103 11.233 ***
IB4 <--- IB .913 .094 9.753 ***
IB5 <--- IB .854 .090 9.462 ***

After three times of running Regression Weights in SEM testing, all the P-
values were smaller than the conventional value 0.05. This showed that the all the
relationships between IB and HP, IB and CSC as well as IB and CNU were significant.
This results brought the Structural Equation Modelling to the final result as Figure 12.

104
Figure 12: Final Structural Equation Model

105
 Figure 12 illustrated the final result of this research study, which included
only four factors left. Again, it is important to check the model fit as well as the
covariances to examine whether the relationships among factors were significant.
Table 35: Model Fit of SEM (Third Round)

Fit Indices Thresholds Current Fit Indices

CMIN/DF < 3 good; < 5 sometimes permissible 1.927: good
CFI > 0.95 great; > 0.9 traditional; > 0.8 0.919: traditional
sometimes permissible
GFI > 0.95 great; > 0.9 traditional; > 0.8 0.833:
sometimes permissible sometimes permissible
TLI ≥ 0.9 0.911
RMSEA < 0.06: good fit 0.055: good fit
0.06 – 0.08: acceptable fit
0.08 – 0.1: mediocre fit
≥ 0.1: poor fit

According from Table 35, all measurement values satisfied the criteria for the
measurement model fit. Chi-squared = 1.927 (<3); CFI =0.919 (>0.9); TLI = 0.911 (>0.9)
and RMSEA = 0.055 (<0.06) implied that the model fitted well with the data collected.
Though GFI still reached at “sometimes permissible” level (0.833 >0.8) after three times
running SEM testing, it can be accepted and the model was in an acceptable fit in general
speaking.
Table 36: Covariances: (Group number 1-Default model) (Third Round)

Estimate S.E. C.R. P Label

CNU <--> OSL .232 .044 5.343 ***
HP <--> OSL .270 .045 5.970 ***

106
 Estimate S.E. C.R. P Label
HP <--> CNU .260 .046 5.614 ***
As can be seen clearly, all the P-values were below 0.05. Thus, it can be
concluded that all the relationships between factors were well significant.
Last but not least, the table of Standardized Regression Weights was displayed
to give estimate values among relationships between dependent factor and independent
factors.
Table 37: Standardized Regression Weights
(Group number 1-Default model)

Estimate
IB <--- HP .385
IB <--- OSL .131
IB <--- CSC -.295
IB <--- CNU .281

Finally, a table of hypotheses testing in SEM was presented to provide the test
result.
4.6.2. Hypotheses Testing in SEM

107
 Table 38: Hypotheses Testing in SEM

The SEM Result

Causal Path Estimate Test Result
H1 Optimum Stimulation Level 0.131 Supported
 Level of Impulse Buying
H2 Hedonic Purchase 0.385 Supported
 Level of Impulse Buying
H3 Consumer’s Need for Uniqueness 0.281 Supported
 Level of Impulse Buying
H5 Consumer Self-Spending Control -0.295 Supported
 Level of Impulse Buying

The final SEM testing reveals that all four hypotheses were supported or this
model could help explain the percentage of impact level of each factor on impulse
buying. In details, hedonic purchase, consumer’s need for uniqueness and optimum
stimulation level had positive influences on impulse buying, in which, hedonic purchase
factor caused the greatest impact level on this consumer behavior with 38.5%. On the
contrary, consumer self-spending control caused negative effect on impulse buying with
the level of impact was 29.5%, which just right after the level of influence of hedonic
purchase. Thus, it can be concluded that consumer’s need for uniqueness had stronger
influence on impulse buying than optimum stimulation level but weaker than hedonic
purchase and consumer self-spending control in terms of impact level. Last but not least,
it is essential to revise the research model after discarding NFC (novelty-fashion
consciousness). The new theoretical framework was performed as follows:
4.6.3. Revised Research Model

108
 Figure 13: The Revised Research Model
In short, the hypotheses which were kept to the final research model were H1 –
“Optimum Stimulation Level has significant positive influence on Level of Impulse
Buying”; H2 – “Hedonic Purchase has significant positive influence on Level of Impulse
Buying”; H3 – “Consumer’s Need for Uniqueness has significant negative influence on
Level of Impulse Buying” and H5 – “Consumer’s Need for Uniqueness has significant
negative influence on Level of Impulse Buying”. These hypotheses were expected to

109
have value for both fashion brands/retailers and researchers in investigating and
developing impulse buying behavior.

110
 CHAPTER V
CONCLUSIONS AND RECOMMENDATIONS
This chapter provides a summary of the whole research process as well as answers
for the research questions proposed in the first chapter. Besides, this paper also gives
recommendations for fashion brands and retailers to improve factors that have impacts
on the level of impulse buying, therefore enhance impulse purchase among consumers as
well as develop business performances. Lastly, this research helps figure out limitations
of the research and recommends new solutions for future researches in the field of
consumer behavior.
5.1. Conclusion
Since there was almost no previous research studies before found out factors
causing more than 30% of impact on level of impulse buying, this research study
continued to conduct research on this topic on purpose of seeking factors that truly have
strong influences on impulse purchase. In summary, the proposed theoretical framework
has confirmed the important roles of optimum stimulation level, hedonic purchase,
consumer’s need for uniqueness, and consumer self-spending control on impulse buying
in the field of research studies as well as practical implementations. In specific, this paper
revealed that hedonic purchase had highest positive level of influence on impulse buying,
in which the impact level nearly reached to 40% (38.5%). On the contrary, consumer
self-spending-control caused the negative impact on this consumer behavior with around
30% level of influence (29.5%). It indicated that individuals having strong interest in
shopping activities are likely to indulge in impulse buying, and these kinds of consumers
were viewed as the most potential impulse buyers. Another explanation is that, hedonic
people will purchase more than normal ones to seek for seek for excitement and satisfy
themselves. In other words, shopping can be used as a way to relax and enjoy joyfulness

111
in life to those individuals having hedonic characteristics, thus lead to increasing in
impulse buying with less or no control.
In contrast, individuals with high self-control will tend to have tight spending
and thus reduce the level of impulse buying naturally. Obviously, if people have high
consciousness about their financial budgets, they will not easy to have unplanned
purchasing while individuals with low self-spending control will feel free to shop at any
time.
About consumer’s need for uniqueness, it can be explained that distinctness is
an important attribute which led consumers to their impulse buying. In other words,
people with high demand for exclusiveness will spend lots of time and money to seek for
items which can differentiate them from others, and that made them become impulse
buyers since they will buy products spontaneously as long as these items perform their
images a unique way. Besides, this kind of customers will acquire new products or
fashion brands faster than those having low need for uniqueness because they think that
owning new arrivals products sooner than others will make them look more exclusive and
trendy. In short, consumers with high need for uniqueness may become impulse buyers,
and vice versa.
Next, optimum stimulation were explored to cause the weakest influence on
impulse buying, and the reason may come from the differences in sampling locations and
sample size, in which this research was conducted in Vietnam, and most of participants
were students – the middle to low class of the society. To their point of views, optimum
stimulation may not an important factor when coming to impulse purchase decisions for
some reasons. First, it is not necessary to increase stimulation by buying things
spontaneously, which costs lots of money and can harm to the long-term financial
budgets. They are still able to achieve optimum stimulation level by doing exciting things
such as changing routine activities or seeking for new experiences in games or jobs.
Besides, to people who want to reach a higher level of stimulation, sometimes shopping
activities are not enough. Regarding above explanations, it can be concluded that

112
optimum stimulation level did affect impulse buying but the level of influence did not
account for a large proportion. Thus, fashion brands or retailers should not invest too
much on increasing optimum stimulation level among consumers to reach a higher degree
of impulse buying behavior.
Last but not least, novelty-fashion consciousness was the only factor among
the hypotheses that caused no impact on impulse buying. The reason may come from the
point of view of consumers towards “fashion”. In their opinions, people with high fashion
consciousness are not necessarily involved in impulsive purchase to build their images or
create their own styles. These kind of consumers still can be “fashionista” (a new term
that describes a devoted follower of fashion) by wearing trendy costumes and utilizing
their own resources without spending too much on shopping. In fact, they usually spend
time on seeking for those items which can combine together to design their brand images
or able to make the most out of their clothes instead of buying with less or no intentions
as impulse buyers. To sum up, people with high novelty-fashion consciousness will tend
to have well-planed purchasing to seek for items which are necessary for their suits rather
than buying with no purposes. Therefore, retailers or fashion brands should not be aware
of this kind of customer if they want to improve the impulse behavior.
In summary, the results of this study found out four factors causing effect on
impulse buying, which included optimum stimulation level, hedonic purchase, consumer
need for uniqueness and consumer’s self-spending control. It can be said that the result
was partially matched with the hypotheses and the theoretical framework proposed in
chapter two. Furthermore, the research’s outcome also contributes some practical
implications for retail industry and theoretical implementations for further research study.
Thus, the next part will give recommendations for retailing strategies as well as future
investigations5.2. Recommendations
5.2.1. Recommendations for Retailing Strategies
On the aspect of practical applications, retailers as well as fashion brands need
to focus most on factors which cause strongest impact on impulse buying motives to

113
enhance the sales and revenues. As can be inferred from the research result, consumer’s
need for uniqueness and hedonic purchase are explored as factors that play main role to
consumer’s impulsiveness. Thus, it would lead to the following issues for the retail stores
to be concentrated on:
Firstly, to enhance the spontaneous buying in hedonic consumers, fashion
brands should launch many “Induction Events” (an event which is held to introduce new
collections of the brand) or Golden Day (an occasion aims to express deep gratitude to
loyal customers by inviting them to private dinners as well as giving gifts or offering
discount V.I.P cards for the next purchasing). Likewise, individuals with high level of
hedonic consumption will easily attracted to the seasonal sales off or occasional discount
due to the high demand for dressing smartly in such special events. Consequently,
organizing events in a suitable timeline combined with releasing new collections
regularly will encourage hedonic customers to shop more than their actual needs.
Secondly, to attract people with high demand for uniqueness, the retailers/
fashion brands should investigate in designing and decorating the store environment in
the most distinct way, which can easily differentiate from other fashion stores. Besides,
clothing and accessories should be combined together to make out the most of the styles
as well as encourage the consumers to buy the whole set instead of one shirt or jeans.
More important, people with high need for distinction will not accept to buy cloths
produced in a series and ready-to-wear fashion, thus, the fashion brands need to create
unique designed items and produce in a very limited amount. These kind of customers
will feel urge to buy “limited edition” products to fulfill their needs of uniqueness.
Finally, it can be said with some degree of confidence that these
recommendations above would useful for those fashion brands or retailers if they applied
them properly and smartly.
5.2.2. Recommendations for Future Research Study
Future research study should investigate into other factors predicted might
cause positive effect on level of impulse buying such as price, shopping enjoyment, store

114
reputation or brand awareness, etc. to explore the factors which have bigger impacts on
impulse buying than those conducted in this paper. Besides, making more examinations
of seeking for new influenced factors on impulse purchase would give a general overview
about impulse buying behavior in many aspects, thus help researchers in proposing more
applicable theoretical frameworks. Additionally, researchers ought to do analysis on the
moderating effect between the novelty-fashion consciousness and the level of impulse
due to the less influence of novelty- fashion consciousness. The moderating effect
suggested is hedonic purchase, thus, the further research topic may include testing the
moderating effect of hedonic purchase on the relationship of novelty-fashion
consciousness and level of impulse buying to provide a full understanding about
shopping behavior among consumers in Ho Chi Minh City for the retailers or fashion
brands.
Furthermore, the future research should extend the sample size to decrease the
bias and increase the validity as well as the reliability in the data collected. In addition,
the further investigation also gets advantage from this study by replicating it in the new
research setting or another industry in Vietnam market. It would be interesting if the
replication in another industry or market brought a reversed results and gave different
point of views among consumers about factors affect impulse buying.
To sum up, extra investigations following this concept should be conducted in
the near future to give the answer for the new objectives as well as fulfill the
understanding about the impulse buying concept for the researchers and the businessman.
5.3. Limitations
Similar to any previous research studies, this paper cannot escape from
drawbacks caused by many reasons. First, the method of gathering responses discussed in
chapter 3 is viewed as one of the limitations of this study. As stated above, convenient
sampling method was applied in this paper, which means respondents were opportunely
chosen from International University, thus lead to bias results. Additionally, lacking of
time and budget prevents the author from choosing larger sample size to gain more

115
valuable results. Therefore, small sample size cannot represent the population so the main
findings may be inaccurate somehow.
Next, the scope of the research study, which actually mainly focused on
shopping malls and retail shops around the center of Ho Chi Minh City, is the second
limitation. Geographical problems are great barriers to broaden the market in collecting
data. However, Ho Chi Minh City was chosen for a reason. As explained above, since Ho
Chi Minh City is the economic center of Vietnam with the highest per capita income
compared to the national average, it is expected to have more impulse buyers due to their
high financial budgets as well as living condition. Nonetheless, to diversify the research
results, future research should vary the sample in other great urban centers such as Hanoi
or Da Nang city.
Last but not least, the highest factor affecting level of impulse accounted for
40%, which means the implication of the findings may limit within a few contexts.
Finally, the drawbacks may come from the questionnaires. Although the pilot test was
conducted before delivering final version, it still caused misunderstanding for some of
respondents in choosing suitable answers. The reason may come from the translating
from English to Vietnamese, which already met some difficulties before.

116
 REFERENCES
Abratt, R., & Goodey, S. D. (1990). Unplanned Buying and In-Store Stimuli in
Supermarkets. Managerial and Decision Economics, 11(2), 111-121.
Abdullah J. S. & Jeff J. & David E. S. (2011). Building consumer self-control: The effect
of self-control exercises on impulse buying urges.
Aluja, A., Garica, O. and Garcia, L.s.F (2003). Relationships among extraversion,
openness to experience, and sensation seeking. Personality and Individual Differences,
35 (3), 671-680.
Anderson, J., Gerbing, W., (1988). Structural equation modelling in practice: A review
and recommended two stage approach. Psychological Bulletin 27(1), 5-24.
Babin, B. J., Darden, W. R., & Griffin, M. (1994). Work and/or fun: Measuring hedonic
and utilitarian shopping value. Journal of Consumer Research, 20(4), 644-656.
Bae, S. (2004). Shopping pattern differences of physically active Korean and American
university consumers for athletic apparel. Doctoral thesis. Florida: The Florida State
University College of Education.
Bagozzi, Richard P. (1981). An examination of the Validity of Two Models of Attitude.
Multivariate Behavioral Research, 16, 323-359.
Bartlett, M. S. (1954). A note on the multiplying factors for various chi square
approximations. Journal of Royal Statistical Society, 16(Series B), 296-298.
Batra, R., & Ahtola, O. (1990). Measuring the Hedonic and Utilitarian sources of
consumer attitude.Marketing Letters, 2(2), 159-170.
Baumeister, R. F. (2002). Yielding to Temptation: Self-control failure, impulsive
purchasing, and consumer behaviour. Journal of Consumer Research, 28(4), 670-676.
Baumeister Roy F. Yielding to temptation: self-control failure, impulsive purchasing and
consumer behavior. Journal Consumer Research 2002; 28(4):670–6.

117
Baumeister, R. F., Gaillot, M., DeWall, C. N., & Oaten, M. (2006). Self-regulation and
personality: how interventions increase regulatory success, and how depletion moderates
the effects of traits on behavior. Journal of Personality, 74(6), 1773–1802.
Bayley, G., & Nancarrow, C. (1998). Impulse Purchasing: A Qualitative Exploration of
the Phenomenon. Qualitative Market Research: An International Journal, 1(2), 99-114
Beatty, S. E., & Ferrell, M. E. (1998). Impulse Buying: Modeling Its Precursors. Journal
of Retailing, 74(2), 169-191.
Bearden, W.O., Netemeyer, R.G., (1998). Handbook of Marketing Scales: Multi-Item
Measures for Marketing and Consumer Behaviour Research, Sage Publications,
Thousand Oaks, USA.
Baumgartner H., Steenkamp J.B.E.M. (1996).Exploratory consumer buying behavior:
conceptualization and measurement. International Journal Research Marketing; 13,121–
37.
Bhatnagar, A., & Ghosh, S. (2004). A latent class segmentation analysis of E-Shoppers.
Journal of Business Research, 57(7),758-767.
Browne, Michael W. and Robert Cudeck (1993). Alternative Ways of Assessing Model
Fit," in Testing Structural Equation Models, K. A. Bollen et al. eds, Newbury Park CA:
SAGE Publications.
Cavana, R., Delahaye, B., and Sekaran, U. (2001). Applied Business Research:
Qualitative and Quantitative Methods, (3ed.).John Wiley & Sons Australia, Ltd.
Chang, H. J., Eckman, M., & Yan, R. N. (2011). Application of the Stimulus-Organism-
Response model to the retail environment: the role of hedonic motivation in impulse
buying behavior. The International Review of Retail, Distribution and Consumer
Research, 21(3), 233-249.
Chang, E., Burns, L. D., & Francis, S. K. (2004). Gender differences in the dimensional
structure of apparel shopping satisfaction among Korean consumers: The role of hedonic
shopping value. Clothing and Textiles Research Journal, 22(4), 185-199.

118
Childers, T. L., Carr, C. L., Peck, J., & Carson, S. (2001). Hedonic and Utilitarian
motivations for online retail shopping behavior. Journal of Retailing, 77, 417-424.
Churchill, G. A. & Iacobucci, D. (2006). Marketing Research: Methodological
Foundations. 9ed. Mason: South-Western College Pub.
Clover, V. T. (1950). Relative importance of impulse-buying in retail stores. The Journal
of Marketing, 15(1), 66-70.
Cobb, C. J., & Hoyer, W. D (1986). Planned versus Impulsive Behavior, Journal of
Retailing, 62 (4), 384-409.
Comfrey, A. L., & Lee, H. B. (1992). A First Course in Factor Analysis. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Cooper, D. R., & Schindler, P. S. (2006). Business Research Methods (9ed). USA:
McGraw-Hill.
Deloitte, Vietnam’s total personal disposable income and total household consumption in
2008-2017. Retail in Vietnam, Emerging Market, and emerging growth. Retrieved from:
http://www2.deloitte.com/content/dam/Deloitte/jp/Documents/consumer-business/dis/jp-
dis-vietnam-retail-en.pdf
Deloitte, Vietnam’s retail sales in 2008-2017. Retail in Vietnam, Emerging Market, and
emerging growth. Retrieved from:
http://www2.deloitte.com/content/dam/Deloitte/jp/Documents/consumer-business/dis/jp-
dis-vietnam-retail-en.pdf
Deloitte, Ho Chi Minh’s retail sales in 2009-2013. Retail in Vietnam, Emerging Market,
emerging growth. Retrieved from:
http://www2.deloitte.com/content/dam/Deloitte/jp/Documents/consumer-business/dis/jp-
dis-vietnam-retail-en.pdf
Dickman Scott J. Impulsivity, arousal and attention. Personality Individual Differences
2000;28. 563–81.
Engel, J., & Blackwell, R. (1982). Consumer Behavior. Chicago: Dryden Press.

119
Eysenck H. J. (1993). The impulsive client: Theory, research and treatment. American
Psychological Association, 57–70.
Faber, R. J., & Vohs, K. D. (2004). To buy or not to buy: self-control and self-regulatory
failure in purchase behavior. Handbook of self-regulation: research, theory, and
applications. 509–524.
Field, A. (2000). Discovering Statistics using SPSS for Windows. London – Thousand
Oaks – New Delhi: Sage publications.
Fornell, C., and Larcker, D. F. (1981). Evaluating Structural Equation Models with
Unobservable Variables and Measurement Error. Journal of Marketing Research 18(1),
39-50
Forsythe, S., Liu, C. L., Shannon, D., & Gardner, L. C. (2006). Development of a scale to
measure the perceived benefits and risks of online shopping. Journal of Interactive
Marketing, 20(2), 55-75.
Fornell, Claes and Larker. D. F. (1981). Evaluating Structural Equation Models with
Unobservable Variables and Measurement Error. Journal of Marketing Research, 18, 39-
50.
Ghiselli, E. E. (1955). The measurement of occupational aptitude. Berkeley, CA:
University of California.
George, D., & Mallery, P. (2003). SPSS for Windows step by step: A simple guide and
reference. 11.0 update (4ed.). Boston: Allyn & Bacon.
Gorusch, R. L. (1983). Factor Analysis (2ed.). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Green, P. E., & Rao, V. R. (1970) Rating scales and informational recovery – how many
scales and response categories to use? Journal of Marketing, 34, 33-39.
Gross, Engel. H. (1977). Micro and Macro Level Implications for a Sociology of Virtue:
The Case of Draft Protesters to the Vietnam War. Sociological Quarterly, 18, 319–339.
Hair, J. F., Anderson, R. E., Tatham, R. L. & Black, W. (1998). Multivariate Data
Analysis. 5ed. Upper Saddle River: Prentice Hall.

120
Hair, J. F., Barry J.B., Arthur H. M., Philip S. (2003). Essentials of business research
methods (International ed). John Wiley & Sons, Inc.
Hair, J. F., Bush, P. R. & Ortinau, J. D. (2006). Marketing Research: Within a Changing
Information Environment. Revised International Edition. Boston: McGraw-Hill.
Hair, J., Black, W., Babin, B., and Anderson, R. (2010). Multivariate data analysis
(7ed.). Prentice-Hall, Inc. Upper Saddle River, NJ, USA. Retrieved from:
http://saeedsharif.com/tag/discriminant/
Hatcher, L. (1994). A Step-by-Step Approach to Using the SAS® System for Factor
Analysis and Structural Equation Modeling. Cary, N.C.: SAS Institute, Inc.
Hausman, A. (2000). A multi-method investigation of consumer motivation in impulse
buying behavior. Journal of Consumer Marketing, 17(5), 403-419.
Haws. K.L., Bearden W.O. and Nenkov. G. Y., Consumer Spending Self-Control
Effectiveness and Outcome Elaboration Prompts, Journal of Academy of Marketing
Science, forthcoming, 1, 3-4.
Heckert, Druann Maria (1989). The Relativity of Positive Deviance: The Case of the
French Impressionists. Deviant Behavior, 10, 131–144.
Hirschman, E. C., & Stern, B. B. (2001). Do consumers' genes influence their behavior?
Findings on novelty seeking and compulsive consumption. Advances in Consumer
Research, 28, 403-410.
Hoch, S. J., & Loewentein, G. F. (1991). Time-inconsistent Preferences and Consumer
Self-control. Journal of Consumer Research, 17 (4), 492-507.
Holbrook, M. B., & Hirschman, E. C. (1982 b). Hedonic consumption: Emerging
concepts, methods and propositions. Journal of Marketing, 46, 92-101.
Holt, Douglas B. (1995). How Consumers Consume: A Typology of Consumption
Practices. Journal of Consumer Research, 22, 1–16.
Hopkinson, C. G., & Davashish, P. (1999). A factor analytic study of the sources of
meaning of Hedonic consumption. European Journal of Marketing, 33(3/4), 273-290.

121
Hyo Jung Chang, Ruoh-Nan Yan and Molly Eckman (2013). Moderating effects of
situational characteristics on impulse buying, Journal of Emerald Insight. 298-341.
Jöreskog, K. G. (1970). A General Method for Analysis of Covariance Structures.
Biometrika, 57, 239-251.
Jöreskog, K. G. (1971). Statistical analysis of sets of congeneric tests. Psychometrika, 36,
109-133.
Kaiser, H. F. (1974). An index of factorial simplicity. Psychometrika, 39, 31-36.
Kim J. (2003). College students’ apparel impulse buying behaviors in relation to visual
merchandising. Under the Direction of Dr. Brigitte Burgess (BFA, American
Intercontinental University, 2000), 40-47.
Kish, G. B. and Donnenwerth, G. V. (1972). Sex differences in the correlates of stimulus
seeking. Journal of Consulting and Clinical Psychology 38 (1), 42-49.
Knight, D.K., & Kim, E. Y. (2007). Japanese consumers' need for uniqueness: Effects on
brand perceptions and purchase intention. Journal of Fashion Marketing and
Management, 11(2), 270-280.
Likert R. Biography (2005). Retrieved from:
http://www.bookrags.com/biography/rensis-likert-soc/#gsc.tab=0
Lundahl, U., Skärvad, P-H. (1999) Utredningsmetodik för samhällsvetare och ekonomer,
Lund: Studentlitteratur.
Mahalanobis, P. C. (1936). On the generalised distance in statistics. Proceedings of the
National Institute of Sciences of India 2 (1), 49–55.
Maslach, Christina, Joy S., and Richard T. S. (1985). Individuation: Conceptual Analysis
and Assessment. Journal of Personality and Social Psychology, 49, 729–738.
McAlister, Leigh and Edgar Pessemier (1982). Variety Seeking Behavior: An
Interdisciplinary Review. Journal of Consumer Research, 9, 311–322.
Menon, S. and Kahn, B. E. (1995). The impact of context on variety seeking in product
choices. Journal of Consumer Research, 22 (3), 285-295.

122
Mika B. (1995) ,Optimum Stimulation Level and Recreational Shopping Tendency, in E -
European Advances in Consumer Research Volume 2, eds. Flemming Hansen, Provo, UT
: Association for Consumer Research, 372-380.
Miller, Christopher M., Shelby H. McIntyre, and Murali K. Mantrala (1993). Toward
Formalizing Fashion Theory. Journal of Marketing Research, 30, 142–157.
Mittelstaedt, R. A., Grossbart, S. L., Curtis, W. W. and DeVere, S. P. (1976). Optimal
stimulation level and the adoption decision process. Journal of Consumer Research, 3
(2), 84-94.
Neuman, W. L. (2006). Social Research Methods: Qualitative and Quantitative
Approaches 6th Edition, Pearson International Edition, USA.
Nguyen T.T.M. (2003). An Exploratory Investigation into Impulse Buying Behavior in a
Transitional Economy: A Study of Urban Consumers in Vietnam. Journal of
International Marketing. 11 (2), 13-35, 21-25.
Prof. Dr. Aisha M. E. M. (2012) “Personal characteristics and situational influences
affecting consumers adoption of impulse buying behavior”, International conference of
business and economic research, 3274-3276.
Pire (2007), Structural Equation Modelling Workshop. Retrieved from:
http://medresearchconsult.com/docs1/SEM1.pdf
Raju, P. S., and Mittelstaedt. V. (1980). Exploratory behavior in the consumer context: A
state of the art review, 7, 258-63.
Raju, P. S. (1980). Optimum stimulation level: Its relationship to personality,
demographics, and exploratory behavior. Journal of Consumer Research, 7 (3), 272-282.
Rietveld, T. & Van Hout, R. (1993). Statistical Techniques for the Study of Language and
Language Behaviour. Berlin – New York: Mouton de Gruyter.
Romal, Jane B. and Barbara J. K. (1995). Difference in Self Control among Spenders and
Savers. Psychology: A Quarterly Journal of Human Behavior, 32 (2), 8-17.
Rook, D. W. (1987). The Buying Impulse. Journal of Consumer Research, 14(2), 189-
197.

123
Rook, D., & Hoch, S. (1985). Consuming impulses. Advances in Consumer Research,
7(1), 23-27.
Rook, D. W., & Fisher, R. J. (1995). Normative influences on impulsive buying behavior.
The Journal of Consumer Research, 22(3), 305-313.
Sharif S.P. (2013). Data screening and CFA. Structural Equation Modeling (SEM) and
AMOS Workshop. Taylor’s Graduate School. Retrieved from:
http://saeedsharif.com/wp-content/uploads/2013/08/Data-Screening.pdf
Sharma, P., Sivakumaran, B., & Marshall, R. (2010). Impulse buying and Variety
seeking: A trait-correlates perspective. Journal of Business Research, 63, 276-83.
Simonson, I., & Nowlis, S. M. (2000). The role of explanations and need for uniqueness
in consumer decision making: Unconventional choices based on reasons. Journal of
Consumer Research, 27, 49–68.
Slater, S. (1995). Issues in Conducting Marketing Strategy Research. Journal of Strategic
Marketing, 3(4), 257-270.
Snyder, C. R. (1992), “Product Scarcity by Need for Uniqueness Interaction: A
Consumer Catch-22 Carousel?” Basic and Applied Social Psychology, 13, 9–24.
Snyder, C. R.; Fromkin, Howard L. (1977) Abnormality as a positive characteristic: The
development and validation of a scale measuring need for uniqueness. Journal of
Abnormal Psychology, 86(5), 518-527.
Sogunro, O. A. (2001). Selecting a quantitative or qualitative research methodology: An
experience. Educational Research Quarterly, 26 (1), 3-10.
Spencer L., Ritchie J, O’Connor W. (2004). Analysis: practices, principles and
processes. Qualitative research practice, 199-218.
Sproles G.B., (1985), “From perfectionism to Dadaism: measuring consumers' decision-
making styles”, in Proceedings American Council on Consumer Interest Conference,
Columbus, OH, 79–85.
Sproles, G.B., Kendall, E.L., (1986), “A methodology for profiling consumer decision
making styles”, The Journal of Consumer Affairs, 20 (2), 67-79.

124
Steenkamp J.B.E.M, and Baumgartner H. (1992). The role of optimum stimulation level
in exploratory consumer behavior. Journal of Consumer Research; 19, 434–48.
Steenkamp J.B.E.M, and Baumgartner H. (1995) Development and cross-cultural
validation of a short form of CSI as a measure of optimum stimulation level.
International Journal Research Marketing; 12, 97–104.
Steenkamp, J.-B. E. M. and Burgess, S. M. (2002). Optimum stimulation level and
exploratory consumer behavior in an emerging consumer market. International Journal
of Research in Marketing, 19 (2), 131-150.
Soares, Ana M. (2004). The influence of culture and consumers: exploratory and risk
taking behavior. Doctoral thesis. Universidade do Minho.
Strauss, A., Corbin, J. (1998). Basics of Qualitative Research – Techniques and
Procedures for Developing Grounded Theory, second edition, London: Sage
Publications.
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5ed.). Boston:
Allyn and Bacon.
Tangney, J. P., Baumeister, R. F., & Boone, A. L. (2004). High self-control predicts good
adjustment, less pathology, better grades, and interpersonal success. Journal of
Personality, 72, 271–324.
Thompson, Craig J., William B. L., and Howard P. (1990). The Lived Meaning of Free
Choice: An Existential- Phenomenological Description of Everyday Consumer
Experiences of Contemporary Married Women. Journal of Consumer Research, 17, 346-
361.
Tian, K.T., Bearden, W.O. and Hunter, G.L. (2001), Consumers’ need for uniqueness:
scale development and validation, Journal of Consumer Research, 28 (1), 50-66.
Tian, K.T. and Mckenzie, K. (2001), “The long-term predictive validity of consumers’
need for uniqueness”, Journal of Consumer Psychology, 10(3), 1971-3.
Tull, D. S., Hawkins, I. D. (1993). Marketing research: measurement and method (6ed).
New York: Macmillan.

125
Ullman. J. B. (2006). Structural Equation Modeling: Reviewing the Basics and Moving
Forward. Department of Psychology California State University, San Bernardino. Journal
of Personality Assessment, 87(1), 35–50.
Vietnam General Statistics Office, 2014. Gross Domestic Product per Capita of Vietnam.
Retrieved from: http://www.gso.gov.vn/default_en.aspx?tabid=487&ItemID=4327
Vries, R. E. d., Vries, A. d. and Feij, J. A. (2009). Sensation seeking, risk-taking, and the
HEXACO model of personality. Personality and Individual Differences, 47 (6), 536-540.
Yin, R. K. (1994). Case study research: design and methods (2ed).Thousand Oaks, CA:
Sage.
Youn, S., & Faber, R. J. (2000). Impulse buying: its relation to personality traits and
cues. Advances in consumer research, 27, 179-185.
Yu, C., & Bastin, M. (2010). Hedonic shopping value and impulse buying behavior in
transitional economies: A symbiosis in the Mainland China marketplace. Journal of
Brand Management, 18(2), 105-114.
Zhou, L., & Wong, A. (2004). Consumer impulse buying and in-store stimuli in Chinese
supermarkets. Journal of International Consumer Marketing, 16(2), 37-53.
Zuckerman M. (1993). Sensation seeking and impulsivity: a marriage of traits made in
biology. The impulsive client: Theory, research and treatment. Washington, DC:
American Psychological Association; 71–92.

World Bank (2014), Vietnam GDP Per capita. Retrieved from

http://www.tradingeconomics.com/vietnam/gdp-per-capita

126
 APPENDIX A
SURVEY QUESTIONNAIRE
 Questionnaire in English

IDENTIFYING FACTORS AFFECTING THE LEVEL OF

IMPULSE BUYING IN FASHION MALLS
IN HO CHI MINH CITY
My name is Ta Bao Tran and I am a senior student of International University,
Vietnam National University. Currently, I am conducting my graduation thesis with the
research’s topic “Factors affecting the level of impulse buying in fashion industry of
consumers in Ho Chi Minh City”. It would be my pleasure if you could spend your
valuable time on filling this survey with purpose of collecting sufficient and precise data
to achieve the research results. There is no right or wrong answer in this research, instead,
each answer may bring value in creating practical perspective to conduct a research.
Besides, all of the respondents’ information collected in this survey will be
kept confidential.
Finally, I would like to give a big thanks to all of you for your contribution.
Wish you happiness and success in life.
PART I: MAIN QUESTIONS
In this context, a research design was formed to measure the level of impulse
by some factors which were predicted to have impact on. Please indicate your level of
agreement for each question by choosing from 1 to 5, which was coded as follows:
1: Totally Disagree – 2: Disagree – 3: Neutral – 4: Agree – 5: Totally Agree.
OPTIMUM STIMULATION LEVEL

127
  To what extent do you agree that the following statements describe you?
1.1 I like to experience novelty and change in daily routine. 1 2 3 4 5
1.2 I am continually seeking new ideas and experiences. 1 2 3 4 5
1.3 I like continually changing activities. 1 2 3 4 5
1.4 When things get boring, I like to try something different. 1 2 3 4 5
HEDONIC PURCHASE
2.1 Shopping to me is truly a joy. 1 2 3 4 5
2.2 I shop not because I have to, but because I want to. 1 2 3 4 5
2.3 The time spent in shopping is truly enjoyable to me. 1 2 3 4 5
2.4 I enjoy being immersed in exciting new products while shopping. 1 2 3 4 5
2.5 I enjoy shopping for its own sake and not because of that I need to purchase 1 2 3 4 5
something.
2.6 While shopping I can feel the excitement of the hunt. 1 2 3 4 5
2.7 While shopping, I am able to forget my other problems. 1 2 3 4 5
2.8 While shopping I feel a sense of adventure. 1 2 3 4 5
2.9 Any shopping is a very nice time out to me. 1 2 3 4 5
CONSUMER’S NEED FOR UNIQUENESS 1 2 3 4 5
 Avoidance of similarity
3.1 I stop wearing fashions when they become popular with the general public. 1 2 3 4 5
3.2 I dislike brands bought by everyone. 1 2 3 4 5
3.3 When a sport-wear brand becomes too popular, I wear it less. 1 2 3 4 5
 Unpopular choice
3.4 I look for one-of-a-kind products to create my own style. 1 2 3 4 5
3.5 The thing that I buy shapes a more unusual personal image. 1 2 3 4 5
3.6 I buy unusual brands to create a more distinctive personal image. 1 2 3 4 5
 Creative choice

128
3.7 An important goal is to find a brand that communicates my uniqueness. 1 2 3 4 5
3.8 The brands that I like best are the ones that express my individuality. 1 2 3 4 5
CONSUMER SPENDING SELF – CONTROL
4.1 I closely monitor my spending behavior. 1 2 3 4 5
4.2 I am able to work effectively toward long term financial goals. 1 2 3 4 5
4.3 I carefully consider my needs before making purchases. 1 2 3 4 5
4.4 I often delay taking action until I have carefully considered the consequences of 1 2 3 4 5
my purchase decisions.
4.5 When I go out with friends, I keep track of what I am spending. 1 2 3 4 5
4.6 I am able to resist temptation in order to achieve my budget goals. 1 2 3 4 5
4.7 I know when to say when regarding how much I spend. 1 2 3 4 5
4.8 In social situations, I am generally aware of what I am spending. 1 2 3 4 5
4.9 Having objectives related to spending is important to me. 1 2 3 4 5
NOVELTY – FASHION CONSCIOUS
5.1 I usually have one or more outfits of the very newest style. 1 2 3 4 5
5.2 I keep my wardrobe up –to-date with the changing fashions. 1 2 3 4 5
5.3 Fashionable, attractive styling is very important to me. 1 2 3 4 5
5.4 To get variety, I shop different stores and choose different brands. 1 2 3 4 5
LEVEL OF IMPULSE BUYING
6.1 When go shopping, I buy things that I had not intended to purchase. 1 2 3 4 5
6.2 I am a person who makes unplanned purchases. 1 2 3 4 5
6.3 It is fun to buy spontaneously. 1 2 3 4 5
6.4 When I bought (the item), I felt a spontaneous urge to buy it. 1 2 3 4 5
6.5 When I saw (the item), I just couldn't resist buying it. 1 2 3 4 5

129
PART II: DEMOGRAPHIC QUESTIONS
1. Your Gender
⎕ Male ⎕ Female
2. Your Age
⎕ 15 – 18 years old ⎕ 31 – 40 years old
⎕ 19 – 30 years old ⎕ Over 40 years old
3. Your Occupation
⎕ Students ⎕ Executive (Engineer, Doctor, Teacher,
etc.)
⎕ Office Executive ⎕ Others
⎕ Retailer/ Housewife
4. Your Income (VND)
⎕ < 3 million ⎕ 8 – 15 million
⎕ 3 – 7 million ⎕ > 15 million

 Questionnaire in Vietnamese
BẢNG KHẢO SÁT NHỮNG NHÂN TỐ ẢNH HƯỞNG ĐẾN
MỨC ĐỘ MUA HÀNG NGẪU HỨNG TẠI CÁC TRUNG TÂM
MUA SẮM THỜI TRANG Ở THÀNH PHỐ HỒ CHÍ MINH
Tôi tên là Tạ Bảo Trân, hiện đang là sinh viên năm cuối của trường Đại học
Quốc Tế - ĐHQG TPHCM. Hiện tôi đang làm khóa luận tốt nghiệp với đề tài nghiên cứu
về những yếu tố ảnh hưởng đến mức độ mua hàng ngẫu hứng của người tiêu dùng Việt
Nam. Vì tính chất của đề tài, thị trường mục tiêu được chọn là thị trường quần áo tại Việt
Nam. Rất mong Anh/Chị dành ít thời gian trả lời các câu hỏi được đưa ra một cách
KHÁCH QUAN. Các câu trả lời đều không có tính đúng hay sai mà có giá trị như nhau
trong nghiên cứu, nhằm tạo nên cái nhìn thiết thực về thực tiễn.

130
 Thông tin thu được từ việc khảo sát sẽ được bảo mật hoàn toàn và chỉ sử dụng
cho mục đích nghiên cứu.
Cuối cùng, cho phép tôi được dành sự biết ơn sâu sắc và cảm ơn chân thành
đến những đóng góp của anh/chị trong việc hỗ trợ trả lời bảng khảo sát này.
PHẦN I: CÂU HỎI CHÍNH
Dưới đây là những câu hỏi đo lường những yếu tố quyết định mức độ mua
hàng ngẫu hứng của khách hàng. Xin vui lòng cho biết mức độ đồng ý của anh/ chị đối
với mỗi yếu tố bằng việc chọn từ 1 đến 5, trong đó:
1: Hoàn toàn không đồng ý – 2: Không đồng ý – 3: Trung lập – 4: Đồng ý – 5: Hoàn
toàn đồng ý
MỨC ĐỘ TỐI ƯU KÍCH THÍCH TIÊU DÙNG
 Anh/chị đồng ý những miêu tả sau đây về mình ở mức độ nào?
1.1 Tôi thích trải nghiệm cái mới và sự thay đổi trong đời sống hàng ngày. 1 2 3 4 5
1.2 Tôi tiếp tục tìm kiếm những ý tưởng và trải nghiệm mới. 1 2 3 4 5
1.3 Tôi thích thay đổi hoạt động liên tục. 1 2 3 4 5
1.4 Tôi thích thử những thứ mới. 1 2 3 4 5
TẬN HƯỞNG MUA SẮM
2.1 Mua sắm đối với tôi thật sự là một niềm vui. 1 2 3 4 5
2.2 Tôi mua sắm không phải vì tôi cần mua sắm, mà vì tôi thích như thế. 1 2 3 4 5
2.3 Thời gian dành cho việc mua sắm thật sự rất thú vị đối với tôi. 1 2 3 4 5
2.4 Tôi tận hưởng cảm giác được ngập trong vui sướng khi mua sắm. 1 2 3 4 5
2.5 Tôi thích mua sắm vì tôi thích việc mua sắm chứ không phải vì tôi cần 1 2 3 4 5
mua sắm thứ gì đó.
2.6 Tôi có thể cảm nhận được cảm giác thích thú khi mua sắm. 1 2 3 4 5
2.7 Khi mua sắm, tôi có khả năng quên hết mọi vấn đề. 1 2 3 4 5
2.8 Tôi có cảm giác thám hiểm khi mua sắm. 1 2 3 4 5
2.9 Bất cứ khi nào mua sắm cũng là quãng thời gian tuyệt vời với tôi 1 2 3 4 5

131
 NHU CẦU ĐỘC NHẤT CỦA KHÁCH HÀNG
 Tránh sự giống nhau 1 2 3 4 5
3.1 Tôi ngưng mặc những sản phẩm khi những sản phẩm đó trở nên phổ biến 1 2 3 4 5
với cộng đồng.
3.2 Tôi không thích những nhãn hiệu có quá nhiều người sử dụng. 1 2 3 4 5
3.3 Khi một nhãn hiệu quần áo trở nên quá thông dụng, tôi mặc chúng ít hơn. 1 2 3 4 5
 Lựa chọn không phổ biến 1 2 3 4 5
3.4 Tôi tìm kiếm những sản phẩm có 1-0-2 để tạo nên phong cách riêng cho 1 2 3 4 5
bản thân.
3.5 Những thứ mà tôi mua tạo cho tôi một hình tượng cá nhân khác lạ. 1 2 3 4 5
3.6 Tôi sử dụng những nhãn hiệu ít thông dụng để tạo nên hình tượng cá nhân 1 2 3 4 5
khác biệt.
 Lựa chọn sáng tạo 1 2 3 4 5
3.7 Mục đích quan trọng là tìm kiếm một nhãn hiệu có thể truyền tải được cá 1 2 3 4 5
tính riêng của tôi.
3.8 Những nhãn hiệu mà tôi thích nhất là những nhãn hiệu mà thể hiện được 1 2 3 4 5
cá tính riêng của tôi.
KIỂM SOÁT TIÊU DÙNG CỦA KHÁCH HÀNG
4.1 Tôi kiểm soát chặt chẽ hành vi tiêu dùng của mình. 1 2 3 4 5
4.2 Tôi có khả năng hoạch định hiệu quả những mục tiêu tài chính dài hạn. 1 2 3 4 5
4.3 Tôi thường cân nhắc cẩn thận nhu cầu bản thân trước khi mua sắm. 1 2 3 4 5
4.4 Tôi chỉ quyết định khi đã cân nhắc cẩn thận hậu quả của quyết định mua 1 2 3 4 5
sắm của mình.
4.5 Khi đi chơi với bạn bè, tôi kiểm soát được chi tiêu của mình. 1 2 3 4 5
4.6 Tôi có khả năng cưỡng lại cám dỗ để không chi tiêu quá ngân sách dự 1 2 3 4 5
định.
4.7 Tôi nhớ hết các khoản chi tiêu của mình. 1 2 3 4 5

132
4.8 Trong các tình huống, tôi thường nhận biết được tôi đang chi tiêu cái gì. 1 2 3 4 5
4.9 Việc có những mục tiêu liên quan đến tiêu dùng quan trọng đối với tôi. 1 2 3 4 5
NHẬN THỨC KHÁCH HÀNG VỀ TÍNH MỚI LẠ VÀ THỜI TRANG
5.1 Tôi thường xuyên có một hoặc nhiều hơn bộ cánh cập nhật mốt mới nhất. 1 2 3 4 5
5.2 Tôi luôn giữ cho tủ quần áo của mình luôn theo kịp với sự thay đổi của 1 2 3 4 5
thời trang
5.3 Phong cách thời trang sành điệu, cuốn hút luôn quan trọng đối với tôi 1 2 3 4 5
5.4 Tôi mua sắm ở nhiều cửa hàng và nhãn hiệu khác nhau để tìm kiếm sự đa 1 2 3 4 5
dạng.
MỨC ĐỘ MUA HÀNG NGẪU HỨNGNGẪU HỨNG
6.1 Khi đi mua sắm, tôi thường mua những thứ tôi không định mua từ trước. 1 2 3 4 5
6.2 Tôi là loại người mua sắm không có kế hoạch. 1 2 3 4 5
6.3 Mua sắm một cách tự phát rất thú vị. 1 2 3 4 5
6.4 Khi mua một món hàng, tôi cảm thấy một sự thôi thúc tự phát để mua nó. 1 2 3 4 5
6.5 Khi trông thấy một món hàng, tôi không thể kiềm chế được mong muốn 1 2 3 4 5
mua nó.

PHẦN II: THÔNG TIN CÁ NHÂN

Câu 1: Giới tính:
⎕ Nam ⎕ Nữ
Câu 2: Độ tuổi:
⎕ Từ 15 – 18 ⎕ Từ 30 – 40
⎕ Từ 19 – 30 ⎕ Trên 40
Câu 3: Xin vui lòng cho biết nghề nghiệp hiện tại của anh/chị
⎕ Học sinh/ Sinh viên ⎕ Chuyên gia (bác sĩ, kĩ sư, kiến trúc sư, giáo viên, etc.)
⎕ Nhân viên văn phòng ⎕ Nghề nghiệp khác
⎕ Nội trợ, buôn bán lẻ

133
Câu 4: Xin vui lòng cho biết thu nhập hàng tháng của anh/chị
⎕ Dưới 3 triệu VND ⎕ 8 triệu – 15 triệu VND
⎕ 3 triệu – 7 triệu VND ⎕ Trên 15 triệu VND

Rất cảm ơn anh/chị đã dành thời gian để hoàn thành bảng khảo sát này.
Chúc anh/chị luôn hạnh phúc, vui vẻ và thành công trong cuộc sống.

134
 APPENDIX B
RELIABILITY TEST
1. Optimum Stimulation Level

2. Hedonic Purchase

135
3. Consumer’s Need for Uniqueness

136
4. Novelty – Fashion Consciousness

137
5. Consumer Self- Spending Control

138
6. Level of Impulse Buying

139
140
 APPENDIX C
FACTOR ANALYSIS
Factor Analysis for Independent Variables (Optimum Stimulation Level,
Hedonic Purchase, Consumer’s Need for Uniqueness, Consumer Self-Spending Control,
and Novelty – Fashion Conscious).

141
142
 APPENDIX D
CFA RESULTS
Covariances: (Group number 1 – Default model)
M.I. Par Change
e39 <--> 6 6.745 -.100
e39 <--> 1 5.799 .078
e38 <--> e39 8.994 .130
e37 <--> 6 6.895 .066
e37 <--> 4 4.512 -.057
e36 <--> e39 7.944 -.102
e36 <--> e38 13.144 -.120
e36 <--> e37 7.872 .065
e35 <--> 4 4.098 -.069
e35 <--> e36 6.108 .081
e34 <--> e39 5.575 .089
e34 <--> e35 13.538 .124
e33 <--> 1 6.168 .067
e33 <--> e35 4.637 -.074
e32 <--> 3 7.212 -.089
e32 <--> 1 11.548 -.104
e32 <--> e34 15.567 -.140
e31 <--> 4 5.605 .075
e31 <--> e35 4.242 -.071
e31 <--> e34 5.745 -.074
e31 <--> e32 28.842 .195
e30 <--> 3 4.831 .059
e29 <--> e35 4.477 .073
e29 <--> e34 6.494 -.079
e29 <--> e32 4.925 .081
e29 <--> e30 11.659 .100
e28 <--> e35 4.081 -.055
e28 <--> e34 5.926 .060
e28 <--> e30 5.127 -.052
e27 <--> e31 4.466 .050

143
 M.I. Par Change
e26 <--> e39 4.909 .080
e25 <--> e36 7.842 -.072
e25 <--> e35 9.117 -.090
e25 <--> e34 4.415 .056
e25 <--> e31 4.890 .061
e25 <--> e26 11.253 .086
e24 <--> 6 4.609 .087
e24 <--> e37 6.338 .086
e24 <--> e35 6.045 .108
e24 <--> e29 7.782 .112
e23 <--> 1 4.156 -.059
e23 <--> e36 5.704 .078
e23 <--> e34 6.000 -.083
e23 <--> e26 5.514 -.076
e22 <--> e24 4.606 .094
e22 <--> e23 13.920 .141
e21 <--> e23 8.422 -.089
e20 <--> e30 6.828 .064
e20 <--> e25 5.792 .054
e20 <--> e24 10.201 -.105
e20 <--> e23 8.183 -.081
e20 <--> e22 4.779 -.062
e20 <--> e21 12.124 .080
e19 <--> e35 4.323 .074
e19 <--> e31 5.916 -.079
e19 <--> e27 4.233 -.050
e19 <--> e25 12.978 -.100
e19 <--> e23 11.168 .118
e18 <--> e36 4.917 -.062
e18 <--> e35 4.438 .068
e18 <--> e26 6.263 -.069
e18 <--> e25 5.671 -.060
e18 <--> e24 6.138 -.093
e18 <--> e22 5.775 -.078
e18 <--> e19 10.189 .096
e17 <--> e39 9.799 .134

144
 M.I. Par Change
e17 <--> e38 7.782 .110
e17 <--> e36 4.942 -.074
e17 <--> e34 5.549 .081
e17 <--> e26 4.475 .070
e17 <--> e23 7.422 -.104
e17 <--> e20 5.177 .066
e16 <--> 4 6.394 .088
e16 <--> e39 11.773 .147
e16 <--> e38 9.885 .125
e16 <--> e37 5.627 -.071
e16 <--> e26 5.089 .075
e16 <--> e23 8.210 -.110
e16 <--> e20 5.607 .069
e16 <--> e17 103.655 .397
e15 <--> e39 15.205 -.147
e15 <--> e36 9.651 .091
e14 <--> 4 4.384 .065
e13 <--> e34 4.849 .070
e13 <--> e26 4.399 .064
e13 <--> e14 27.545 .169
e12 <--> 4 9.148 -.098
e12 <--> e28 5.251 -.059
e12 <--> e17 18.421 -.155
e12 <--> e16 14.768 -.139
e12 <--> e14 6.510 -.083
e11 <--> 4 6.417 -.086
e11 <--> 2 6.299 .072
e11 <--> e26 6.271 -.081
e11 <--> e23 6.242 .094
e11 <--> e19 4.429 .074
e11 <--> e17 11.070 -.127
e11 <--> e16 21.987 -.180
e11 <--> e14 9.172 -.104
e11 <--> e13 5.197 -.080
e11 <--> e12 66.408 .289
e10 <--> 6 6.576 .096

145
 M.I. Par Change
e10 <--> e38 9.004 -.126
e10 <--> e37 4.038 .064
e10 <--> e36 7.023 .094
e10 <--> e33 6.382 -.096
e10 <--> e26 5.502 -.083
e10 <--> e17 19.954 -.185
e10 <--> e16 13.219 -.151
e10 <--> e14 7.600 -.102
e10 <--> e13 7.976 -.108
e10 <--> e12 17.930 .163
e10 <--> e11 32.895 .233
e9 <--> 4 6.303 -.087
e9 <--> 2 4.573 .063
e9 <--> e35 12.080 .133
e9 <--> e27 4.546 -.056
e9 <--> e24 10.019 .141
e8 <--> 2 27.263 .178
e8 <--> e38 5.765 .109
e8 <--> e33 4.948 -.091
e8 <--> e21 7.534 -.100
e8 <--> e15 6.435 .100
e8 <--> e14 7.650 .111
e8 <--> e9 42.344 .290
e7 <--> 2 4.189 .068
e7 <--> e38 5.512 .104
e7 <--> e19 4.274 .083
e7 <--> e9 23.889 .213
e7 <--> e8 52.734 .366
e6 <--> e24 13.610 -.126
e6 <--> e17 8.978 .089
e6 <--> e8 6.381 -.086
e5 <--> 5 7.485 .086
e5 <--> e39 4.235 -.091
e5 <--> e36 7.000 .091
e5 <--> e35 4.021 .080
e5 <--> e32 4.916 .093

146
 M.I. Par Change
e5 <--> e25 12.458 -.110
e5 <--> e19 17.908 .157
e5 <--> e6 7.431 -.084
e4 <--> e8 11.637 -.122
e4 <--> e7 5.514 -.082
e3 <--> e39 8.578 .097
e3 <--> e37 6.320 -.059
e3 <--> e35 20.197 -.134
e3 <--> e34 6.984 .071
e3 <--> e32 10.080 -.100
e3 <--> e29 6.374 -.069
e3 <--> e27 5.065 .046
e3 <--> e25 4.645 .051
e3 <--> e19 5.264 -.064
e3 <--> e17 7.250 .081
e3 <--> e15 5.025 -.060
e3 <--> e11 4.885 -.066
e3 <--> e9 7.851 -.084
e3 <--> e7 8.404 -.098
e3 <--> e6 12.784 .082
e3 <--> e5 4.808 -.068
e2 <--> e38 5.730 -.090
e2 <--> e36 4.026 .063
e2 <--> e32 7.378 .104
e2 <--> e25 6.821 -.075
e2 <--> e19 14.318 .129
e2 <--> e17 7.802 -.103
e2 <--> e15 4.652 .070
e2 <--> e8 4.116 .086
e2 <--> e6 17.900 -.119
e2 <--> e5 86.222 .354
e2 <--> e3 4.847 -.063
e1 <--> 4 6.482 .066
e1 <--> 2 4.666 -.048
e1 <--> e39 6.729 .083
e1 <--> e32 6.050 -.075

147
 M.I. Par Change
e1 <--> e27 5.038 .044
e1 <--> e19 5.816 -.065
e1 <--> e15 4.694 -.056
e1 <--> e9 10.563 -.094
e1 <--> e8 24.978 -.168
e1 <--> e7 11.083 -.109
e1 <--> e6 17.309 .092
e1 <--> e5 11.600 -.103
e1 <--> e4 5.322 .053
e1 <--> e3 22.800 .107
e1 <--> e2 5.660 -.066

Correlation: (Group number 1 – Default model)

Estimate
HP <--> CNU .427
HP <--> CSC -.160
HP <--> OSL .423
HP <--> IB .605
HP <--> NFC .586
CNU <--> CSC .145
CNU <--> OSL .403
CNU <--> IB .463
CNU <--> NFC .603
CSC <--> OSL .082
CSC <--> IB -.278
CSC <--> NFC -.079
OSL <--> IB .379
OSL <--> NFC .262
IB <--> NFC .522
e2 <--> e5 .543
e7 <--> e8 .435
e8 <--> e9 .392
e11 <--> e12 .418
e10 <--> e11 .233

148
 Estimate
e16 <--> e17 .566
e31 <--> e32 .379
e1 <--> e8 -.143
e1 <--> e3 .225
e7 <--> e9 .325
e11 <--> e16 -.083
e13 <--> e14 .301

Squared Multiple Correlations: (Group number 1- Default model)

Estimate
NFC4 .339
NFC3 .515
NFC2 .793
NFC1 .697
IB5 .443
IB4 .589
IB3 .633
IB2 .407
IB1 .474
OSL4 .375
OSL3 .404
OSL2 .664
OSL1 .772
CSC9 .377
CSC8 .497
CSC7 .343
CSC6 .488
CSC5 .398
CSC4 .578
CSC3 .634
CSC2 .446
CSC1 .532
CNU8 .353
CNU7 .445
CNU6 .662

149
 Estimate
CNU5 .431
CNU4 .595
CNU3 .552
CNU2 .517
CNU1 .438
HP9 .508
HP8 .314
HP7 .373
HP6 .707
HP5 .453
HP4 .774
HP3 .698
HP2 .530
HP1 .679

150
 APPENDIX E
SEM RESULTS
Covariances (Group number 1 – Default model) (First Round)
M.I. Par Change
OSL <--> NFC 12.532 .162
CNU <--> NFC 78.546 .386
CNU <--> OSL 35.861 .231
CNU <--> CSC 5.072 .076
HP <--> NFC 79.639 .414
HP <--> OSL 43.446 .270
HP <--> CSC 6.361 -.091
HP <--> CNU 42.798 .256
e39 <--> OSL 15.709 .176
e39 <--> CNU 6.796 .111
e39 <--> HP 17.958 .191
e38 <--> OSL 7.556 .113
e38 <--> CNU 4.570 .084
e38 <--> HP 10.542 .135
e38 <--> e39 15.657 .178
e37 <--> OSL 8.846 -.088
e36 <--> OSL 6.108 .084
e36 <--> CNU 16.454 .131
e36 <--> HP 4.312 .072
e35 <--> e36 6.143 .080
e34 <--> e39 7.056 .098
e34 <--> e38 4.420 .072
e32 <--> CSC 5.103 -.078
e32 <--> HP 5.511 -.094
e32 <--> e35 4.872 .082
e31 <--> OSL 4.382 .072
e31 <--> e36 4.815 -.063
e30 <--> NFC 4.757 .083
e30 <--> CSC 4.145 .060

151
 M.I. Par Change
e30 <--> HP 7.730 .095
e30 <--> e39 4.647 .080
e29 <--> e35 7.931 .096
e29 <--> e34 6.331 -.075
e29 <--> e30 14.047 .112
e28 <--> e35 5.674 -.064
e28 <--> e34 8.766 .070
e27 <--> e37 5.008 -.044
e27 <--> e36 5.319 .052
e27 <--> e31 4.792 .050
e26 <--> OSL 10.140 .107
e26 <--> HP 6.548 .087
e26 <--> e39 5.846 .090
e25 <--> e36 7.254 -.070
e25 <--> e35 12.316 -.102
e25 <--> e34 7.467 .070
e25 <--> e32 5.865 -.072
e25 <--> e31 8.143 .074
e25 <--> e26 10.908 .084
e24 <--> NFC 7.887 .144
e24 <--> CNU 5.272 .100
e24 <--> e37 6.222 .083
e24 <--> e35 6.198 .107
e24 <--> e29 8.027 .115
e23 <--> NFC 4.327 -.092
e23 <--> HP 8.398 -.115
e23 <--> e36 5.320 .076
e23 <--> e34 5.299 -.075
e23 <--> e26 5.035 -.073
e22 <--> e24 4.700 .096
e22 <--> e23 14.580 .145
e21 <--> e29 4.148 -.058
e21 <--> e23 7.683 -.086
e20 <--> e37 5.308 -.050
e20 <--> e30 6.229 .061
e20 <--> e25 5.383 .052

152
 M.I. Par Change
e20 <--> e24 10.029 -.105
e20 <--> e23 7.063 -.076
e20 <--> e22 4.796 -.063
e20 <--> e21 11.724 .079
e19 <--> e35 4.463 .073
e19 <--> e31 4.663 -.067
e19 <--> e25 13.882 -.103
e19 <--> e23 11.652 .121
e18 <--> e36 4.269 -.058
e18 <--> e26 6.312 -.070
e18 <--> e25 5.989 -.062
e18 <--> e24 5.927 -.091
e18 <--> e22 5.682 -.077
e18 <--> e19 10.073 .096
e17 <--> e34 7.862 .076
e16 <--> OSL 5.409 .076
e16 <--> e39 4.933 .080
e16 <--> e38 5.342 .077
e15 <--> e39 13.018 -.139
e15 <--> e36 7.366 .080
e14 <--> OSL 7.489 .093
e13 <--> NFC 4.679 .087
e13 <--> e39 5.003 .088
e13 <--> e35 4.118 -.068
e13 <--> e34 7.151 .079
e12 <--> NFC 5.120 .087
e12 <--> e28 6.267 -.060
e12 <--> e15 5.951 -.070
e11 <--> e19 8.184 .086
e10 <--> NFC 11.805 .160
e10 <--> e38 7.894 -.117
e10 <--> e36 5.422 .081
e10 <--> e33 5.370 -.085
e10 <--> e12 17.241 .144
e9 <--> e35 11.411 .116
e9 <--> e24 10.909 .134

153
 M.I. Par Change
e8 <--> CNU 13.398 .138
e8 <--> e33 6.653 -.090
e8 <--> e32 4.938 .085
e7 <--> CSC 4.728 -.077
e7 <--> e19 5.336 .084
e6 <--> OSL 4.505 .063
e6 <--> e24 11.868 -.115
e6 <--> e17 5.690 .057
e5 <--> K 5.536 .064
e5 <--> e25 5.894 -.065
e5 <--> e19 6.523 .082
e4 <--> e9 11.698 .096
e4 <--> e8 5.780 -.073
e3 <--> e39 5.439 .075
e3 <--> e37 6.653 -.055
e3 <--> e35 20.015 -.123
e3 <--> e34 11.017 .080
e3 <--> e32 6.642 -.073
e3 <--> e17 7.595 .065
e3 <--> e11 8.597 -.070
e3 <--> e8 4.130 .058
e2 <--> e32 4.227 .065
e2 <--> e15 4.467 .060
e2 <--> e8 7.655 .088
e2 <--> e6 7.181 -.064
e1 <--> OSL 5.644 .066
e1 <--> e6 5.109 .045

Covariances: (Group number 1 – Default model) (Second Round)

Estimate S.E. C.R. P Label
HP <--> NFC .447 .059 7.528 ***
CNU <--> NFC .423 .061 6.977 ***
OSL <--> NFC .188 .049 3.858 ***
CNU <--> OSL .236 .044 5.353 ***
HP <--> OSL .270 .045 5.964 ***

154
 Estimate S.E. C.R. P Label
HP <--> CNU .265 .047 5.646 ***
e2 <--> e5 .380 .050 7.642 ***
e7 <--> e8 .402 .061 6.636 ***
e8 <--> e9 .313 .053 5.914 ***
e11 <--> e12 .246 .044 5.640 ***
e10 <--> e11 .155 .041 3.836 ***
e16 <--> e17 .382 .049 7.714 ***
e31 <--> e32 .196 .050 3.922 ***
e1 <--> e8 -.076 .028 -2.731 .006
e1 <--> e3 .074 .025 2.916 .004
e7 <--> e9 .257 .051 5.035 ***
e11 <--> e16 -.053 .028 -1.926 .054
e13 <--> e14 .159 .039 4.079 ***
e34 <--> e35 .152 .041 3.690 ***
e33 <--> e34 .085 .041 2.059 .039

Covariances: (Group number 1 – Default model) (Final Round)

Estimate S.E. C.R. P Label
CNU <--> OSL .232 .044 5.343 ***
HP <--> OSL .270 .045 5.970 ***
HP <--> CNU .260 .046 5.614 ***
e2 <--> e5 .388 .050 7.703 ***
e7 <--> e8 .403 .061 6.634 ***
e8 <--> e9 .311 .053 5.879 ***
e11 <--> e12 .244 .044 5.548 ***
e10 <--> e11 .159 .041 3.878 ***
e16 <--> e17 .376 .049 7.608 ***
e31 <--> e32 .194 .050 3.871 ***
e1 <--> e8 -.078 .028 -2.783 .005
e1 <--> e3 .071 .026 2.794 .005
e7 <--> e9 .260 .051 5.061 ***
e11 <--> e16 -.057 .028 -2.047 .041
e13 <--> e14 .151 .039 3.864 ***
e34 <--> e35 .155 .041 3.750 ***

155
 Estimate S.E. C.R. P Label
e33 <--> e34 .084 .041 2.021 .043

Standardized Regression Weights of SEM testing:

(Group number 1 - Default model)
Estimate
IB <--- HP .385
IB <--- OSL .131
IB <--- CSC -.295
IB <--- CNU .281
HP1 <--- HP .824
HP2 <--- HP .721
HP3 <--- HP .839
HP4 <--- HP .879
HP5 <--- HP .666
HP6 <--- HP .845
HP7 <--- HP .605
HP8 <--- HP .559
HP9 <--- HP .713
CNU1 <--- CNU .651
CNU2 <--- CNU .724
CNU3 <--- CNU .739
CNU4 <--- CNU .774
CNU5 <--- CNU .664
CNU6 <--- CNU .816
CNU7 <--- CNU .667
CNU8 <--- CNU .595
CSC1 <--- CSC .729
CSC2 <--- CSC .669
CSC3 <--- CSC .797
CSC4 <--- CSC .761
CSC5 <--- CSC .631
CSC6 <--- CSC .693
CSC7 <--- CSC .584

156
 Estimate
CSC8 <--- CSC .707
CSC9 <--- CSC .615
OSL1 <--- OSL .879
OSL2 <--- OSL .814
OSL3 <--- OSL .638
OSL4 <--- OSL .611
IB1 <--- IB .721
IB2 <--- IB .669
IB3 <--- IB .791
IB4 <--- IB .685
IB5 <--- IB .620

157