Factors Influencing the Consumers' Behavior Intention to Use Mobile

Authors Tran Kim Thuy Pailin Kunnawat

Submitted to Institute of Retailing, Sales and Marketing

Thesis Supervisors FACTORS INFLUENCING THE Assoz.Univ.-Prof.inDr.in CONSUMERS’ BEHAVIOR Katharina Hofer

INTENTION TO USE MOBILE December 2020 PAYMENT: SCOPE IN VIETNAM AND THAILAND.

Master’s Thesis to confer the academic degree of Master of Science in the Master’s Program Management

JOHANNES KEPLER UNIVERSITÄT LINZ Altenberger Straße 69 4040 Linz, Österreich jku.at

SWORN DECLARATION

I hereby declare under oath that the submitted Master’s Thesis has been written solely by Pailin Kunnawat and Tran Kim Thuy without any third-party assistance, information other than provided sources or aids have not been used and those used have been fully documented. Sources for literal, paraphrased and cited quotes have been accurately credited.

The submitted document here present is identical to the electronically submitted text document.

Place, Date Linz, 16.12.20 Signature

(Pailin Kunnawat)

(Tran Kim Thuy)

Table of Contents

A. Lists of Table ...... 6 B. Lists of Figure ...... 10 C Lists of Abbreviation ...... 11 D Distribution in Pair Thesis ...... 13 E Executive Summary ...... 14 1. Introduction ...... 16 2. Problem Statement ...... 17 3. Objectives ...... 18 4. Research questions ...... 19 5. Literature review ...... 19 Conceptual foundations ...... 20 Mobile payment ...... 20 Internet payment method ...... 20 Electronic payment (E-payment) ...... 21 Mobile commerce (M-Commerce) ...... 23 Development of mobile payment and its challenges ...... 25 Mobile payment in an international scope ...... 27 Mobile payment in an emerging market ...... 28 Mobile payment in Vietnam ...... 31 Mobile payment in Thailand ...... 33 Mobile payment systems and its threats ...... 36 Mobile payment as marketing strategy ...... 39 Covid – 19 crises ...... 40 Theoretical foundations ...... 41 Basics of technology acceptance research ...... 41 Technology Adoption research ...... 41 Technology Diffusion research ...... 47 Technology Acceptance research ...... 49 Development of Technology Acceptance Models ...... 53 Innovation Diffusion Theory (IDT) ...... 55 Theory of Reasoned Action (TRA) ...... 57 Theory of Planned Behavior (TPB)...... 58 Decomposed Theory of Planned Behavior (DTPB) ...... 59 Technology Acceptance Model (TAM) ...... 60

Combined TAM and TPB (C-TAM-TPB) ...... 62 Technology Acceptance Model 2 (TAM2) ...... 63 Technology Acceptance Model 3 (TAM3) ...... 65 Unified Theory of Acceptance and Use of Technology (UTAUT) ...... 67 Unified Theory of Acceptance and Use of Technology 2 (UTAUT2) ... 69 Discussion ...... 70 Research gaps in mobile payment research ...... 72 Research gaps in general scope ...... 72 Findings & Research gaps: Vietnam ...... 75 Findings & Research Gaps: Thailand ...... 77 Summarization table of research gaps between Vietnam and Thailand ...... 80 5.4. Research Model ...... 85 Hypotheses ...... 85 Variable definitions ...... 86 Behavioral intention ...... 86 Attitude ...... 86 Perceived usefulness ...... 87 Perceived ease of use ...... 88 Subjective norm ...... 89 Perceived cost ...... 90 Perceived risk ...... 91 Covid-19 ...... 91 6. Research Methodology ...... 92 Research process ...... 93 Measurement scale ...... 93 Research sample ...... 96 Target population...... 96 Target sample ...... 96 Regions...... 96 Gender and Age group ...... 97 Sample size ...... 98 Sampling technique ...... 99 7. Empirical findings ...... 102 Samples and demographic data ...... 102 Handling of missing values ...... 111

Descriptive Analysis ...... 112 Combination of two countries dataset ...... 113 Vietnam ...... 113 Thailand ...... 114 Reliability Test ...... 116 Cronbach’s Alpha Test ...... 116 Combination of two countries dataset ...... 116 Vietnam...... 121 Thailand ...... 125 Exploratory Factor Analysis (EFA) ...... 130 Combination of two countries dataset ...... 131 Vietnam...... 133 Thailand ...... 136 Average Variance Extracted (AVE) and Composite Reliability (CR)...... 139 Combination of two countries dataset ...... 139 Vietnam...... 140 Thailand ...... 142 Summarization of AVE and CR between all datasets ...... 143 Discriminant Validity Test ...... 143 Combination of two countries dataset ...... 144 Vietnam ...... 145 Thailand ...... 145 Summarization of discriminant validity tests between all datasets ...... 146 Normality Test ...... 147 Confirmatory Factor Analysis (CFA) ...... 147 Combination of two countries dataset ...... 148 Vietnam ...... 151 Thailand ...... 155 Summarization of CFA between all datasets ...... 158 Multicollinearity Test ...... 159 Combination of two countries dataset ...... 160 Vietnam ...... 161 Thailand ...... 161 Summarization of multicollinearity tests between all datasets ...... 162 Regression analysis with control variables ...... 162

Combination of two countries dataset ...... 165 Education level...... 165 Age group ...... 168 Vietnam ...... 171 Education level...... 171 Age group ...... 174 Thailand ...... 178 Education level...... 178 Age group ...... 181 Summarization of regression analysis for control variables between all datasets ...... 185 Multiple Regression Analysis ...... 185 Combination of two countries dataset ...... 185 Vietnam ...... 187 Thailand ...... 189 Summarization of multiple regression analysis between all datasets ...... 191 8. Discussions ...... 191 Discussions based on the factors of research model ...... 191 Discussions between countries ...... 191 Discussions in the context of Vietnam ...... 192 Discussions in the context of Thailand ...... 193 9. Conclusion ...... 194 Theoretical implications ...... 194 Theoretical implications in general ...... 194 Theoretical implications for Vietnam ...... 195 Theoretical implications for Thailand...... 196 Practical implications ...... 198 10. Limitations ...... 199 11. Direction for future study ...... 200 12. References ...... 201 13. Appendix: Questionaire...... 216

A. Lists of Table

Table 1: Theories and constructs used in recent mobile payment research. (Source: self-edited) ...... 45 Table 2: Summarization of Technology Acceptance and Adoption Models and Theories (Source: Self- Edited) ...... 55 Table 3: Research gaps between Vietnam and Thailand from previous findings.Source: Self-edited ...... 85 Table 4: Measurement scale for Behavioral Intention. Source: Lu, Y., Yang, S., Chau, P. Y.K., & Cao, Y. 2011, Adapted from Taylor, S. et al., 1995...... 94 Table 5: Measurement scale for Attitude. Source: Luna, I. et al. 2019, Adapted from Davis, 1989; Kim et al., 2010; Yang & Yoo 2004...... 94 Table 6: Measurement scale for Perceived Usefulness. Source: Shankar, A., & Datta, B. 2018, Adapted from Davis, 1989...... 94 Table 7: Measurement scale for Perceived Ease of Use: Source: Shankar, A., & Datta, B. 2018, Adapted from Venkatesh, V. 2003...... 94 Table 8: Measurement scale for Subjective Norm. Source: Shankar, A., & Datta, B. 2018, Adapted from De Sena Abrahão et al., 2016; Luarn et al, 2005; Wei et al., 2009...... 95 Table 9: Measurement scale for Perceived Cost. Source: Lu, Y., Yang, S., Chau, P. Y.K., & Cao, Y. 2011, Adapted from Featherman et al., 2003; Luarn and Lin 2005, Wei et al 2009...... 95 Table 10: Measurement scale for Perceived Risk: Source Lu, Y., Yang, S., Chau, P. Y.K., & Cao, Y. 2011, Adapted from Venkatesh, V. 2003...... 95 Table 11: Measurement scale for the impact of Covid-19. Adapted from Girish et al., 2020 ...... 95 Table 12: Adapted from Bhat (2020) ...... 100 Table 13: Table of Quota Plans: Mobile Payment Users in Vietnam ...... 101 Table 14: Table of Quota Plans: Mobile Payment Users in Thailand ...... 102 Table 15: Summarization Table of demographic data between Vietnam and Thailand ...... 103 Table 16: Descriptive Statistics of respondents from Thailand and Vietnam ...... 113 Table 17: Descriptive Statistics of respondents from Vietnam ...... 114 Table 18: Descriptive Statistics of respondents from Thailand ...... 115 Table 19: Reliability Statistics of Behavior Intention in Thailand and Vietnam ...... 116 Table 20: Item-Total Statistics of Behavior Intention in Thailand and Vietnam ...... 117 Table 21: Reliability Statistics of Attitude in Thailand and Vietnam ...... 117 Table 22: Item-Total Statistics of Attitude in Thailand and Vietnam ...... 117 Table 23: Reliability Statistics of Perceived Usefulness in Thailand and Vietnam ...... 118 Table 24: Item-Total Statistics of Perceived Usefulness in Thailand and Vietnam ...... 118 Table 25: Reliability Statistics of Perceived Ease of Use in Thailand and Vietnam ...... 118 Table 26: Item-Total Statistics of Perceived Ease of Use in Thailand and Vietnam ...... 118 Table 27: Reliability Statistics of Subjective Norm in Thailand and Vietnam ...... 119 Table 28: Item-Total Statistics Subjective Norm in Thailand and Vietnam ...... 119 Table 29: Reliability Statistics of Perceived Cost in Thailand and Vietnam ...... 119 Table 30: Item-Total Statistics of Perceived Cost in Thailand and Vietnam ...... 119 Table 31: Reliability Statistics of Perceived Risk in Thailand and Vietnam ...... 120 Table 32: Item-Total Statistics of Perceived Risk in Thailand and Vietnam ...... 120 Table 33: Reliability Statistics of Covid-19 in Thailand and Vietnam ...... 120 Table 34: Item-Total Statistics of Covid-19 in Thailand and Vietnam ...... 121 Table 35: Reliability Statistics of Behavior Intention in Vietnam ...... 121 Table 36: Item-Total Statistics of Behavior Intention in Vietnam ...... 121

Table 37: Reliability Statistics of Attitude in Vietnam ...... 121 Table 38: Item-Total Statistics of Attitude in Vietnam ...... 122 Table 39: Reliability Statistics of Perceived Usefulness in Vietnam ...... 122 Table 40: Item-Total Statistics of Perceived Usefulness in Vietnam ...... 122 Table 41: Reliability Statistics of Perceived Ease of Use in Vietnam...... 123 Table 42: Item-Total Statistics of Perceived Ease of Use in Vietnam ...... 123 Table 43: Reliability Statistics of Subjective Norm in Vietnam ...... 123 Table 44: Item-Total Statistics of Subjective Norm in Vietnam ...... 123 Table 45: Reliability Statistics of Perceived Cost in Vietnam ...... 124 Table 46: Reliability Statistics of Perceived Cost in Vietnam ...... 124 Table 47: Reliability Statistics of Perceived Risk in Vietnam ...... 124 Table 48: Item-Total Statistics of Perceived Risk in Vietnam ...... 124 Table 49: Reliability Statistics of Covid-19 in Vietnam ...... 125 Table 50: Item-Total Statistics of Covid-19 in Vietnam ...... 125 Table 51: Reliability Statistics of Behavioral Intention in Thailand ...... 125 Table 52: Item-Total Statistics of Behavioral Intention in Thailand ...... 125 Table 53: Reliability Statistics of Attitude in Thailand ...... 126 Table 54: Item-Total Statistics of Attitude in Thailand ...... 126 Table 55: Reliability Statistics of Perceived Usefulness in Thailand ...... 126 Table 56: Item-Total Statistics of Perceived Usefulness in Thailand ...... 127 Table 57: Reliability Statistics of Perceived Ease of Use in Thailand ...... 127 Table 58: Item-Total Statistics of Perceived Ease of Use in Thailand ...... 127 Table 59: Reliability Statistics of Subjective Norm in Thailand ...... 128 Table 60: Item-Total Statistics of Subjective Norm in Thailand ...... 128 Table 61: Reliability Statistics of Perceived Cost in Thailand ...... 128 Table 62: Item-Total Statistics of Perceived Cost in Thailand ...... 128 Table 63: Reliability Statistics of Perceived Risk in Thailand ...... 129 Table 64: Item-Total Statistics of Perceived Risk in Thailand ...... 129 Table 65: Reliability Statistics of Covid-19 in Thailand ...... 129 Table 66: Item-Total Statistics of Covid-19 in Thailand ...... 129 Table 67: KMO and Bartlett's Test for Thailand and Vietnam dataset ...... 131 Table 68: Extraction Method: Principal Component Analysis of Thailand and Vietnam dataset ...... 132 Table 69: Rotated Component Matrix of Thailand and Vietnam dataset ...... 133 Table 70: KMO and Bartlett's Test for Vietnam dataset ...... 133 Table 71: Total Variance Explained of Vietnam dataset ...... 134 Table 72: Rotated Component Matrix for Vietnam dataset ...... 135 Table 73: KMO and Bartlett's Test for Thailand dataset ...... 136 Table 74: Total Variance Explained for Thailand dataset ...... 137 Table 75: Rotated Component Matrix for Thailand dataset ...... 139 Table 76: Summarization of Standardized loading, Cronbach’s α, AVE and CR – analysis of Thailand and Vietnam. Source: Self-calculated ...... 140 Table 77: Summarization of Standardized loading, Cronbach’s α, AVE and CR – analysis of Vietnam. Source: Self-calculated ...... 142 Table 78: Summarization of Standardized loading, Cronbach’s α, AVE and CR – analysis of Thailand. Source: Self-calculated ...... 143 Table 79: Factor AVE- Correlation Matrix for Thailand and Vietnam dataset. Source: Self-edited ...... 144 Table 80: Factor AVE- Correlation Matrix for Vietnam dataset. Source: Self-Edited ...... 145

Table 81: Factor AVE- Correlation Matrix for Thailand dataset. Source: Self-Edited ...... 146 Table 82: Model Fit Summary for Thailand and Vietnam dataset ...... 150 Table 83: Model Fit Summary for Vietnam dataset ...... 153 Table 84: Model Fit Summary for Thailand dataset ...... 157 Table 85: Multicollinearity Test result for Thailand and Vietnam dataset...... 161 Table 86: Multicollinearity Test result for Vietnam dataset ...... 161 Table 87: Multicollinearity Test result for Thailand dataset ...... 162 Table 88: Regression analysis about education level for Thailand and Vietnam: Behavioral Intention .... 165 Table 89: Regression analysis about education level for Thailand and Vietnam: Attitude ...... 165 Table 90: Regression analysis about education level for Thailand and Vietnam: Perceived Usefulness . 166 Table 91: Regression analysis about education level for Thailand and Vietnam: Perceived Ease Of Use ...... 166 Table 92: Regression analysis about education level for Thailand and Vietnam: Subjective Norm ...... 167 Table 93: Regression analysis about education level for Thailand and Vietnam: Perceived Risk ...... 167 Table 94: Regression analysis about education level for Thailand and Vietnam: Covid-19 ...... 168 Table 95: Regression analysis about age group for Thailand and Vietnam: Behavioral Intention ...... 168 Table 96: Regression analysis about age group for Thailand and Vietnam: Attitude ...... 169 Table 97: Regression analysis about age group for Thailand and Vietnam: Perceived Usefulness ...... 169 Table 98: Regression analysis about age group for Thailand and Vietnam: Perceived Ease Of Use ...... 170 Table 99: Regression analysis about age group for Thailand and Vietnam: Subjective Norm ...... 170 Table 100: Regression analysis about age group for Thailand and Vietnam: Perceived Risk ...... 171 Table 101: Regression analysis about age group for Thailand and Vietnam: Covid19 ...... 171 Table 102: Regression analysis about education level for Vietnam: Behavioral Intention ...... 172 Table 103: Regression analysis about education level for Vietnam: Attitude ...... 172 Table 104: Regression analysis about education level for Vietnam: Perceived Usefulness ...... 172 Table 105: Regression analysis about education level for Vietnam: Perceived Ease of Use ...... 173 Table 106: Regression analysis about education level for Vietnam: Subjective Norm ...... 173 Table 107: Regression analysis about education level for Vietnam: Perceived Risk ...... 174 Table 108: Regression analysis about education level for Vietnam: Covid-19 ...... 174 Table 109: Regression analysis about age group for Vietnam: Behavioral Intention ...... 175 Table 110: Regression analysis about age group for Vietnam: Attitude ...... 175 Table 111: Regression analysis about age group for Vietnam: Perceived Usefulness ...... 176 Table 112: Regression analysis about age group for Vietnam: Perceived Ease of Use ...... 176 Table 113: Regression analysis about age group for Vietnam: Subjective Norm ...... 177 Table 114: Regression analysis about age group for Vietnam: Perceived Risk ...... 177 Table 115: Regression analysis about age group for Vietnam: Covid-19 ...... 178 Table 116: Regression analysis about education level for Thailand: Behavioral Intention ...... 178 Table 117: Regression analysis about education level for Thailand: Attitude ...... 179 Table 118: Regression analysis about education level for Thailand: Perceived Usefulness ...... 179 Table 119: Regression analysis about education level for Thailand: Perceived Ease Of Use ...... 180 Table 120: Regression analysis about education level for Thailand: Subjective Norm ...... 180 Table 121: Regression analysis about education level for Thailand: Perceived Risk ...... 180 Table 122: Regression analysis about education level for Thailand: Covid-19 ...... 181 Table 123: Regression analysis about age group for Thailand: Behavioral Intention ...... 182 Table 124: Regression analysis about age group for Thailand: Attitude ...... 182 Table 125: Regression analysis about age group for Thailand: Perceived Usefulness ...... 183 Table 126: Regression analysis about age group for Thailand: Perceived Ease Of Use ...... 183

Table 127: Regression analysis about age group for Thailand: Subjective Norm ...... 184 Table 128: Regression analysis about age group for Thailand: Perceived Risk ...... 184 Table 129: Regression analysis about age group for Thailand: Covid-19 ...... 185 Table 130: Model Summary for dataset of Thailand and Vietnam ...... 186 Table 131: ANOVA test result for dataset of Thailand and Vietnam ...... 187 Table 132: Coefficients result for dataset of Thailand and Vietnam ...... 187 Table 133: Model Summary for dataset of Vietnam ...... 187 Table 134: ANOVA test result for dataset of Vietnam ...... 188 Table 135: Coefficients result for dataset of Vietnam ...... 189 Table 136: Model Summary for dataset of Thailand ...... 190 Table 137: ANOVA test result for dataset of Thailand ...... 190 Table 138: Coefficients result for dataset of Thailand ...... 190

B. Lists of Figure

Figure 1: Framework of factors impacting the mobile payment services market (Dahlberg et al. 2007 pp. 3) ...... 26 Figure 2: Payments Trends (J.P. Morgan, 2019) ...... 35 Figure 3: Mobile Payment Process (Wang, Y. et al. 2016) ...... 37 Figure 4: Frequency of technology adoption in m-commerce (Chhonker, et al. 2017, pp. 893) ...... 44 Figure 5: Frequency of technology adoption frameworks constructs in mobile commerce (Chhonker, et al. 2017, pp. 893) ...... 44 Figure 6: Chronological Graph for the evolution of Technology Acceptance Theory, source: Momani, A. & Jamous, M. (2017)...... 54 Figure 7. Theory of Reasoned Action (TRA) Source: Ajzen & Fishbein (2010) edited by Otieno, O. 2016 58 Figure 8: Theory of Planned Behavior (TPB) Source: Ajzen, 1991 edited by Li, L. 2010 ...... 59 Figure 9: Decomposed Theory of Planned Behavior (Source: Taylor, S. &Todd, P. 1995a pp. 146, self- edited) ...... 60 Figure 10: Technology Acceptance Theory (TAM) Source: Davis (1989) ...... 61 Figure 11: Combined TAM and TPB (Source: Taylor, S. & Todd, P. 1995b, self-edited) ...... 62 Figure 12: TAM2 Source: Venkatesh and Davis (2000) pp. 188 ...... 64 Figure 13: TAM3 Source: Venkatesh, V. & Bala, H. 2008 pp. 280) ...... 66 Figure 14: Unified Theory of Acceptance and Use of Technology Model (UTAUT) source: Venkatesh et al. (2003). pp.447 ...... 67 Figure 15: UTAUT2 Source: Venkatesh, et al. 2012 pp. 160 ...... 70 Figure 16: Hofstede’s Cultural Dimension Index Comparison , retrieved from https://www.hofstede- insights.com/country-comparison/china,thailand,the-usa,vietnam/ ...... 79 Figure 17: Research Model. Source: Self-edited ...... 85 Figure 18: Model Fit figure for Thailand and Vietnam dataset. Source: Self-edited ...... 150 Figure 19: Model fit figure for Vietnam dataset. Source: Self-edited ...... 153 Figure 20: Model fit figure for Thailand dataset ...... 156 Figure 21: Evaluation model between two countries. Source: Self-edited ...... 191 Figure 22: Evaluation model of Vietnam. Source: Self-edited ...... 192 Figure 23: Evaluation model of Thailand. Source: Self-edited ...... 193

C. Lists of Abbreviation

Abbreviation Explanation AGFI Adjusted Goodness of Fit Index AT Attitude AU Actual Use AVE Average Variance Extracted B Behavior BI Behavioral Intention C Covid-19 factor CFA Confirmatory Factor Analysis CFI Comparative Fit Index COMP Compatibility CPX Complexity CR Composite Reliability CRC Computing Resource Center C-TAM-TPB Combined TAM and TPB dF Degrees of Freedom DTPB Decomposed Theory of Planned Behavior EE Effort Expectancy EFA Exploratory Factor Analysis FC Facilitating Conditions GDP Gross Domestic Product GOF Goodness of Fit IBPG Internet Banking Payment Gateway IDT/DOI Innovation of Diffusion Theory/ Diffusion of Innovations IM Image INN Innovativeness JR Job Relevance KMO The Kaiser Meyer Olkin test MAPS The Model of Acceptance with Peer Support MM The Motivational Model MNOs Mobile Network Operators MPCU The Model of PC Utilization MPSPs Mobile Payment Service Providers MSV Maximum Shared Variance NFC Near Field Communication NNFI or NFI Non-Normed Fit Index OB Observability OQ Output Quality OTC Over the Counter P.ENJ Perceived Enjoyment P2P Peer to Peer Payment PC Perceived Cost PDA Personal Digital Assistant 12

PEOU Perceived Ease of Use PLS-SEM Partial Least Squares Structural Equation Modelling POS Point of Sale PR Perceived Risk PU Perceived Usefulness RA Relative Advantage RMSEA The Root Mean Square Error of Approximation SCT The Social Cognitive Theory SEA Southeast Asia SEM Structural Equation Modelling SET Secure Electronic Transaction SI/SN Social Influence/Subjective Norm SRMR Standardized Root Mean Square Residual TAM Technology Acceptance Model TAM2 Technology Acceptance Model 2 TAM3 Technology Acceptance Model 3 TLI Tucker-Lewis Index TPB Theory of Planned Behavior TRA Theory of Reasoned Action TRL Trialability TTF Task-Technology Fit UTAUT Unified Theory of Acceptance and Use of Technology Model UTAUT2 Unified Theory of Acceptance and Use of Technology 2 X2 Model Chi Square

D. Distribution in Pair Thesis

The table below was the distribution of work in the pair thesis.

Miss Pailin Kunnawat Miss Tran Kim Thuy Both Lists of Abbreviation Table of content Executive summary Chapter 3 Lists of Table Chapter 5.4 Chapter 4 Lists of Figure Chapter 7, 7.1 and 7.2 Chapter 5: Formatting Chapter 9.2 5.1.4, 5.1.5, 5.1.9, 5.1.10, 5.2, Chapter 1 Chapter 12: References 5.3.1, 5.3.3, 5.3.4 Chapter 2 Chapter 6: Chapter 5: 6.3.2.1, 6.3.2.2 5.1.1, 5.1.2, 5.1.3, 5.1.6, 5.1.7, Chapter 7: 5.1.8, 5.1.11, 5.1.12, and 5.3.2 7.3.1, 7.3.3, Definition in 7.4 Chapter 6: ,7.4.2 ,7.4.2.3, Definition in 6.1, 6.2, 6.3, 6.3.1, 6.3.2, 7.4.3, 7.4.3.3, 7.4.3.4, 6.3.2.3 and 6.3.2,4 Definition in 7.5, 7.5.3, 7.5.4, Chapter 7: 7.6, Definition in 7.7, 7.7.3, 7.3.2, 7.4.1.1, 7.4.1.2. ,7.4.2.1, 7.7.4, Definition in 7.8, 7.8.3, 7.4.2.2, 7.4.3.1, 7.4.3.2, 7.5.1, 7.8.4, Definition in 7.9, 7.9.1, 7.5.2, 7.7.1, 7.7.2, 7.8.1, 7.8.2, 7.9.3, 7.9.4, Definition in 7.10, 7.9.2, 7.10.1 and 7.10.2 7.10.3, 7.10.4 Chapter 8: Chapter 8: 8.1, 8.1.2 8.1.1, 8.1.3 Chapter 9.1.2 Chapter 9: Chapter 10 9.1.1 ,9.1.3 Chapter 11 Proofreading

E. Executive Summary

This research paper aims to discover the significant factors that have influenced the behavioral intention to use mobile payment between Vietnam and Thailand. We have fulfilled the research gaps regarding to the current state of art in both countries. Therefore, we have used the quota samplings as our methodology for collecting our aimed samples. The online surveys were distributed across the target samples in both countries to both non-users and the current users of mobile payment. The target samplings were set to 200 respondents in each country, in total of 400 respondents are expected to be reached. Eventually, the actual number of respondents are 289 in Vietnam and 292 in Thailand. We have mainly used SPSS and other statistical methods to analyze our data and interpret the results.

As the result, we found that Attitude was the strongest predictor for the consumer intention to use mobile payment in both Vietnam and Thailand. Following, Perceived Usefulness also impact on this behavioral intention to use this service in both countries. On the theorical perspectives, we could once again confirm that Subject Norm – factor had many contrast ideas in the previous researches - had no influence to the behavioral intention to use mobile payment in both Vietnam and Thailand. Besides, the country and cultural context were also added in our research to fill the gap of studies. On the practical perspective, the management levels should develop the marketing strategies in two directions: one focused on retention and another for new users. For the loyal customer, the mobile payment companies could remain their retention by serving the good services, always enhancing application quality to satisfy them. For the new users, the marketers should focus more on above the line marketing activities to raise customer awareness about mobile payment usage by its convenience, usefulness and simplicity. Although this research has interpreted in details about behavioral intention of mobile payment’s current users, the non-user portion of mobile payment in Vietnam and Thailand were still large. Therefore, the future studies were encouraged to explore more about non-user by different research method approaches to understand much more about non-user insights then potentially switched this target into new user.

1. Introduction

The combination between payment methods and smartphones are generating the new opportunity which is mobile payment method (Sharma et al., 2018). By using the mobile wallet, the users can proceed the online payment via their mobile devices without physical presences as the normal wallet (Sharma et al., 2018). The awareness of mobile payment services has been becoming more popular due to its lower cost and conveniences which are not only benefit for the users but also for the enterprises (Nguyen et al., 2016). From 2016 to 2017, the world witnessed the highest volumes of 530 billion dollars non-cash transaction in the last two decades and the dominant of this growth is emerging Asian market which accounted for 32% globally. With this fast growth, emerging market was expected to contribute as the half of global non-cash transactions market in the future ("World Payments Report 2019", 2020). According to the Global Consumer Insights Survey in 2019 by PwC, Vietnam was the fastest growing country in mobile payment sector globally and Thailand was also ranked as the second most quick adapting to mobile payment country in emerging market ("Mobile payments in Vietnam fastest growing globally, Thailand emerges second in Southeast Asia", 2020).

It is known that emerging markets are the new potential and dynamic markets when it brings many attractive offers for the enterprises. In the household perspective, these markets are a main source of reasonable consumer goods. For the information technology sector, this is a good place for technical support outsourcing. With the multinational companies, emerging markets are the main growth drivers for developed countries during the amid stagnation and financial crisis (Fornes & Mendez, 2014).

The development of dynamic cashless revolution in such an attractive market like emerging market (Capgemini Research Institute, 2019) have promoted us to conduct the research about the mobile payment in this area; especially Vietnam and Thailand where the growth rate is highest. Due to the promising of these markets, it is crucial to investigate the most influential factors towards consumers intention which will maintain and stabilize this fast growth and development of mobile payment. The cross-country research also helps us considering different user behaviors since each market is unique and there are cultural factors involved, comparing to single-country samples where it will further weaken generalizability of findings (Zhang, Y. 2018). Hence our topic is diverse and interesting to see user behavior throughout different regions across Vietnam and Thailand.

2. Problem Statement

By going through some research papers regarding the factors influencing the consumer behavior, this topic of mobile payment caught our interests due to the fact as mentioned above about the extensive use of mobile phone and the new trends which has an obvious evidence to Asian regions about its increase of usage lately. The trends of using and how a user have decided to choose mobile payment could vary between countries and regions and under different factors. Therefore, we could use this chance to integrate current factors to the changes in culture and trends that has affected people in Thailand and Vietnam. In a normal consumer behavior theory, cultural trends, social trends, life style and motivation play a significant role to the consumer buying behavior (Yuvaraj, S. 2016). To be precise, each research contains different perspective of results, however, some of them are difficult to be compared and some of them are still lack in cultural context, the data is still not distributed equally to each region, lack of comparison to the other countries in the same region.

Moreover, there are many great factors in order to understand the behavioral intention to use mobile payment. There are many research papers referred about the factors influencing the usage of mobile payment; however, they have used different independent variables to determine what influence the use of mobile payment the most based on the theory they have chosen, some could be a mixture of many theories altogether. It is crucial that we will be using factors only what is necessary and know what is not relevant. The research results could help many firms to predict the behavior of customer and know who to target and what strategies to work on.

In addition, we would like to know if the social influence, perceptions and the attitude influence the behavioral intention of user in each country or not, if so, how strong of this influence would be in comparison to both countries between Vietnam and Thailand and from this point, we could compare and see the trends in each country in order to see the difference of internal factors and external factors and figure out why the results are different and why it’s similar and then develop practical solution for mobile payment firms or industries.

During the time we conducted this research, there has been a serious pandemic called Covid- 19. And the reason why we mention this pandemic here because it impacts not only human life but also the global business. This pandemic has infected millions of people worldwide and caused more than 400,000 deaths (WHO, 2020). The World Health Organization (WHO) estimates the case fatality rate to be around 2% (WHO, 2020), but Covid-19 's overall burden remains uncertain, and it is still unclear to state when and how regular economic and social life will be returned. According to Gupta et al. (2020), many countries usually declared their emergency in the crisis to convey a sense of urgency about the situation. Then, depends on the extent of situation, 17

government will apply the most optimal solution to avoid viral transmissions. And the most common solution is known as restricting physical contact between individuals. Due to social distancing and lockdown policies, we would like to test the influences of this crisis towards the usage of mobile payment method in Vietnam and Thailand. For example, mobile payment shares in Vietnam is still not significant. There is a possibility that the Vietnamese may change their cash- based habits into mobile payment usage during this pandemic to ensure the hygiene issues. Therefore, this research will largely contribute to mobile payment industry in general; especially Vietnam and Thailand when it tests about a new influential element such as Covid-19.

3. Objectives

Although the mobile payment is developing continuously, the cash circulation (CIC) has remained stable or slightly increased from 4% to 7% annually over the last five years ("World Payments Report 2019", 2020). It illustrated the market share for mobile payment can be extended continuously by persuading cash payment users change habits into mobile payment. This study will be conducted on both mobile payment services users and non-users to explore what are their preferred factors for using this payment method. As the result, the enterprises in Vietnam and Thailand can be aware of the most influential factors impact on the usage of mobile payment services and proceed it as the practical implementations on their marketing strategies and business tactics. From that, the research results will enable the enterprises in this sector enlarge their market shares and switch the cash payment users into mobile payment users.

Moreover, conducting surveys in two emerging countries like Vietnam and Thailand allow us to have an overview about the variances between these two markets. The study will indicate the most important factors towards to consumer intention of mobile payment services of each country. From that, we can conclude what are the differences and similarities between two markets. Also, it will be a qualified reference for international enterprises to learn about market insights before entering.

Hence, our main objectives are: • To examine factors that effect on the behavioral intention to use mobile payment between Vietnam and Thailand. • To find out which factors are similar and different on behavioral intention to use mobile payment between Vietnam and Thailand in order to develop suitable marketing strategies and approach to reach different groups of people.

4. Research questions

According to research objectives and problem statement, it’s essential to develop the research question which will enable to explore the most influential factors to the mobile payment usage in Vietnam and Thailand.

What are factors influencing consumer behavioral intention of mobile payment in Vietnam and Thailand?

We tend to use the applied theories such as theory of planned behavior, innovation diffusion theory, technology acceptance theory, and theory of reasoned action to determine our fundamental factors to the behavioral intention to use mobile payment. It is possible that we will find some extra relevant theories regarding to the topic we are doing the research. After the data analysis, we will then be able to see which factors are the most influential to the behavioral intention by using a statistical tool to assist us with that, therefore we can answer the next research question on the country basis about the differences and similarities between these factors.

We would like to compare the differences (if any) the crucial factors in usage of mobile payment between both countries to see why it’s different and why things are developed in a different way, possibly we need to refer to those internal and external factors to consumer buying behavior and also cultural context in each country. Additionally, the comparison between a market from Vietnam and Thailand should be proceeded to illustrate the crucial variables among these countries. As mentioned in the problem statement, we would like to see some relevant factors whether they have any significant influence on the behavioral intention between the two countries or not. Furthermore, we will also test those factors with the moderators such as age groups and education levels to see how this could be relevant to the behavioral intention.

5. Literature review

In the literature review, we have divided the main parts into conceptual foundations, theoretical foundations, research gaps, research models and its hypotheses. In the conceptual foundations, the main objective is to know the definition, the development of the mobile payment and how it can be used for the future challenges and marketing strategies. Furthermore, in the theoretical foundation, it is important to investigate the discussions and the arguments between the current research and also the knowledge of the original theoretical definition and how it has been used from the past to the present. At the end, we have integrated these knowledges and constructs used into the current state of art of the literatures in the research gaps. We then created the research models and hypotheses based on the research gaps.

Conceptual foundations

Mobile payment The invention of mobile payment services has gradually replaced the traditional payment methods such as cash or banking system. Mobile payment can be described as the payment method which is conducted via handheld device such as mobile device or personal digital assistant (PDA) and wireless communication technologies (Valcourt et al., 2005). In this payment method, the direct or indirect exchange of monetary values between the parties will be involved (Kim et al., 2010). And it allows handheld devices to proceed the payment at real point of sale (POS), e- commerce platforms, m-commerce platforms (Kim et al., 2010). Mobile payment is operated by the mobile devices, the devices will access to mobile payment service then conduct authentication and authorization processes to complete the transaction (Kim et al., 2010). In this payment method, the user can pay for different type of transactions such as flight ticket, parking fee, bus, tram, train, and taxi fares via mobile devices (Kim et al., 2010). Mobile payment service continuously proposed many different solutions to obtain the best quality for cost, functionalities, scalability, and security (Kim et al., 2010). Furthermore, mobile payment service involves the credible parties which ensure the security and perform unique value-added roles in the mobile payment delivery chain (Kim et al., 2010). Normally, payments for purchases and payments for bills/invoice are the main transaction conducted via mobile payment service (Kim et al., 2010).

The evolution of payment method has been developed from physical exchange of notes and coins to writing checks then bank transferring either person or distance; and finally, through phone devices or Internet as today (Valcourt et al., 2005). This evolution has turned the physical transference of tangible assets into information exchange between different parties (Valcourt et al., 2005). Due to the development of e-commerce, the payment method should be digitalized as well to quickly adapt with the common trend. The e-commerce platform allows the buyer and the seller conduct all the payment over the open networks and without physical contact (Valcourt et al., 2005).

Three main types of mobile payment methods are known as internet payment method, point of sales mobile payment (POS) and payment for mobile commerce applications.

Internet payment method

Internet payment method is also widely known as a type of electronic payment system (Shon, T. et al., 1998). By this payment method, the users will give their cell phone number then the transaction will be charged via mobile carrier phone bill (Valcourt et al., 2005). The advantages

of internet payment method are its speed and security while it allows the users to pay without credit card information and no special components or equipment needed (Valcourt et al., 2005). Also, according to Valcourt et al. (2005), the three steps of Internet payment method are (1) input mobile phone number, (2) receive a transaction code via text message. (3) Then input that transaction code on the site; the user will receive a text message about transaction confirmation. The amount of this transaction will be automatically added to their mobile phone bill later. And it takes only 20 seconds to conduct the whole process of this transaction (Valcourt et al., 2005). The advantages of this method are its speed and security when it pays without credit card requirement (Valcourt et al., 2005). Also, the users don’t need to invest any money for the merchant, special devices or components (Valcourt et al., 2005).

POS mobile payment method

This method of mobile payment allows the user to pay at the point of sales by mobile devices. In POS, the users have to synchronize with the merchant system to complete the transaction (Valcourt et al., 2005). It is very advantageous with the micro payment when users don’t have enough coins or available coins, they can easily pay by this method (Valcourt et al., 2005). However, it will be obstacle when this method asks the users to modify and install the payment system from the merchant (Valcourt et al., 2005).

Payment for mobile commerce applications

Mobile commerce is the use of portable wireless devices such as mobile phones and tablets to conduct the online commercial transactions consisting of product purchases and sales, online banking, and bill payments (Bloomenthal, 2020). The usage of mobile commerce is increasing which is estimated up to $207.2 billion in 2017 in US market (Bloomenthal, 2020). The advantage of this method is that the customers can flexibly purchase goods at anytime, anywhere at a virtual point of sale (Valcourt et al., 2005). It is considered as an extension of electronic commerce to mobile phones (Valcourt et al., 2005). But the design mobile phone sometimes does not match 100% with mobile commerce which cause some incontinence to the users (Valcourt et al., 2005). It is expected that the development of mobile device can improve this problem and gain the market share for this mobile payment service (Valcourt et al., 2005).

Electronic payment (E-payment)

Due to the development of e-commerce platforms and many different exchanges among business partners here, traditional cash-based system has been gradually altered by electronic

payment systems (Masihuddin et al., 2017). This payment method can also be considered as the most successful application of technology in bank industry when it can maximize the efficiency and performance of bank services (Fatonah et al., 2018). The productivity of banking service is significantly enhanced due to the higher speed and accuracy of e-payment system (Fatonah et al., 2018). According to Fatonah et al., (2018), e-payment is defined as a payment mechanism using electronic media without cash involvement. Besides, it is also known as a type of inter- organizational information system (IOS) for related money transactions which connects different associations and individual customers (Masihuddin et al., 2017). Therefore, the complicated interconnections between parties such as partners, environment and technology may be compulsory (Masihuddin et al., 2017). As the Federal Financial Institution Examination Council’s identification (2010), e-payment is a “new payment practice for retail where a merchant retrieves payment information for goods and services and places this information in an electronic template that creates electronic files for processing over the network” (Fatonah et al., p.2, 2018).

Particularly, the non-cash transaction market via e-payment and m-payment (mobile payment) has been fluctuated significantly over the years. In 2011-2012, this the annual growth rate dropped from 8.6% to 7.7% (Masihuddin et al., 2017). However, in 2014, the market of global non-paper exchange witnessed the most remarkable growth rate since the first publication of World Payments Report when it peaked 8.9% reaching 387.3 billion (World Payments Report, 2014). The growth rate was expected to continuously increase in 2015 since it obtained 10.1% approaching 426.3 billion (World Payments Report, 2016). Gradually, e-payment has proved it vital roles towards both individuals and organizations by its security and conveniences. It also reduces the fraud rates resourcefulness in world payment systems.

The numerous e-payment services have been generated globally which include electronic cheques, e-cash, credit cards and electronic fund transfer. But generally, the two main types of online payment method are known as Internet Banking Payment Gateway (IBPG) and outsider payment platform (Masihuddin et al., 2017). First, IBPG is a direct form of online payment when the users can directly proceed the payment in e-business framework linking to banking system (Masihuddin et al., 2017). Besides, in the outsider payment form, the money will be transferred from buyer’s account to merchant’s account through another third-party payment platform (Masihuddin et al., 2017).

In the research of Yu et al. (2002), four groups of e-payment systems were mentioned as electronic cash, online credit card payment, small payments and electronic cheques. Obviously, each method has its own advantages and disadvantages. And the qualities of these e-payment method can be assessed by following four criteria: technological aspect, economic aspect, social

aspect and institutional and law aspects (Masihuddin et al., 2017). However, only two kinds of payment system will be mostly applied in the reality which are internet-based payment system and electronic transaction-based payment system (Singh et al., 2013). There are four modes under internet-based payment system including debit card, credit card, smart card and e-cash (Masihuddin et al., 2017). Besides, the four modes of electronic transaction-based payment system are also known as secure electronic transaction (SET), cyber cash, net bill and first virtual holdings (Masihuddin et al., 2017). The appearance of e-payment system has brought various advantages to the users. First, it helps the users avoid the threats and frauds (Fiallos et al., 2005). Normally, the customers have to insert the credit card number in an unencrypted form over the system when they purchase online goods which easily leads to frauds and threats (Fiallos et al., 2005). With e-payment, it can offer the users plenty of secure payment forms which will protect the important data of users more efficiently (Masihuddin et al., 2017). Second, the e-payment service helps the users save their time when they only need to input their account related information and address once then all the information will be automatically saved in the systems for their following transactions (Hord, 2005). Besides, e-payment system has brought various advantages to both vendors and consumers when it can help them reduce the transaction costs (Masihuddin et al., 2017).

Mobile commerce (M-Commerce)

Mobile commerce is characterized by convenience and ubiquity (Du, S., & Li, H. 2019). The differences between electronic commerce and mobile commerce are the fact that mobile commerce happened after E-commerce (Du, S., & Li, H. 2019). There are still researches that are related to m-commerce but are still lack of standards in terms, concepts and theories (Okazaki, S. 2005). However, according to Durlacher 1999 as cited by Okazaki 2005, m-commerce has defined as “any transaction with a monetary value that is conducted via a mobile telecommunication network” In addition, mobile payment is considered to be a subset part of E-commerce.

Moreover, the framework of mobile commerce has explained by the article from Vashney, U. & Vetter, R. 2001, that there are total four different levels of framework, M-commerce has been divided into 1.M-commerce applications 2. User infrastructure 3. Middleware and 4. Network infrastructure. It has been explained in a simple way as the applications will be taking user infrastructure’s capabilities as a consideration (mobile devices). The mobile middleware is playing an important role of developing the new mobile commerce applications. Lastly, the network infrastructure also counted as an important role since it is considered one of the factors from the users as they perceived service quality depends on available resources (Vashney, U. & Vetter, R. 2001). From this point, not only the application itself, but the network infrastructure that might

relates to the decision whether people will use m-commerce services or not. Taking a closer look on what factors played an important role for the success of m-commerce, according to Du, S., & Li, H. 2019, there are three different factors including 1. Security and trust 2. Personalization and localization 3. User’s convenience. Hence, we assume that these factors are important to be considered in our research. Some factors that have been mentioned before are related to the Technology Acceptance Model by Davis (1989) as TAM is the basic model for the development of our research and the model itself involves the factors such as perceived usefulness and perceived ease of use from the point of view of the consumers.

In the perspective of the m-commerce users, people who are the current users of m- commerce tend to enjoy service because m-commerce is timeless, and they can use it without position restrictions. The fast development of mobile devices makes this possible for the m- commerce to be simple enough to be operated by the users and considered as cost-effective (Du, S., & Li, H. 2019). As the popularity of m-commerce arose, it is important to know what makes the users continue to use or stop using m-commerce, as there are expectations that relates to the user satisfactions and the user satisfactions related to the future use of m-commerce (Xu, C., Mongo, P., & Ganiyu, S. 2020).

In our research, we tend to focus on the factors that influence the uses of m-payment. However, we need to consider pre and post use of the service as many experts do believe that user satisfaction is not only got effected by complex factors but also constantly changing. According to the research from Xu, C., Mongo, P., & Ganiyu, S. 2020, user satisfaction of m- commerce applications and services is a process that can changed overtime, and has considered in three different stages where the satisfaction could change in 1. Pre-transaction stage 2. In- transaction stage and 3. Post – transaction stage. We considered this part important to develop our understanding when collect the information.

There are many classes of m-commerce application, according to Vashney, U. & Vetter, R. 2001, the Mobile Financial Application is considered as the most important part when focusing on the m-commerce topic. The examples are mobile banking and brokerage service, mobile money transfer and mobile micro payments. The services could turn the mobile devices into a business tool, replacing ATM, banks and credit cards by allowing the users to conduct the transaction by themselves with mobile money (Vashney, U. & Vetter, R. 2001). There are several important requirements for using m-commerce. The important factors are the security and privacy, atomic transaction, supports which relate to the use of Mobile Financial Applications (Vashney, U. & Vetter, R. 2001).

Development of mobile payment and its challenges

We would mainly focus the articles that has been written about the past, present and future development of m-payment in order to understand the key idea of how it is developed and what involved in its process. This includes the findings and the challenge for the future research, and what should be considered when doing the research relates to this topic.

In early 1990s, oil companies in the US started the use of RFID chips for the customers to wave at the pump to purchase gasoline, later the text message donations and similar trends were catching up fast (Rao, S.V. 2015). Bank institutions are also a part of this technology where banking systems tries to introduce the definition of cashless transaction (Rao, S.V. 2015). According to an article from Dahlberg et al. (2007), looking through the past between late 1990s and early 2000s mobile payment is a hot topic for the research since the introduction of internet era, after those times, hundreds of services from mobile payment services have appeared widely around the world. Back in those time, some of the services were not successful, for example, most of the mobile payment services in EU have been discontinued, meanwhile PayPal and Visa cards are growing successfully. Dahlberg et al. (2007) has explained that the mobile technologies were not growing mature enough to handle such big data and handle security. In 2011, Google has introduced Google Wallet where the Google Wallet could pay in stores, redeem coupons and earn loyalty points (Rao, S.V. 2015). Later the technology called (NFC) Near Field Communication is well accepted worldwide, however the availability is restricted with one phone model (Rao, S.V. 2015). In an emerging economy, the adoption of cashless payment comes at the same time where smartphones are well affordable by many people, the E-commerce has also played a big role for the increased use of the cashless transactions (Rao, S.V. 2015).

Nowadays, the technology has been developed fast and the key components from the basic mobile payment ecosystem would be identified with the three-party scheme including the merchant, the customer and the service provider (Lu, L. 2019). What truly gives these m-payment providers/developers a great success, are another aspect we should come across is the business model they are using. According to an article from Lu, L. 2019, there are three questions to be mentioned

1. What features make it attractive? 2. What benefits can it bring to market? 3.How to profit from it? (Lu, L. 2019).

To answer question 1, the value proposition is the key to attract customers, there are convenience when using it, technology itself – easy to use and simple, integration – by turning a mobile phone into a business tool conducting transactions by oneself, cost saving – many transactions are cheap especially the person-to-person services (transfers between accounts), lastly safety – the transactions are encrypted during the use (Lu, L. 2019). To answer the second question, market contribution such as E-commerce, m-payment usually gives benefit and increase the value to the market in general, therefore, many industries and businesses got influenced by the way people use m-payment quite a lot. Lastly, to answer to question 3, revenue sources is the way m-payment companies survived, according to Lu, L. 2019, there are two types of revenue for m-payment, this includes transaction depended and transaction un-depended. There are features in the applications such as embedded games or in app purchase which keeps the user participated and active from time to time (Lu, L. 2019).

Besides, there are many factors that will determine the trend of m-payment development for the future and its challenge in the market. We should also consider how our research should pay a closer look to these factors that has an impact on the changes of m-payment and its users.

Figure 1: Framework of factors impacting the mobile payment services market (Dahlberg et al. 2007 pp. 3)

According to figure 1 above includes both market factors and contingency factors. We intend to use this framework as a part of guiding theories and according to Dahlberg et al (2007), this framework helps to bring clarity to vague terminology presented in academic mobile payment literatures. The four outer dimensions are the contingency factors – this includes the factors in

social/cultural environment, commerce environment, technological environment and lastly change in legal, regulatory and standardization environment.

The first factor, changing in social/cultural environment includes the consumers’ consumption habit, buying behavior, the need for new payment system. Bohle and Krueger (2001) analyzed the influence of culture on development of payment systems (Dahlberg et al ,2007), as well as research from Mahmood et al. (2004), the author examined the influence of culture on online shopping behavior, factors that were discussed are demographics and lifestyle characteristics. Changes in commerce environment is the next factor, this factor means the introduction and internet have increased automation and self-service orientation of payment services, these changes drive the need of payment services (Dahlberg et al ,2007).

The research results from Dahlberg et al ,2007 shown that changes in commerce environment may push the development of enhanced mobile payment services. Changes in technological environment is a topic of different research and it is the most popular ones. The research on technology is meant to analyze and find limitations of various technology and propose new advancement to the current mobile payment services. The proposals are mostly addressing the architecture, security and trust issues of the payment systems (Dahlberg et al ,2007).

Lastly changing in the legal, regulatory and standardization environment may drive or hinder the development of mobile payment services and these areas could be one of the topics for the future research (Dahlberg et al ,2007). In our research, it is important to realize the inner facet of this model in figure 1 since this inner facet mentioned the Porter’s five forces model into the context of mobile payment, there are the forces from consumer, merchant, new E-payment services, traditional payment services and the competition between mobile payment providers. In our analysis from our empirical results, it is necessary to be aware of these factors in each of the country since the results could be varied and different based on the factors involved in this figure.

Mobile payment in an international scope

The large number of mobile devices around world have created many opportunities for the development of cashless market; especially, mobile payment service (Clement, 2020). Recently, this payment method has been rapidly grown in many markets such as Africa, Asia and Latin America (Clement, 2020). Thanks to mobile network operators, mobile payment service allows customers to send, receive and store the money by their mobile devices (Clement, 2020). It is known that Apple Pay, Google Pay and Samsung Pay are the three biggest mobile payment

providers globally (Enberg, 2019). However, China is dominated by local services such as Alipay and WeChat Pay and Paytm is the leader in Indian market (Enberg, 2019).

In 2018, more than one third of global internet users used the mobile payment services and most users were from Thailand and China (Clement, 2020). Following by 2019, 36.6% smartphone users were expected to use mobile payment methods at least one time per six months and the main the users located in Asia-Pacific: especially China (Enberg, 2019). Besides, this payment method is also widely used in India, Denmark, Sweden, and South Korea (Enberg, 2019). Although the mobile payment usage in US is far behind from Asia and Northern Europe, the volume and average transaction size for mobile payments and peer to peer payment (P2P) also play a vital role in this country (Wurmser, 2019). In 2019, 29% of smartphone users (64 million people) accessed to mobile payment services (Wurmser, 2019). And obviously, Apple Pay has been the leading provider of mobile payment service in US which accounts for half of users in this market (Wurmser, 2019).

And the most notable market of mobile payment usage should belong to China when it obtains the majority of users over the world (Cheung, 2019). In 2019, proximity mobile payment users increased 10% which peaked 577.4 million users (Cheung, 2019). It is known that 81.1% smartphone users in this country are also the users of mobile payment method (Cheung, 2019). Following by China, Denmark is in the second position when the mobile payment users account for 41% of smartphone users in this country (Clement, 2020). Many supports for mobile payment from global merchants such as mobile applications for online shopping, point of sale, mobile wallet, etc. have significantly enhanced the usage and proximity of this payment method (Clement, 2020). Currently, point of sale mobile transactions have been the most used service (Clement, 2020).

Mobile payment in an emerging market

Over decades, the cash or traditional payment has been affected by the new means of payment – namely mobile payment. The emergence of smart phones, internet banking and internet have increased the change of user behavior of using paper-based payments (e.g. cash and cheques) (Mondego, D. 2018). Online banking activities have influenced the behavior of consumer behavior and emerged as an evolutionary path to the payment process in order to respond to the growth of online transactions, of course, this improvement is surely convenient for users and the experience they will gain from purchasing goods from their home (Mondego, D. 2018). Mobile devices have acted in the way that connects users and the online banking possible, the main use of online banking would be to purchase, pay for services and goods, transfer of money, pay bills online or at point of sales (POS) at anytime and anywhere (Mondego, D. 2018).

There is the introduction of the mobile payment applications in emerging markets, the mobile money services or mobile money systems are considered being deployed rapidly across emerging markets (Lal, R. et al, 2015). In the article Mobile Money Services – Design and Development for Financial Inclusion has been analyzed about which characteristics are critical for success for a mobile money service, their research covered five successful mobile money deployments and also the five least successful mobile money deployments, the research has taken those samples from the emerging market countries, the key successes are associated to the structure of mobile money service, the typical structures are for example, owner/operator or namely the financial institution (Banks), bank account operators, service offers (peer to peer money transfer, remittances both domestic and international, all sort of payments and receipts), service delivery method (mobile money services are delivered in two ways : 1.directly through customer’s mobile phone or 2. Over the counter (OTC)), distribution network, customers served and fee structure. The key successes over this type of payment are 1. building an effective working relationship with regulators 2. Building trust in service 3. Safekeeping of customer funds 4. Facilitating cash-in and cash-out 5. Liquidity management (Lal, R. et al, 2015).

In the aspect of customers who use the mobile payment methods, many researches have shown the impact of customer trust on the users’ acceptance of using new payment systems, the strongest element that affects the intention of people to utilize their smartphones to make payment is trust (Duane, O’Reilly, & Andreev, 2014 as cited by Mondego, D. 2018). Moreover, the studies from Yan and Yang (2015) stated that perceived ease of use, perceived usefulness, structure assurance and ubiquity have significant effect on user’s trust (Mondego, D. 2018). There is of course the reason why trust is the most important factors to customer’s satisfaction and intention to use mobile payment service is that there are difficulties in maintaining the service when the users and the providers do have less face to face contacts (Bourreau & Valetti, 2015 as cited by Mondego, D. 2018). In addition to the trust factors, there are more key drivers and the barriers to the adoption of m-payment, according to an article “Trends in Mobile payment research: A literature review” the authors Dennehy and Sammon (2015) have described the key drivers into 1. Offering added value for consumers, merchants, mobile operators, financial institutions, and other participants in the ecosystem, 2. User experience (ease of use), moreover, the barriers are 1. Complex value chain with lack of co-operation 2. Financial regulation 3. Security/Risks 4. Cost 5. Unavailability of a broad range of mobile payment and lastly 6. Lack of interoperability / lack of technology standards. Those general drivers and barriers are used as key variables to the users’ intention to use mobile payment in several articles. The findings will be discussed after the theories background.

In Southeast Asia (SEA), about 50% of the population in six major Southeast Asian Nations use smartphones according to Euromonitor as cited by Yukako Ono. The number is expected to grow to 70% in 2021. The reason why people choose this option in mobile payment is the great convenience for people outside the conventional banking system, by only 30% of the Indonesian and Filipinos have bank accounts (World Bank Data, 2014 as cited by Yukako Ono). In Singapore, Oversea-Chinese Banking Corp (OCBC) has launched mobile payment applications that can be used over 1000 merchants, including Robinson department stores, the customers will be using the QR code scanner to scan the code which is linked to the mobile application so the customers could pay directly and the money will be deducted from OCBC accounts (Ono, Y. 2017). In Indonesia, the company Go-Jek launched Go-pay, Go-Pay can be used for growing number of Go-Jek’s ancillary operations such as food deliveries and personal shopping services (Ono, Y. 2017).

There is also a great factor from China to SEA, the mobile payment is still dominating the trends in using mobile payment, by 430 billion dollars of mobile transaction in 2016 (Ono, Y. 2017). The mainland market is dominated by Alibaba Group’s Holding’s Alipay and Tencent Holdings’ WeChat Pay (Ono, Y. 2017). These two companies are exporting the business models to Southeast Asia by partnering with local conglomerates (Ono, Y. 2017). In Singapore and Thailand, the payment available options are mostly for hotel as well as convenience and department stores (Ono, Y. 2017). According to Nikkei Asian Review (2017), the mobile payment became a strong trend in South East Asia, for example, in Thailand, the Rabbit transit cards (to be used with the BTS (Bangkok Mass Transit System) sky trains) owned by 7 million domestic users to commute around Bangkok every day, the Rabbit transit cards are accepting the smart phone to be used as the Rabbit tickets instead of the real tickets (Ono, Y. 2017). The BTS teamed up with Line, a Japanese messaging program, to develop application Rabbit Line Pay to be used together with many shops and restaurants as they offered attractive privileges and promotions. The Line not just only be paid for restaurants and some shops, it is expecting to expand further to the retail outlets (Ono, Y. 2017).

Thailand and Vietnam are almost becoming mobile payment markets as the two country’s governments encourage a push towards cashless transaction (Mobilepaymentstoday, 2019). Many of Vietnam’s carriers offer e-wallets and the percentage of using mobile payment jumped from 37% in 2018 to 61% in 2019 (Mobilepaymentstoday, 2019). Thailand even has larger mobile payment penetration rate; the region’s highest rate is at 67% in 2019. Banks play a key role as they push to cashless society by the government believing that cashless could also help with reducing corruption and tax avoiders (Mobilepaymentstoday, 2019).

Mobile payment in Vietnam

Similar to Thai market, it is better to have some general insights about Vietnam market before conducting the research about it. First of all, an overall information about Vietnam’s demographics should be mentioned. In 2020, the total population of Vietnam is 96.9 million inhabitants which 36% of them concentrates on urban areas (Kemp, 2020). There is an ideal balance in gender of Vietnam’s population when the half of population is male and another half is female. The median age of Vietnamese is 32.5 which represents that most of the Vietnamese are in the working age (Demography - Elderly population - OECD Data, 2020). The average of 313.9 people lives in every squared kilometer (Kemp, 2020). And the over literacy rate of the country is 95% when female literacy rate is 94% and male is 96% (Kemp, 2020). According to World Bank and Trading Economics, the gross domestic product (GDP) of Vietnam in 2019 was 255 million US dollars which contributed 0.21 percent of the global economy (“Vietnam GDP | 1985-2019 Data | 2020-2022 Forecast | Historical | Chart | News", 2020).

Then, it is essential to indicate some general information about mobile and internet connection market in Vietnam. There are large numbers of 145.8 million mobile connections in Vietnam which are even more than its total population (Kemp, 2020). And the 68.17 million internet users which also penetrate for 70% of the whole population (Kemp, 2020). Also, 65 million of active social media users (Kemp, 2020) which indicates this is a dynamic and potential market for all new technologies. And this market still maintains its potential when the growth rate of mobile phone connection increase 1.9%, internet users raise 10% and active social media users increase 9.6% annually (Kemp, 2020). To support for the background of this study, it is known that 93% of Vietnam’s population owns smart phone and 32% owns tablet device (Kemp, 2020). In the survey about daily time consumption for media, it is illustrated that the Vietnamese spend an average of more than 6 hours for using internet, over 2 hours for social media per day (Kemp, 2020). Some most common social media platforms in Vietnam are Facebook, YouTube, Zalo, Facebook messenger, Instagram (Kemp, 2020).

Next, it’s also important to the observe the trends of mobile usages in Vietnam market. First, there are some statistics about mobile application usages by Vietnamese. The most used mobile application by Vietnamese are chat apps (93%), social media platforms (94%), entertainment apps (85%), game apps (58%), shopping apps (55%) and banking apps (36%). Furthermore, the statistics about mobile actions among Vietnamese also indicate that most of them use mobile device for search tools or services (57%), transfer money (46%) and the rest for entertainments and other activities.

Moreover, there is a strong connection between e-commerce and mobile payment market so it will be useful to have an overall view about e-commerce market of Vietnam. In general, around 30% of Vietnamese has a financial account with financial institution. Among that, 4.1% owns a credit card in which 3.7% is female and 4.6% is male (Kemp, 2020). And the percentage of mobile money account is only 3.5% which is significant low compared to banking users (Kemp, 2020). There is 21% of online purchasing or online payment (Kemp, 2020). Vietnam can be considered as potential market for e-commerce when we have a look on their spending on e- commerce. There was more than half of Vietnam’s population purchased consumer online in 2019 which penetrated the value of 2.96 billion dollar for this new digital evolution (Kemp, 2020). These spending was known as $717 million dollars on fashion and beauty, $716 million dollars on electronic and physical media, $517 million dollars on food and personal care, $526 million dollar for furniture and appliances, etc. (Kemp, 2020). In Vietnam, the average annual revenue for online consumer goods was $54 dollars in 2019 (Kemp, 2020). Although the Vietnamese are well adapted towards e-commerce market, the percentage of e-wallet usages for those transactions are still low. Most of the transactions in e-commerce are particularly through credit card (37%), cash (17%) or bank transfer (30%). Only 11% of consumers use mobile payment method to pay for e-commerce transactions (Kemp, 2020).

Lastly, the brief insight about digital payment market should be considered when we conduct the research about mobile payment in Vietnam. In 2020, there are more 51 million of Vietnamese people making digital payment transactions. It means that more than half of population have digital transactions in Vietnam and the value for those transactions are $8.52 billion dollars. Although the many advantages that mobile payments have brought, this method has not widely applied in Vietnam market. One of the main reasons here is that there are only 30% of people concentrate on urban area where the technologies are highly developed. Therefore, to utilize the population advantages and take the opportunities from digital revolution, many enterprises about non-bank payment services were established. In 2019, there were 30 licensed non-bank payment services were registered in which 20 of them offered e-wallets. (Overview of Vietnam’s Major E-Wallet and Mobile Payment Players - Fintech Singapore, 2020). Currently, Vietnam has four main methods of mobile payment which are SMS, Wap charging, E-wallet and E-banking service (Nguyen et al., 2015). And most used mobile payment apps in Vietnam are known as MoMo, Viettel Pay, Zalo Pay, Moca (Overview of Vietnam’s Major E-Wallet and Mobile Payment Players - Fintech Singapore, 2020).

There is some evidence which can illustrate that mobile payment will be the potential payment type in this country. Vietnam is one of the most six attractive retail market in the world. In 2017, the number of 105 million bank cards were issued in this country which mainly aimed to young

consumers (Nga et al., 2020). The Vietnamese are opened minded to the new payment technologies when 88% of surveyed people have confirmed that they are willing to pay via smartphone (Nga et al., 2020). Hence, it is supposed that Vietnam would have a fast growth in mobile payment market. However, Vietnam market are still facing with many challenges in non- cash payment methods. It was found that 65% of Vietnamese still used the normal financial systems to transfer money (Nga et al., 2020). The explanation for this problem can be the consumer behavior and the habits of using cash over the long time in Vietnam (Demirguc-Kunt et al., 2015). Furthermore, there are only 36% of Vietnam’s inhabitants in urban areas (Kemp, 2020) and new technologies have not approached to rural or remote areas.

Mobile payment in Thailand

First of all, the demographics in Thailand is important to this research. The population in 2020 is around 69,71 Million people. Female population is calculated as 51% of the whole population and the 49% of the whole population are male (United Nation, 2020 as cited by Kemp, S. 2020). The median age is at 40,1 and average age is at 38,1 year. The total population density is around 136.6 people per squared kilometer and the overall literacy rate is at 93% by divided into female, the literacy rate is at 91% and male with 95%. The gross domestic product (GDP) is at 455,3 billion dollars (J.P. Morgan Global Payment Trends ,2019). Moreover, there are internet users of 52 Million users, it has been counted as the penetration rate of 75% in 2020 (Global Web Index Q3/2019 as cited by Kemp, S. 2020). The smartphone devices ownership is at 94% of the whole internet users age between 16 to 64 years old. 50% of them owns laptop and 33% of them owns tablet device, 15% owns smart watch and 12% owns game console (Global Web Index Q3/2019 as cited by Kemp, S. 2020). In addition, Thai people who count as internet users spend around 4 hours 57 minutes per day using internet on their smartphone. The mobile internet users are counted as 97% of total internet users while internet users are around 50,8 million people (Global Web Index Q3/2019 as cited by Kemp, S. 2020).

Going into details about the mobile usages, there are the uses of mobile applications taken as the percentage from the internet users the age between 16 to 64, by 95% of most used application is chat applications, 97% is social networking applications, 89% is the entertainment applications, 67% is the game applications, shopping applications is 58%, music applications is at 65% as well as mobile banking applications of 65% (Global Web Index Q3/2019 as cited by Kemp, S. 2020).

There are the rankings of mobile applications by the average active users throughout 2019, the first ranking is Line, chat application from Japan, rank number 2 is Facebook and number 3 is Facebook Messenger, number 4 is Instagram and number 5 is Lazada, the mobile shopping

application from Alibaba Group, number 6 is Shopee which is also online shopping from SEA company, rank number 7 is K plus from Kasikorn Bank (mobile bank), ranking number 8 is SCB Easy which is also from Siam Commercial Bank (mobile bank), rank number 9 is My AIS which is internet and phone carrier company, number 10 is Twitter (App Annie, as cited by Kemp, S. 2020).

According to this ranking, we can also say that four of our tops using applications in mobile phone in Thailand is mobile shopping and also the mobile banking or mobile payment, this reflects that almost half of the active users are involved in the mobile payment. The evidence has shown, that from the mobile actions, again taken the percentage from the internet users age of 16 to 64 years old who perform each action using their mobile phone each month, the highest of all, 69% is the transfer money, 60% is to use an image search tool or service, 54% is to use or scan QR codes, following by 27% watch content on a TV by casting from mobile phone, and lastly, 15% use a mobile phone as tickets or boarding pass when travelling or using public transportations (Global Web Index Q3/2019 as cited by Kemp, S. 2020). The expansion of the e-payment acceptance point has become popular because the government has supported to install over 800.000 EDC machines, the Thai QR payment has also installed more than 3 million physical merchant points (Bank of Thailand, 2018).

Taking a look at E-commerce Use in smart phones for Thai people, according to World Bank Global Financial Inclusion Data (January 2020) as cited by Kemp, S. 2020, the percentage of the population age above 15 years old that reports owning or using each financial product or service, by 81% has an account with a financial institution, average of 19% has make online purchase or pay bills online, with 19% of women and men making online transactions equally. All in all, the users who have made an online purchase via mobile devices is counted as total 69% from all the internet users between age of 16 to 69 years old (Global Web Index Q3/2019 as cited by Kemp, S. 2020). Moreover, the average transactions of E-payment per person per capita is jumping from 49 transactions per person in year 2016 and up to 89 transactions per person in year 2018 (Bank of Thailand, 2018).

Taking a look at online purchases of consumer goods, there total of 34,80 Million people who purchases goods online in 2019 (Statista Market Outlook as cited by Kemp, S. 2020) and the value of purchase in U.S. dollars are in total of 4,31 billion dollars and the mobiles’ share of B2C e-commerce transaction value is counted as 48% (E-Commerce Report, 2019 as cited by Kemp, S. 2020). According to PPRO as cited by Kemp, S. 2020, there are percentages of e-commerce transactions completed by using each method of payment, the highest percentage would be 32% the credit card use, 25% E-wallet uses (in mobile payment applications), 20% is the banking transfer, 12% is cash.

Referring to the data above, it has been proven that the payment behavior of Thai people is changing dramatically from the past to present. The use of smart phones and mobile payments including mobile payment applications and mobile banking applications are widely used and very popular nowadays. The introduction of 5G, is a push to the E-Commerce to growth making Thailand to be considered a mobile first country (J.P. Morgan Global Payment Trends ,2019). In February 2019, Thailand and Huawei began trialing 5G speeds as a part of the plan to roll out this service until 2040 (J.P. Morgan Global Payment Trends ,2019).

Figure 2: Payments Trends (J.P. Morgan, 2019)

According to the methodologies in Thailand from online shoppers, J.P Morgan has illustrated the figure 2 above, that Thai people are preferred to use E-Commerce or go online shopping if all the processes are done within mobile device, the transactions worth from this activity is around 13.6 billion dollars, meanwhile the second popularity method is to use in-app payment in the mobile and the transaction worth is around 8.9 billion dollars, following by the mobile commerce that the payment is done on a browser by 4.7 billion dollars’ worth of transaction.

Why are the mobile payments and cashless methods in Thailand such a trend? To answer this question, as mentioned earlier about the subsidy from government, Thai government tries to push the card use when spending online, also the government incentivizes the sellers via tax

rebates to adopt the services (J.P. Morgan Global Payment Trends ,2019). The government sector has provided personal income tax returns over 2 million people – making this subsidy calls for more people to use less cash and switch to ‘PromptPay’ (Bank of Thailand, 2018). Not only supporting cashlessness, but the government has also introduced PromptPay which allows registered users to transfer funds between consumers and businesses using their mobile phone number or the citizen ID (Bank of Thailand, 2018). The PromptPay has over 46.5 million PromptPay numbers around the country – as of December 2018 and average of 4.5 Million transaction per day (Bank of Thailand, 2018). In the aspect of E-wallet or digital wallet adoption. The current digital wallets are used to pay 23% of E-commerce transactions (tie with bank transfers), the growth rate of digital payment is expected to be 18% in 2021 (J.P. Morgan Global Payment Trends ,2019). The most popular brand for digital wallet in Thailand is PayPal, TrueMoney and AirPay (J.P. Morgan Global Payment Trends ,2019).

Mobile payment systems and its threats

By the recent research, there are the publications that have associated to the threats and the recent technologies of the mobile payment services. It is important to know the latest trends since those risks which might be one of our factors that have an impact on the consumer’s decision of using mobile payment services.

There are reasons behind consumers’ decision of using mobile payment services as they give their trusts and there is convenience of using it. It is under the same topics since security is encapsulated into convenience, the art of strong authentication and highly secured process together give the consumers a fast and smooth customer journey and it is the overall definition of the convenience (Bader, J. 2019). To explain this in easier way, mobile wallets acted as intermediaries within the financial eco system. It connects customers and merchants at the point of transactions (Bader, J. 2019). Mobile payment technology provides merchants to be able to drive financial ends of a business without the need of massive changes in their infrastructures (Bergthaler, M. 2019).

In the past, the cost of purchase is charged on the mobile subscriber’s monthly phone bill, for the example consumers use mobile payment via SMS. Laster NFC (Near Field Communication) has introduced, this encourages the increased use of mobile payment systems (Wang, Y. et al. 2016). Apple pay program has also use this NFC technologies, allowing people to use their mobile devices to make a payment at contactless POS (Wang, Y. et al. 2016). In the traditional card payment systems, there are 5 key players including consumers, merchants (retailers), acquiring bank, issuing bank and card associations (Visa, MasterCard etc.) (Wang, Y.

et al. 2016). For the mobile payment process, there are two additional key players that are mobile network operators, (MNOs), and mobile payment service providers (MPSPs).

Figure 3: Mobile Payment Process (Wang, Y. et al. 2016)

According to figure 3, as MNOs and MPSPs has introduced, a mobile payment user can send money to another user within the mobile payment system. The process itself can only involve card holder and a mobile payment service provider. There is no merchant, acquirer, card associations and issuers involved in the process, the example of this would be WeChat red envelope, it has been launched in 2014, it has function such as delivering virtual money, withdrawing cash etc. (Wang, Y. et al. 2016). Moreover, there are an introduction of mobile wallet which has similar options and functions. The mobile wallets may contain digital coupons, digital money, digital cards (Bezhovski, Z. 2016). Mobile wallet service allows use to download and install their application in the mobile devices of the user (Bezhovski, Z. 2016). The user could use these applications for the online and offline purchase activities (Bezhovski, Z. 2016). Besides NFC technologies, there are further technologies such as sound waves, QR codes could-based solution. Lastly, mobile wallets are believed to provide more convenient for the users (Bezhovski, Z. 2016).

There are in total five different mobile payment systems and we will explain its terminologies in a short and easy way for our research since we need to understand how it works for the mobile payment systems, as a consequence, we could use those terms without getting confused of which systems we are discussing in the paper. First, Mobile payment at POS, this method enables the customers to pay with a mobile phone at the POS, some of the methods has 37

built-in NFC technologies such as Apply Pay and Google Wallet (Wang, Y. et al. 2016). Secondly, mobile payment as POS, this method allows merchant to use mobile phone as the POS and process the card payments, this method requires a mobile application downloaded to a mobile device and allow a credit card reader to connect to the mobile device (Wang, Y. et al. 2016). To setup, it is easy, quick, and convenient. Third, mobile payment platform, this method allows online payment services on a mobile device, again, this requires application installed in the mobile device – this method is well known for mobile wallet as explained earlier. Forth, independent mobile payment system, this method is similar to the platform but the company itself decide to develop its own online payment service to support mobile devices such as Amazon, Starbucks etc. (Wang, Y. et al. 2016). Last one, direct carrier billing, this method allows user to purchase products and services through mobile devices, the costs are tied up to the monthly phone bill (Wang, Y. et al. 2016). In addition, it is important to notice whether there are threats in using mobile payment systems or not. Consumer might feel unsafe while using the mobile payment method, instead, some recent research confirmed that there are still frauds and threats of using mobile payment systems, as well as other data stolen issues, this topic is still broadly discussed. However, according to Bosamia, M. & Patel, D. (2019), the research paper discussed about mobile wallet threats model, we will only focus on the threats that could happen from the mobile application providers, payment service providers and not so much about users’ threats. The user threats could be very broad to be discussed, such as mobile permissions (allowing others to obtain data), the stolen of mobile device, accidentally install rogue and malware applications which is one part of the threat’s issues (Bosamia, M. & Patel, D. 2019). However, there are also mobile wallet application threats that effect directly to the transaction such as reverse engineering – attack on hardcoded password and encryption keys, weaknesses in biometric identification for initial authorization of transactions, and weaknesses in payment authorizations (Bosamia, M. & Patel, D. 2019). Moreover, there are also threats from payment service providers such as insecure point to point connections between merchant POS server to Payment Service Provider (PSP) and between PSP to acquirers (Bosamia, M. & Patel, D. 2019).

Mobile payment securities need to be improved to prevent such threats as much as possible and develop suitable cautions for the users to feel safe. Some possible security challenges are Malware detection, many cautions are used to detect those malwares, but the malware still find its way to propagate on mobile devices (Wang, Y. et al. 2016). Next, multi-factor authentication, this way users are required to enter an authentication code. Lastly, fraud detection and protection, when fraudulent transactions occur, it must be detected and prevented (Wang, Y. et al. 2016).

Mobile payment as marketing strategy

The proximity of mobile device has opened many potentials for itself to become a new marketing tool (Smutkupt et al., 2010). In the near future, it’ is assumed that mobile device will mostly alter the traditional advertising channels such as TV or newspaper (Smutkupt et al., 2010). Although the large potentials of mobile marketing, it hasnot been utilized by the marketers (Smutkupt et al., 2010).

Today, we would like to focus on the capability of marketing on mobile payment services which will potentially dominate the future transaction market. It is necessary to analyze the competitive factors of mobile payment first. These competitive factors include consumer power and merchant power will give the marketers the most precise directions for further applications on mobile payment method (Dahlberg et al., 2006). Consumers are considered as the crucial element to drive mobile payment’s success when all the main demands of mobile payment are from the consumers (Dahlberg et al., 2006).

According to the research of Faber et al. (2003), the factors will significantly impact on consumer of mobile payments service including trust, ease of use, cost, and reach. The ease of use can be improved by the process of mobile payment usage; for instance, Mobipay allows consumer to use this payment method without any complicated registrations (Faber et al., 2003). Furthermore, providers can also extend the mobile payment capabilities throughout the consumer lifecycle and simplify the checkout process (Porcellana, 2016). Time consuming procedure will discourage the consumer usage so many mobile payment providers have motivated the users by creating many convenient options like special counters or codes of Starbucks. Thanks to this quick option, 21% of Starbuck’s customers in US has conducted their payment via mobile devices (Porcellana, 2016).

Besides, trust is also especially important toward customers when the decide to use mobile payment. In the case of Mobipay in Finland, they allow the users to freely choose their trusted home banks. The credibility of mobile payment can also be enhanced by mobile transaction authentications, fraudulent activities tracking, suspicious activities reporting to the authorities, etc. (Porcellana, 2016). Additionally, the advantage of mobile solution is also known as cost reduction and providing more convenience for consumers (Smutkupt et al., 2010). Base on the mobile marketing services, the customers easily approach to their necessary products or services at many different the point of sale which will give the customers many price comparisons and then reducing searching costs. Therefore, the cost structure is also a crucial concern for the mobile payment providers when they build up mobile marketing campaigns (Smutkupt et al., 2010).

On the other hand, merchant power is also another important factor when they are the main market of mobile payment services. Some failures of mobile payment service were recorded as lacking merchant involvement in developed and proceeding phases (Dahlberg et al., 2006). In fact, some merchants have created their own mobile payment service; for instance, Octopus public transportation in Hongkong and IKEA are the specific examples for this case (Dahlberg et al., 2006). In the past, some research from Ondrus el al. (2005) and from Ondrus el al. (2006b) were conducted to understand more about the potential risk of merchants becoming mobile payment providers. The adoption factors by merchants were also explored by the research of Mallat el al. (2005), Teo et al., (2005) to understand better about their role in the mobile payment market.

Covid – 19 crises

During the current situation of crisis of spreading of the deadly virus, Covid-19 has impacts around the world especially the lock down situation, all the manufactures closed temporary, many bank branches are also closed (Chawla, A. 2020). It is also relevant to our topic of how people avoid using cash and this could be another push to the increase in using more of e-wallet, or mobile payment. In the UK, many stores selling goods and services are trying to avoid cash transactions (for transactions below 40 GBP), most stores can accept the contactless payment by credit or debit card and mobile payment (Chawla, A. 2020). In Australia, basic amenities outlets are also asking customers to use contactless technology to avoid touching EPOS machines (Chawla, A. 2020). In emerging regions such as Africa, is known for the largest unbanked population are implementing measures to shift a great amount of transactions to mobile money and away from cash (Chawla, A. 2020).

According to Johannes Knobloch, a microbiologist at University of Hamburg who studies how microbes interact with surfaces, his research has revealed that microbes interact better on bills than coins made of metal (Betuel, E. 2020), this proves that cash is one part of the spreading of covid-19 viruses. New research also suggested that US. Consumers are becoming more interested in contactless payments. While in Germany, more than half of payments currently made by card are contactless, since digital payment and contactless payment methods certainly require less physical interaction and are more secure (Mercury-processing.com, 2020). More evidence has shown that Covid-19 has probably changed German payment behavior faster than any single technology ever has – says Goerg Hauer the General Manager at N26 (Arneson, K. 2020). Moreover, a survey in the UK shown that 75% of people tend to use less cash because of the pandemic (Arneson, K. 2020). Bundesbank Survey has also shown that 57% of Germans use

more of debit and credit cards, and the bank recorded 56% fewer withdrawals from ATMs in the first month of Germany’s lockdown compared to the previous month (Arneson, K. 2020).

Although Covid-19 pandemic negatively impact the whole world, it has accidentally created more potential opportunities for some high technological industries, especially mobile payment services. In Vietnam, the limited physical interactions during lockdown period generate more chances for mobile payment growth (Lee, 2020). In the context of Covid-19 pandemic, the Vietnamese government has required the State Bank of Vietnam to speed up research and the use of mobile payment methods to reduce cash transactions (Ha, 2020). Since the outbreak, the total number of non-cash payment transactions through the State Bank's financial switching system has increased by 76% compared to last year's same period (Lee, 2020). Within only two months during the crisis, the growth rate of Ngan Luong e-wallet has increased 30%. Also, ZaloPay mentioned that they obtain a significant increase in remittance and purchase transactions estimated at 36% (Lee, 2020). According to Bangkok post (2020), during the restriction by coronavirus, the number of transactions through PromptPay platform in Thailand was also doubled increased compared to last year.

Theoretical foundations

Basics of technology acceptance research

We intentionally divide the basics of each part into three different sections, this includes the Technology Adoption Research, Technology Diffusion Research and Technology Acceptance Research. Each of the sub chapter will describe the general understanding of the theories and the development of constructs used in the previous research, the credibility of theories. Hence, we will only focus more into the general basics and not going through pure theory and its definitions since the explanation of theories and definitions will be in another chapter under ‘Technology Acceptance Models’ where we refer to primary references, original quotes and model figures. Furthermore, the context below will be about how the theories were used from the past to present and the current state of art, how technological relevant topics use these constructs and formulate the theories. Some of the research topics were related directly to mobile payment and others are about the innovation of new technologies.

Technology Adoption research

In this chapter we tend to focus on the main theories explained and used in the research papers from the past to the present and also the direction of the future research of technology adoption. We will go through about models and theories briefly since these topics will be deeply discussed under chapter 5.2.5.

Chhonker, et al. (2017) reviewed the highlight on the growth of mobile commerce adoption literatures as this topic were conducted by many researches to explore the significant growth of mobile commerce adoption and factors affecting behavioral intention, attitude and usage for the past years (Chhonker, et al. 2017). However, many of these studies were not investigated much about the state of art specifically based on technology adoption frameworks in the last five years (Chhonker, et al. 2017). In addition, it is crucial to explain fundamental and background of theories related to technology adoption and what constructs were used in those studies and also how often were these constructs used. Two main theoretical frameworks Technology Acceptance Model (TAM) – developed by Davis (1989) and Theory of Planned Behavior (TPB) – developed by Ajzen (1985, 1991) were used as a foundation of technology adoption studies within various context (Oliveira, T. & Martins, N. 201); Kohl, S. & Eydgahi, A. 2017; Chhonker, et al. 2017) , where TAM is widely used for assessment of how people make decisions regarding new technology adoption (Chhonker, et al. 2017).

Moreover, TAM is used to predict user acceptance with two constructs, Perceived Usefulness (PU) and Perceived Ease Of Use (PEOU), Davis defined PU as “the degree to which a person believes that using a particular system would enhance his or her job performance” (Davis, D. 1989 p.320) and PEOU is defined as “the degree to which a person believes that using a particular system would be free of effort” (Davis, D. 1989 p.320).

TAM was originally implemented with the support from IBM Canada and is rooted from the psychological theory called Theory of Reasoned Action (TRA) – developed by Ajzen and Fishbein (1980) and it has explained the ‘attitude’ as individual’s evaluation of an object and defined ‘beliefs’ as a link between an object and some attribute, ‘behavior’ is defined as intention or result (Lai P.C., 2017). Ajzen (1991) has developed Theory of Planned Behavior (TPB) and is explained about one factor that determine behavioral intention of the persons’ attitudes towards that behavior (Lai P.C., 2017).

Furthermore, TAM2 is developed by Venkatesh and Davis (2000), TAM2 has introduced cognitive and social influence processes as a way of measuring usage intentions and perceived usefulness (Kohl, S. & Eydgahi, A. 2017). In TAM2, the subjective norm has direct relations with perceived usefulness and intention of use, its relation with perceived usefulness is moderated by user experiences while its intention of use is moderated by the user experience and voluntariness of use (Momani, A. & Jamous, M. 2017).

Innovation of Diffusion Theory/ Diffusion of Innovations (IDT/DOI) is also a theory which is mainly used in technology adoption studies and information technology adoption, it has developed by Rogers (1995). It is a theory of how, why and what rate new ideas and technology spread through cultures both applied in an individual level and firm level (Oliveira, T. & Martins, N. 2011). Moreover, from studies and constructs from previous models (TRA, TAM, TPB), the Unified Theory of Acceptance and Use of Technology Model (UTAUT) has developed from Venkatesh et al. (2003) and it has four predictors of users’ behavioral intention, plus there are performance expectancy, effort expectancy, social influence and facilitating conditions (Lai P.C., 2017). All in all, TAM, TRA, TPB and the extension of TAM (TAM2 and TAM3), UTAUT were used over years by researchers to explain adoption technology systems (Lai P.C., 2017).

Chhonker, et al. (2017) has collected and analyzed literatures related to the growth of mobile commerce studies based on technology adoption framework from year 2008 to 2016. The author has collected 201 studies related to TA framework and found out that 182 studies (91%) were using quantitative method – the rest of articles used mixed methods between qualitative and quantitative and only qualitative methodology.

Moreover, 53 articles out of 201 articles are published from China alone, there is an assumption that mobile commerce has been enormously developed in East Asia. In addition, cross-country study is gaining more popularity (Chhonker, et al. 2017). In terms of factors, there was an increased attention of scholars and academicians towards behavioral intention because of the increase of studies in the areas of mobile banking and mobile payment technology (Chhonker, et al. 2017).

According to Chhonker, et al. (2017), out of 201 articles, 138 studies were using TAM, the reasons were explained that TAM does not include any external variables and demographic and it has its own limitation to explain behavioral intention. As supported by Lai P.C., (2017), TAM attempts to help researchers and practitioners to distinguish why a particular system might be accepted or unaccepted and it takes up suitable measures by explanation besides prediction (Lai P.C., 2017). Besides TAM, there are recent consumer behavior theories that have been mostly discussed in articles related to mobile payment topics, such as risk perception, five factor model of personal traits, uses and gratification theory and most recent would be task-technology fit (TTF). Briefly, TTF has been developed by Goodhue et al. (1995), the author assumed that the good fit between task and technology is to increase the likelihood of utilization and also increase the performance impact (Goodhue et al. 1995; Lai P.C., 2017). The result from Chhonker, et al. 2017 has shown the overall frequencies of technology adoption theories in m-commerce in the figure below, as mentioned earlier, TAM is predominantly used by most of the articles, following by the mixed of 2 models (any theories mixed) and UTAUT.

Figure 4: Frequency of technology adoption in m-commerce (Chhonker, et al. 2017, pp. 893)

As supported by Kohl, S. & Eydgahi, A. (2017), both TAM and TPB will continue serve as the fundamental theories for researchers seeking to study the factors influencing consumers’ adoption intentions of various technologies.

Furthermore, there are constructs that has mostly used in the topic of mobile commerce under the technology adoption framework, from the collected of 201 articles, the figure below has shown the number of frequency of constructs with its year of publication.

Figure 5: Frequency of technology adoption frameworks constructs in mobile commerce (Chhonker, et al. 2017, pp. 893)

Notes to fig. 5: BI: Behaviour Intention; PU: Perceived Usefulness; PEOU: Perceived ease of use; SI/SN: Social Influence/Subjective Norm; ATT: Attitude P.ENJ: Perceived enjoyment; EE: Effort Expectancy; COMP: Compatibility; PE: Performance Expectancy; FC: Facilitating Conditions; INN: Innovativeness; AU: Actual Use; RA: Relative Advantage; TRL: Trialability; CPX: Complexity; B: Behaviour ; OB: Observability; JR: Job Relevance; IM: Image; OQ: Output Quality.

As shown the figure 5, the constructs were based on the models and theories used by the most articles, for example, the constructs of Perceived Usefulness (PU) and Perceived Ease of 44

Use (PEOU) were related to TAM and of course TAM is used the most in the 201 articles – that is why those PU and PEOU were used the most in those articles. Moreover, Behavior Intention (BI) was mentioned very often and used as dependent variable in many articles related to m- commerce, this gives the evidence that our research should also take BI as one of a consideration when developing our hypothesis. According to Lai P.C. (2017). TAM has become so popular that has been cited in most of the research; TAM has also been tested widely with different samples in different situations and proved to be valid and reliable model (Lai P.C., 2017).

To illustrate the technology adoption frameworks focusing on the topic of mobile payment or mobile commerce, in the table below, we have collected some of the articles that has used the constructs and factors based on technology adoption frameworks. This way we see how those articles have used their models and how they used the constructs in their research as this will help us understand what constructs were missing and what theories/models were used and tested by recent articles.

Table 1: Theories and constructs used in recent mobile payment research. (Source: self-edited)

Theories and Constructs and Methods Data and Authors Models variables1 Context

TAM PU, PEOU, SE, Structural 381 potential m- Shankar, A., & Datta, SN, INN, BI equation payment user B. (2018) modelling (SEM) respondents

Online & offline surveys

Traditional PU, PR, PEmo Structural 484 respondents Wu, J., Liu, L., & decision theoretic (perceived equation Huang, L. (2017). 2 set of online approach, emotion), BI modelling (SEM) surveys at 2 Consumer diffusion stages response system of WeChat model application

UTAUT BI, PE, PR, PC Structural 605 respondents Abrahão, R. d. S., (perceived cost), equation Moriguchi, S. N., & online and offline SI, modelling (SEM) Andrade, D. F. (2016) surveys

Extended TAM SN, PR, P.ENJ, Partial least 263 respondents Driediger, F., & Visibility, BI squares online and offline Bhatiasevi, V. (2019) Structural surveys equation

modelling (PLS- SEM)

TAM, UTAUT HEDONIC Structural 480 respondents Boonsiritomachai, W. (hedonic equation & Pitchayadejanant, online and offline motivation), PE, modelling (SEM) K. (2017) surveys EE, FC, SEC (security), BI

TAM PT (perceived Structural 604 responses Liu, G.-S., & Tai, P. T. trust), PU, PEOU, equation only online (2016) PR, COMP, BI, modelling (SEM) surveys Mobility Exploratory factor analysis (EFA)

TPB AT, SN, PBC Partial least 311 respondents Tan, K.-L., Memon, (perceived squares online and offline M. A., Sim, P.-L., behavioral Structural surveys Leong, C.-M., control), BI equation Soetrisno, F. K., & modelling (PLS- Hussain, K. (2019) SEM)

TAM, UTAUT PR, SN, PU, Partial least 298 respondents Zhang, Y., Sun, J., PEOU, INN, BI squares online and offline Yang, Z., & Wang, Y. Structural surveys (2018). equation modelling (PLS- SEM)

Extended TAM SN, AT, PU, Structural 287 respondents Luna, I . et al. (2019) PEOU, SEC equation online and offline (security), BI modelling (SEM) surveys

TAM, TPB PU, PEOU, P, Structural 489 respondents Nguyen, T. N., Cao, ENJ, PT equation online and offline T. K., Dang, P. L., & (perceived trust), modelling (SEM) surveys; 16 Nguyen, H. A. (2016). SN, PBC structured (perceived interviewees behavioral control), BI

TAM PU, PEOU, SN, Structural 256 respondents Phonthanukitithaworn, PT (perceived equation ;surveys & C. Sellitto, C. & trust), COMP, modelling (SEM) Fong, M. (2015) PR, BI

UTAUT2 HEDONIC Structural 243 respondents; Kim, S.H, Yoo, B.K. (hedonic equation surveys (2020) motivation), PC modelling (SEM) (perceived cost), PR, SI, EE, FC, PE, BI

According to table 1 above, we can briefly discuss that the recent research still uses TAM (8 articles), UTAUT (4 articles) and mixed theories (4 articles). Meanwhile they use constructs that have been mentioned in both theories such as BI, PU, PEOU, SN, INN, PR, and PC and other constructs such as PBC, EE. Moreover, these articles mostly used surveys and they have used Structural Equation Modelling (SEM) as their methods. We tend to focus on the used theories as our base to this research and also the constructs which has been mentioned in the research gaps.

Technology Diffusion research

In this chapter we tend to focus on the Innovation Diffusion Theory (IDT) by Rogers (1983). The focus will mostly be related to the general topics, definition, factors and then focus more onto the scope of research in NFC technology/mobile payment by applied by the IDT theory. The knowledge of IDT is important for our research since there are some constructs used by the previous research related to mobile payment.

Innovation Diffusion Theory provides well-developed concepts for the study of technology evaluation, adoption and implementation (Fichman, R.G., 1992). Moreover, diffusion theory helps identifying numerous factors that facilitate or hinder technology adoption and implementation (Fichman, R.G., 1992). According to Rogers (1983), diffusion refers to “The process by which an innovation is communicated through certain channels over time among the members of a social system.” (Rogers, E. 1983 pp.5). Moreover, he defined communication as “A process in which participants create and share information with one another in order to reach a mutual understanding” (Rogers, E. 1983 pp.5). It is how the technology has been spreading from non- users to the current users of a new technology. Innovation has also defined as “An idea, or object that is perceived as new by an individual or other unit of adoption” (Rogers, E. 1983 pp.11). More

of the definition of diffusion he added “Diffusion is a kind of social change, defined as the process by which alteration occurs in the structure and function of a social system” (Rogers, E. 1983 pp.6). The diffusion of technology creates change in the social system, how we live our lives. To summarize this, the diffusion process has been explained in four main elements which are 1. The innovation 2. Communication channels 3. time 4. social system (Rogers, E. 1983).

Some of the constructs used by some research, are taken by the IDT theory and are obviously from the characteristics of innovation. According to Rogers (1983). The characteristics of innovations can be perceived by individuals as it helps to explain the different rate of adoption, there are:

1. Relative advantage: is the degree to which an innovation is perceived as better than idea it supersedes, and it can be measure in economic terms

2. Compatibility: is the degree to which an innovation is perceived as being consistent with the existing values, past experiences, and needs of potential adopters

3. Complexity: is the degree to which an innovation is perceived as difficult to understand and use

4. Trialability: is the degree to which an innovation may be experimented with on a limited basis

5. Observability: is the degree to which the results of an innovation are visible to others

(Rogers, E. 1983 pp. 15; pp.16)

All in all, IDT theory from Rogers (1983) is the most cited work, it provides a synthesis of over 3000 previous studies of adoption and diffusion (Fichman, R.G., 1992). According to Oh, J.S. et al. (2014), Innovation Diffusion Theory or IDT helps explain the acceptance and diffusion of innovative services such as NFC technology from an academic perspective. Under the similar topic, an article from Longyara, T. & Van, H.T. (2015) used IDT to utilize the study of NFC mobile payment between South Korea and Thailand which is a cross-countries research. The result has shown that when customer perceived clear advantages offered by mobile NFC payment, they are likely to have positive attitude toward adopting (Longyara, T. & Van, H.T. 2015). Moreover, the user’s perceptions about compatibility of mobile payment with the experiences, ability, and needs, appear to be a predictor of attitude of using mobile payment. As confirmed by Jabri I.M & Sohail, M.S (2012), similar results and similar constructs were used, it has shown that relative advantage, compatibility, and observability have impacts on mobile banking adoption. In addition, a construct ‘Trialability’ has been tested in the hypothesis towards the attitude for using mobile payment by the research paper from Dash, M. et al (2014), as the article has used this construct with the early adopters, it has positive impact on the attitude of using the mobile payment. 48

However, our research will mainly focus on the current users which is not counting as the early adopters; therefore, we will consider this construct as not suitable for the current users. In the article from Dash, M. et al (2014), the compatibility, relative advantage and observability has a positive impact on the attitude of using mobile payment. All constructs from IDT such as relative advantage, compatibility, complexity, trialability and observability were all used as constructs in an article from Yeh, H. (2020), all of the IDT constructs have impact on the usage intention of using mobile payment. The interpretation of the results from the research paper of Yeh, H. (2020) has clarified that at the current level of development of mobile payment, the potential adopters were motivated by others in their social networks. Therefore, the social influence has a role in technological adoption. In addition to the use of IDT constructs, the research about the intention to use online financial transaction from Syahadiyanti. L, & Subriadi A.P. (2018) has shown that the relative advantages, compatibility and trialability influenced the intention to use the online financial transaction and meanwhile those three constructs are influenced by the cultures.

However, in order to explain how an individual decides to use mobile payment, instead of mentioning only how one gets motivated by others and this impact leads to the decision making of using mobile payment, there was an explanation from Rogers (1983), he stated “the innovation- decision process is the process through which an individual passes from first knowledge of an innovation to forming an attitude toward the innovation” (Rogers E. 1983 pp.20). No matter if it is the decision of reject, adopt or implement the decision, Rogers (1983) conceptualized five main steps in the innovation-decision process including 1. Knowledge 2. Persuasion 3. Decision 4. Implementation and 5. Confirmation. These steps could be helpful to help us understand the decision process of the mobile payment users and how these steps will guide us when we analyze the result and adapt suitable suggestions to the mobile payment developers in order to maintain the usage of the current users and attract more non-users.

Technology Acceptance research

In this chapter, we will concentrate on the summarization of articles in technology acceptance research, and the discussion of using Technology Acceptance Model (TAM) for IT relevant topics. It is necessary to mention how could Technology Acceptance Model be applied for different fields of research especially on the mobile payment topics.

Technology Acceptance Model or TAM is developed by Davis (1989) to explain the potential of user’s behavioral intention to use technological invention (King W.R & He, J. 2006). Originally, TAM is based on the Theory of Reasoned Action, a psychological theory which explains the behavior. Furthermore, TAM involved two predictors Perceived Ease of Use (PEOU) and

Perceived Usefulness (PU) (King W.R & He, J. 2006). TAM is one of the most widely used models in technology related topics and informatics because of its simplicity and easiness of use (King W.R & He, J. 2006; Yousafzai, S. et al. 2007a; Chhonker, et al. 2017). The popularity of TAM is broadly attributable to three factors including: 1. IT-specific, 2. It has strong theoretical base and validity in terms of psychometric measurement scales 3. It has accumulated the strong empirical supports (Yousafzai, S. et al. 2007b). As discussed by an article from Patil P.P. et al. (2018), the summarizes articles of 14 studies were using TAM and its extension. These studies adopted, adapted, and extended the TAM through variety of contexts (Patil P.P. et al. 2017).

Moreover, Yousafzai, S. et al. (2007a) clarified that the original TAM explained a person’s acceptance of technology is determined by his or her voluntary intentions toward using the technology. The intention is determined by a person’s attitude towards the use of technology and the users’ perception of its usefulness. Attitudes are formed from the beliefs of a person holds about technology. The PU was originally designed to be use in the organizational context such as the topics related to job performance but later it has been used for non-organizational contexts such as online shopping. It has been found that PU is influenced by PEOU (Yousafzai, S. et al. 2007a). In addition to the meta-analysis of Yousafzai, S. et al. 2007b, the findings from a collected 95 TAM studies shown that PU is the most found to be significant determinant of usage (82% of studies) , intention ( 90% of studies) and attitude (96% of studies). Similar to the research from Patil P.P. et al. (2017), the findings have supported the PU to be the most significant determinant of consumers’ behavioral intention to use mobile payments. However, the findings on PEOU are mixed, around 59% of the studies found PEOU to be significant determinant of usage, intention (67% of studies) and attitude (82% of studies).

A study from King W.R & He, J. (2006) has tested the construct reliabilities and the correlation of the relationship between constructs in TAM studies. The research has conducted the total of 88 TAM studies and it has found that all constructs from TAM are highly reliable (all constructs reliability is above 0.8). They are also strong correlation between PU and Behavioral Intention (BI), but less between PEOU and BI.

The findings from Yousafzai, S. et al. (2007b) has concerned about the impact of subject types and study setting on the attitude-usage relationship. Many of research measured the self- reported usage but not many of them will measure the actual usage (Yousafzai, S. et al. 2007a). Similar to the studies from King W.R & He, J. (2006), the research has focus on the types of usage and has categorized into 1. Job-related 2. Office 3. General (email and telecom) and 4. Internet and E-commerce. Across all the types of usage, it has been found that PEOU and BI effect is quite consistent, and the influence of PU on BI is profound.

In order to explain which factors should be considered when using TAM in mobile payment topics, according to the research from Yousafzai, S. et al. (2007b), there two different settings, the

individual and organizational settings. The findings have shown that individual has develop the intention to use technology because they perceived it as useful for their job performance, convenient, but it does not mean that they enjoy using those technologies – hence they might possess the negative attitude and not positive. The effect between attitude and mandatory usage would be more useful for the future research. It is suggested that in the organizational settings, it is necessary to include the attitude for the accuracy of data interpretation.

There are many fields that TAM and its constructs were used to explain the user’s behavior towards the innovation or technology. Research from Beldad, A.D & Hegner, S.M (2018) used TAM to see German users’ intention to continue use of a specific fitness application on mobile phone. Additional to PU and PEOU, the social norm and trust were used to see the relationship to the behavioral intention to use such applications. Moreover, besides only the applications on mobile phone, there are the adapted use of TAM on the topic of travelers’ adoption of technology.

According to the research from Diop, E.B. et al. (2019), they examined the understanding of travelers’ adoption of variable message signs, and the extension of TAM has applied, the amount of 761 drivers were interviewed, the original constructs, PU, PEOU, and Attitude have been tested along with other additional constructs such as information quality. Another field that has use TAM as the base theory is the educational topics. However, the topics are relevant to the technological adaptation such as E-learning and E-assessment, a research from Imtiaz, M.A & Maarop, N. (2014) examined that many of E-learning and E-assessment research did not use the original constructs from TAM but many of them mixed the constructs related to the topics such as Computer Self-Efficacy, Perceived Playfulness, Goal Expectancy and Content were used in the field of education based on TAM. Therefore, the main use of TAM and its constructs depend on the specific topics and background of each study.

Taking a consideration in our topic, the mobile payment acceptance research, we have summarized the use of TAM in those research papers. We observe how these research papers tried to use other constructs related to their topics, how do they use original TAM constructs with the other relevant constructs and what is the result of the research.

The research from Gbongli, K. et al. (2019) used extended TAM and its constructs to study the adoption of mobile-based money services and sustainability in developing countries. The total of 539 respondents has answered the surveys and it was found that PEOU is the most significant factor affecting consumers’ attitudes to mobile based money services while PU and personal innovativeness affect adoption decisions. More specific topic towards mobile payment such as the research from Cobanoglu, C. et al. (2015). The research topic is about the consumer acceptance of mobile payment technology in restaurant industry, the authors used TAM and its additional predictors which are relevant to the restaurant industry. The study from Cobanoglu, C. et al. (2015) revealed that compatibility with lifestyle was the strongest predictor of consumers’ intention to

adopt mobile payment technology in restaurants, following by PU, subjective norm, security and previous experience.

Another recent research from Wong W.H. & Mo. W.Y (2019) had studied consumer intention to use mobile payment in Hong Kong based on TAM, the results have shown that perceived risk, perceived trust, perceived security and all constructs from TAM have impacted on the consumer intention. With the support from the research paper from Shankar, A. & Datta, B. (2018), they have use the TAM constructs such as PU, PEOU, trust, self-efficacy to see the intention to use mobile payment in India, and the result has shown that all constructs were having positive impact on intention to use, only subjective norm and personal innovativeness are found to have no significant impact on the intention to use mobile payment.

In addition to the specific mobile payment application (more narrowed scope), a research from Li, J. et al. (2019) had examined about the mobile payment with Alipay. In this research paper, the authors used the extended TAM and have investigated using the users’ risk perception besides the normal constructs to original TAM (PU, PEOU, Attitudes and Intentions). At the end, the total users of 491 respondents reflected the result as their risk perception to the use of Alipay is high. The risk perception has a direct effect on PEOU and PU respectively. From this, it shows that many studies nowadays used extended TAM and other constructs related to their topics more than only considered the focus on original TAM.

Furthermore, the research from Lwoga E.T. & Lwoga, N.B. (2017) investigated the effects of user-centric, security and system characteristics and genders on behavioral intention to use mobile payment (the 292 mobile payment users were collected), the study revealed that compatibility, social influence, mobile payment knowledge determined PU, while compatibility, trust and mobile payment knowledge predicted PEOU. This study has revealed that genders played roles in the behavioral intention, where men are more intentionally wanted to use the mobile payment than women. Besides the quantitative research papers about user acceptance to use mobile payment, there is also some with focus group interviewed. The study from Dahlberg, T. et al. (2003) interviewed 61 users of mobile payment, TAM is applied to the study. Since the research is held in qualitative structure, making TAM unable to be explained in a statistical way, however, users found PU and PEOU relevant. The additional factors such as risk and trust were discussed that it has impacts to the attitudes in the studies.

Similar to the research from Nguyen, et al. (2016). The research investigated the consumer intention to use mobile payment services in Vietnam, the authors had chosen TAM and the evidence has shown that 489 consumers reflected that perceived trust is the strongest predictor of intention to use mobile payment services followed by PEOU, perceived enjoyment, perceived behavioral control, PU and subjective norm. In the study from Nguyen, et al. (2016), they have used other constructs other than the original TAM constructs, however, they have fulfilled the

previous research gaps and discussed that there is the gap to the future research with other moderators and also the connection of intention and the actual usage.

According to the research paper from Zhang, Y. et al. (2018), it tried to see the connection between the behavioral intention of the mobile payment with the actual usage across countries (between the USA and China). This study revealed that the extrinsic factors such as cultures, socioeconomic status, personality traits, demographics and genders have direct effects on mobile payment acceptance and usage. The authors used the TAM and UTAUT for this study but they have also used broader factors such as moderators and the other perspectives such as cultures and characteristics of people in different countries. The result has been discussed that the people from the USA are more risk-averse than the Chinese, making the result of constructs from TAM different from one another. The PU and PEOU had stronger effects on behavioral intention in the USA samples than the China samples. The social influence on behavioral intention was stronger in China than the USA samples.

All in all, the extended TAM is widely used in many studies depending on the topics and each construct were formed according to the objectives and what the authors wanted to do with the research topics and methods. Moreover, TAM is proved to be highly credible, but some studies still proved that there is still less connection between the intention to use and the actual usage of the new technology. Therefore, it is necessary for us that we will collect the actual usage of real users along with the intention to use mobile payment.

Development of Technology Acceptance Models

According to Momani, A. & Jamous, M. (2017), all technology acceptance theories are designed to measure the degree of acceptance and satisfaction to the individuals with the new technologies or information system, but each theory is different because each construct reflect on the different point of view and structure. However, there are many theories which are suitable to use between the technology acceptance and technology adoption. Many of them were used in the same way. We tend to focus more on the technology acceptance models since it is associate to our topic. The term diffusion of technology, according to Rogers (2003), the term diffusion refers to the stage of technology spreads to general use by an individual or an organization. Therefore, we should look at each theory and the extension of them.

The most important and famous used theories are as follows: The Diffusion of Innovation Theory (IDT) was introduced by Rogers (1960). The Theory of Reasoned Action (TRA) – this theory was developed by Ajzen and Fishbein (1975). TRA has later extended to the Theory of Planned Behavior (TPB) – developed by Ajzen (1985, 1991), which later had also extended to 53

Decomposed Theory of Planned Behavior (DTPB) – developed by Taylor and Todd (1995).The information system had a contribution to the Technology Acceptance Model (TAM), which is an extension of TRA, the extension of TAM is known as TAM2 and it has been developed by Venkatesh and Davis (2000). The extension of TAM2 has been introduced as Unified Theory of Acceptance and Use of Technology (UTAUT) by Venkatesh et al. (2003). TAM3 was also extended again by Venkatesh and Bala (2008). Moreover, there is the combination of TAM and TPB called C-TAM-TPB. The Model of PC Utilization (MPCU) was introduced by Thompson et al. (1991). The Motivational Model (MM) was developed by Davis et al. (1992). The Social Cognitive Theory (SCT) was introduced by Bandura (1989). Lastly, the Model of Acceptance with Peer Support (MAPS) was introduced by Sykes et al. (2009) (Lai, P.C. 2017; Momani, A. & Jamous, M. 2017; Sharma, R. & Mishra, R. 2014).

To illustrate the development of relevant theories, the figure 6 below shows the chronological path of the psychological studies and social studies area. This chronological path did not include other theories that we have mentioned above, however, in addition to the UTAUT model, it is the extended theory which should be in the same path but after the TAM2 in figure below. According to this path (from TRA to TAM2), we will use these theories to develop our suitable models to our research.

Figure 6: Chronological Graph for the evolution of Technology Acceptance Theory, source: Momani, A. & Jamous, M. (2017)

Year Theory Developed by

1960 IDT Rogers

1975 TRA Ajzen and Fishbein

1985, 1991 TPB Ajzen

1986 TAM Davis

1989 SCT Bandura

1991 MPCU Thompson et al.

1992 MM Davis et al.

1995 DTPB Taylor and Todd

C-TAM- 1985 TPB Taylor and Todd

2000 TAM2 Venkatesh and Davis

2003 UTAUT Venkatesh et al.

2008 TAM3 Venkatesh and Bala

2009 MAPS Sykes et al.

Table 2: Summarization of Technology Acceptance and Adoption Models and Theories (Source: Self-Edited)

According to the table 2 above, we have summarized all relevant technological adoption and acceptance theories from the past to the present. We will refer to these theories and explain the constructs and definition of each theory in the following sub-chapters. According to Chhonker, et al. 2017, most of the theories used to form the constructs in many researches related to mobile payment topics refers to the green highlighted rows above. Therefore, the theories which were highlighted will be explained in our next sub chapters. The discussion of which theories would be the suitable theories to develop our constructs and hypotheses will be based on the research gaps and findings.

Innovation Diffusion Theory (IDT)

We will take look into the constructs in Innovation Diffusion Theory (IDT) or so-called Diffusion of Innovation Theory (DOI) first since it’s relevant to technology adoption and

acceptance. The IDT has originally been introduced in 1962 by Everett Rogers, it has focuses on the understanding of how, why and at what rate innovative ideas and technologies spread in a social system (Rogers, 1962 as cited by Wani, T. 2015).

There are characteristics of an individual whether the technology or new innovation could be adopted or not, by refer to Rogers (1983), there are five adopter categories (the classification of members of social system based on their innovativeness), this includes 1. Innovators 2. Early adopters 3. Early Majority 4. Late Majority 5. Laggards. The innovators are active information seekers about new ideas. (Rogers,1983 pp.23). Later Rogers and Shoemaker (1971) has observed that there are five attributes of an innovation are largely involved to influence the adoption of an innovation it is

1. Relative advantage – this refers to a powerful concept for diffusion of an innovation meaning that a person is willing to compelled to adopt to new innovation when it’s more advantageous to use.

2. Compatibility – is the extent to which adopting the innovation is compatible with what people do or the degree to which an innovation is perceived as consistent with consumer needs, values, and beliefs

3. Complexity – this refers to an innovation perceived as difficult to understand and use

4. Observability – it is the easiness of new innovation to the perspective users or put in easy words, the current users have given reviews on the new innovation and products

5. Trainability – the possibility to which the products could be tested before purchasing or using such as test-drive for cars (Wani, T. 2015). Those are found to be the constructs when doing the research relates to the new innovation.

According to Rogers (1983), diffusion refers to “The process by which an innovation is communicated through certain channels over time among the members of a social system.” (Rogers, E. 1983 pp.5). In addition, he defined communication as “A process in which participants create and share information with one another in order to reach a mutual understanding” (Rogers, E. 1983 pp.5). Innovation has also defined as “An idea, or object that is perceived as new by an individual or other unit of adoption” (Rogers, E. 1983 pp.11). There are in total four elements in the IDT theory, 1. Innovation 2. Communication channels 3. Time 4. Social System. The characteristics of innovation as mentioned above, could be perceived by the perceivers as having more relative advantage, compatibility, trialability, observability and less of complexity. With this format, the perceivers tend to adopt the innovation more rapidly than other innovations. (Rogers, E. 1983).

When a person passes from first knowledge of an innovation to form an attitude towards new technology or new innovation, there is a decision to make whether it is to adopt or to reject it. The decision process involves implementation of new ideas and the confirmation of the decision. Rogers (1983) conceptualized the five main steps of the decision process, there are

1. Knowledge: It occurs when an individual is exposed to the innovation’s existence and gains some understanding of how it functions.

2. Persuasion: This happens when a person formed favorable or unfavorable attitude towards an innovation.

3. Decision: This leads to when a person engaged in activities that lead to a choice to adopt or reject an innovation.

4. Implementation: This happens when a person puts an innovation into use.

5. Confirmation: It happens when a person seeks reinforcement of an innovation decision that has already been made, however, he or she can reverse the previous decision if exposed to conflicting messages about innovation. (Rogers, E. 1983 pp.20,21)

Theory of Reasoned Action (TRA)

First of all, by initially taking a look at theory of reasoned action (TRA) in figure 7 below, the model has developed by Ajzen and Fishbein in 1975;1980. The TRA is an explanation of human’s attitude towards the behavior that produces outcome and the evaluation of outcome. Together behavior and the evaluation outcomes became the sum of behavioral beliefs. According to Ajzen (1991), people tend to hold several behavioral beliefs in relation to any given behavior. In general, TRA assumed that those behavioral beliefs and outcome evaluations combined to produce positive or negative attitude toward the behavior (Ajzen, I. 1991).

Moreover, Ajzen (1991) refer to the ‘Subjective norm’ as in the same construct from Dulany’s (1968) behavioral hypotheses term ‘Normative Beliefs’. In TRA, the definition of subjective norm is described by overall perceived social pressure which is the sum of strength of each normative belief weighed by the motivation to comply (Ajzen, I. 1991, pp. 443). The subjective norm could be influenced by many factors for example, economy, politics, society, and demographic factors (Ajzen, I. 1991).

The TRA has been later widely used to see the prediction of attitude towards behavior in the many areas for example to see attitudes in job satisfaction and performance, the attitude towards ethnic group, institutions, racial and policies (Ajzen, I. 1991). TRA also went beyond an impact of personal attitude by considering the role of perceived social norm in relation to behavior 57

interests. However, the TRA model is not able to explain behaviors which are spontaneous, impulsive, or habitual or the results due to cravings since it is counted as unconscious decision (Pathirana, P.A. & Azam, S.M. 2017). The model tends to work under some circumstances and not to all situation.

Figure 7. Theory of Reasoned Action (TRA) Source: Ajzen & Fishbein (2010) edited by Otieno, O. 2016

Theory of Planned Behavior (TPB)

There are many theories that are related and has been developed to explain human behavior regarding to the topic of human behavior adopting technology and intention to use. According to the theory of planned behavior (TPB) by Icek Ajzen (1991) has been used to know what perceived behavioral controls that influence the behavioral performance (Ajzen, I. 1991). The TPB has been developed from the reason actioned assumptions since 1966 and most of theories as the bases, for example, Bandura’s social cognitive theory (1986, 1997), Triantis’s theory of subjective culture and interpersonal relations (1972), the health belief model by Rosenstock 1994, goal setting theory by Locke and Latham, 1994, the information-motivation-behavioral skills model by Fisher and Fisher (1992) and the technology acceptance model (TAM) by Davis et al. 1989.

The behavioral control definition, according to Ajzen (1991), has referred to “Extent to which people possess the requisite information, mental and physical skills and abilities” (Ajzen, I. 1991 pp.446). Moreover, he claimed that the degree of actual behavioral control is expected to moderate the effects of intentions on behavior. When an individual intends to perform the behavior and that person has a high level to control it, he or she is most likely to perform it (Ajzen, I. 1991).

Besides the understanding of behavioral control, the perceived behavioral control is mentioned in the theory of planned behavior (TPB) as a new factor, it’s the “extent to which that people believe they perform a given behavior if they are inclined to do so” (Ajzen, I. 1991 pp.446). The perceived behavioral control in TPB relates to the work of Bandura’s cognitive theory about 58

self-efficacy. The term perceived behavioral control is focusing on the extent to which people believe that they are capable of or have control over, performing a given behavior, which is similar to self-efficacy. (Ajzen, I. 1991).

To summarize the TPB, according to Ajzen (1991) is a general model designed to be applicable to any behavior, and not just only for an individual motivated to perform. The theory is containing 3 constructs, which are, attitude towards behavior, subjective norm and lastly the perceived behavioral control to moderate the behavioral intention and the perceived behavioral control to moderate to the actual behavior. Moreover, Theory of planned behavior (TPB) as seen in figure 8 below has applied for the understanding of individual acceptance and usage of many different technologies (Momani, A. & Jamous, M. 2017).

Figure 8: Theory of Planned Behavior (TPB) Source: Ajzen, 1991 edited by Li, L. 2010

Decomposed Theory of Planned Behavior (DTPB)

In addition to TPB, the extended theory towards the planned behavior of human using technologies has been named as Decomposed theory of planned behavior (DTPB) in the figure 9 as shown below where the studies by Taylor and Todd (1995) have decompose the attitude toward behavior, subjective norm and perceived behavioral control (basically all factors from TPB) into the multi-dimensional belief constructs within technology adoption context (Momani, A. & Jamous, M. 2017). The DTPB has expanded the TPB by adding another 3 factors from Innovation Diffusion Theory (IDT) viewpoints which are relative advantage, compatibility and complexity (Momani, A. & Jamous, M. 2017). Taylor and Todd (1995) stated that “the DTPB model draws upon constructs from the innovations characteristics literature and more completely explores the dimensions of subjective norm (e.g. social influence)” (Taylor, S. &Todd, P. 1995a. pp.147). The combination of IDT, TAM and TPB gives the DTPB model many advantages similar to those models combined

especially when the topic of research is associated to specific beliefs that may influence IT usage (Taylor, S. &Todd, P., 1995a).

According to their studies from Taylor and Todd (1995), the TRA and TPB are good in predating the behavior, however, the DTPB is proved to be more effective in order to explain about behavior. Another study by Taylor and Todd (1995) was to compare the TPB and DTPB to the Technology Acceptance Model (TAM) to gain more effectiveness in application of DTPB in technology usage. For example, they took an exchange of complexity from IDT by the ease of use from TAM, it has also exchanged the relative advantage from IDT by perceived usefulness from TAM (Momani, A. & Jamous, M. 2017).

In addition to the DTPB model by Taylor and Todd (1995), they claimed that the model itself should become clearer and more readily understood because the decomposition can provide a stable set of beliefs which can be applied across variety of settings. The model itself focus on the specific beliefs and become more managerially relevant, more complex, pointing to specific factors and larger factors that may influence technology adoption and usage (Taylor, S. &Todd, P. 1995a).

Figure 9: Decomposed Theory of Planned Behavior (Source: Taylor, S. &Todd, P. 1995a pp. 146, self-edited)

Technology Acceptance Model (TAM)

Technology Acceptance Model (TAM) has been widely used and most cited in the studies on information system (IS) as well as ICT with respect to adoption of a new innovation due to its simplicity and its easiness of use (King W.R & He, J. 2006; Yousafzai, S. et al. 2007a; Chhonker, et al. 2017; Momani, A. & Jamous, M. 2017). Davis (1989) had studied and presented TAM to the

objective of predict and explain the usage behavior of ICT users. To be precise, TAM is the extended theory after the TRA again, TAM replaced attitude factor from TRA by the two factors which are perceived usefulness (PU) and perceived ease of use (PEOU) as seen in figure 10 below and it didn’t include the subjective norm in its theory (Momani, A. & Jamous, M. 2017).

Figure 10: Technology Acceptance Theory (TAM) Source: Davis (1989)

According to Davis (1989), among many variables that determine whether people will reject or accept the information technology, there are two determinants that are useful, the perceived usefulness which refers to “the degree to which a person believes that using a particular system would enhance his or her job performance” (Davis, F.D. 1989. pp.320). Moreover, he clarified that people tend to use or not use an application is totally related to the idea where people find it helpful for their job or not. The second variable, the perceived ease of use (PEOU) has defined as “the degree to which a person believes that using a particular system would be free of effort” (Davis, F.D. 1989. pp.320), for the PEOU, he clarified that if a potential users believe that a given application is useful, however, at the same time they finds it hard to use and that his or her effort of using it to increase the performance are not worth what they are doing, it could have an effect on the attitude towards the usage.

The development of TAM makes this theory less general to TPB and TRA, and more focused on technology. The development of TAM are including 3 different phases, including adoption, validation and extension, the adoption is where it has been tested and adopted through huge numbers of applications , in the validation phase, many researchers are noting that TAM uses accurate measurement of user’s acceptance behavior in different technologies, the third phase is the extension where many researchers introduce many variables and factors and relationships between the TAM constructs (Momani, A. & Jamous, M. 2017). Furthermore, there is an extension of TAM, it is the TAM2 where it tries to explain the perceived usefulness and the perceived ease of use by the social influence and cognitive instrumental processes viewpoints, the social influence processes are referring to the subjective norm, voluntariness, and image while cognitive instrumental processes are referring to job relevance, output quality, result as demonstrability and perceived ease of use (Momani, A. & Jamous, M. 2017).

Combined TAM and TPB (C-TAM-TPB)

C-TAM TPB is the combination of factors from TAM and TPB model, this model has been formed and tested in the study about the usage of computing resource center (CRC) by business school students from Taylor and Todd (1995). The importance which makes C-TAM TPB different from a normal research is that they tried to divide the subgroup of technological users into two categories including the experienced users and inexperience users (Taylor, S. & Todd, P. 1995b). It is implying that IT usage may be more effective modeled for the experienced users, moreover, they assumed that there may be differences between experienced and inexperienced users in the relative influence of the various determinants of IT usage (Taylor, S. & Todd, P. 1995b). Based on the constructs of C-TAM TPB in figure 11 below, Taylor and Todd (1995) have anticipated the differences of experienced and inexperienced group based on the moderation of each pair of constructs.

Figure 11: Combined TAM and TPB (Source: Taylor, S. & Todd, P. 1995b, self-edited)

It is assumed that the direct experience will result in a stronger and stable behavioral intention to actual behavior relationship (Ajzen and Fishbein, 1980 as cited by Taylor, S. & Todd, P. 1995b). It is expected that for experienced users, behavioral intention will fully mediate the relationship between perceived behavioral control and actual behavior (Taylor, S. & Todd, P. 1995b). Moreover, it is expected that attitudes correlate more strongly with behavior for people who had direct experience with an object, similar to the relationship between subjective norm to behavioral intention, where the relative influence of subjective norm on intentions is expected to be stronger with inexperienced users since they are willing to rely on the reaction of others (Taylor, S. & Todd, P. 1995b). Lastly, the perceived ease of use and perceived usefulness to attitude may have different relative influences, depends on the experience of the users. It is suggested that the experience users might overcome the perceived ease of use and focus on the attention on perceived usefulness and the perceived ease of use to attitude would have stronger influence

especially for the inexperience users since they intend to use technology when they find it easy to use (Taylor, S. & Todd, P. 1995b).

Technology Acceptance Model 2 (TAM2)

TAM2 is the model that has extended the original TAM model. TAM2 has introduced by Venkatesh and Davis (2000) and they claimed that TAM2 encompasses social influence process including subjective norm, voluntariness, image and cognitive instrumental processes which are job relevance, output quality, result demonstrability and perceived ease of use as determinants of perceived usefulness and usage intention (Venkatesh, V. & Davis, F.D.2000).

The constructs are shown in figure 12 below, the voluntariness has shown that in the computer usage context, the direct effect of subjective norm on intention over perceived usefulness and perceived ease of use will happen in mandatory and not voluntary (Venkatesh, V. & Davis, F.D.2000). Therefore, the voluntariness is a construct which determine the differences between voluntary and mandatory in the usage. To compared the relationship between subjective norm and intention, from the point of view in TRA and TPB model, it is based on compliance meanwhile in TAM2, it is encompassing two additional mechanisms which means subjective norm can influence intention indirectly through perceived usefulness in two ways, internalization and identification (Venkatesh, V. & Davis, F.D.2000). The internalization means if the superiors believe that a particular system is useful, a person may get convinced that it is useful (Venkatesh, V. & Davis, F.D.2000).

For the identification, it is associated to the ‘image’ construct, where TAM2 claimed that subjective norm will positively influence image because it is important to the person’s social group at work, this is one-way people can show their identity to fit in the social group (Venkatesh, V. & Davis, F.D.2000). For the construct experience, it is suggested that an increased system experience will directly affect the subjective norm on intentions and subjective norm on perceived usefulness (Venkatesh, V. & Davis, F.D.2000).

Figure 12: TAM2 Source: Venkatesh and Davis (2000) pp. 188

According to Venkatesh and Davis (2000), they theorize the four cognitive instrument determinants of perceived usefulness which are job relevance, output quality, result demonstrability and perceived ease of use. The job relevance, output quality and result demonstrability are expected to have influence on the perceived usefulness in the model (Venkatesh,V. & Davis,F.D.2000).

There are reasons why they introduced these new constructs except the perceived ease of use which is taken from original TAM. Firstly, job relevance is defined as “individual’s perception regarding the degree to which the target system is applicable to his or her job” (Venkatesh,V. & Davis,F.D.2000 pp.191). Which means the level that the system will support a worker. For the output quality, they explained over and above considerations of what tasks a system is capable of performing and the level that those tasks match the job goals – people will see how well the systems could perform and this perception is known as ‘output quality’ (Venkatesh,V. & Davis,F.D.2000).

Lastly, result demonstrability which expected to influences perceived usefulness, it is explained that if people find it hard to see the gains in their job performance by using the system, they will reject the use of it (Venkatesh, V. & Davis, F.D.2000). All in all, TAM2 proved detailed account of key forces underlying judgments of perceived usefulness. TAM2 extends TAM by showing that subjective norm has direct effects on usage intentions over and above perceived ease of use and perceived usefulness for mandatory systems (Venkatesh. & Davis,F.D.2000).

Technology Acceptance Model 3 (TAM3)

TAM3 model was introduced by Venkatesh and Bala (2008). The model itself derived from TAM2 which has proposed by Venkatesh and Davis (2000) with more constructs and details. TAM2 theorized about the general determinants of perceived usefulness which are subjective norm, image, job relevance, output quality, result demonstrability and perceived ease of use, with additional of two moderators that is experience and voluntariness (Venkatesh, V. & Bala, H. 2008). TAM2 presents a complete network of determinants of individuals’ IT adoption and usage (Venkatesh, V. & Bala, H. 2008). Moreover, TAM3 is the combination of TAM2 and model of determinants of perceived ease of use (Venkatesh, V. & Bala, H. 2008). The determinants of perceived ease of use contain 6 constructs including computer self-efficacy, perception of external control, computer anxiety, computer playfulness, perceived enjoyment, and objective usability as seen in figure XX below.

Computer self-efficacy is explained by Comeau & Higgins (1995) as cited by Venkatesh and Bala (2008) as the level of beliefs of an individual to perform a specific job using computer. Next, the perception of external control is the way an individual believes that organizational/technical resources exist to support the use of a system (Venkatesh, V. & Bala, H. 2008). The third construct, computer anxiety refers to the fear of an individual when he or she faced with the possibility of using computers (Venkatesh, V. & Bala, H. 2008). The construct computer playfulness refers to the level of individual’s cognitive interaction with computer (Venkatesh, V. & Bala, H. 2008). Fifth construct is the perceived enjoyment which referred to the activity of using a specific system is perceived to be enjoyable besides the performance resulting from the system use (Venkatesh, V. & Bala, H. 2008). Lastly, objective usability refers to the comparison of systems regarding the performance to complete a specific task (Venkatesh, V. & Bala, H. 2008).

In TAM3 model, they claimed that the determinants of perceived usefulness will not influence perceived ease of use and determinants of ease of use will not influence perceived usefulness (Venkatesh, V. & Bala, H. 2008). Therefore, it is stated that TAM3 has no cross-over effect. The thick lines that have shown below in figure 13 indicates the new relationship proposed in TAM3.

Figure 13: TAM3 Source: Venkatesh, V. & Bala, H. 2008 pp. 280)

TAM3 model posits three relationships that were not tested in the TAM2 model before, Venkatesh and Bala (2008) suggested that the experience will moderate the relationships between 1. Perceived ease of use and perceived usefulness 2. Computer anxiety and perceived ease of use and 3. Perceived ease of use and behavioral intention. For the first relationship, it is expected that when an individual has an increasing hands-on experience in a system, he or she will have more information on how easy or hard the system is to use (Venkatesh, V. & Bala, H. 2008). They based this argument on action identification theory from Vallacher & Kaufman (1996) from the high- and low-level action identities, where high level identity relates to an individual’s goal and plans, low level identity referred to the means to achieve goals and plans (Venkatesh, V. & Bala, H. 2008). Furthermore, they suggested that with an increasing experience, the influence of

perceived ease of use (considered low-level identity) on perceived usefulness (high level identity) will be stronger (Venkatesh, V. & Bala, H. 2008). Next relationship, computer anxiety to perceived ease of use, moderated by experience. They have expected that with increasing experience, system-specific beliefs will be stronger determinants of perceived ease of use of a system (Venkatesh, V. & Bala, H. 2008). Lastly, the relationship of perceived ease of use to behavioral intention moderated by experience, they have expected that experience will moderate the effect of perceived ease of use on behavioral intention and the effect will be weaker with an increase of experience (Venkatesh, V. & Bala, H. 2008).

Unified Theory of Acceptance and Use of Technology (UTAUT)

Figure 14: Unified Theory of Acceptance and Use of Technology Model (UTAUT) source: Venkatesh et al. (2003). pp.447

Many articles have been cited using the Unified Theory of Acceptance and Use of Technology Model (UTAUT) developed from Venkatesh et al. (2003). The UTAUT has being an extension of many theories combined together namely TRA, TAM, Motivational Model, TPB, and a combination of TPB and TAM. However, UTAUT tries to challenge the contextuality by adding new independent variables such as individual characteristics, situational variables, and organizational characteristics (Otieno, O. 2016).

The figure 14 above shown the UTAUT model which explains about the behavioral intention and intention to use new technologies and it’s widely used to test the hypotheses with many new trends of technology, UTAUT consists of 4 constructs that are playing as a significant role as direct determinant of user acceptance and usage behavior, this includes performance expectancy, effort expectancy, social influence and facilitating conditions (Venkatesh et al. 2003).

The dependent variables are the behavioral intention and use behavior, other moderators are gender, age, experience, and voluntariness of use which have been influencing directly to the relationship between dependent variables and independent variables (Pathirana, P. & Azam, S. 2017).

The first construct, performance expectancy is defined as “the degree to which an individual believes that using the system will help him or her to attain gains in job performance” (Venkatesh et al. 2003, pp. 447). The performance expectancy has combined 5 constructs altogether from different models which includes perceived usefulness from TAM/TAM2 and C- TAM-TPB, extrinsic motivation from MM, job fit from MPCU, relative advantage from IDT, and outcome expectations (SCT) (Venkatesh et al. 2003).

Next construct, effort expectancy, has referred to “the degree of ease associated with the use of the system” (Venkatesh et al. 2003, pp. 450). The concept of effort expectancy was captured by three constructs from different models, the perceived ease of use from TAM/TAM2, complexity from MPCU, the ease of use from IDT (Venkatesh et al. 2003). The social influence is the third construct and is defined as “the degree to which an individual perceives that important others believe he or she should use the new system” (Venkatesh et al. 2003, pp. 451). Social influence is represented as subjective norm in TRA, TAM2, TPB/DTPB and C-TAM-TPB, social factors in MPCU and image in IDT. Last construct of UTAUT called facilitating conditions, where it referred to “the degree to which an individual believes that an organizational and technical infrastructure exists to support the use of system” (Venkatesh et al. 2003, pp. 453). The concept of facilitating conditions embodied by three constructs namely perceived behavioral control from TPB, DTPB, C-TAM-TPB, facilitating conditions from MPCU and compatibility from IDT (Venkatesh et al. 2003).

However, the findings from Williams et al. (2015) shows that the most frequently used external variables are self-efficacy following by attitude and trust. There are most used variables across UTAUT and TAM which are: self-efficacy, personal innovativeness, subjective norms, voluntariness, computer anxiety, compatibility and relative advantage (Williams et al. 2015). UTAUT model provides a broader view of the user acceptance of information systems and subsequent usage behavior (Pathirana, P. & Azam, S. 2017). UTAUT is widely used due to it can be applied to the most important factors which could be used when convincing customers to adopt any new technology (Pathirana, P. & Azam, S. 2017). The weakness of using UTAUT is when there are so many variables due to many extensions of model it has been used and also many combination – this leads to the theory to lose its strength or to state that the theory is becoming weaker in its clarity (Otieno, O. 2016).

Unified Theory of Acceptance and Use of Technology 2 (UTAUT2)

The Unified Theory of Acceptance and Use of Technology 2 (UTAUT2) was an extended model from original UTAUT. The differences are that the UTAUT is mostly focus on the context of the technological use in both organizational and non-organizational setting. Where UTAUT2 added more constructs and more relationships aimed to suit the consumer use context (Venkatesh, et al. 2012). There are three additional constructs identified from the prior research both general adoption and use of technologies, and consumer adoption and use of technologies (Venkatesh, et al. 2012). The UTAUT2 has altered some of existing relationships in the original concept of UTAUT and introduce new relationship (Venkatesh, et al. 2012) – this will be explained below.

The objective of the extension version of UTAUT was to pay attention to the consumer use context and develop UTAUT2. The new constructs, hedonic motivation, price value, and habit brings new mechanism such as affect monetary constraints and automaticity and had tied to the new constructs into the largely cognition and intention based UTAUT (Venkatesh, et al. 2012).

The first new construct in UTAUT2 is called hedonic motivation, it is defined as fun or pleasure derived from using technology (Venkatesh, et al. 2012). According to Brown and Venkatesh 2005 as cited by Venkatesh, et al. (2012), hedonic motivation plays an important role in determining technology acceptance and use. Next construct, price value, it is said to be important to consider the different contexts between consumer and technological settings where consumers will bear the cost of using such technology by themselves and the employees do not (Venkatesh, et al. 2012). Therefore, the cost and pricing structure may have a significant impact on consumers’ technology use (Venkatesh, et al. 2012). The third construct, experience and habit, experience reflects on an opportunity to use a target technology and operationalized as the passage of time from the initial use of a technology by an individual (Venkatesh, et al. 2012).

According to the figure 15 below we will refer to the new established relationships of UTAUT2. There are in total five different relationships. Firstly, the impact of facilitating conditions moderated by age, gender, and experience is the first change from the original UTAUT. There is direct relationship from facilitating conditions to behavioral intention over and above the existing relationship between facilitating conditions and technology use (Venkatesh, et al. 2012). It is expected that the effect of facilitating conditions on behavioral intention to be moderated by age, gender and experience (Venkatesh, et al. 2012). Moreover, the age, gender and experience have joint impact on the linkage between facilitating condition and intention to use (Venkatesh, et al. 2012).

Figure 15: UTAUT2 Source: Venkatesh, et al. 2012 pp. 160

The second relationship is the impact of hedonic motivation is moderated by age, gender and experience. It is expected that effect of hedonic motivation on behavioral intention to be moderated by age, gender, and experience because of the differences in consumers’ innovativeness, novelty seeking and perceptions of novelty of a target technology (Venkatesh, et al. 2012). Next relationship is the impact of price value moderated by age and gender. It is expected of the price value on behavioral intention to be moderated by age and gender, because the authors draw from theories of social roles from Bakan (1966) and Deaux and Lewis (1984), it is theorizing about the differential importance of price value among men and women and among younger and older person (Venkatesh, et al. 2012). Next relationship is impacts of habit moderated by age, gender, and experience. It is stated that individual differences in information processing and association in memory may play an important role in moderating the effect of habit (Venkatesh, et al. 2012). Lastly, the impact of behavioral intention moderated by experience. IT is expected that the effect of behavioral intention on technological use will decrease as experience decreases (Venkatesh, et al. 2012).

Discussion

All the technology acceptance theories and extended models are designed to predict the individuals’ behavior intention to use and accept the new technology or information system, the prediction and measurement have been researched through different theories and different

models as mentioned above, by using different variables depending on which problems that those theories and models wanted to answer (Momani, A. & Jamous, M. 2017). For the Theory of Reasoned Action (TRA), the original theories predicted the specific behavior, attitude and intention to be agreed on action, target, context and time frame – this is still limited due to no mention about other variables that affected in behavioral intention such as fear, mood, previous experience and the context is still too broad or too general for the technological issues (Momani, A. & Jamous, M. 2017). In TPB, it has been argued that since it’s the extension of TRA, the TPB still lack of the variables that’s affected the behavioral intention and also TPB does not take economic, or environmental factors into account – similar weaknesses apply to the extended TPB or DTPB (Momani, A. & Jamous, M. 2017).

For the most relevant theory to technological adoption to behavioral intention would be Technology Acceptance Theory (TAM), there are many points to be considered when applying for this model in our thesis. The limitations of TAM are that it does not provide the feedback on factors that could enhance adoption such as integration, flexibility, completeness of information and information currency (Momani, A. & Jamous, M. 2017). Another limitation is that the TAM does not considered number of different subjective factors such as values of the societies, personal attributes and personal traits (Ajibade, P. 2018). Therefore, the argument that a relative and friends could influence the use of technology through social pressure is highly falsifiable (Ajibade, P. 2018). TAM and TAM2 do not specify how expectancies are influencing on behavior some arguments shown that the findings of TAM is sometimes opposite to the original findings (Momani, A. & Jamous, M. 2017).

According to Legris et al. (2003) as cited by Long Li (2010), performed meta-qualitative analysis and they found that TAM fails to predict in many studies for example, Perceived Usefulness, which indicates from Davis (1989) that it’s the strongest predictor of an individual’s intention to use an information technology. However, Jackson et al. (1997) found no relation between Perceived Usefulness and Attitude (Li, L. 2010). Jackson et al. (1997) and Lucas & Spitler (1999) find no empirical evidence to support the relation between Perceived Usefulness and Behavior Intention (Li, L. 2010). The Perceived Ease Of Use is suggested by TAM that it has significant influence on perceived usefulness, attitude, intention and actual use (Davis, F. 1989). In UTAUT, Venkatesh (2003) use the construct of effort expectancy to capture the concepts of perceived ease of use (TAM/TAM2), complexity and ease of use (Li, L. 2010). However, many researchers found no empirical evidence to support to relation between perceived ease of use and actual use (Chau and Hu ,2001; Bajaj and Nidumolu, 1998; Hu et al., 1999; Jackson et al. 1997; Subramanian, 1994 as cited by Li, L. 2010).

There are more arguments about the attitude towards using technology, in some context for example, in the organization, the attitude simply wouldn’t affect the technological use since

there is a new system and also the technology that has just purchased, staffs are required to update the IT knowledge and ability to use new system (Ajibade, P. 2018). Obviously, the use of IT isn’t based on the perceived ease of use or attitude but the organizational culture and policies which encourage the training to staffs to increase the ability to use new system (Ajibade, P. 2018). The context of TAM and relevant theories could not apply totally to the organizational context. In our case, the challenge of the thesis would be the right samples that do not have company policies and other constraints involved in the use of individual mobile payment. This is also a challenge when it could be possible if one would want to use mobile payment application but the company culture and policies that they are working in requires different payment methodologies.

Another relevant factor, the social influences or namely subjective norm is in the TAM and TAM2 are stated to have an effect on the intention to use technology (Venkatesh and Davis ,2000 as cited by Li, L. 2010). Some researchers find that social influences have inconsistent roles in the empirical studies, to support this argument, Lewis et al. 2003 find no empirical support to say that subjective norms are important in understand of individuals’ choice to use IT (Li, L. 2010). The social norms could work for some people, but not for others and this is still a research gap for any future studies (Li, L. 2010). There are many limitations to the research related to TAM and UTAUT, Williams et al. (2015) stated that the limitations are that TAM and UTAUT only focus on a few variables and lack of the focus on variables should be broader and related to differences such as community, culture, organization, agency, department, person, age group. It is totally understandable that many regions and many countries tend to have different beliefs, social structures, subjective norms and cultures that are not the same and therefore the result could also vary and difficult to compare. In our research topic, it is necessary to consider cultures as it relates to each country (between Vietnam and Thailand), also the background and lifestyle of people in each country in order to understand the models and theories in the country basis.

Research gaps in mobile payment research

Research gaps in general scope

In general findings, there are several results appeared from the relevant articles, those articles have focused on the scope of behavior intention to use mobile payment, there are interesting and different factors, differences and aspects concerning to the intention to use mobile payment in the several articles, we will compare and discuss over the cross countries findings to see what makes those factors similar or different. Later we will look through the recent findings based on each country including Vietnam and then Thailand.

Initially, we would like to have an overall view on the relevant research in emerging market, especially Asian area. The research from Tan, K. et al. (2019) compared the factors influencing on two ethnic groups of people in Malaysia, namely the Malay and the Chinese to see whether the ethnicity will make any difference towards the intention to use mobile payment because they stated that ethnicity is predicted to significantly affect the level of participation in the use of technology. In addition, there are significant shortages in the literature that articulates intention towards mobile payment systems across ethnic groups and culturally diverse country (Mohd Suki and Ramayah ,2010 as cited by Tan, K. 2019). The sample size of 311 respondents have completed the surveys, the results have shown that Chinese and Malay people are motivated differently towards the adoption of mobile payment system (Tan, K. 2019). Friends and community (social influences/ subjective norm) have little effect within the Malay community but the perceived behavioral control is being the strongest predictor to intention (Tan, K. 2019). Comparing to mainland China, the subjective norms influence behavior intention stronger than the USA (Zhang, Y. et al ,2018). In Qatar, social influence also effects the intention to use mobile payment especially female users and younger generation tend to have stronger impact on social influences more than male users (Khan, H. et al. 2015). Similar to South Korea, social influences also effect the usage intention for mobile payment (Lin, X. et al., 2019).

Chinese consumers are willing to use mobile payment systems if they hold positive attitude, therefore the attitude is the strongest predictor to intention (Tan, K. 2019). This is the Chinese consumers who live in Malaysia, however, what about the mainland Chinese? To compare this, an article from Teng, P. et al (2018) shows the findings about factors influence customer intention to use mobile payment service in Nanjing, China, they have used the qualitative methodology (interviews) to gather information over 612 respondents, the TRA and TAM theories were applied in the research paper (Teng, P. et al ,2018). The findings show that attitude is the main factor which influence customer to adopt to mobile payment service (Teng, P. et al ,2018). This support the findings from Chinese ethnicity in Malaysia which attitude is the strongest predictor to the intention to use mobile payment. More interesting result shows that consumers are more likely to use mobile payment service because there’s existence intention which affects their behavior. (Teng, P. et al ,2018). Moreover, most respondents perceive mobile payment as useful tool since it is more convenient more than cash (Teng, P. et al ,2018).

The perceived risk is another result in China shows that most of consumers are confident toward protecting their privacy information by the providers, more over people are easily influenced by society such as government, social media, friends and family (Teng, P. et al ,2018) – which is opposite to the USA where people are concerned more about their personal privacy, their perceived risk effects on the behavior intention to use mobile payment (Zhang, Y. et al ,2018).

In cross-country context of behavioral intention to use mobile payment, Zhang, Y. et al (2018) collected the samples from China (n= 162) and the USA (n= 136). The research has concerned about the cultural difference that has effect on the research findings when comparing to both countries.

The results shown that perceived usefulness and perceived ease of use effects the behavior intention stronger in the USA than in China, plus, from the research paper in Germany samples of 1104 respondents also proved that Perceived Usefulness and Perceived Ease Of Use are the strongest predictors to intention to use mobile payment (Zhang, Y. et al ,2018 ; Pousttchi, K. & Wiedemann, D. 2007). Similar to India, Perceived Ease Of Use is considered the strongest predictor for Behavior Intention (Shankar, A. & Datta, B., 2018). Perceived usefulness and Self- efficacy also effect in Behavior Intention in India (Shankar, A. & Datta, B., 2018). In Singapore, perceived usefulness, and transaction security influence strongly to intention to use mobile payment (Seetharaman, A. et al, 2017). Personal innovativeness’s also effects on Behavior Intention in China and USA (Zhang, Y. et al ,2018).

Cost also plays significant role in their intention to use mobile payment where users are willing to switch to another system for financial transaction if they perceived extra benefits in less or similar costs (Shankar, A. & Datta, B., 2018). In Singapore, transaction cost factor hasn’t been looked at before, making the findings more interesting, the results from this research taken over 300 respondents, shows that transaction cost has very strong influence over perceived usefulness and perceived ease of use and transaction security (Seetharaman, A. et al, 2017).

Moreover, consumer trust is considered essential factor when it comes to the intention to use mobile payment and to adopt any technology-enabled service (Shankar, A. & Datta, B., 2018). Performance expectancy plays a crucial role and influence to consumer intention to use mobile payment in Qatar and in both South Korea and China (Khan, H. et al., 2015; Lin, X. et al., 2019). Another cross-country research done between South Korea and China (samples of Chinese n=467, Korean n=441), the results reflect on many new interest aspects, for example, Chinese users play most crucial role in online shopping more than South Korean (Lin, X. et al., 2019).

New variable such as user satisfaction is being used for the cross-country research between China and south Korea, the result also shows that the user satisfaction significantly influences usage intention to mobile payment (Lin, X. et al., 2019).

Findings & Research gaps: Vietnam

The study of Liang (2016) focused on the influences of Subjective Norm and customer adoption toward to mobile banking in Taiwan and Vietnam. In this research, the main theories about TAM and UTAUT model were applied as the references for research model development. All the independent variables were known as performance expectancy, effort expectancy, subjective norm, facilitating conditions, attitude toward use and the dependent variable was intention to adapt (behavioral intention). The sample size was 440 surveys in Vietnam and 372 surveys in Taiwan. It is very qualified when this research could be tested in the large sample size so the credibility would be gained. However, the conceptual foundation of this research was insufficient when it only mentioned about theories background, general information of mobile banking and theories background. And this research only discussed about mobile banking in general not mobile payment in specific. The illustrations of the cultural context of each country, behavioral intention and all the conceptual about other tested variables were not mentioned. Besides, the overall view about emerging market would be necessary if they conducted the research in this market. As the result, it is found that performance expectancy, effort expectancy and facilitating conditions strongly impacted on consumer intention of mobile payment in Taiwan market. On the other hand, facilitating conditions were insignificant in explaining mobile banking adoption in Vietnam market. One similarity was indicated that effort expectancy significantly impacted on consumer intention of mobile payment in both countries. This can be considered as the most relevant articles for our research when it mentioned the consumer intention of mobile payment in two countries in Asia. Base on this research, we can enhance about the deliverables or add the necessary missing to the previous findings.

Another research related to mobile payment adoption in Vietnam was conducted by Dao (2019). This research was mainly built on the revised model of UTAUT framework. The main tested variables were behavioral intention and use behavior. The other factors which were used to test the behavioral intention and use behavior towards mobile payment were known as performance expectancy, effort expectancy, social influence (subjective norm) and trust. As the result, performance expectancy, effort expectancy and social influence had positive influences on consumer intention of mobile payment. On the other hand, trust was the only factor has negative influence. Although this is the most updated research about consumer intention of mobile payment in Vietnam, some more important factors will be tested in our research which was not included here. For example, subjective norm was illustrated by Liang (2016) as an important factor has positive impact on both Vietnam and Taiwan market. In the collectivism society like Vietnam and Thailand (figure Hofstede) customers will be significantly influenced by their environment (Liang, 2016). Hence, subjective norm is a substantial factor when we conduct the study about consumer

intention in these countries. It also supports for our research about country context in the initial phase of this research.

In 2016, Dao and Nguyen conducted a research to predict consumer intention towards mobile payment before this new technology entered to Vietnam market. This study referred to the combination of TAM and TPB model. The six independent tested variables included perceived usefulness, perceived ease of use, perceived enjoyment, perceived trust, subjective norm, enjoyment, and perceived behavioral control. After the test, it was concluded that all the variables except for perceive behavioral control had significant influences on the consumer intention to mobile payment service. The study was tested on the sample size of 489 respondents but there was important shortage that they did not mention about the surveyed regions. It is essential to survey in various regions because there are many differences about cultural context and perception in different parts of Vietnam and it will significantly impact to the quality of your study. For example, it is more traditional in the North but more open minded and more modern in the Southern area. Besides, this study might cause some bias because the respondents all assumed their answers about mobile payment when they had no experiences about this service before. The first mobile payment service penetrated into Vietnam market in 2018 (Dougn, 2020) and this research was conducted in 2016 so this research was only to predict the intention of consumer before the entrance of this method. In our research, we are able to test the experienced consumers then comparing our findings to this prediction to illustrate the changes and differences of influential factors towards mobile payment usage.

Nguyen et al., (2015) also conducted an empirical study about factors affecting to consumer intention of mobile payment use in Vietnam. This research only based on the framework of TAM model so the main factors of perceived usefulness and perceived ease of use were tested. The survey of 365 respondents was collected in this study. In the findings, they indicated that two factors of trust and compatibility in perceived usefulness strongly affected to consumer intention of mobile payment service. In contrast, the factor of perceived ease of use negatively impacted on it. However, it was pointed in the research of Dao (2019) that the “trust” had negative relation with consumer intention of mobile payment service. This research possibly had the same problem with the research of Dao and Nguyen (2016) when they were both conducted the research before the technology of mobile payment entered to Vietnam market. Therefore, a new up to date research should be conducted to test again the reliability and consistency of the influential elements so we can deliver the most valuable and credible information to this potential new sector in Vietnam and Thailand.

The research of Phan et al., (2020) discussed about the determinants influencing customers' decision to use mobile payment services which was built on the extended framework of UTAUT model. In this study, all the independent factors were known as performance expectation, effort expectation, social influence, security and privacy, perceived cost and reputation of supplier. Target sample of this research only focused on Ha Noi (North of Vietnam) which meant that the qualification of diversified region base was not achieved in this study. To collect the sample, the surveys was distributed to 223 respondents in Ha Noi. As the result, all the factors positively impacted on the consumer intention of mobile payment; except for “perceived cost”. It’s said that social influence was the factor has the least influence but Liang (2016) indicated that this factor played an important role toward consumer intention in mobile payment sector. As the information from Hofstede model, Vietnam has a high collectivism index, and it is supposed that Vietnamese should be significantly impacted by their social environment. Therefore, we would like to test this factor again in our research to give the better conclusion about this unclear factor.

In summary, after going through almost the academic research about mobile payment in Vietnam, some crucial research gaps can be illustrated. First of all, the English research about mobile payment in Vietnam is very limited. We could only find less ten academic literature about this topic and half of them mainly discussed about digital banking. Second, most of the articles were conducted in last 5 years and even the mobile payment services were unavailable in Vietnam market at that time. Mobile payment segment first entered Vietnam market in 2018. Next, currently, there is no specific research including the factors of “attitude” and “perceived risk”. Although there are some researchers have the factors of “perceived risk”, all of them mainly focused on digital banking not specific on mobile payment. And none study about mobile payment applied the combination of all TAM, TPB and UTAUT models. Moreover, there are many conflicts among the tested variable in the previous researches so our research aims to handle the conflicts in the previous research and to deliver the most up to date and useful information for mobile payment segment in Vietnam and Thailand as well.

Findings & Research Gaps: Thailand

We will now focus into country specific: Thailand. An Article from Boonsiritomachai, W. & Pitchayadejanant K. (2017) has focused on the mobile banking usage based on UTAUT and concept from TAM. In total, 480 respondents were participated in surveys, the demographic information were collected such as gender, age, education level, occupation and monthly income, the variables that influence the behavior intention are self-efficacy, facilitating condition, security and hedonic motivation (enjoyment) (Boonsiritomachai, W. & Pitchayadejanant K. ,2017). The social influence/ subjective norms and mobile banking expectancy did not have any significant

influence on the intention to use mobile banking (Boonsiritomachai, W. & Pitchayadejanant K. ,2017), which is in contrast to the general findings for the social influences/subjective norms from Tan, K. 2019; Zhang, Y. et al ,2018; Khan, H. et al. 2015; Lin, X. et al., 2019, where they found out that social influences/ subjective norms effects strongest on intention to use mobile payment in China, Qatar, South Korea and USA. Moreover, the article from Boonsiritomachai, W. & Pitchayadejanant K. (2017) did not test on other moderators collected such as demographic backgrounds to other factors. However, this article is the most recent ones among all the articles found online based on the country specific (Thailand).

Moreover, there are only the mobile banking users which is insufficient for us to compare with other general findings since other articles, for example, Lin, X. et al., 2019 mentioned about Kakaopay application on mobile phone in South Korea and China and it has nothing to do directly to online banking. Similar to article from Seetharaman, A. et al, 2017 where they are talking about E-wallet in the mobile phone as mobile payment application and it was less to do with only banking application. In addition, this article claims that they are using the model from UTAUT and TAM where the important variables in TAM such as perceived usefulness and perceived ease of use have not mentioned and tested in their hypotheses.

Another article related to mobile payment in Thailand from Phonthanukitithaworn, C. et al., (2015) focused on the early adopter in mobile payment in Thailand by referring to the model of TAM which has modified into cultural context, namely, the authors used Hofstede’s cultural dimensions to be applied, and therefore the findings will be based on the Thai national setting. According to Chau et al. 2002, as cited by Phonthanukitithaworn, C. et al., (2015), the cultural environment in different countries may influence the way people behave and play significant role in the way people use technology. That is why it this research has pointed out that it is essential to mentioned and adapt the Hofstede’s cultural dimensions including power distance index, individualism vs. collectivism, country’s uncertainty avoidance index, masculinity vs. femininity, and whether a culture has long term orientation.

Figure 16: Hofstede’s Cultural Dimension Index Comparison , retrieved from https://www.hofstede- insights.com/country-comparison/china,thailand,the-usa,vietnam/

According to Hofstede-insights.com, in figure 16, Thailand is considered high in power distance, Phonthanukitithaworn, C. et al., (2015) considered high power distance as the society that needs to wait until the authority and opinion leaders to give a signal of when can people use those new technologies, therefore, the initiatives are very low when it comes to early adopters in technology. The author considered subjective norms as a cultural factor. Next, as shown in the figure 16 , Thailand, China and Vietnam are counted as collectivist country, therefore, it is stated that collectivist culture related to the notion of subjective norms (Lee and Green, 1991, as cited by Phonthanukitithaworn, C. et al., 2015). Next dimension, masculinity vs. femininity, Thailand is considered more feminine, meaning that Thais value relationships, caring, preserving environments and service-oriented (Phonthanukitithaworn, C. et al., 2015). The face to face service is very important to any types of service in Thailand, therefore, mobile payment is more of reducing face to face contact between other humans which is inconsistent to Thai lifestyle – making this factor considered to be compatibility – a variable from innovation diffusion theory (Phonthanukitithaworn, C. et al., 2015). Next dimension is uncertainty avoidance, according to the figure 7 above, Thailand is considered very high in this, meaning that Thais value security, clear rules, formality to the structure of life, making mobile payment seems to be high-risk payment method, hence, the factor perceived risk is used (Phonthanukitithaworn, C. et al., 2015). Last dimension by this research paper, the long-term/short-term orientation, Thailand is considered long-term orientation which is opposite to the recent data from figure 7, long-term orientation tend to emphasize thrift and savings for future and prefer investment in projects with long-term benefits, therefore, the perceived cost is considered in their research (Phonthanukitithaworn, C. et al., 2015). However, since the recent index number is different from their research paper, Thailand is now considered short-term orientation, the changes of findings are expected to be different in our thesis. Lastly, indulgence dimension has not been mentioned in their research paper.

In total, there are 256 early adopters of mobile payment respondents to questionnaires were collected in their research, findings showed that perceived usefulness, perceived ease of use and perceived risk had no significant effect on behavior intention to use mobile payment. Compatibility is found to be strongest predictor to intention to use mobile payment, followed by perceived trust, subjective norms, and perceived costs. We can see opportunities for us, this research is only focused on early adopters in mobile payment and the research paper is not up to date much since the research was in year 2015. According to Samarcom (2020), the introduction of GrabPay Wallet was in 2019, RabbitLine Pay was introduced in 2017, AirPay from Garena was introduced in 2018 and Lazada Wallet was also introduced in 2018, PromptPay as mentioned

earlier (government’s introduction to E-Wallet) was introduced in 2017. This makes the research from Phonthanukitithaworn, C. et al., (2015) not so up to date and even the Hofstede’s long-term and short-term orientation was completely changed in Thailand from long-term orientation to short- term orientation dimension within 5 years after their publication. We can also focus in our research on the current users and not early adopters. Moreover, some of independent variables such as attitude, user satisfaction etc. and other moderators such as age, genders, education level were not tested in their hypotheses. The UTAUT model which considered the most used and most up to date model was not used previously in their research paper.

Another research paper from Chansaenroj, P. & Techakittiroj, R. (2015) also has focus on mobile banking usage, their independent variables are in in total 4 of them including perceived usefulness, perceived ease of use, perceived costs and perceived risk. They referred to TAM model and the total respondents are 400. Their findings shown that the strongest predictor is the perceived usefulness of mobile banking to the intention to use mobile banking. There is only negative relationship between perceived costs and intention to use mobile banking, perceived risk and perceived ease of use also have positive relationship with intention to use mobile banking. The demographic data is collected but they did not test it in the hypotheses. Similar to the article from Boonsiritomachai, W. & Pitchayadejanant K. ,2017, they only focused on the mobile banking users and not the mobile payments by mobile application since mobile applications are very new and up to date (mobile payment e.g. GrabPay Wallet, Lazada Wallet, PromptPay, RabbitLine Pay, AirPay were introduced from 2017 onwards). We can also see this that their hypotheses are only focusing on TAM and not UTAUT which is considered more recent to use as model (which also includes demographic moderators such as age, genders etc.). Cultural back grounds and other cultural dimensions were not discussed in the research paper from both Boonsiritomachai, W. & Pitchayadejanant K., (2017) and Chansaenroj, P. & Techakittiroj, R. (2015). Therefore, this give us a great opportunity to do the research further using these research gaps as mentioned above

Summarization table of research gaps between Vietnam and Thailand

To illustrate the research gaps between both countries, what was fulfilled and what was missing in those 3 relevant research papers about mobile payment/banking in Thailand, we came up with diagram which reflects on what could we do to improve the hypotheses or specific scopes (based on Thailand’s current state of findings and Vietnam’s current state of findings).

Research Findings Shortages/Limitations Issues for future of this research research Article 1: Positively - Not the context of - Combination of Determinants affecting influence on mobile payment, but TAM TPB, and mobile banking behavioral focus on the mobile UTAUT adoption by generation intention: banking - Cultural context Y based on the Unified - self-efficacy - Article is too old, most - Regions based Theory of Acceptance - facilitating condition of m-payments were data collection and Use of Technology - security introduced in year - Test of Model modified by the - hedonic motivation 2017 – 2019 onwards moderators Technology - Not include cultural (demographic Acceptance Model No influence on factors data on our concept. behavioral - Not mention about research) intention: taking samples from - Prove -Subjective Boonsiritomachai, W. - social influence different regions norms have & Pitchayadejanant K. (subjective norm) - Refer to TAM but significant (2017) - Expectancy didn’t mention about influence to the perceived ease of behavioral use and perceived intention to use usefulness as mobile payment independent variables Article 2: Positively - Not use UTAUT - Combination of User Intentions to influence on (recent model) TAM, TPB and Adopt Mobile Payment behavioral - Article is too old – in UTAUT models Services: A Study of intention: 2015 and m-payment - Focus on current Early Adopters in applications were users and non- Thailand. - compatibility introduced mostly in users of mobile - perceived trust 2017 - 2019 payment Phonthanukitithaworn, - subjective norms - Not focus on current - Regions based C. et al., (2015) - perceived cost users and non-users data collection in m-payments - Test moderators No influence on - Didn’t test moderators to variables to see behavioral on variables relationship intention: - Didn’t talk about 6th - Links indulgence dimension of dimension to - perceived ease of Hofstede variables that we use (Indulgence) are using (find - perceived articles) usefulness - The short/long - perceived risk term orientation country index is so

- Linked Hofstede’s different now and cultural dimensions then Thailand to the choice of change from long- variables term to short term - Focus on early orientation country adopters in M- (perceived cost payment might not have significant influence on behavior intention) Article 3: Factors Positively - Context on mobile - Retest their Influencing The influence on banking and not hypotheses, why Intention to Use Mobile behavioral mobile payments by is perceived Banking Services in intention: application usefulness, Bangkok, Thailand. - perceived - Article is too old perceived ease of usefulness - Cultural backgrounds use, perceived Chansaenroj, P. & - perceived ease of and context weren’t risk are found to Techakittiroj, R. (2015) use discussed in the have significant - perceived risk paper influence on - Variables perceived behavior intention No influence on usefulness, perceived and the article behavioral ease of use, from intention: perceived risk have Phonthanukitithaw - perceived costs significant influence orn, C. et al., on behavior intention, (2015) says the which is totally opposite opposite to research - Combination of from TAM, TPB, and Phonthanukitithaworn UTAUT , C. et al., (2015) - Cultural context - Didn’t test the - Regions based moderators to data collection variables - Only data collection in - Test of Bangkok moderators (demographic data on our research) Article 4: Vietnam market - Focus more on mobile - Combination of Subjective Norms and Positively banking; not TAM TPB, and Customer Adoption of influence on UTAUT

Mobile Banking: behavioral especially for mobile - Cultural context Taiwan and intention: payment - Academic Vietnam (Liang, C. C., - subjective norm - Countries contexts in literature view 2016) - effort expectancy general and mobile about relevant No influence on payment in specific variables behavioral were not discussed - Test of intention: - The conceptual moderators - facilitating conditions foundation didn’t (demographic cover emerging data on our Taiwan market market, behavioral research) Positively intention and other - Specific research influence on variables on mobile behavioral payment intention: - performance expectancy - effort expectancy - facilitating conditions

Article 5: Positively - The model is so - Combination of Mobile payment influence on simple, only apply TAM TPB, and adoption in Vietnam behavioral UTAUT model which UTAUT (Dao, 2019) intention: is insufficient in such a - More inputs of complex and flexible relevant factors to - performance environment like match with the expectancy emerging market country context - effort expectancy - Culture is an - Background about - social influence important element in country context Asian countries, but - To point out No influence on this study lacks of whether in the behavioral important factors short time, the intention: relevant to country influential factors - trust context like subjective towards mobile norm payment usage - Not mention about are fluctuated or country context not.

Article 6: Positively - Not tested in different - Combination of Predicting Consumer influence on regions TAM TPB, and Intention to Use Mobile behavioral - Research before the UTAUT Payment Services: intention: mobile payment - More inputs of Empirical Evidence - perceived entered Vietnam relevant factors to from Vietnam (Cao et usefulness - Lack of country match with the al., 2016) - perceived ease of context country context use - Background about - perceived enjoyment country context - perceived trust - Up to date - subjective norm research - enjoyment delivering the most valuable and No influence on credible behavioral information for intention: mobile payment - perceived behavioral sector in Vietnam control

Article 7: Positively - Research before the - Combination of An Empirical Study on influence on mobile payment TAM TPB, and Factors Affecting behavioral entered Vietnam UTAUT Customer Intention to intention: - Lack of country - More inputs of Use Mobile Payment - perceived context relevant factors to System in Vietnam usefulness - The contradiction to match with the (Nguyen et al., 2015) - trust another research country context - compatibility about the influence of - Background about “perceived of trust” country context No influence on towards consumer - Up to date behavioral intention research intention: - The application of delivering the - perceived ease of only TAM model most valuable and use credible information for mobile payment sector in Vietnam Article 8: Positively - Limited testing in one - Combination of Determinants influence on region of Vietnam TAM TPB, and influencing customers' behavioral - The application of UTAUT decision to use mobile intention: only UTAUT model - More inputs of payment services: The relevant factors to

case of Vietnam (Phan - performance - Research before the match with the et al., 2020) expectation mobile payment country context - effort expectation entered Vietnam - Background about - social influence - Lack of country country context - security and privacy context - reputation of supplier - The conflict with Liang (2016) about social No influence on influence factor behavioral intention: - perceived cost

Table 3: Research gaps between Vietnam and Thailand from previous findings.Source: Self-edited

5.4. Research Model

Figure 17: Research Model. Source: Self-edited

Hypotheses

According to our conceptual model in figure 17, we have designed this model to use in our thesis, there are in total six different variables namely attitude, perceived usefulness, perceived ease of use, subjective norms, perceived cost and perceived risk. Those independent variables were chosen based on the relevant theories and research gap in our research gap and finding section. The attitude and subjective norm are two of the variables used in the Theory of Planned 85

Behavior (TPB). The perceived usefulness and perceived ease of use are the most use variables from Technology Acceptance Model (TAM) and also the subjective norm is used as independent variable in Unified Theory of Acceptance and Use of Technology Model (UTAUT), UTAUT has moderators such as gender, age, education level and previous experience which will also be collected in our questionnaires. The perceived risk and perceived cost are also tested in our hypotheses, the research gap is one of the obvious reasons in the previous findings where we would like to choose those two independent variables in our hypotheses. The behavioral intention to use mobile payment is our dependent variable and it is used by our referred theories such as TAM, TPB, and UTAUT. Our hypotheses will be discussed about its definition, the reason why do we choose each independent variable, and previous findings.

Variable definitions

Behavioral intention

According to Theory of Reasoned Action (TRA), the theory itself can be extended to conceptualize the human behavioral pattern in the decision-making strategy on the utilization of a new innovation or technology, moreover, it is able to explain individuals’ behavioral intention to utilize the new technology (Otieno, O. et al., 2016). The behavioral intention is defined as a function of an individual’s attitude towards the behavior (Otieno, O. et al., 2016). In addition, all models and theories including TAM, TPB, TAM2, DTPB, UTAUT are also used behavioral intention as dependent variable to predict why people are willing to make a decision (Momani, A. & Jamous, M., 2017; Venkatesh, V. et al. 2003). Moreover Ajzen (1991) has explained clearly that the TPB and other reasoned action models seek idea that behavior is guided by intention.

Many studies have shown that behavioral intentions account for a considerable proportion of variance in behavior (Ajzen, I. 1991). Moreover, the role of intention as a predictor of behavior (e.g. usage) is critical to use as dependent variable in technological context (Venkatesh, V. et al. 2003). All in all, our decision in using behavioral intention to use mobile payment as the dependent variable is relevant to those theories mentioned above. Our research topics are relevant to TAM, TPB and UTAUT, therefore, some independent variables are used to test our hypotheses and make assumption to our research question, those independent variables are explained below.

Attitude

Attitude is defined as an individual’s positive or negative feelings about performing the target behavior (Momani, A. & Jamous, M. 2017). Many theories and models used attitude as the main independent variables. The formulation of Theory of Reasoned Action (TRA) was after trying 86

to estimate the discrepancy that existed between attitude and behavior (Otieno, O. et al., 2016). Furthermore, attitude is used in Technology Acceptance Model (TAM) and TAM is the extended theory after the Theory of Reasoned Action (TRA) again, TAM replaced attitude factor from TRA by the two factors which are perceived usefulness and perceived ease of use – which we will use those two variables as well. In addition, Theory of Planned behavior (TPB) and Decomposed theory of planned behavior (DTPB) have used attitude as independent variable as a predictor to see the intention to use new technology (Momani, A. & Jamous, M. 2017). In the previous findings, out research gap found that the research from Tan, K. (2019), the attitude is the strongest predictor to intention for the Chinese ethnic group in Malaysia, also in mainland China, where the research from Teng, P. et al (2018) shows the findings from constructing the qualitative methodology (interviews) to gather information over 612 respondents, the findings show that attitude is the main factor which influence customer to adopt to mobile payment service in China (Teng, P. et al ,2018). For the reason why do we want to use attitude as independent variable, in the 3 research papers about factors influencing intention to use mobile banking and mobile payment in Thailand found that none of them use attitude as independent variable. Currently, there is no specific research about mobile payment in Vietnam included “attitude” factor so it will be new findings for mobile payment market in Vietnam if we test this factor in our research. Since in China, attitude plays significant role and important predictor, therefore, it would be interesting to see if attitude would be the strongest predictor between Vietnam and Thailand or not. Therefore, we assume that attitude has significant influence on behavioral intention to use mobile payment

H1: Attitude has significant influence on behavioral intention to use mobile payment

Perceived usefulness

Perceived usefulness is defined as “the degree to which a person believes that using a particular system would enhance his or her job performance” (Davis, D. 1989 p.320). Most of time, perceived usefulness and perceived ease of use are indicated as fundamental and distinct constructs that are influential in decisions to use information technology (Davis, D. 1989). Perceived usefulness is one of the main independent variables to be used to predict the attitude towards the behavioral intention in Technology Acceptance Model (TAM). There is argument to the use of perceived usefulness claimed by Jackson et al. (1997) and Lucas & Spitler (1999), they found no empirical evidence to support the relation between perceived usefulness and behavior intention (Li, L. 2010). However, the recent findings, it has shown that perceived usefulness and perceived ease of use effects the behavior intention in the USA, China, India and also Germany, the samples from the research paper in Germany 1104 respondents also proved that perceived

usefulness and perceived ease of use are the strongest predictors to intention to use mobile payment (Shankar, A. & Datta, B., 2018; Zhang, Y. et al ,2018 ; Pousttchi, K. & Wiedemann, D. 2007). As mentioned earlier, in Singapore, perceived usefulness, and transaction security influence strongly to intention to use mobile payment (Seetharaman, A. et al, 2017). The recent findings from Thailand also has a big clash between article from Chansaenroj, P. & Techakittiroj, R. (2015), where the perceived usefulness has significant influence on behavioral intention to use mobile payment and article from Phonthanukitithaworn, C. et al., (2015) stated that the findings are the opposite, perceived usefulness has no significant influence on behavioral intention to use mobile banking. Almost the study in Vietnam illustrated that perceived usefulness positively impacted on the usage of mobile payment services. To prove that perceived usefulness has significant influence on behavioral intention in mobile payment between Thailand and Vietnam, we come up with the hypothesis H2

H2: Perceived usefulness has significant influence on behavioral intention to use mobile payment

Perceived ease of use

Perceived ease of use is defined as “the degree to which a person believes that using a particular system would be free of effort” (Davis, D. 1989 p.320)’. Perceived ease of use is one of the main independent variables to be use to predict the attitude towards the behavioral intention in Technology Acceptance Model (TAM) along with perceived usefulness (Davis, D. 1989). Many recent findings use perceived ease of use to the behavioral intention to use mobile payment system, for example, in India, perceived ease of use is considered the strongest predictor for behavior intention (Shankar, A. & Datta, B., 2018). According to the previous findings, perceived ease of use effects the behavior intention in the USA, China, India and Germany (Shankar, A. & Datta, B., 2018; Zhang, Y. et al ,2018; Pousttchi, K. & Wiedemann, D. 2007). However, from the research gap we have found that in the research articles from Thailand, the findings from Phonthanukitithaworn, C. et al., (2015) proved that perceived ease of use has no significant influence on the behavioral intention to use mobile payment. On the other hand, an article from Chansaenroj, P. & Techakittiroj, R. (2015) found that perceived ease of use has significant influence on behavioral intention to use mobile payment. This contrast gives us an opportunity to use perceived ease of use as another independent variable. Moreover, in Vietnam, there are two conflicted findings about the factor perceived ease of use toward mobile payment. And both researches were conducted before the mobile payment methods were available in Vietnam market. While Nguyen et al., 2015 concluded that there was no influence from perceived ease of

use, Cao et al., 2016 indicated a totally contrast findings on it. Therefore, it is very necessary to conduct an update research and test this factor with the experienced market. Since there are more current users in mobile payment both in Vietnam and Thailand, we would also like to use this to see the relationship to the intention to use mobile payment. Therefore, we have come up with our third hypothesis H3

H3: Perceived ease of use has significant influence on behavioral intention to use mobile payment

Subjective norm

Subjective norms are used in many theories that we will be using including theory of reasoned action (TRA). TRA has mention about the ‘Subjective norm’ as one of the independent variables, the definition of subjective norm is described by overall perceived social pressure which is the sum of strength of each normative belief weighed by the motivation to comply (Ajzen, I. 1991). TRA goes further through an impact of personal attitude by considering the role of perceived social norm in relation to behavior interests (Ajzen, I. 1991) – in this case it is the intention to use mobile payment. Moreover, subjective norm is used in Theory of Planned Behavior (TPB) from Ajzen, I. (1991) as one of independent variable to predict the behavioral intention. Furthermore, Unified Theory of Acceptance and Use of Technology Model (UTAUT) developed from Venkatesh et al. (2003) also used subjective norm as independent variable – the UTAUT is the extended model after TAM and TPB by adding other demographic variable as moderators.

Subjective norm is considered one of the most use independent variable across the studies from TAM and UTAUT (Williams et al. ,2015). The subjective norm found to be working for some group of people, but not for others and this is still a research gap for any future studies (Li, L. 2010). Therefore, we would like to use this to fill the research gap by using subjective norm to understand the cultural context or impact of gender, age group to the behavioral intention to use mobile payment between Thailand and Vietnam. The recent findings as mentioned before shown that subjective norm has little effect within the Malay community (Tan, K. 2019).

Comparing to China, the subjective norms influence behavior intention and found stronger than the USA (Zhang, Y. et al ,2018). In Qatar, subjective norm also effects the intention to use mobile payment especially female users and younger generation (Khan, H. et al. 2015). Similar to South Korea, subjective norm also effects the usage intention for mobile payment (Lin, X. et al., 2019). In Thailand, there is a contrast of two literatures, from Boonsiritomachai, W. & Pitchayadejanant K. (2017) found that subjective norm has no significant influence on behavioral

intention to use mobile payment while another article from Phonthanukitithaworn, C. et al., (2015) found subjective norm has significant influence on behavioral intention to use mobile payment. This is another research gap that we would like to test our hypothesis. Subjective norm has an important role in Vietnam when this country has high index of collectivism. There are two research mentioned about subjective norm or social influence but a little conflict was raised among them. In the research of Liang (2016), she found that subjective norm significantly influences on consumer intention of mobile payment usage. However, in the research of Phan et al., (2020), they indicated that subjective has the least influence. Therefore, subjective norm is important to use as our independent variable in our hypothesis H4:

H4: Subjective norm has significant influence on behavioral intention to use mobile payment

Perceived cost

Perceived cost is defined as “overall expenses associated with the adoption of particular technology platform” (Pathirana, P. & Azem, S. 2017, p.70). There are many researches which use the perceived cost as one of the factors to see the relationship between the perceived cost and the behavioral intention to use mobile payment. According to Luarn and Lin (2005), found that perceived cost can act as a factor inhibiting consumer decision – making (Phonthanukitithaworn, C. et al., 2015). Moreover, Cheong and park (2005) found perceived cost to be an influential factor in predicting behavioral intention to use mobile commerce in South Korea (Phonthanukitithaworn, C. et al., 2015). There are further findings shown that cost also plays significant role in their intention to use mobile payment where users are willing to switch to another system for financial transaction if they perceived extra benefits (Shankar, A. & Datta, B., 2018). Moreover, according to Phonthanukitithaworn, C. et al., 2015 many researchers found that perceived cost can be used as the extended construct of TAM and especially when Thailand is considered the cost-conscious country. There are contrast of result between Phonthanukitithaworn, C. et al., 2015 where perceived cost found to have impact on behavioral intention to use m-payment and opposite to article from Chansaenroj, P. & Techakittiroj, R. (2015) where the research found that perceived cost has no significant influence to behavioral intention to use mobile payment, this could be our opportunities to see how perceived cost will actually have any relationship with the behavioral intention to use mobile payment in both Vietnam and Thailand. For Vietnam, it’s found that the only study of Phan et al., (2020) mentioned about perceived cost and they concluded that this factor had no influence to the consumer intention of mobile payment service. Hence, our findings can give more argumentations about this factor which will be the good contribution for mobile

payment market and the later researchers as well. All in all, it is important to count perceived cost as one of our independent variables, hence, our hypothesis H5 is as followed:

H5: Perceived cost has significant influence on behavioral intention to use mobile-payment

Perceived risk

Perceived risk is defined as feeling of uncertainty regarding possible negative consequences of using product or service (Bauer, 1967 as cited by Phonthanukitithaworn, C. et al., 2015). In China, the perceived risk is used to predict the trust that consumers have towards the privacy when using mobile payment application, where in the USA people are concerned more about their personal privacy, their perceived risk effects on the behavior intention to use mobile payment (Teng, P. et al ,2018; Zhang, Y. et al ,2018). Moreover, Zhang, Y. et al (2018) collected the samples from China (n= 162) and the USA (n= 136).

The research has concerned about the cultural difference that has effect on the research findings when comparing to both countries. In addition, in the cross-country context, perceived risk might vary and the result could refer to how people are willing to use those technologies or not, it relates to their trust to the mobile payment system, where in Thailand, the result from Phonthanukitithaworn, C. et al., (2015) shown that perceived risk has no significant influence on behavioral intention to use mobile payment because Thailand is considered high in risk avoidance, whereas Chansaenroj, P. & Techakittiroj, R. (2015) shows that perceived risk influence to behavioral intention to use mobile payment. This again shows the contrast between two studies across one country. Therefore, this is our research gap to find out whether current users truly have trust to their privacy or financial security to use mobile payment or not. As we mentioned before, there is no specific research about perceived risk in the previous study. This will become the first research about mobile payment in Vietnam has the test on influential level of perceived risk toward consumer intention. Therefore, we assume that perceived risk has significant influence on behavioral intention to use mobile payment in our hypothesis H6:

H6: Perceived risk has significant influence on behavioral intention to use mobile payment

Covid-19

As mentioned earlier in the literature review parts, Covid-19 seems to be relevant to the growth of mobile payment, e-wallet, debit and credit cards in many countries such as Germany, UK, USA (Chawla, A. 2020; Betuel, E. 2020). In Thailand and Vietnam, there are also a significant

growth of mobile payment during the pandemic because of the switching from cash-based habit into non-cash transactions to ensure hygiene issues.

Moreover, pandemic Covid- 19 has fanned the public concerns that the corona virus could be transmitted by cash, however, some researchers have revealed that the probability of transmission via banknotes is low compared to the frequently-touched objects such as credit card terminals or PIN-pads (Auer, R. et al, 2020). Moreover, central banks are also concerning the banknotes used, they are thinking about sterilizing the notes or quarantine them, some encourage to use contactless payment (Auer, R. et al, 2020). There are of course arguments from some of the articles, in the US consumers still prefer physical transactions over contactless methods, just only 5% increase of consumers report using Apple Pay at the POS in early – mid March (Auer, R. et al, 2020). According to the argument, we would like to test the dependent variable whether Covid-19 is relatively has impact on the behavioral intention to use mobile payment in the two countries or not. Since this is very recent and new. Therefore, this is a great chance to add this issue into our conceptual model.

H7: Covid-19 has significant influence on behavioral intention to use mobile payment.

6. Research Methodology

In this research, quantitative method will be applied to answer the research questions and find out the most important factors to consumer intention of mobile payment. For quantitative method, Metler, C 2016 has stated in his book about the quantitative research method that the goal of quantitative research method is to gain a better understanding of situation or event. When conducting the quantitative research studies, the researchers are required to see the relationship of current situations and establish relationships between variables. According to the book ‘introduction to educational research’, the outlines have guided us to understand why do we need to conduct the survey, in order to conduct a quantitative research, we are allowed to choose some approaches to collect sample size. There are non-experimental research designs, observation research, and survey research (many different types of survey).

In the book of “Doing Quantitative Research in Education”, Muijis explained detail about quantitative methods and how it would be applied in research. He mentioned that quantitative research “explaining phenomena by collecting numerical data that are analyzed using mathematically based methods (in particular statistics)” (Muijs., 2004, p.1). In quantitative research, the examined data will be in the numerical form. Therefore, qualitative research collects numerical data to give the explanation for specific phenomenon, specific questions which are only

appropriate to being clarified by using quantitative method. Quantitative method is evaluated as flexible method because there is unlimited phenomenon can be tested in this way. There are four types of research questions that usually find the suitable answers by using quantitative method. First, when the researcher expects for quantitative answer. Second, the numerical change can only be accurately studied by quantitative application. Third, when the researcher wants to illustrate something or explain some phenomenon. The last one, it will be suitable for testing hypothesis. The main two designs of quantitative research are known as experiment and non- experiment. In the experimental method, the test will be under the strict controlled conditions, so this kind of design is normally found in laboratories.

In contrast, the non-experimental study is not able to control the external influences. Therefore, in this method, the control can be enhanced by using predictor variables to suppose the influential factors toward the outcome. It’s known that the most common quantitative research in social science is survey research. Survey research is a method to collect data by using questionnaire forms. This method is very flexible and can be conducted in different approaches like telephone, face to face, postal pencil and paper questionnaires but the most common recent methods are web based and email forms.

Research process

Measurement scale

To conduct the research questionnaire, it’s essential to identify the measurement scales for all defined the research hypotheses which are Behavioral Intention (BI), Attitude (AT), Perceived 93

Usefulness (PU), Perceived Ease of Use (PE), Subjective Norm (SN), Perceived Cost (PC) and Perceived Risk (PR). Based on the relevant refences about mobile payment, the Likert Scales of 5 will be applied for all of the variables

• Measurement scale for Behavioral Intention of mobile payment

Table 4: Measurement scale for Behavioral Intention. Source: Lu, Y., Yang, S., Chau, P. Y.K., & Cao, Y. 2011, Adapted from Taylor, S. et al., 1995.

Items Question Item BI1 Assuming I have access to the mobile payment services, I intend to use it BI2 Given that I have access to the mobile payment services, I predict that I would use it

• Measurement scale for Attitude of Behavioral Intention on mobile payment

Table 5: Measurement scale for Attitude. Source: Luna, I. et al. 2019, Adapted from Davis, 1989; Kim et al., 2010; Yang & Yoo 2004.

Items Question Item AT1 The use of mobile payment is a good idea AT2 The use of mobile payment is convenient AT3 The use of mobile payment is beneficial AT4 The use of mobile payment is interesting

• Measurement scale for Perceived Usefulness of Behavioral Intention on mobile payment

Table 6: Measurement scale for Perceived Usefulness. Source: Shankar, A., & Datta, B. 2018, Adapted from Davis, 1989.

Items Question Item PU1 Using mobile payment would enable me to pay more quickly PU2 Using mobile payment would make it easier for me to conduct transactions PU3 Using mobile payment would be advantageous PU4 I would find mobile payment a useful tool for paying

• Measurement scale for Perceived Ease of Use of Behavioral Intention on mobile payment

Table 7: Measurement scale for Perceived Ease of Use: Source: Shankar, A., & Datta, B. 2018, Adapted from Venkatesh, V. 2003.

Items Question Item

PE1 I believe that when I use mobile payment, the process will be clear and understandable PE2 I believe that it is easy for me to become skillful at using mobile payment PE3 I believe that mobile payment is easy to use

• Measurement scale for Subjective Norm of Behavioral Intention on mobile payment

Table 8: Measurement scale for Subjective Norm. Source: Shankar, A., & Datta, B. 2018, Adapted from De Sena Abrahão et al., 2016; Luarn et al, 2005; Wei et al., 2009.

Items Question Item SN1 People who are important to me think that I should use mobile payment SN2 People whose opinions I value are prefer me to use mobile-payment SN3 People who are important to me (e.g., family members, close friends, and colleagues) support me to use mobile payment

• Measurement scale for Perceived Cost of Behavioral Intention on mobile payment

Table 9: Measurement scale for Perceived Cost. Source: Lu, Y., Yang, S., Chau, P. Y.K., & Cao, Y. 2011, Adapted from Featherman et al., 2003; Luarn and Lin 2005, Wei et al 2009.

Items Question Item PC1 It would cost a lot to use mobile payment services. PC2 There are financial barriers (e.g. having to pay for handset and communication time) to my using mobile payment services

• Measurement scale for Perceived Risk of Behavioral Intention on mobile payment

Table 10: Measurement scale for Perceived Risk: Source Lu, Y., Yang, S., Chau, P. Y.K., & Cao, Y. 2011, Adapted from Venkatesh, V. 2003.

Items Question Item PR1 I would not feel totally safe providing personal private over the mobile payment system PR2 I’m worried about using mobile payment services, because other people may be able to access my account. PR3 I would not feel secure sending sensitive information across the mobile payment system.

• Measurement scale for the impact of Covid-19 on mobile payment

Table 11: Measurement scale for the impact of Covid-19. Adapted from Girish et al., 2020

Items Question Item C1 Covid-19 has increased my use of mobile payment services C2 Mobile payment services are more convenient during Covid-19 lock down C3 Mobile payment service helped to maintain social distance while making payment or receipt during Covid-19 lock down C4 COVID-19 has not spread through mobile payment as it doesn’t involve touching of currencies C5 I will always use mobile payment applications even after COVID-19 lock down

Research sample

Target population

The main target group of this study will aim to both mobile payment users in Vietnam and Thailand market. The potential respondents can be all smartphone users because they will have high possibility of using mobile payment methods or become potential customers of this method in the future. In 2020, there are more than 38 million smartphone users in Vietnam which accounts for almost 40% of Vietnamese population (Doan, 2020). In Thailand, more than 28 million smartphone users in 2020 which is also more than 40% Thai population ("Smartphone users in Thailand 2013-2022 | Statista", 2019). This huge number of smartphone users in these markets will be a big advantage for this study.

Target sample

Sample is understood as “subset of your population by which you select to be participants in your study”. It’s very important to have an appropriate sample because the selection of the sample will significantly impact on the quality of the research (Morse, 1991).

Regions

For each country, the biggest and most developed cities in main regions will be selected to distribute the surveys. In Vietnam, Ho Chi Minh (South) and Ha Noi – capital (North) are the places that most of mobile payment users concentrate. Firstly, Ho Chi Minh is known as the largest city in Vietnam which has more than 8 million inhabitants ("Ho Chi Minh City, Vietnam Population (2020) - Population Stat", 2020). This is also economic and financial central of Vietnam when it contributes 22% national GDP and 29% financial capital (Tran, 2019) and being the fastest

growing market for technology and manufacturing in the country ("Ho Chi Minh - The dynamic economic city and financial hub of Vietnam", 2020). Next, Ha Noi – the capital of Vietnam with almost 7 million inhabitants which is also one of the most important Vietnam’s economic keys (Leducq & Scarwell, 2018).

In Thailand, we will collect the data and samples based on main regions including the central region, north eastern regions (U.S. Library of Congress, 2020). Bangkok belongs to the central regions and is a capital city of Thailand, Bangkok has over 16 million inhabitants (Citypopulation.de, 2019). Opposite to Northern region is the Northeastern Region (Isan) is considered the largest regions, it occupied over 20 provinces and bordered Laos and Cambodia (Rodgers, G. 2019), the cultures are shared with Laos including language and Khmer cultures, lastly, the southern regions, this region bordered Malaysia so the cultures are shared with Malay. The data collection will be based on three different cities including Bangkok, Ubon Ratchathani Province (in the North Eastern region) and also Pattaya City in Chonburi province.

Gender and Age group

There are several subgroups that we will focus to distribute the questionnaires, which are genders and age groups

Vietnam

In Vietnam, according to countrymeter.info, as per 23 July 2020, the ratio of gender between male and female population are 49,4% and 50,6% respectively making this ratio similar to Thailand. However, according to statista.com (2019), the mobile payment users are 51,6% male and 48,4% female, therefore, we expected the number of male respondents to be a little higher.

In Vietnam, according to countrymeter.info the percentage of population age of 15 – 64 years old is counted as 69,3% of a whole population which is very similar to Thailand’s structure. In Vietnam, there are 51% mobile payment users are around 22-29 years old and 40% of them are 30-44 years old so we will mainly focus on these age groups to obtain the most effective result (Zalopay,2020). In addition, according to statista.com (2019), the most used mobile payment age groups were shown that the age group between 25 – 34 years old are the most used group by 35,3% and second most used groups are 35 – 44 years old (23,5%) and 18-24 years old (18,6%), 45 – 54 years old (15,8%) and lastly 55 – 64 years old (6,8%). Therefore, the target will be the age group of between 18 to 44 years old in 200 respondents.

Thailand

In Thailand, according to countrymeter.info. The population as in live-counting per 23 July 2020 shows that the ratio of gender between male and female population are 49,1% and 50,9% respectively. According to statista.com (2019), the mobile payment by genders in Thailand has shown that 46,5% are male users and 53,5% is female users. Therefore, we expected that our quota plan to go for approximately more female respondents than male respondents.

For the age group, according to countrymeter.info,in Thailand, the age range between 15- 64 years old is counted as the biggest group of 70% of the whole age population. Moreover, the age group that has used mobile payment the most is around 25 - 34 years old (31,3%) and 35 – 44 years old (31,1%) and the second biggest group is between 18 – 24 years old (20,4%) and 45 – 54 years old (17,3%) for a total user. Therefore, we expected to have respondents of age between 18 - 44 years old the most in 200 respondents (statista.com, 2019).

Sample size

The total of 27 variables are generated in this research which are 2 dependent variables and the rest 25 variables are from independent variables. The sample size will largely depend on the number of all variables. It’s stated by Roscoe et al., (1975) that the most sufficient sample size should be over 30 and under 500. On the other hand, Black el al., (2019) found that the ideal sample size has to be over 100 and the minimum sample would depend on the desired ratio of 5 observations per variable. Therefore, some calculations should be conducted to give the most appropriate sample size for this research. We supposed that N will represent for the sample size:

According to the statement of Black et al., (2019): N > 100 N= 5K (K is the number of variables) The minimum sample size is N= 5*26 = 130

For standard multiple regression analysis, Tabachnick et al., (1991) proposed that the desired level is: N > 50 + 8M (M is number of independent variables) Hence, the required sample is: N > 50 + 8*22 = 226

After all, we decide to select the sample size of 400 for our research because it will be conducted in two countries; and the equal 200 surveys will be distributed in each country. From that we can qualifies that standard of ideal sample size and it’s also more reasonable for us to collect the data within the timeline framework

Sampling technique

Sample is understood as “subset of your population by which you select to be participants in your study”. It is very important to have an appropriate sample because the selection of the sample will significantly impact on the quality of the research (Morse, 1991).

According to Landreneau & Creek (2009), the sampling strategies possibly estimate the margin of error that data obtained from the samples. Therefore, it is important to decide the most appropriate sampling design which brings least possible error in the data set. There are two main sample designs which are probability sampling and non-probability sampling. In probability sampling, each sampling unit has an equal chance to be selected from the target population. The main strategy of this sampling is random selection which means that the subjectivity will be removed during sample selection process (Meadows, 2003). Normally, this method has high possibility to produce the representativeness of target population in the appropriate level of confidence. Particularly, probability sampling is highly evaluated because of its cost effectiveness and simplicity. This method not only save time consumption but also maximize cost efficiency and the technical knowledge is not required when applying this method as well (Fleetwood, 2020). There are several ways of conducting probability sampling such as:

• Simple random sampling: the completely random selection process will be conducted like drawing numbers from a hat. The researchers only need to assign the number of samples then selecting those samples through an automatic process. As the result, all the selected options will become the sample (Fleetwood, 2020). • Stratified random sampling: the whole population will be classified into different smaller groups according to some main domains like gender, age, ethnicity, etc. There will be a distinct among members in each group to ensure that they have same opportunity to be selected (Fleetwood, 2020). • Cluster sampling: the random respondent selection when the population is allocated in different places (Fleetwood, 2020). • Systematic sampling: this is the best practice of old probability technique when each member of group will have equal opportunity to be selected in regular periods to develop the sample (Fleetwood, 2020).

On the other hand, the elements in non-probability method which are selected by non-random methods, will structure the sample. Contrast to probability sampling, this sampling seems to less produces representative samples (Landreneau et al., 2009). Normally, the non-probability method depends on the observations of researchers and it is widely used in qualitative research. There are three main methods for this sampling:

• Convenience sampling: the sample is only selected from the population due to the conveniences toward the researchers (Bhat, 2020). Besides, it is defined as “using a group of individuals that are readily available and are willing to be surveyed” (Meadows, 2003, p.522) • Quota sampling: this method will select the sample by dividing target population into different subgroups according to their characteristics. After that, the selection from each subgroup will be implemented to obtain the sample (Meadows, 2003). • Snowball sampling: the researchers will apply this method in case that the sample is too small and hard to identify. This method is another form of referral program where researchers can collect the relevant samples based on references of the current respondents (Bhat, 2020)

The most common differences between non-probability and probability sampling can be illustrated in the below table:

Table 12: Adapted from Bhat (2020)

Non-probability sampling Probability sampling Sample selection based on the subjective The sample is selected at random judgment of the researcher Not everyone has an equal chance to Everyone in the population has an equal participate chance of getting selected The researcher does not consider sampling Used when sampling bias has to be reduced bias Useful when the population has similar traits Useful when the population is diverse

The sample does not accurately represent the Used to create an accurate sample population Finding respondents is easy Finding the right respondents is not easy

100

After going through all the sampling strategies, we decide quota sampling technique is the most suitable method for our research. Our total sample size is 400 surveys in both Vietnam and Thailand and it will be collected in six largest and most developed cities of these countries. Most of these cities are located in different regions of Vietnam and Thailand. And it is also necessary for us to collect the sample based on the different ages and genders to illustrate the differences about mobile payment usages among these elements. Therefore, quota sampling method is the one that satisfies almost the criteria about diversified locations, ages, gender, and large scale of sample in this research (Mertler, C. 2016). This method also supports for the diversified locations in our research. In this research, we will mainly focus on the gender, location and age groups that concentrate most mobile payment users of Thailand and Vietnam so it will be costly and time consuming if we apply other sampling strategies into this research. When we have obtained the target sample, we will use SPSS statistical software to help us to interpret the data and help us to analyze them. This method will ensure the credibility due to its large sample size and also give an overall insight about demographic of users in these countries. The Structural Equation Modeling (SEM) will be used in our research to combine our factor analysis and regression analysis. As seen below in our table 13, we have collected the data and percentage of the current mobile payment users in both country, Vietnam and Thailand from Statista.com, 2019. Then we created the table showing the age group against gender in both countries to see the minimum number of the quota sample that we aim to collect.

Table 13: Table of Quota Plans: Mobile Payment Users in Vietnam

Female Male Total Total 96.80 103.20 200 respondents Current Users Current Users Total % Current users Age Group Female 48,40% Male 51,60% samplings Age group <18 0.00 0.00 0.00 0.00% 18 - 24 18.00 19.20 37.20 18.60% 25 - 34 34.17 36.43 70.60 35.30% 35 - 44 22.75 24.25 47.00 23.50% 45 - 54 15.29 16.31 31.60 15.80% 55 - 64 6.58 7.02 13.60 6.80% > 65 0.00 0.00 0.00 0.00%

Total 97 103 200 100,10%

101

Table 14: Table of Quota Plans: Mobile Payment Users in Thailand

Female Male Total Total respondents 107 93 200 Current Users Current Users Total % Current users Age Group Female 53,5% Male 46,5% samplings Age group <18 0,00 0,00 0 0,00% 18 - 24 21,83 18,97 41 20,40% 25 - 34 33,49 29,11 63 31,30% 35 - 44 33,28 28,92 62 31,10% 45 - 54 18,51 16,09 35 17,30% 55 - 64 0,00 0,00 0 0,00% > 65 0,00 0,00 0 0,00% 0 Total 107 93 200 100,10%

7. Empirical findings

In this section, we will focus on the quantitative analyses and its results by using the tools such as SPSS program to illustrate the result in tables, the initial research model will be used to show whether any hypotheses could be accepted or rejected. Later we will also answer to the research question on “What are factors influencing consumer behavioral intention of mobile payment in Vietnam and Thailand?”

Quantitative analyses and results Under this chapter, the demographic data and relevant information will be illustrated in graphs and charts. It is explained in detail as well as the handling of any missing values.

Samples and demographic data

According to our quota plans, the plan was to see our minimum collected samples. Our minimum quota is 200 respondents each country. The focus of our respondents is only the current users of the mobile payment since there was not many of non-user ‘respondents who took the test. However, it is necessary to mention all relevant data and number of respondents related to each of our questions such as age group, what applications of mobile payment are they using, how often do they use these applications or services, etc.

102

Using status Vietnam Thailand Current user 255 275 Non-current user 34 17 Gender

Female 123 150

Male 132 125 Age < 18 1 13 18 -24 62 86 25 - 34 92 64

35 - 44 51 63

45 - 54 35 35 55 - 64 14 6 > 65 0 1 Educational level No education 0 1 Primary 5 0

Secondary 48 33 Undergraduate 129 211 Postgraduate 59 25 Phd or higher 14 2 Region

North 96 14

South 134 9 Central 25 154

Northeast 0 25 East 0 62

West 0 4

Table 15: Summarization Table of demographic data between Vietnam and Thailand

Vietnam

In total, we were successfully distributed the survey to 289 people in Vietnam. As the below chart shows that there are 255 current users and only 34 non-users of mobile payment service in this survey. However, we only proceed the analysis for the current users so the non-user data will be totally subtracted. The questionnaire form was designed by Google form. And this survey was distributed to various channels such as Facebook groups, chat applications (Zalo, Messenger, WhatsApp, Skype, etc.) to obtain the goal of data collection. The data collection was relied on quota plan as the indication, so the result was quite close to the initial plan.

103

Are you currently using mobile payment method? n=289

No 12%

Yes 88%

Chart 1: Ratio of mobile payment users and non-users of respondents from Vietnam

The below chart can show that MoMo is the most-used mobile payment service in Vietnam leading by 130 users. Next, Airpay and ZaloPay can be considered as top favorite for users when there are 74 and 66 respondents are using these applications accordingly. Moca with 47 users and VNPay with 28 users are less preferred by the users while they are only new players in the market. The rest of respondents used other applications which are not popular or just recently launching in Vietnam.

The use of mobile payment services (multiple selection is possible) n=255

Others 66 VNPay 28 Airpay 74 Moca 47 MoMo 130 Zalo Pay 66

0 50 100 150

Chart 2: The most used mobile payment applications of respondents from Vietnam

According to chart 3, most of the respondents (90 people) use mobile payment service a few times a month. Besides, there are also a large number of people (64 people) said that they used this service at least once a week. The next significant group was people using mobile

104

payment service everyday which accounted for 17% of total users. The rest 23% of respondents were rarely used the mobile payment service while they only used it a few times a year, less than once a year or never.

How often do you use mobile payment ? n=255

10% 17% 7%

25%

35%

Almost everyday At least once a week A few times a month A few times a year Chart 3: Frequent of mobile payment used of respondents from Vietnam

According to the result, the rate between male and female in this survey are quite balance while there are 131 males and 124 females. In the reality, we even obtained the higher numbers than what we planned in the quota plan which could ensure the quality for the next step of data interpretation. As we could see from the chart, there are 62 respondents among 18 – 24 years old, 92 people in 25 – 34 years old, 50 people in 35 – 44-years old, 35 people in 45 – 54 years old, 15 people belong to 55 – 64 years old and no one is over 65 years old. According to the age group data, we completely adapted the initial quota plan when we compared it once again.

Gender n=255

Male Female 51% 49%

Chart 4: Gender ratio of respondents from Vietnam

105

Age Group n=255 0%0% 6%

14% 24%

20% 36%

< 18 years 18 - 24 25 – 34 35 - 44 45 – 54 55 - 64 > 65 years

Chart 5: Age group ratio of respondents from Vietnam

As the initial plan, the survey would be distributed nationwide from North to South. However, we would like to focus on two main regions which were North (capital city) and South (economic center). Hence, the chart indicated that the 52% and 38% of respondents were from the South and North accordingly. And the rest respondent (10%) would belong to center.

Regions n=255

38%

52%

10%

Chart XX:North Regions of CentralrespondentsSouth from Vietnam

Chart 6: Allocation of repsondents by regions in Vietnam

The chart indicated that most of the respondents were graduated from the University. Besides, it is also a significant number when there are 23% respondent achieved postgraduate level. The next group is all the respondents who have secondary school educational level

106

accounting for 19% of total sample. There are a small percentage of respondents obtained Ph.D. or higher level of education. And no respondent does not have educational level.

Education Level n=255

0% 2%

5% 19%

23%

51%

No education Primary (elementary/middle school) Secondary University (undergraduate)

Postgraduate Ph.D or higher level

Chart 7: Education level of respondents from Vietnam

Thailand

In total of 292 respondents have answered the questionnaires. According to chart 8 below, among the 292 respondents, we have found that only 275 respondents are the current users of mobile payment in Thailand, meanwhile only 17 respondents are the non-users. Therefore, in order to make an analysis focusing on only the current users, we then need to subtract the non- users out of the total amount. The distribution of the questionnaires was sent online via chat applications called Line. The questionnaire itself were made in Google Form. The link of questionnaires was distributed to college students, office workers, and different chat groups. In these chat groups, the respondents were asked if they were in our quota plans or not. Hence, we only communicated the exact required age groups and genders.

107

Are you currently using any mobile payment method? n=292 No

Yes 94%

Chart 8: Ratio of mobile payment users and non-users of respondents from Thailand

According to our data collection, the chart 9 below has shown the answers from 292 respondents. The mobile payment applications that most Thai people use are PromptPay which is an application introduced by Thai government, the responses of 132 people claimed that they use the PromptPay. Second most popular application is called TrueMoney which 120 respondents have claimed that they use it. The third top used application according to the data collection is ‘Others’ with 86 responses and ‘AIS’ with 85 responses, respectively. The 66 responses have chosen ‘Airpay’ application, 38 responses have chosen ‘Rabbit LINE pay’ and the least numbers go to PaySocial with 11 responses.

Chart 9: The most used mobile payment applications of respondents from Thailand

Next, our data collection out of 290 respondents replied to the question of how often do they use the mobile payment method. According to our chart 10 below, most of the respondents

108

44% or 128 respondents claimed that they use it almost every day, following by at least once a week by 29% or 84 people. Third biggest group is a few times a month with 21% or 60 respondents. The frequent of ‘A few times and year’ and ‘Never’ has 3% which are 9 and 8 respondents, respectively. Only 1 person from respondents has chosen ‘Less than once a year’ which has percentage below 1 or counted as 0%.

How often do you use mobile payment method for your transactions? n=290

0% 3% 3%

21% 44%

29%

Almost everyday At least once a week

A few times a month A few times a year

Less than once a year Never

Chart 10: Frequent of mobile payment used of respondents from Thailand

Moreover, in our data collection as illustrated in chart 11 below, there are female respondents of 150 people, which is counted as 55% and male respondents of 125 people, which is counted as 45%. The ratio has achieved the minimum quota plans as in number of respondents, female of at least 107 people and male respondents of at least 93 people, respectively.

Gender n= 275

Male Female 45% 55%

Chart 11: Gender ratio of respondents from Thailand

109

In our age group information as shown in the chart 12 below, there are in total 268 respondents who have responded to our surveys. We aimed to reach the minimum requirement for our quota plans and at the end we have achieved the minimum number. The age group of less than 18 years old in our quota plans does not require any respondents, however, we can still collect the data in this age group, and we had 13 respondents in this group. Second, the age group of 18-24 years old required minimum of 41 respondents but we have collected more than the requirement which is in total 86 people in this age group. Third, the age group of 25-34 years old required minimum 63 people and we have collected 64 people. Next, the age group of 35-44 years old required minimum of 62 people and we have collected 63 people. The age group of 45-54 years old we need minimum of 35 people and we have collected 35 people. In our quota plans, the age group of 55-64 years old and over 65 years old required 0 respondents but we have managed to have more respondents. We have 6 respondents in the age group of 55-64 years old and 1 respondent in the age group of over 65 years old.

Age group n=268 0% 2% 5%

13%

32% 24%

24%

< 18 years 18 - 24 25 – 34 35 - 44 45 – 54 55 - 64 > 65 years

Chart 12: Age group ratio of respondents from Thailand

According to our chart 13 below, this shows the ratio of respondents in Thailand’s region. We do not need the specific requirement of which regions most respondents should fill up the survey but there are main regions that we have aimed for example, 58% of respondents or 154 respondents out of 268 respondents are from central part of Thailand (this includes Bangkok). The second biggest group of respondents are from East of Thailand, and it is counted as 23%. Following by the Northeastern region with 9% or 25 respondents, the Northern region has 5% or 14 respondents, the least is Southern and West region which has 3% (9 person) and 2% (4 person) respectively.

110

Regions n=268

2% 5%

23%

9% 58% 3%

North Central South Northeast East West

Chart 13: Regions of respondents from Thailand

Last demographic information is the education level of respondents from Thailand, out of total 272 respondents, the most respondents are in undergraduate level (78% or 211 respondents) and following by secondary school level of 12% or 33 respondents. The postgraduate level is counted as 9% or 25 respondents. Lastly, PhD level is counted as 1% or 2 person, and 0% or 1 person for no education.

Education level n=272 0% 1% 0%

9% 12%

78%

No education Primary (elementary/middle school) Secondary University (undergraduate) PostgraduateChart xx: Education level of respondentsPh.D or from higher Thailand level

Chart 14: Education level of respondent in Thailand

Handling of missing values

Vietnam

111

In Vietnam, our questionnaire was designed as two options (1) for the users: can continue with their survey until the final question, (2) for non-users: their survey would be automatically ended when they ticked to non-user box. And for the users, all filled questions are compulsory so we have no missing value in the collection stage. In total we have 255 respondents in Vietnam survey.

Thailand

Originally, we have in total of 292 respondents in Thailand, only 275 respondents are the current users of mobile payment services. We have subtracted the 292 respondents by 17 non-users out left with 275 samples. Then, we have in total 27 missing values in demographic sections. Therefore, we use this number of respondents for our SPSS analysis. There is no missing value from gender section, we have finally in total 248 current user respondents in the SPSS.

Combination Data Set

For our combination set of data, we combine the Vietnam current users and Thailand current users of mobile payment with excluding the missing value. In total, we have 503 respondents from both countries.

Descriptive Analysis

In the initial stage of data analysis, the demonstration of the quantitative information will be done through our descriptive analysis. It is a good idea to begin with the straightforward statistics and then move through the more complex statistics (Taylor, L. & Churchill, B. 2013). Moreover, it is crucial to know the general pattern of variables since the overall results should be understood and interpreted (Taylor, L. & Churchill, B. 2013). The descriptive analysis enables data to be summarized into some values that make available a statistic to be discussed. A common descriptive analysis includes measures of central tendency such as median, mean and measures of dispersion such as range, variance, frequencies, and standard deviation (Taylor, L. & Churchill, B. 2013). In our descriptive statistics, we decided to compute our variables by the mean of items. The items will be grouped together in order to generate the single variable. We chose to see minimum, maximum, mean, and standard deviation of each construct.

According to Sykes, L M, Gani, F, & Vally, Z. (2016), mean is the most common measure of central tendency and refers to the average value of a group of numbers. Standard deviation

112

and variance are both measures of variability (Sykes, L M, Gani, F, & Vally, Z. 2016). The variance describes how much each value in the data set deviates from the mean (the spread of responses) and it is a squared value, for standard deviation, it also describes variability and it is the squared root of the variance. Low in standard deviation means the data spreads quite close to the mean and high standard deviation means the data spread over a wide range of values (Sykes, L M, Gani, F, & Vally, Z. 2016). In our case, the likert scale is used between 1 to 5. 1 means strongly disagree, 3 means neutral and 5 means strongly agree.

Combination of two countries dataset

According table 16, our descriptive statistics between the combination of the two countries Thailand and Vietnam has shown the total number of participants of 503, the minimum and the maximum is according to our Likert scale from 1 to 5. In addition to the mean, Behavioral Intention, Attitude, Perceived Usefulness and Perceived Ease of Use and Covid-19 factors are considered having the mean above 4 which people tend to agree to the statements in those factors. Moreover, the standard deviation shows us how much the distribution has been dispersed from the mean, by looking at the Subjective Norm, Perceived Cost and Perceived Risk tend to have standard deviation of above 1 or near 1, which reflects that the data has been distributed further away from the mean.

Descriptive Statistics

N Minimum Maximum Mean Std. Deviation Behavioral Intention 503 1.00 5.00 4.1561 .87858 Attitude 503 1.00 5.00 4.2525 .72269 Perceived Usefulness 503 1.00 5.00 4.3054 .72836 Perceived Ease Of Use 503 1.00 5.00 4.1604 .76724 Subjective Norm 503 1.00 5.00 3.6418 1.06000 Perceived Cost 503 1.00 5.00 3.1322 1.21409 Perceived Risk 503 1.00 5.00 3.6667 .97948 Covid19 503 1.00 5.00 4.0481 .89393 Valid N (listwise) 503

Table 16: Descriptive Statistics of respondents from Thailand and Vietnam

Vietnam

To summarize the descriptive data from the below table, we have total 255 respondents who are current users of mobile payment services in Vietnam. For all the variables including both

113

independent variables and dependent variable, we all obtain the minimum of 1 and the maximum of 5. Then, we will discuss more detail about the indication of each variable. First, the mean of behavior intention is 3.88 which indicates that most of the respondents have neutral opinion on the statement. And the standard deviation indication of this variable is quite high which shows that the observations of this variable are highly spread out. Second, the mean of Attitude is approximate to 4 indicating that many respondents agree on these statements. And the standard deviation is comparative, but the spread of this variable is still lower than the Behavior Intention. Third, the Perceived Usefulness and Perceived Ease of Use variables obtains the mean of 4.1 and 4.0 respectively which demonstrate that the majority of respondents agree with questions belong to these variables and the standard deviation are also considered as high extent. For the Subjective Norm and Perceived Risk, their means are 3.2 and 3.4 respectively means that many respondents say that they have the neutral answers for these questions and the spread of the answers are these variables are quite high when they all achieve the significant standard deviation numbers. For the Perceived Cost, many respondents tend to disagree with the statement due to its mean of 2.67. However, the spread of these answers is also high when it obtains the large standard deviation number. The last discussion will be Covid-19, it has mean of 3.7 indicating that many respondents have neutral answers on these questions. And the standard deviation is also relatively high.

Descriptive Statistics N Minimum Maximum Mean Std. Deviation Behavior Intention 255 1.00 5.00 3.8765 .91221 Attitude 255 1.00 5.00 4.0069 .77052 Perceived Usefulness 255 1.00 5.00 4.0863 .81322 Perceived Ease of Use 255 1.00 5.00 3.9176 .80284 Subjective Norm 255 1.00 5.00 3.2065 1.14750 Perceived Cost 255 1.00 5.00 2.6667 1.09341 Perceived Risk 255 1.00 5.00 3.4327 1.02150 Covid-19 255 1.00 5.00 3.7075 .96223 Valid N (listwise) 255

Table 17: Descriptive Statistics of respondents from Vietnam

Thailand

According to the table 18 below, the descriptive analysis from the respondents in Thailand has shown that the total numbers of the current users are in total 248 people. The construct of Behavioral Intention has minimum of 1.5 and maximum of 5. The mean is quite high with the

114

average of 4.44 – this reflect on most respondents agreed on the statement and the standard deviation is slightly high. This shows that the spread of data of across respondents who has an opinion of strongly disagree and strongly agree is still spread widely from the mean. Next is the Attitude where the mean is at 4.5, shows that many respondents agreed on the statement. The standard deviation is considered low, this shows that the data spreads are not so deviate from the mean.

Therefore, we look at the table below and found out that the minimum of 2.75 tells us that the data is leaning against the direction of agreement. Similar to the construct Perceived Usefulness and Perceived Ease of Use where the minimum stays above 2 with 2.75 and 2.33 respectively. The mean of PU and PEOU is 4.53 and 4.4 respectively. The standard deviation is quite low comparing to other constructs. This shows that the distribution of data is quite close to the mean and the mean shows that many people tend to agree on the statements. Next is the Subjective Norm where the minimum is 1.67 and the mean is 4.09, standard deviation is 0.73 shows that there are more spreads of the data between agree and disagree across our data set. Next construct is the Perceived Cost and Perceived Risk where both of them tend to have the lowest minimum of 1 and the mean of 3.6 and 3.9 respectively, the standard deviation is high, with the value of 1.14 and 0.87 respectively. This shows that many respondents feel neutral to the statements in the survey. Many of them believed in strongly disagree and strongly agree, therefore, the deviation of the data is quite far from the mean. Lastly, the Covid-19 construct shows that minimum of 1.2 and the mean of 4.4 which is in the direction of ‘agree’ from the respondents. The standard deviation is considered low comparing to other constructs. The deviation of the data is quite close to the mean.

Descriptive Statistics

N Minimum Maximum Mean Std. Deviation Behavioral Intention 248 1.50 5.00 4.4435 .74124 Attitude 248 2.75 5.00 4.5050 .56975 Perceived Usefulness 248 2.75 5.00 4.5373 .54212 Perceived Ease Of Use 248 2.33 5.00 4.4099 .64048 Subjective Norm 248 1.67 5.00 4.0981 .73027 Perceived Cost 248 1.00 5.00 3.6109 1.14567 Perceived Risk 248 1.00 5.00 3.9140 .87590 Covid19 248 1.20 5.00 4.4024 .63920 Valid N (listwise) 248

Table 18: Descriptive Statistics of respondents from Thailand

115

Reliability Test

Cronbach’s Alpha reliability is one of the most widely used measures of reliability in social and organizational sciences (Bonett, D. & Wright, T. 2014). The Cronbach’s Alpha reliability describes the reliability of a sum (or average) of q measurement (Bonett, D. & Wright, T. 2014). The q measurements may represent q raters, occasions, questionnaire/test items – in our case the q measurement applied for the questionnaire items, which is the most common application. Cronbach’s Alpha is referred to as a measure of ‘internal consistency’ reliability. As suggested by Nunnally (1978), the reliability of 0.7 or higher is considered acceptable depending on the number of items in the scale, as indicated by Lance, C. E., Butts, M. M., & Michels, L. C. (2006), many researches have a higher cut off of 0.8 for Cronbach’s Alpha. When there are a small number of items in the scale (less than 10), Cronbach’s Alpha can be small. In our reliability test, we aim for the value of Cronbach’s Alpha above 0.7.

Cronbach’s Alpha Test

Combination of two countries dataset

Because we made the research in two countries as we mentioned from very beginning, it is very crucial for us to ensure the reliability and balance the biases among the data set between Thailand and Vietnam. Therefore, we decide to combine the data set in both country and conduct the essential reliability tests for it before we interpret the data in each country separately. From that, we are able to establish standard scales for Vietnam and Thailand in later separate interpretation. This combination interpretation not only enhances the reliability for the data in each country but also avoids the biases when we made the detail analyses for these countries separately.

Behavior Intention

First of all, we test the reality of this data set with Cronbach's Alpha standard. As we can see from the below table, Cronbach’s Alpha of Behavior Intention is 0.816 which is higher than the acceptable scale of 0.7 so all the answers for this variable is considered as credible. We decide to keep all the items in this variable for the upcoming test.

Reliability Statistics Cronbach's N of Items Alpha .816 2

Table 19: Reliability Statistics of Behavior Intention in Thailand and Vietnam

116

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted BI1 4.18 .890 .689 . BI2 4.13 .939 .689 .

Table 20: Item-Total Statistics of Behavior Intention in Thailand and Vietnam

Attitude

Similarly, we can see the result of reliability test for Attitude by considering Cronbach's Alpha indicator. As we can see from the below table, the Cronbach's Alpha for Attitude is 0.854 which is significantly higher than the acceptable scale 0.7 so we can conclude that all the answers among Attitude variable is quite reliable and we won’t eliminate any item in this variable after this test.

Reliability Statistics Cronbach's N of Items Alpha .854 4

Table 21: Reliability Statistics of Attitude in Thailand and Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted AT1 12.70 5.058 .706 .811 AT2 12.64 5.159 .669 .826 AT3 12.85 4.724 .692 .817 AT4 12.84 4.775 .721 .804

Table 22: Item-Total Statistics of Attitude in Thailand and Vietnam

Perceived Usefulness

Refers to the table below, the Cronbach’s Alpha of Perceived Usefulness is 0.893 which is over the acceptable scale and considered as reliable. In the column ‘Cronbach’s Alpha If Item Deleted’, all the items are also qualified so we decide to keep all of them for the next tests.

Reliability Statistics

117

Cronbach's N of Items Alpha .893 4

Table 23: Reliability Statistics of Perceived Usefulness in Thailand and Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted PU1 12.8459 4.957 .772 .860 PU2 12.8260 5.004 .772 .860 PU3 13.0875 4.893 .728 .877 PU4 12.9056 4.924 .787 .854

Table 24: Item-Total Statistics of Perceived Usefulness in Thailand and Vietnam

Perceived Ease of Use

According to the below table, the Cronbach’s Alpha of Perceived Ease of Use is 0.855 which is qualified compared to our standard so we conclude that all the answers among this variable is quite reliable. In the column of Cronbach's Alpha if Item Deleted, there is no value higher than Cronbach's Alpha so we won’t eliminate any item in this variable.

Reliability Statistics Cronbach's N of Items Alpha .855 3

Table 25: Reliability Statistics of Perceived Ease of Use in Thailand and Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted PE1 8.36 2.562 .726 .799 PE2 8.27 2.494 .741 .785 PE3 8.33 2.519 .716 .809

Table 26: Item-Total Statistics of Perceived Ease of Use in Thailand and Vietnam

Subjective Norm

118

According to the below table, the Cronbach’s Alpha of Subjective Norm is 0.895 which also higher than the acceptable standard so we conclude that all the answers among this variable are reliable and qualified enough to proceed for the next tests.

Reliability Statistics Cronbach's N of Items Alpha .895 3

Table 27: Reliability Statistics of Subjective Norm in Thailand and Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted SN1 7.267 4.536 .807 .838 SN2 7.305 4.738 .809 .836 SN3 7.278 4.918 .764 .874

Table 28: Item-Total Statistics Subjective Norm in Thailand and Vietnam

Perceived Cost

Comparing to other variables, Perceived Cost has lowest Cronbach's Alpha but it also meets the minimum requirement. Therefore, we can conclude that this variable is also qualified enough for the next tests.

Reliability Statistics Cronbach's N of Items Alpha

.791 2

Table 29: Reliability Statistics of Perceived Cost in Thailand and Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted PC1 3.12 1.703 .655 . PC2 3.15 1.861 .655 .

Table 30: Item-Total Statistics of Perceived Cost in Thailand and Vietnam

119

Perceived Risk

Refers to the table below, the Cronbach’s Alpha of Perceived Risk is 0.888 which means that all the answers refer to this variable is reliable enough for the upcoming tests. In the column ‘Cronbach’s Alpha If Item Deleted’, it does not give us any indication that we should delete any item.

Reliability Statistics Cronbach's N of Items Alpha .888 3

Table 31: Reliability Statistics of Perceived Risk in Thailand and Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted PR1 7.419 4.024 .765 .856 PR2 7.308 4.073 .786 .837 PR3 7.272 4.059 .794 .830

Table 32: Item-Total Statistics of Perceived Risk in Thailand and Vietnam

Covid-19

As we can see from the result of previous Cronbach’s Alpha reliability test of other variables, it can be said that Covid-19 has the highest Cronbach's Alpha indication which mean the answers among this variable has the highest reliability compared to others. And we will not remove any item in this variable when all the values in “Cronbach's Alpha if Item Deleted” column is lower than Cronbach's Alpha.

Reliability Statistics Cronbach's N of Items Alpha .917 5

Table 33: Reliability Statistics of Covid-19 in Thailand and Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted C1 16.38345 12.894 .743 .908

120

C2 16.19657 12.771 .822 .890 C3 16.16774 12.756 .850 .885 C4 16.10313 13.435 .740 .907 C5 16.11034 13.405 .781 .899

Table 34: Item-Total Statistics of Covid-19 in Thailand and Vietnam

Vietnam

Behavior Intention

From the below statistics, we can see that the Cronbach’s Alpha for Behavior Intention is 0.790 which is acceptable, and we can consider that the answers for this variable are reliable.

Reliability Statistics Cronbach's N of Items Alpha .790 2

Table 35: Reliability Statistics of Behavior Intention in Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted BI1 3.93 .960 .653 . BI2 3.82 1.054 .653 .

Table 36: Item-Total Statistics of Behavior Intention in Vietnam

Attitude

From the reliability test for Attitude variable, we can see that the Cronbach’s Alpha is 0.861 which is higher than 0.7 and no item in Corrected Item Total Correlation column is less than 0.3. Hence, we conclude that the data for this variable is quite reliable and we can keep all the items in this variable.

Reliability Statistics Cronbach's N of Items Alpha .861 4

Table 37: Reliability Statistics of Attitude in Vietnam

Item-Total Statistics

121

Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted AT1 11.88 5.632 .744 .808 AT2 11.90 5.462 .737 .810 AT3 12.13 5.625 .655 .845 AT4 12.17 5.647 .696 .827

Table 38: Item-Total Statistics of Attitude in Vietnam

Perceived Usefulness

Relatively, the Cronbach’s Alpha of Perceived Usefulness is 0.914 which is considered as quite high; and no item in Corrected Item-Total Correlation column is lower than 0.3. Therefore, the reliability of this variable is qualified, and we will keep all the items in this variable since they all meet the qualification.

Reliability Statistics Cronbach's N of Items Alpha .914 4

Table 39: Reliability Statistics of Perceived Usefulness in Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted PU1 12.19 6.179 .793 .893 PU2 12.22 6.078 .798 .891 PU3 12.39 6.191 .770 .901 PU4 12.24 6.039 .858 .871

Table 40: Item-Total Statistics of Perceived Usefulness in Vietnam

Perceived Ease of Use

From the below tables, we can withdraw some conclusions for Perceived Ease of Use variable. First, its Cronbach’s Alpha is considered as high when it obtains 0.874 and we don’t delete any item in this variable since all of their values in Corrected Item-Total Correlation column are larger than 0.3

Reliability Statistics Cronbach's N of Items Alpha

122

.874 3

Table 41: Reliability Statistics of Perceived Ease of Use in Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted PE1 7.85 2.847 .731 .847 PE2 7.78 2.660 .778 .805 PE3 7.88 2.714 .767 .814

Table 42: Item-Total Statistics of Perceived Ease of Use in Vietnam

Subjective Norm

From the below statistics, we can see that the data collected from Subjective Norm is quite reliable when it has the significant Cronbach's Alpha of 0.913. And no item among is deleted as well because we have all values in Corrected Item-Total Correlation are over 0.3. We can conclude that the collected data from this variable is quite reliable and we can use all values from this for the upcoming tests

Reliability Statistics Cronbach's N of Items Alpha .913 3

Table 43: Reliability Statistics of Subjective Norm in Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted SN1 6.45 5.398 .830 .870 SN2 6.43 5.482 .841 .861 SN3 6.36 5.609 .804 .892

Table 44: Item-Total Statistics of Subjective Norm in Vietnam

Perceived Cost

From the below tables, we can withdraw some conclusions for Perceived Cost variable. First, its Cronbach’s Alpha is considered as high when it obtains 0.819 and we decide to keep all the items in this variable for the next tests.

123

Reliability Statistics Cronbach's N of Items Alpha .819 2

Table 45: Reliability Statistics of Perceived Cost in Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted PC1 2.73 1.393 .694 . PC2 2.60 1.430 .694 .

Table 46: Reliability Statistics of Perceived Cost in Vietnam

Perceived Risk

For the Perceived Risk as well, we highly apricate the reliability of this variable when it has significant number of Cronbach's Alpha (0.91). And as the result, we won’t remove any item in this variable since all of the items are larger than 0.3 in the Corrected Item-Total Correlation column and no value in Cronbach's Alpha if Item Deleted is over the Cronbach's Alpha. Reliability Statistics Cronbach's N of Items Alpha .910 3

Table 47: Reliability Statistics of Perceived Risk in Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted PR1 6.93 4.452 .803 .884 PR2 6.85 4.364 .827 .865 PR3 6.82 4.269 .829 .863

Table 48: Item-Total Statistics of Perceived Risk in Vietnam

Covid-19

The last discussed variable will be Covid-19. The reliability of this variable is also highly evaluated when its Cronbach's Alpha value is far away from the acceptable value of 0.7. Besides,

124

we will keep all the items in this variable for the next analysis when most of them are able to qualify standard.

Reliability Statistics Cronbach's N of Items Alpha .922 5

Table 49: Reliability Statistics of Covid-19 in Vietnam

Item-Total Statistics Scale Mean if Scale Variance Corrected Item- Cronbach's Item Deleted if Item Deleted Total Alpha if Item Correlation Deleted C1 15.11 15.213 .730 .919 C2 14.85 14.691 .827 .899 C3 14.78 14.975 .860 .893 C4 14.67 15.214 .780 .908 C5 14.74 15.415 .803 .904

Table 50: Item-Total Statistics of Covid-19 in Vietnam

Thailand

Behavioral Intention

According table 51, our Cronbach’s Alpha of Behavioral Intention is 0.805 which is considered reliable. In table 52, it does not suggest us to delete any item.

Reliability Statistics Cronbach's Alpha N of Items .805 2

Table 51: Reliability Statistics of Behavioral Intention in Thailand

Item-Total Statistics Corrected Item- Cronbach's Scale Mean if Scale Variance Total Alpha if Item Item Deleted if Item Deleted Correlation Deleted BI1 4.44 .692 .674 . BI2 4.45 .621 .674 .

Table 52: Item-Total Statistics of Behavioral Intention in Thailand

125

Attitude

In table 53, this shows the Cronbach’s Alpha value of the ‘Attitude’ construct. The value of 0.791 is considered high and reliable. In table 54, the column of ‘Cronbach’s Alpha If Item Deleted’ has shown that if we delete the item number AT2, we will get a higher value of Cronbach’s Alpha of 0.806 which is even more reliable. Therefore, before we make a further test, we will delete the item AT2

Reliability Statistics Cronbach's Alpha N of Items .791 4

Table 53: Reliability Statistics of Attitude in Thailand

Item-Total Statistics Corrected Item- Cronbach's Scale Mean if Scale Variance Total Alpha if Item Item Deleted if Item Deleted Correlation Deleted AT1 13.54 3.083 .615 .733 AT2 13.40 3.715 .451 .806 AT3 13.58 2.738 .675 .700 AT4 13.53 2.963 .674 .702

Table 54: Item-Total Statistics of Attitude in Thailand

Perceived Usefulness

Refers to table 55, the Cronbach’s Alpha of Perceived Usefulness is 0.804 which is considered reliable and in table 56 below, in the column ‘Cronbach’s Alpha If Item Deleted’ does not give us any indication if we should delete any item.

Reliability Statistics Cronbach's Alpha N of Items .804 4

Table 55: Reliability Statistics of Perceived Usefulness in Thailand

Item-Total Statistics

126

Corrected Item- Cronbach's Scale Mean if Scale Variance Total Alpha if Item Item Deleted if Item Deleted Correlation Deleted PU1 13.54 2.800 .656 .736 PU2 13.48 3.117 .637 .754 PU3 13.81 2.553 .605 .768 PU4 13.62 2.803 .606 .760

Table 56: Item-Total Statistics of Perceived Usefulness in Thailand

Perceived Ease of Use

Refers to table 57, the Cronbach’s Alpha of Perceived Ease of Use is 0.783 which is considered reliable enough and in table 58 below, in the column ‘Cronbach’s Alpha If Item Deleted’ does not give us any indication if we should delete any item.

Reliability Statistics Cronbach's Alpha N of Items .783 3

Table 57: Reliability Statistics of Perceived Ease of Use in Thailand

Item-Total Statistics Corrected Item- Cronbach's Scale Mean if Scale Variance Total Alpha if Item Item Deleted if Item Deleted Correlation Deleted PE1 8.89 1.725 .666 .655 PE2 8.78 1.823 .627 .700 PE3 8.79 1.910 .571 .758

Table 58: Item-Total Statistics of Perceived Ease of Use in Thailand

Subjective Norm

Refers to table 59, the Cronbach’s Alpha of Subjective Norm is 0.771 which is considered reliable and in table 60, in the column ‘Cronbach’s Alpha If Item Deleted’ does not give us any indication if we should delete any item.

127

Reliability Statistics Cronbach's Alpha N of Items .771 3

Table 59: Reliability Statistics of Subjective Norm in Thailand

Item-Total Statistics Corrected Item- Cronbach's Scale Mean if Scale Variance Total Alpha if Item Item Deleted if Item Deleted Correlation Deleted SN1 8.13 2.286 .634 .658 SN2 8.23 2.382 .627 .667 SN3 8.22 2.465 .555 .746

Table 60: Item-Total Statistics of Subjective Norm in Thailand

Perceived Cost Refers to table 61, the Cronbach’s Alpha of Perceived Cost is 0.702 which is considered reliable enough and in table 62, in the column ‘Cronbach’s Alpha If Item Deleted’ does not give us any indication if we should delete any item.

Reliability Statistics Cronbach's Alpha N of Items .702 2

Table 61: Reliability Statistics of Perceived Cost in Thailand

Item-Total Statistics Corrected Item- Cronbach's Scale Mean if Scale Variance Total Alpha if Item Item Deleted if Item Deleted Correlation Deleted PC1 3.52 1.716 .541 . PC2 3.71 1.690 .541 .

Table 62: Item-Total Statistics of Perceived Cost in Thailand

Perceived Risk

Refers to table 63, the Cronbach’s Alpha of Perceived Risk is 0.841 which is considered reliable and in table 64, in the column ‘Cronbach’s Alpha If Item Deleted’ does not give us any indication if we should delete any item.

128

Reliability Statistics Cronbach's Alpha N of Items .841 3

Table 63: Reliability Statistics of Perceived Risk in Thailand

Item-Total Statistics Corrected Item- Cronbach's Scale Mean if Scale Variance Total Alpha if Item Item Deleted if Item Deleted Correlation Deleted PR1 7.94 3.122 .700 .788 PR2 7.80 3.392 .702 .783 PR3 7.74 3.423 .720 .767

Table 64: Item-Total Statistics of Perceived Risk in Thailand

Covid-19 Refers to table 65, the Cronbach’s Alpha of Covid-19 construct is 0.852 which is considered reliable and in table 66, in the column ‘Cronbach’s Alpha If Item Deleted’ does not give us any indication if we should delete any item.

Reliability Statistics Cronbach's Alpha N of Items .852 5

Table 65: Reliability Statistics of Covid-19 in Thailand

Item-Total Statistics Corrected Item- Cronbach's Scale Mean if Scale Variance Total Alpha if Item Item Deleted if Item Deleted Correlation Deleted C1 17.72 6.746 .635 .829 C2 17.60 6.663 .724 .805 C3 17.60 6.468 .749 .798 C4 17.60 6.954 .587 .841 C5 17.54 7.067 .626 .831

Table 66: Item-Total Statistics of Covid-19 in Thailand

To summarize, only the item AT2 from the reliability test that we decided to delete and will not be included for the further analysis. Overall, the reliability test is considered reliable for all constructs.

129

Exploratory Factor Analysis (EFA)

After our evaluation of the Cronbach’s Alpha reliability analysis, the exploratory factor analysis (EFA) would be used to uncover the complex patterns by exploring the dataset and testing predictions (Child, D. 2006). Therefore, we use EFA as another tool to see which variables we should retain, and which variables should we delete (Yong, A. & Pearce, S. 2013). The adequacy of the samples is measured by The Kaiser Meyer Olkin test of sampling adequacy (KMO) in SPSS (Hadi, N. et a. 2016). According to Kaiser, H. F. (1974), There are several standards applied in this test, if KMO >= 0.90 is considered very good; if KMO >=0.80 is considered good; if KMO >=0.70 is considered acceptable, if KMO >= 0.60 is considered so-so; if KMO >=0.50 is considered bad; if KMO< 0.50 is considered unacceptable (Kaiser, H. F. 1974).

Moreover, the strength of the relationship in SPSS can be measured by Bartlett’s Test of Sphericity. The significant value less than 0.05 indicates that these data do not produce an identity matrix and are approximately multivariate normal and acceptable for further analysis (Field, A. 2000).

In this EFA test have been conducted for independent variables. The total of 26 items will be proved the reliability through Cronbach’s Alpha test. Afterwards, we will considered delete the unreliable items for the EFA test. In the EFA technique, we will be using with “Principal component method” and “Varimax rotation”. According to the rule of thumb, a rotated factor loading for a sample size of at least 300 would need to be at least .32, in order for the variables to be considered meaningful (Tabachnick, B. G., & Fidell, L. S. 2007). In our research, both of the samples are still below 300 respondents but the number of respondents is almost reaching 300, therefore, we will be using this standard for our decision of retaining or deleting our items. In addition to the criteria of deleting or retaining items in EFA, it is recommended that the researchers should delete items with factor loadings less than .32 or cross - loadings less than .15 difference from an item’s highest factor loading (Worthington, R. L., & Whittaker, T. A. 2006).

To be added, it is suggested we will be looking at the Total Variance Explained table to determine the number of significant factors, especially look at the Extraction Sums of Squared Loadings where the cumulative column shows the percentage of all factors are explained by the percentage of data variability (Yong, A. & Pearce, S. 2013). Kaiser, H. F. (1960) suggested the eigenvalues should be more than 1 and the Extraction Sums of Squared Loadings, in cumulative percentage, it should be more than 50% (Kaiser, H. F. 1960; 1974).

130

Combination of two countries dataset

As we can see from table 67, the KMO is 0.935 which even exceeds the good standard. Besides, the significant value of Bartlett's Test falls below 0.05 illustrating that this data set is qualified enough to be proceeded for the upcoming tests.

KMO and Bartlett's Test Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .935 Approx. Chi-Square 8761.928 Bartlett's Test of Sphericity df 276 Sig. .000

Table 67: KMO and Bartlett's Test for Thailand and Vietnam dataset

Then, we look at the Extraction Sums of Squared Loadings where the cumulative column shows 70.519% of all factors are explained by the percentage of data variability. In this test, the acceptable Cumulative % should be higher than 50%. Besides, our Initial Eigenvalue is also qualified when it exceeds 1.

Total Variance Explained Compo Initial Eigenvalues Extraction Sums of Squared Rotation Sums of Squared Loadings nent Loadings Total % of Cumulative Total % of Cumulative Total % of Cumulative Variance % Variance % Variance % 1 9.895 41.228 41.228 9.895 41.228 41.228 7.095 29.564 29.564 2 3.533 14.720 55.947 3.533 14.720 55.947 3.769 15.703 45.267 3 2.059 8.580 64.528 2.059 8.580 64.528 3.052 12.718 57.985 4 1.438 5.991 70.519 1.438 5.991 70.519 3.008 12.534 70.519 5 .699 2.913 73.432 6 .638 2.660 76.092 7 .613 2.555 78.646 8 .485 2.020 80.666 9 .419 1.748 82.414 10 .409 1.706 84.119 11 .379 1.580 85.699 12 .365 1.520 87.219 13 .346 1.442 88.661 14 .328 1.365 90.026 15 .305 1.269 91.295

131

16 .301 1.254 92.549 17 .274 1.140 93.689 18 .253 1.055 94.745 19 .252 1.048 95.793 20 .233 .970 96.763 21 .222 .925 97.688 22 .196 .817 98.504 23 .188 .782 99.287 24 .171 .713 100.000

Table 68: Extraction Method: Principal Component Analysis of Thailand and Vietnam dataset

In the last step of EFA analysis, we will interpret the value of the factor loading. As we already mentioned in the beginning, we would have cross factor loading and factor loading in this Rotated Component Matrix table. For factor loadings, it will be acceptable when it is higher than 0.32. As we can see from this table, all of our factor loadings are over 0.5 which will be completely qualified for this test. And for the cross-factor loadings, we will calculate the difference between largest value and smallest value and this result will be supported if it is higher than 0.15. All results of factor loading will be AT4 (0.427), C3 (0.544), C4 (0.429), C5 (0.294), PC1 (0.072), PC2 (0.139), SN1 (0.486). As we can see from the above results, there are only PC1 and PC2 which have unqualified results so we can conclude that these items have some problems.

Rotated Component Matrixa Component 1 2 3 4 PU1 .817 PU4 .817 PU2 .779 PU3 .771 AT1 .768 AT2 .767 PE1 .753 AT4 .728 .301 PE2 .723 AT3 .704 PE3 .701 C3 .310 .854 C2 .838 C1 .800 C4 .334 .763

132

C5 .447 .741 PR3 .876 PR2 .875 PR1 .875 PC2 .627 .488 SN2 .841 SN3 .803

SN1 .316 .802

PC1 .503 .575 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 6 iterations. Table 69: Rotated Component Matrix of Thailand and Vietnam dataset

Vietnam

As we see from table 70, the KMO indication is 0.906 which is even higher than 0.9 so we can conclude that our data is very good (Kaiser, H. F. 1974). Also, the significant value of Bartlett's Test of Sphericity is less than 0.05 which indicate that this data will not distribute any identity matrix and it’s approximately multivariate normal and acceptable for further analysis (Field, A. 2000).

KMO and Bartlett's Test Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .906 Approx. Chi-Square 5253.465 Bartlett's Test of Sphericity df 325 Sig. .000

Table 70: KMO and Bartlett's Test for Vietnam dataset

Next, from the result of Total Variance Explained table, we can see that the final cumulative value from Extraction Sums of Squared Loadings column is acceptable because it is larger than 50%. Here, our cumulative percentage in Extraction Sums of Squared Loadings is 70.074% which means that all factors are 70.074% explained by this model (Kaiser, H. F. 1960; 1974). Additionally, it shows that the total value in Initial Eigenvalues column is also higher than 1 which is totally qualified the criteria (Kaiser, H. F. 1960; 1974). After we evaluate all the necessary element in first two tables and all the criteria are qualified enough, then we should move to Rotated Component Matrix table.

133

Total Variance Explained

Compone Initial Eigenvalues Extraction Sums of Squared Rotation Sums of Squared Loadings nt Loadings

Total % of Cumulative Total % of Cumulative Total % of Cumulative % Variance % Variance % Variance 1 9.974 38.361 38.361 9.974 38.361 38.361 8.231 31.658 31.658 2 3.959 15.225 53.587 3.959 15.225 53.587 3.920 15.077 46.735 3 2.399 9.227 62.813 2.399 9.227 62.813 3.180 12.231 58.966 4 1.888 7.261 70.074 1.888 7.261 70.074 2.888 11.109 70.074 5 .937 3.604 73.679 6 .803 3.088 76.767 7 .674 2.592 79.359 8 .613 2.359 81.718 9 .493 1.898 83.615 10 .454 1.747 85.362 11 .398 1.531 86.893 12 .390 1.499 88.392 13 .328 1.262 89.654 14 .304 1.170 90.825 15 .277 1.066 91.890 16 .276 1.063 92.954 17 .251 .964 93.917 18 .235 .902 94.819 19 .221 .849 95.668 20 .214 .823 96.491 21 .198 .761 97.252 22 .163 .628 97.880 23 .157 .602 98.482 24 .148 .568 99.050 25 .130 .498 99.549 26 .117 .451 100.000

Extraction Method: Principal Component Analysis. Table 71: Total Variance Explained of Vietnam dataset

To assess the data from table 72 we should first consider whether it is single factor loading or cross factor loading. For the single factor loading, this value will be qualified when it is larger than 0.32; and for the cross factor loading, the difference between the highest value and smallest value should not be less than 0.15 (Worthington et al., 2006). First, the results of the difference

134

between highest value and lowest value in cross factor loading are C4 (0.472), C5 (0.376), PC1 (0.159), PC2 (0.105). As the result, only PC2 has some problem and we will detect this problem in the future test of CFA. Then, we will evaluate the single factor loading. As we can see, all the factor loadings among below table vary from 0.6 to 0.9 which are all qualified to be proceeded for the next tests.

Rotated Component Matrixa Component 1 2 3 4 PU4 .848 PU1 .834 AT1 .812 PU3 .802 AT2 .799 PU2 .795 PE2 .747 BI2 .735 PE1 .729 AT4 .726 PE3 .725 BI1 .698 AT3 .692 C3 .872 C2 .851 C1 .829 C4 .327 .799 C5 .405 .781 SN2 .877 SN3 .847 SN1 .844 PC2 .587 .482 PC1 .555 .396 PR2 .904 PR3 .897 PR1 .882 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 6 iterations. Table 72: Rotated Component Matrix for Vietnam dataset

135

Thailand

According to table 73, we look at the KMO and the Bartlett’s Test, our KMO is .892 which is considered good, moreover, the sig. value in the Bartlett’s Test of Sphericity is very small as shown below .000, this reflects that our dataset is reliable enough to be used for the further analysis.

KMO and Bartlett's Test Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .892 Bartlett's Test of Sphericity Approx. Chi-Square 3016.774 df 253 Sig. .000

Table 73: KMO and Bartlett's Test for Thailand dataset

Next step, we look at the Extraction Sums of Squared Loadings where the cumulative column shows 62.562 % of all factors are explained by the percentage of data variability. The cumulative percentage is over 50% which is acceptable. Moreover, when we look at the Initial Eigenvalues, the row number 4 shows the value above 1, which is 1.191. Therefore, our criteria have been met with the standard and we tend proceed to look at the table “Rotated Component Matrix”.

136

Total Variance Explained Extraction Sums of Squared Rotation Sums of Squared Initial Eigenvalues Loadings Loadings % of Cumulative % of Cumulative % of Cumulative Component Total Variance % Total Variance % Total Variance % 1 8.007 34.814 34.814 8.007 34.814 34.814 5.341 23.221 23.221 2 3.279 14.257 49.072 3.279 14.257 49.072 3.296 14.330 37.552 3 1.912 8.312 57.384 1.912 8.312 57.384 3.111 13.526 51.078 4 1.191 5.179 62.562 1.191 5.179 62.562 2.641 11.485 62.562 5 .865 3.762 66.324 6 .796 3.460 69.784 7 .678 2.948 72.732 8 .641 2.787 75.519 9 .569 2.472 77.991 10 .552 2.401 80.392 11 .524 2.277 82.669 12 .490 2.130 84.799 13 .437 1.900 86.700 14 .431 1.874 88.573 15 .412 1.793 90.366 16 .358 1.558 91.924 17 .344 1.497 93.422 18 .331 1.438 94.859 19 .287 1.247 96.106 20 .260 1.131 97.238 21 .231 1.004 98.242 22 .213 .925 99.167 23 .192 .833 100.000 Extraction Method: Principal Component Analysis.

Table 74: Total Variance Explained for Thailand dataset

In table 75, we tend to look at the factor loadings, whether the value is above 0.32 or not. When the item has cross factor loadings, we need to make a calculation so the difference of the highest factor loading, and the lower ones will be above 0.15. Otherwise, we need to delete the item that has not fit our criteria.

137

By taking a look at the rows that has cross loadings first, the item AT4 has the cross factor loading, the difference is 0.306 (.696-.390) which is acceptable, therefore, we decide to keep this item. Next item is PE2, we decide to delete the item since the difference of cross factor loading is about 0.14 (.583 - .443) which is lower than 0.15. For the PE3, the difference of cross loading factor is about 0.218 (.576-.358) which is higher than the cut off of 0.15, therefore, we decide to keep this item. Next item is the item PC1, where the difference cross loading factor is 0.087 (.591 - .504) which is extremely low, since our cut off is 0.15, therefore, we need to delete this item. Next item is the C5, where the difference of cross loading is 0.193 (.628-.435) which is still higher than our threshold of 0.15, therefore we decided to keep this item.

Besides the cross-factor rows in the table below, we screen through all the factors and found out that many items have the factor loading of between 0.6 to 0.8 which is considered highly usable for the future analysis.

Rotated Component Matrixa Component 1 2 3 4 AT1 .660 AT3 .708 AT4 .696 .390 PU1 .797 PU2 .718 PU3 .683 PU4 .699 PE1 .710 PE2 .583 .443 PE3 .576 .358 SN1 .671 SN2 .778 SN3 .664 PC1 .591 .504 PC2 .751 PR1 .864 PR2 .827 PR3 .811 C1 .743

138

C2 .811 C3 .814 C4 .664 C5 .435 .628 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 6 iterations. Table 75: Rotated Component Matrix for Thailand dataset

To summarize the deletion of the items from the reliability test Cronbach’s Alpha and the EFA, we then conclude that that items that we decided to not proceed further is the item AT2, PE2, and PC1.

Average Variance Extracted (AVE) and Composite Reliability (CR)

The criterion of Fornell - Larcker (1981) has been commonly used to assess the degree of shared variance between the latent variables of the model. According to Fornell & Larcker (1981), the convergent validity of the measurement model can be assessed by the Average Variance Extracted (AVE) and Composite Reliability (CR). It is suggested that the value of AVE should be more than 0.5 and the value of CR should be more than 0.6 (Fornell & Larcker, 1981). However, in case of AVE is less than 0.5 but CR is higher than 0.6, the convergent validity of the construct is still adequate (Fornell & Larcker, 1981).

Combination of two countries dataset

After assessing the reliability of data by considering Cronbach’s α, we keep proceeding other reliability tests with AVE and CR for the data set to ensure the validity of the construct in this research. First of all, the combination data of Thailand and Vietnam will be assessed to establish the common standard for each country later. As we can see in the below table, all variables have their AVE and CR are over 0.5 and 0.6 relatively which mean all of these constructs are normal and qualified enough for the research. However, Perceived Cost is the only variable which has AVE falls below 0.5 and its CR is also less than 0.6 so we can see that this construct is problematic and we will keep testing it in next tests to define these problems.

139

Construct Items Standardized Cronbach’s AVE CR Loading α Behavioral BI1 0.698 0.816 0.5093225 0.67479862 Intention BI2 0.729 Attitude AT1 0.768 AT2 0.767 0.854 0.55092825 0.830528986 AT3 0.704 AT4 0.728 Perceived PU1 0.817 Usefulness PU2 0.779 0.893 0.634065 0.873832876 PU3 0.771 PU4 0.817 Perceived PE1 0.753 Ease of Use PE2 0.723 0.855 0.527046333 0.769597723 PE3 0.701 Subjective SN1 0.802 Norm SN2 0.841 0.895 0.665098 0.856216321 SN3 0.803 Perceived PC1 0.575 0.791 0.361877 0.53097297 Cost PC2 0.627 Perceived PR1 0.875 Risk PR2 0.875 0.888 0.766208667 0.907680542 PR3 0.876 Covid 19 C1 0.800 C2 0.838 C3 0.854 0.917 0.640562 0.898836523 C4 0.763 C5 0.741

Table 76: Summarization of Standardized loading, Cronbach’s α, AVE and CR – analysis of Thailand and Vietnam. Source: Self-calculated

Vietnam

After conducting reliability test with SPSS, we proceed another step with hand-calculated AVE and CR to ensure the better quality of reliability for our data set. In table 77, we will give you the overview about factors loadings, Cronbach’s Alpha, AVE and CR. The qualified value for AVE

140

should be higher than 0.5 and CR should be over 0.6. However, in the situation that AVE is lower than 0.5 but CR is still higher than 0.6, we can accept this variable. As we can see in the below table, only Behavior Intention and Perceived Cost have AVE value lower than 0.5 but CR value of these variables exceeds 0.6 so their constructs are qualified for this model. Similar to the combination data test above, the problem with Perceived Cost is detected when its AVE and CR are both less than 0.5 and 0.6 relatively. In this step, we can conclude that the construct of Perceived Cost is problematic in this model and this problem will be found out in upcoming tests.

Construct Items Standardized Cronbach’s AVE CR Loading α Behavioral Intention BI1 0.662 0.790 0.4551065 0.62544972 BI2 0.687 Attitude AT1 0.798 AT2 0.813 0.861 0.5769685 0.844547296 AT3 0.695 AT4 0.726 Perceived Usefulness PU1 0.840 0.69137825 0.899554455 PU2 0.813 0.914

PU3 0.812

PU4 0.860 Perceived Ease of Use PE1 0.759 0.568632667 0.798135506 PE2 0.764 0.874

PE3 0.739 Subjective Norm SN1 0.847 0.737480667 0.893907778 SN2 0.877 0.913

SN3 0.852 Perceived Cost PC1 0.554 0.819 0.3222385 0.487272236 PC2 0.581 Perceived Risk PR1 0.880 0.804150667 0.924901018 PR2 0.906 0.910

PR3 0.904

Covid 19 C1 0.835 0.915575572 C2 0.854 0.922 0.68483

C3 0.872

C4 0.792

C5 0.781 141

Table 77: Summarization of Standardized loading, Cronbach’s α, AVE and CR – analysis of Vietnam. Source: Self- calculated

Thailand

Under table 78, we have summarized the loadings, Cronbach’s Alpha value from our reliability analysis above and the self-calculation of the AVE and CR. According to our cut off, the AVE should be above 0.5, meanwhile CR should be above 0.6, if the AVE value falls below 0.5, but the CR value is still above 0.6, it is still considered reliable. Therefore, we take a look at the column below and we have found out that all of our constructs contain CR value above 0.6 which is considered reliable, however, there are some constructs that has AVE below 0.5 which are Attitude, Perceived Usefulness, Perceived Ease of Use and Perceived Cost. However, those constructs are still considered reliable by looking the value of CR.

Construct Items Standardized Cronbach’s AVE CR Loading α Behavioral BI1 0.794 0.805 0.623331 0.767959 Intention BI2 0.785 Attitude AT1 0.540 AT2 0.570 0.791 0.358044 0.689336 AT3 0.626 AT4 0.651 Perceived PU1 0.789 Usefulness PU2 0.745 0.804 0.475589 0.781365 PU3 0.603 PU4 0.601 Perceived PE1 0.695 Ease of Use PE2 0.573 0.783 0.383748 0.649522 PE3 0.583 Subjective SN1 0.673 Norm SN2 0.782 0.771 0.500018 0.748886 SN3 0.660 Perceived PC1 0.594 0.702 0.460676 0.627507 Cost PC2 0.754 Perceived PR1 0.866 0.841 0.694482 0.872017 Risk PR2 0.825

142

PR3 0.808 Covid 19 C1 0.736 C2 0.804 C3 0.809 0.852 0.536657 0.851519 C4 0.669 C5 0.627

Table 78: Summarization of Standardized loading, Cronbach’s α, AVE and CR – analysis of Thailand. Source: Self- calculated

Summarization of AVE and CR between all datasets

For the combination dataset the AVE seems to be lower than 0.5 for only the construct perceived cost, the CR for perceived cost is also lower than other constructs, which is similar to Vietnam and Thailand. For Vietnam and Thailand, the AVE seems to be quite low for behavioral intention and perceived cost for Vietnam. The CR values in Vietnam is acceptable except for the perceived cost. For Thailand, the AVE is lower than 0.5 for many constructs such as attitude, perceived usefulness, perceived ease of use and perceived cost. CR in Thailand dataset has no problem comparing to other datasets. This reflects that we might still have an issue about reliability tests and some items or constructs need to be removed or merged, especially for the construct perceived cost across all datasets.

Discriminant Validity Test

According to Hubley A.M. (2014), discriminant validity is the measure of constructs that suggested that it should not be highly related to one another. If the correlation coefficients are high, this shows the lack of discriminant validity (Beck, L. et al. 2004). Another way around, if the correlation coefficients are low, it demonstrates that there is discriminant validity (Beck, L. et al. 2004).

Discriminant validity assessment is generally well-accepted widely for the purpose of analyzing relationships between variables (Henseler, J. et al. 2014). Partial least squares, Fornell- Larcker criterion and examination of cross loadings are the dominant approach for evaluating discriminant validity for the variance-based structural equation modeling (Henseler, J. et al. 2014). As suggested by Fornell and Larcker (1981), the lack of convergent validity is to see if AVE is below the cut off of 0.5. The discriminant validity could be evaluated by comparing the squared root of AVE and the square of the correlation between factors. According to Fornell and Larcker

143

(1981), if the squared root of AVE is higher than the coefficient of the correlation between factors, this provides evidence of discriminant validity. Moreover, it is suggested that the AVE should be greater than maximum shared variance (MSV) (Fornell & Larcker, 1981).

All the criteria mentioned earlier was the traditional discriminant validity assess methods. According to Hair et al. (2013), the square root of each AVE coefficients constructs should be greater than its highest correlation with any other construct to evidence discriminant validity.

Combination of two countries dataset

As we already have AVE from the above calculation, we will continue to self-calculate MSV in this step by getting the square of the highest correlation coefficient in each construct. Firstly, we will compare between AVE and MSV, the results will be highly evaluated when AVE is larger than MSV otherwise these constructs will meet high correlation problems. From the below table, we can illustrate that Attitude, Perceived Usefulness, Perceived Ease of Use have AVE values smaller than MSV values so we can shortly conclude that these constructs have some problems with high correlation. In the next step, we will calculate the square root of AVE and put it in front of each construct then we need to compare this value with the highest correlation to see whether the construct is problematic with high correlation or not. As we can see from the below table, Behavioral Intention and Perceived Usefulness all have square root of AVE higher than highest correlation so we can conclude that these constructs have some issues with discriminant validity.

Perceived Behaviora Perceived Subjectiv Perceive Perceived Covid AVE MSV Attitude Ease of l Intention Usefulness e Norm d Cost Risk 19 Use Behavioral 0.509 0.491 0.714 Intention

Attitude 0.551 0.702 0.701 0.742

Perceived 0.634 0.702 0.664 0.838 0.796 Usefulness Perceived 0.527 0.593 0.581 0.77 0.766 0.726 Ease Of Use Subjective 0.665 0.253 0.327 0.456 0.414 0.503 0.816 Norm Perceived 0.362 0.309 0.1 0.121 0.096 0.181 0.499 0.602 Cost Perceived 0.766 0.309 0.143 0.142 0.086 0.147 0.299 0.556 0.875 Risk

Covid19 0.641 0.343 0.468 0.586 0.562 0.531 0.329 0.11 0.184 0.800

Table 79: Factor AVE- Correlation Matrix for Thailand and Vietnam dataset. Source: Self-edited

144

Vietnam

Then, we will have a look at dataset of Vietnam to define the discriminant validity problem in this dataset. As we can see from the below table, Behavioral Intention, Attitude and Perceived Usefulness tend to have AVE less than MSV so we can shortly suppose that these constructs have some issues with high correlation. The other constructs have no problem with high correlation because their AVE are all larger than MSV. Next, we will compare square root of AVE in each construct and its highest correlation to define the discriminant validity issue. In the below table, we can see that only Perceived Usefulness is problematic with discriminant validity because its square root of AVE higher than highest correlation. However, there is no further evidence to support that Behavioral Intention and Attitude have high correlation issues when their square root of AVE values is both less than their highest correlation. In the next chapter, we are able to define the problem of high correlation in more detail by CFA test.

Perceived Behaviora Perceived Subjectiv Perceive Perceived Covid AVE MSV Attitude Ease Of l Intention Usefulness e Norm d Cost Risk 19 Use Behavioral 0.455 0.487 0.675 Intention

Attitude 0.577 0.709 0.698 0.760

Perceived 0.691 0.709 0.665 0.842 0.831 Usefulness Perceived 0.569 0.546 0.574 0.739 0.77 0.754 Ease Of Use Subjective 0.737 0.182 0.144 0.293 0.27 0.346 0.859 Norm Perceived 0.322 0.189 -0.099 -0.052 -0.064 0.049 0.427 0.568 Cost Perceived 0.804 0.189 0.103 0.094 -0.001 0.085 0.236 0.435 0.897 Risk

Covid19 0.685 0.264 0.395 0.514 0.486 0.459 0.098 -0.129 0.026 0.828

Table 80: Factor AVE- Correlation Matrix for Vietnam dataset. Source: Self-Edited

Thailand

According to table 81, we have the column comparing the AVE and self-calculated MSV. Along with the correlation between each of the constructs. Firstly, according to the AVE as comparing to MSV, we expect the AVE values should be higher than the MSV otherwise these constructs are having problem with highly correlation. As seen below, Perceived Usefulness and Perceives Ease of Use tend to have AVE values lower than MSV which is considered as that they have issues with high correlation between constructs. Then we did self-calculation of the square

145

root of the AVE, therefore we put them above of each column comparing the correlation to its squared root AVE, again, we need to see if any construct has problem with high correlation. In the column of Attitude, the squared root of AVE is 0.5980 which is lower than the highest correlation coefficient of that column which are 0.728 and 0.692 respectively which is the Perceived Usefulness and Perceived Ease Of Use, which means there is no evidence of discriminant validity or that the constructs are highly correlated. The rest of the constructs has higher of squared root of AVE than the highest correlation coefficient, therefore, when we did the CFA, these items in the highly correlated construct will be analyzed whether we need to merge them or delete them in the next chapter.

Perceived Behavioral Perceived Subjective Perceive Perceived Covid AVE MSV Attitude Ease Of Intention Usefulness Norm d Cost Risk 19 Use

Behavioral 0.623 0.624 Intention 0.790

Attitude 0.358 0.317 0.563 0.598

Perceived 0.476 0.530 0.561 0.728 0.690 Usefulness

Perceived Ease Of 0.384 0.454 0.451 0.692 0.674 0.619 Use

Subjective 0.500 0.315 0.383 0.526 0.473 0.561 0.707 Norm

Perceived 0.461 0.176 0.020 0.002 -0.059 -0.051 0.266 0.679 Cost

Perceived 0.694 0.379 0.018 0.025 0.029 0.054 0.202 0.616 0.833 Risk

Covid19 0.537 0.048 0.402 0.521 0.529 0.465 0.419 0.075 0.218 0.733

Table 81: Factor AVE- Correlation Matrix for Thailand dataset. Source: Self-Edited

Summarization of discriminant validity tests between all datasets

According to our data above, when comparing AVE to MSV, the AVE should be more than MSV, otherwise there is the problem with the multicollinearity. Our combination dataset contains three constructs that are problematic including the attitude, perceived usefulness and perceived ease of use. For Vietnam dataset, only behavioral intention, attitude, and perceived usefulness that has an issue with multicollinearity, not the perceived ease of use. For Thailand, the result is quite similar to the combination of dataset, only the constructs attitude, perceived usefulness and perceived ease of use that has problem with multicollinearity. Therefore, we need to proceed to

146

CFA in order to make a good model fit so this will solve our problems with highly correlated items and constructs.

Normality Test

To see whether the populations from which the samples are taken are normally distributed or not, there are several methods to see them, both visually and through normality tests (Ghasemi, A. & Zahediasl, S. 2012). According to Ghasemi, A. & Zahediasl, S. (2012), in the large samples of more than 30 – 40 respondents, the sampling distribution tend to be normal, regardless the shape of data. Refer to the visual methods, data will be represented in graphs such as histogram, stem-and-leaf plot, box plot, P – P plot and Q-Q plot tests (Ghasemi, A. & Zahediasl, S. 2012). There are used for checking the normality tests (Ghasemi, A. & Zahediasl, S. 2012). However, the main tests are for instance Kolmogorov-Smirnov test or K-S test, Lilliefors corrected K-S test, Shapiro-Wilk test and Anderson-Darling test (Razali, M.N, et al. 2011). Among all these tests, K- S is much used test and both K-S and Shapiro-Wilk test can be run by using SPSS (Ghasemi, A. & Zahediasl, S. 2012).

To a contrary, according to Norman, G. (2010) there is an exception when using parametric statistics such as normal distribution tests or normality tests, as data may not be normally distributed under the Likert Scales which are considered ordinal – therefore the parametric statistics cannot be used. To sum up, we have decided not to proceed further with the normality tests due to the fact that this method is not suitable for our rating scales, which is the Likert Scale.

Confirmatory Factor Analysis (CFA)

The confirmatory factor analysis (CFA) is an important tool for structural equation modeling techniques that allows the investigation of casual relations among latent and observed variables (Mueller, R.O. & Hancock, 2001). CFA is well known for the best practice to understand the data- model fit assessment and potential model modification (Mueller, R.O. & Hancock, 2001). Moreover, it is widely used for assessing both convergent and discriminant validity (Shaakumeni, S.N & Csapo, 2019). According to Gallagher M.W. & Brown T.A. (2013), after evaluating the items in EFA, researchers are suggested to move to CFA framework to provide how a theoretical model represents the observed data. CFA also provides superior methods of evaluating reliability than the traditional methods such as Cronbach’s Alpha (Gallagher M.W. & Brown T.A. 2013).

147

By dealing the definition of a goodness of fit (GOF) in the model, many experts recommended many cut-off values (Prudon, P. 2015). It is crucial that we acknowledge the procedures of using CFA, for instance, to test the predicted factor structure with CFA, the predicted correlations and error variances of the factor structure should be translated back to the correlation matrix (Prudon, P. 2015). It is necessary to have a look at the implied matrix and residual matrix and compare them (Prudon, P. 2015). The two matrices when comparing , it is expressed in X2 or so-called Model Chi Square with the degrees of freedom (df), therefore, the X2 should be small enough in relation to df or significant enough (P – value is under 0.05).Next value that is important in running CFA is the GFI or AGFI (Adjusted Goodness of Fit Index) , According to Williams and Holahan, (1994), the GFI as well as the comparative fit index (CFI) and the root mean square error of approximation (RMSEA) were used, and the cut off of both GFI and CFI should be 0.95, which indicates a good fit , if the value is higher than 0.9, it is considered acceptable fit.

To go through more of the meaning and definition, CFI as referred by Bentler (1990), subtracting df from X2 provides some parameter, where values are suggested that CFI should be more or equal to 0.95 as a cut-off value for a good fit (Hu L.T. & Bentler, P.M. 1999). In addition, RMSEA is based on X2, df and N, according to Hu L.T. & Bentler, P.M. 1999, ‘0’ indicates a perfect fit and less than or equal to 0.6 is considered a cut-off value for a good fit. Nest, the SRMR, or so-called standardized root mean square residual, according to Hu L.T. & Bentler, P.M. 1999, ‘0’ indicates a perfect fit and suggest a cut-off value of less than or equal to 0.8 as a good fit. Moreover, another important parameter of CFA is called TLI (Tucker-Lewis index, 1973) or also is known as (N)NFI (non-normed fit index) , according to Hu L.T. & Bentler, P.M. 1999, ‘1’ indicates a perfect fit, it may assumed value less than ‘0’ or more than ‘1’ , however, if the values is higher or equal to 0.95, it is a cut-off for a good fit.

In our research, we will be using AMOS program to run the CFA and we will consider the criteria of good model fit such as X2, dF, RMSEA, CFI, TLI or NFI in our analysis.

Combination of two countries dataset

According to our model fit figure below, we have run the CFA to see the model fit by looking at the improvement of Modification Indices to see if we should merge the items or delete the items in our combination set of data between Vietnam and Thailand, so we can achieve the cut offs of dF , Chi-square, CFI , RMSEA and the NFI as a good fit model for our SEM. In our model figure as shown below, the number between the linkage of the item shows the factor loadings.

148

According to our model fit figure 18 of the CFA as seen below, it shows that for the Attitude, we do not need to delete any items, or merge any items from Attitude. Secondly, For the perceived Usefulness we tend to merge only the items PU1 and PU2 together and name them PU12 in SPSS. For the PU, we did not delete any items to make a model fit. Next is the Perceived Ease of Use where we did not delete any items and we do not need to merge an item to increase the good fit of the model. The subjective Norm, we decided to delete the item SN3 to increase the criteria of the model to be a good fit and then merge item SN1 and SN2 together and name it as SN12. For the Perceived Cost, we decided to delete the whole constructs due to the adjustment for the best model fit. Next construct is he Perceived Risk where we decided to delete one item in the Perceived Risk, which is PR1 and then merge the PR2 and PR3 together. Lastly, for the covid19 factor, we decided to delete the items C1, C3 and C4 and then merge the items of C2 and C5 in order to achieve the good fit model. For the Behavioral Intention, we cannot delete any items since it is our dependent variable.

According to our Model Fit Summary in table 83 below, we would like to see the important criteria, by starting from the Chi-square, at the very beginning our Chi-square tend to have a very high value over 1500, then we improve the Modification Indices so the Chi-square value falls to 487.8 as seen below in our table. Moreover, the dF value is also went down to 148 after the value almost 300. Next criteria is our CRI, where the threshold should be between 0.9 and 0.95. Our model has a CFI value of 0.947 which considered a model fit. Next criteria is the NFI where the NFI value is 0.925 which is still acceptable for this model. For the RMSEA, the value should be as close as zero as possible, and our model has achieved the value of 0.068 and therefore, this is considered a good model fit for the combination set of the data between Vietnam and Thailand.

149

Figure 18: Model Fit figure for Thailand and Vietnam dataset. Source: Self-edited

Table 82: Model Fit Summary for Thailand and Vietnam dataset

CMIN Model NPAR CMIN DF P CMIN/DF Default model 61 487.806 148 .000 3.296 Saturated model 209 .000 0 Independence model 38 6539.394 171 .000 38.242

Baseline Comparison NFI RFI IFI TLI Model CFI Delta1 rho1 Delta2 rho2 Default model .925 .914 .947 .938 .947 Saturated model 1.000 1.000 1.000 Independence model .000 .000 .000 .000 .000

Parsimony – Adjusted Measures Model PRATIO PNFI PCFI Default model .865 .801 .819 Saturated model .000 .000 .000 Independence model 1.000 .000 .000

NCP Model NCP LO 90 HI 90 Default model 339.806 276.647 410.568 Saturated model .000 .000 .000 Independence model 6368.394 6107.290 6635.833

FMIN Model FMIN F0 LO 90 HI 90 Default model .972 .677 .551 .818 Saturated model .000 .000 .000 .000 Independence model 13.027 12.686 12.166 13.219

150

RMSEA Model RMSEA LO 90 HI 90 PCLOSE Default model .068 .061 .074 .000 Independence model .272 .267 .278 .000

AIC Model AIC BCC BIC CAIC Default model 609.806 614.869 Saturated model 418.000 435.344 Independence model 6615.394 6618.548

ECVI Model ECVI LO 90 HI 90 MECVI Default model 1.215 1.089 1.356 1.225 Saturated model .833 .833 .833 .867 Independence model 13.178 12.658 13.711 13.184

HOELTER HOELTER HOELTER Model .05 .01 Default model 183 197 Independence model 16 17

Execution time summary Minimization: .021 Miscellaneous: .469 Bootstrap: .000 Total: .490

Vietnam

After we conduct the CFA test for combination data between Thailand and Vietnam. Now, we will proceed this CFA test for the data of each country relatively. Now, we will conduct this CFA test for Vietnam data set to see if we could merge or delete any item to enhance the result of dF,

151

Chi-square, CFI, RMSEA and NFI to achieve the fittest model for our SEM. This below model will show the number between the linkage of the item shows the factor loadings.

First, the items of AT1 and AT2 should be merged to become AT12 then we will combine AT12 with AT4 and compute these values as “Attitude” for further analysis. In the next construct of “Perceived Usefulness”, we will merge PU3 and PU4 to have a new item of PU34. Then, we will compute this PU34 with PU1, PU2 to obtain the average value for PU. As the result of “Perceived Ease of Use” construct, we decide to merge PE2 and PE3 to have PE23 then we will combine this new item with PE1 and compute the value for PE to support for the upcoming tests.

After we consider all the possible merged items, we evaluate which items should be deleted to enhance our model. First of all, we decide to remove the whole construct of “Perceived Cost” because we are not able to obtain the fit model when we keep this construct. Then, we will take AT3 out of “Attitude”, SN3 out of “Subjective Norm”, PR1 out of “Perceived Risk” and C1, C4, C5 out of “Covid-19”

As the suggestion, we should keep the whole construct of dependent variable “Behavioral Intention”. Base on the below model fit summary table, we will consider all criteria to enhance our model fit. Now, we can see the Chi-square in CMIN column is 289.125 which is significant improved comparing to the initial model of 1200. Next, the degree of freedom now is decreased to 132 but it was as high as around 300 before. Additionally, RMSEA value is 0.068 which is very close to the “good fit” standard of zero. The next criteria should be assessed is CFI, the perfect standard of CFI will range between 0.90 to 0.95. After removing and merging some possible items, we obtain the CFI value of 0.953. Lastly, we want to mention about NFI value, NFI is suggested to be equal or above 0.9. As the result, we have 0.917 as the value of NFI which is completely qualified the suggested standard. And the TLI value is 0.945 which is quite acceptable for the final adjustment of our model.

152

Figure 19: Model fit figure for Vietnam dataset. Source: Self-edited

Table 83: Model Fit Summary for Vietnam dataset

CMIN Model NPAR CMIN DF P CMIN/DF Default model 57 289.125 132 .000 2.190 Saturated model 189 .000 0 Independence model 36 3482.196 153 .000 22.759

Baseline Comparisons NFI RFI IFI TLI Model CFI Delta1 rho1 Delta2 rho2 Default model .917 .904 .953 .945 .953 Saturated model 1.000 1.000 1.000 Independence model .000 .000 .000 .000 .000

153

Parsimony-Adjusted Measures Model PRATIO PNFI PCFI Default model .863 .791 .822 Saturated model .000 .000 .000 Independence model 1.000 .000 .000

NCP Model NCP LO 90 HI 90 Default model 157.125 111.809 210.182 Saturated model .000 .000 .000 Independence model 3329.196 3140.803 3524.902

FMIN Model FMIN F0 LO 90 HI 90 Default model 1.138 .619 .440 .827 Saturated model .000 .000 .000 .000 Independence model 13.709 13.107 12.365 13.878

RMSEA Model RMSEA LO 90 HI 90 PCLOSE Default model .068 .058 .079 .003 Independence model .293 .284 .301 .000

AIC Model AIC BCC BIC CAIC Default model 403.125 412.342 Saturated model 378.000 408.562 Independence model 3554.196 3560.018

ECVI Model ECVI LO 90 HI 90 MECVI Default model 1.587 1.409 1.796 1.623

154

Model ECVI LO 90 HI 90 MECVI Saturated model 1.488 1.488 1.488 1.609 Independence model 13.993 13.251 14.763 14.016

HOELTER HOELTER HOELTER Model .05 .01 Default model 141 152 Independence model 14 15

Minimization: .030 Miscellaneous: .566 Bootstrap: .000 Total: .596

Thailand

According to our model fit figure below, we have run the CFA to see the model fit by looking at the improvement of Modification Indices to see if we merge the items or delete the items, so we can achieve the dF , Chi-square, CFI , RMSEA and the NFI to whether if it falls near any threshold of our cut offs as a good model fit for our SEM. In our model figure as shown below, the number between the linkage of the item shows the factor loadings. According to our final model as seen below, we will merge items AT3 and AT4 together and name them as AT34 as a new item, then merge this AT34 with the AT1, therefore we compute these two items as ‘Attitude’ for our further analysis.

Next construct is the Perceived Usefulness where we merge PU1 and PU2 together, then we will name it as PU12 and then compute this new item to the PU3 and PU4 again, then name this construct as ‘Perceived Usefulness’. For the perceived Ease of Use, we will change nothing about the items for PE. For the ‘Subjective Norm’, the item SN1 and item SN2 will be merged and named as SN12, and hence will be computed with the item SN3. For the Perceived Risk, we will merge the two items. Next is the deletion of the construct ‘Perceived Cost’, since we could not achieve a good model fit, we decided to delete a whole construct and could not proceed further to the other analysis. Lastly, in our covid-19 factor, we decided to delete item C1 and C4, then proceed to merge the items C2 and C3 together and name it as ‘C23’, then compute the item C5 and name it under ‘Covid19’ for the further analysis. 155

For our dependent variable BI, it is necessary that we keep our dependent variables and items. According to the model fit summary table as shown below, we will go through the criteria that we are looking for as our model fit. For example, our Chi-square or CMIN column, which is around 344.83 which is considered lower than the first model (around almost 1300), therefore we can reduce it down as low as 344.83 which is in our satisfaction level. Next criteria is the degree of freedom which is 146 (at the beginning was way higher than 250) which we could lower it down as much as we could. In addition to RMSEA, the value is 0.074 which is very close to zero. Since zero means a perfect fit. We are very close to 0.1 which is considered a good fit. Our next criteria is the CFI where we would want to achieve something in between 0.90 – 0.95 and after deleting, and merging items, our model falls in a satisfied level which is at 0.91. Last criteria is the NFI where the value should be equal to 0.9 or higher, we made it so it is quite close to the 0.9 which is 0.86 and for TLI, it is 0.895. We consider this quite acceptable for the final adjustment of our model

Figure 20: Model fit figure for Thailand dataset

156

Table 84: Model Fit Summary for Thailand dataset

CMIN Model NPAR CMIN DF P CMIN/DF Default model 63 344.833 146 .000 2.362 Saturated model 209 .000 0 Independence model 38 2386.559 171 .000 13.956

Baseline Comparisons NFI RFI IFI TLI Model CFI Delta1 rho1 Delta2 rho2 Default model .856 .831 .911 .895 .910 Saturated model 1.000 1.000 1.000 Independence model .000 .000 .000 .000 .000

Parsimony-Adjusted Measures Model PRATIO PNFI PCFI Default model .854 .730 .777 Saturated model .000 .000 .000 Independence model 1.000 .000 .000

NCP Model NCP LO 90 HI 90 Default model 198.833 148.405 256.970 Saturated model .000 .000 .000 Independence model 2215.559 2061.386 2377.102

FMIN Model FMIN F0 LO 90 HI 90 Default model 1.396 .805 .601 1.040 Saturated model .000 .000 .000 .000 Independence model 9.662 8.970 8.346 9.624

RMSEA Model RMSEA LO 90 HI 90 PCLOSE Default model .074 .064 .084 .000

157

Model RMSEA LO 90 HI 90 PCLOSE Independence model .229 .221 .237 .000

AIC Model AIC BCC BIC CAIC Default model 470.833 481.934 Saturated model 418.000 454.828 Independence model 2462.559 2469.255

ECVI Model ECVI LO 90 HI 90 MECVI Default model 1.906 1.702 2.142 1.951 Saturated model 1.692 1.692 1.692 1.841 Independence model 9.970 9.346 10.624 9.997

HOELTER HOELTER HOELTER Model .05 .01 Default model 126 136 Independence model 21 23

Minimization: .026 Miscellaneous: .483 Bootstrap: .000 Total: .509

Summarization of CFA between all datasets

According to our model fit as seen above, all of the datasets did not contain the construct perceived cost. We decided to delete a whole construct due to the fact that when deleting the construct, we can run CFA properly and therefore, the values of our thresholds for the good model fit got improved afterwards. For the combination dataset, we have no issues regarding the deletion of the items in attitude, perceived usefulness, and perceived ease of use. We decided to delete some items in subjective norm, perceived risk and covid19. For Vietnam, we decided to delete items in attitude, subjective norm, perceived risk and covid19 which is similar to the combination

158

dataset. For Thailand dataset, the attitude was already eliminated from Cronbach’s alpha reliability test, left with the additional deletion of items in perceived ease of use, perceived risk and covid19 which is quite different from both combination dataset and Vietnam dataset. After this we will run the multicollinearity tests to see if there is still problems with multicollinearity or not.

Multicollinearity Test

There is an assumption in logistic regression which refers to the exploratory variables should not be highly correlated with one another (Senaviratna, N. & Cooray, T. 2019). To see whether our variables are having problem with the multicollinearity or not, normally it is suggested that we should do Pearson correlation to see its coefficient to measure the strength of the association between two variables, general rules said that if the coefficient between two variables is greater than 0.8 or 0.9, then there’s a serious problem with multicollinearity (Senaviratna, N. & Cooray, T. 2019).

In addition to the Pearson correlation, multicollinearity can mainly be found with the help of tolerance and its reciprocal called variance inflation factor or VIF where the tolerance is the percentage of variance in a predictor that cannot be explained by other predictors (Senaviratna, N. & Cooray, T. 2019). It is suggested that the value of tolerance should be more than 0.10. Otherwise, if the tolerance value is below 0.10, this indicates the collinearity (Daoud, J.I 2017). VIF in general shows how much the variance of the coefficient is being inflated by multicollinearity and also it has no fix cut off on what value should be the cut-off in determining the detection of multicollinearity. Often many researchers use value of 10 or higher as an indicator of multicollinearity (Senaviratna, N. & Cooray, T. 2019).

To resolve the problem with multicollinearity, it is recommended that the highly correlated variables should be combined with one another through principal component analysis or leave out a variable that highly associated to other variables (Daoud, J.I 2017).

In our research, we are required to run Pearson correlation analysis to see whether the variables are highly correlated or not after the reliability tests and CFA and also we will be looking at the VIF value whether are there any adjustment to the multicollinearity (if there’s any). If we could detect the multicollinearity, we will then proceed to the principal component analysis.

159

What is Pearson Correlation?

Normally correlation analysis aimed for analyzing the strength of the relationship between the two observed variables (Schober, P. et al. 2018). A linear relationship between two variables is a special case of a monotonic relationship, where the term ‘Correlation’ is used in the context of linear relationship between 2 continuous or random variables known as a Pearson product- moment correlation and its abbreviation is “r” (Schober, P. et al. 2018).

The correlation refers to the association between variable where they move together or not (Sykes, L M, Gani, F, & Vally, Z. 2016). The positive correlation indicates that one variable rises or falls, the other does as well, or the two variables have a strong relationship (Sykes, L M, Gani, F, & Vally, Z. 2016). Similar to the negative correlation where it indicates that two variables move in the opposite direction or have a strong unrelated relationship to one another (Sykes, L M, Gani, F, & Vally, Z. 2016).

The interpretation of correlation coefficient was suggested to use such words as “Weak”, “Moderate” or “Strong” relationship (Schober, P. et al. 2018). According to Schober, P. et al. (2018), the correlation coefficient between 0.00 – 0.10 is considered “Negligible correlation”, 0.10 – 0.39 is considered “Weak correlation”, 0.40 – 0.69 is considered “Moderate correlation”, 0.70 – 0.89 is considered strong correlation and lastly 0.90 – 1 is considered very strong correlation. All in all, many researchers tend to use correlation to see how strong the relationship between two variables is (Schober, P. et al. 2018).

Combination of two countries dataset

According to the table below we will be looking at the Tolerance value and the VIF. Refer to Daoud, J.I (2017), the Tolerance value should be more than 0.10 so there is no multicollinearity. Moreover, the VIF value should be below 10, so there is no multicollinearity (Senaviratna, N. & Cooray, T. 2019). All of the independent constructs contain the Tolerance value of above 0.1 and the VIF is way lower than 10. Therefore, we can assume that there is no multicollinearity for the combination dataset between Thailand and Vietnam.

Coefficientsa Unstandardized Standardized Collinearity Coefficients Coefficients Statistics Model B Std. Error Beta t Sig. Tolerance VIF 1 (Constant) .168 .189 .888 .375

160

Attitude .539 .076 .443 7.127 .000 .251 3.985 Perceived Usefulness .301 .072 .252 4.164 .000 .264 3.783 Perceived Ease Of .014 .062 .012 .220 .826 .327 3.061 Use Subjective Norm -.014 .030 -.017 -.462 .645 .694 1.442 Perceived Risk .039 .029 .044 1.352 .177 .909 1.100 Covid19 .064 .037 .068 1.705 .089 .618 1.619 a. Dependent Variable: Behavioral Intention Table 85: Multicollinearity Test result for Thailand and Vietnam dataset

Vietnam

The below table summarize the Collinearity Statistics of Thailand dataset after the model fit. First of all, the Tolerance value should be larger than 0.1 and the VIF value should be smaller than 10 to conclude that there is no multicollinearity (Daoud, J.I 2017; Senaviratna, N. & Cooray, T. 2019). As we can see from this below table, all of Tolerance values are qualified when they are all less than 0.1. Similarly, all the VIF values are less than 10. As the result, we can conclude that there is no multicollinearity problems with the dataset of Vietnam.

Coefficientsa Model Unstandardized Standardized t Sig. Collinearity Statistics Coefficients Coefficients B Std. Error Beta Tolerance VIF

(Constant) .248 .257 .967 .335

Perceived Usefulness .294 .100 .262 2.944 .004 .243 4.112

Perceived Ease of Use .064 .084 .057 .763 .446 .351 2.847

1 Attitude .514 .098 .445 5.263 .000 .270 3.707

Subjective Norm -.067 .037 -.087 -1.818 .070 .837 1.195

Perceived Risk .067 .039 .078 1.727 .085 .946 1.057

Covid-19 .021 .045 .023 .461 .645 .755 1.325 a. Dependent Variable: Behavioral Intention Table 86: Multicollinearity Test result for Vietnam dataset

Thailand

According to the table below, the table shows the Collinearity Statistics of Thailand dataset after the model fit. When we take a look at the Tolerance value, the number should be higher than 0.1 and the VIF value should be less than 10 so there is no multicollinearity (Daoud, J.I 2017;

161

Senaviratna, N. & Cooray, T. 2019). All the independent variables contain Tolerance value above 0.1 and the VIF values are all below 10, therefore we can conclude that there is no multicollinearity problems in the Thailand dataset.

Coefficientsa Unstandardized Standardized Collinearity Coefficients Coefficients Statistics Model B Std. Error Beta t Sig. Tolerance VIF 1 (Constant) .650 .351 1.853 .065 Attitude .360 .096 .310 3.768 .000 .372 2.689 Perceived Usefulness .399 .106 .305 3.766 .000 .386 2.590 Perceived Ease Of -.041 .085 -.038 -.486 .627 .424 2.359 Use Subjective Norm .067 .064 .067 1.055 .293 .631 1.585 Perceived Risk -.008 .042 -.010 -.190 .849 .936 1.069 Covid19 .073 .071 .067 1.035 .302 .608 1.645 a. Dependent Variable: Behavioral Intention Table 87: Multicollinearity Test result for Thailand dataset

Summarization of multicollinearity tests between all datasets

Refer to the tables above, all of our tolerance values are all above 0.1 which is quite satisfied for our data and the VIF values which are all less than 10 which means there is no multicollinearity problems with our all datasets.

Regression analysis with control variables

The regression analysis was built on many statistical concepts such as sampling, probability, correlation, distribution, hypotheses testing etc. (Dhakal, C. 2018). We use the multiple regression analysis because we have a dependent variable that was measured by a continuous scale and there two or more predictors or independent variables in the model (Dhakal, C. 2018). Moreover, we would like to see the linear relationship between the dependent variable and independent variables (Dhakal, C. 2018).

In order to interpret whether the model has a good fit or not, we need to have a look at the R value and R square value. R value is one measure of the quality of the prediction of the dependent

162

variable, the higher the value, the better the level of prediction (Dhakal, C. 2018). Moreover, R square value is the proportion of variance in the dependent variable that can be explained by the independent variables, for example, if the R square is .85, it means that the independent variables explain 85% of variability of the dependent variable and the rest (100% - 85% = 15%), therefore 15% is the other factors that explain the variability of the dependent variable (Dhakal, C. 2018). According to Cohen, J. (1988), if R-square value equals to 0.02 (or 2% or the variance), it means a “Small” effect size and if R-Square equals to 0.13, it is considered “Medium” effect size and lastly R- square equals to 0.26, it means the “Large” effect size.

According to Frost (2017), the small number of R square is not a problem and the high R square is not necessarily good. Therefore, we need to have a look at other values as well. In addition, the adjusted R square indicates the percentage of the variation in the outcome variable is explained by the predictors which are to keep in the model (Dhakal, C. 2018). In the ANOVA table, it is important to see if the significance level is below 0.05 because it can tell us if the regression model has a good fit for the data or not (Dhakal, C. 2018). The good fit means whether the independent variables statistically significantly predict the dependent variable (Dhakal, C. 2018).

In addition to the Coefficient table, it is important for us to be able to see and interpret the result, in the column of “Unstandardized coefficients” indicate how much the dependent variable varies with the independent variable when all other independent variables are held constant (Dhakal, C. 2018). For the standardized coefficient, it is called Beta weights, it measures how much the outcome variable increases in standard deviation when the predictor variable is increased by one standard deviation assuming other variables in the model are held constant (Dhakal, C. 2018). In other words, Beta coefficient compares the strength of the effect of each individual independent variable to the dependent variable, the higher the absolute Beta, the stronger the effect (Dhakal, C. 2018).

Our study aims to see the relationship between independent variables and dependent variables; therefore, it is considered ‘alternative hypothesis’ and also the null hypothesis (H0) is the opposite of the stated hypothesis – for example there is no relationship between variable A and variable B (Sykes, L M, Gani, F, & Vally, Z. 2016). The test of the result determines the probability of seeing the results if H0 were true (Sykes, L M, Gani, F, & Vally, Z. 2016). The p- value indicates the likeliness and unlikeliness of the probability to occur by chance. There are usually a set with a cut-off point of 0.05 (5%) or 0.01 (1%) (Sedgwick, P. 2014). If any p value falls below 0.05 or 0.01, it is considered statistically significant and means we will reject the null hypothesis in favor of alternative hypothesis (Sedgwick, P. 2014). If the p value large, this usually means that there is evidence to support the null hypothesis (Sedgwick, P. 2014). 163

Regression with control variables

Our control variables will be divided into two different groups including the educational level and also the age group. By starting from the education level, we will only look at our dataset based on the majority, the education level of the most respondents are including Secondary level, Undergraduate level, and Post graduate level in both countries. We will be coding the group of Secondary as code number 3, Undergraduate level number 4 and Postgraduate as number 5. Then we will proceed to the linear regression analysis. Our dependent variables will be all of our used variables including Behavioral Intention, Attitude, Perceived Usefulness, Perceived Ease Of Use, Subjective Norm, Perceived Risk and Covid19. If there is any statistic significant level, then it means there is the mean difference between the variables and our dummy variables. In addition to our Age group, we will focus onto the groups that are in our quota plans, this includes the age group between18 – 24 years old , which will be coded as number 2, the age group between 25 – 34 will be coded as number 3, the age group between 35 – 44 will be coded as number 4, the age group between 45 – 54 will be coded into number 5 and lastly our age group between 55 – 64 will be coded as number 6. According to our summarization points below, there is the group names and also the codes explained which we use for our regression analysis with dummy variables.

Education level

No education = 1 Primary = 2 Secondary = 3 Undergraduate = 4 Postgraduate = 5 PhD or higher = 6

Age group

Under 18 = 1 18 – 24 = 2 25 – 34 = 3 35 – 44 = 4 45 – 54 = 5 55-64 = 6 Over 65 = 7

164

Combination of two countries dataset Education level

Behavioral Intention

Refer to our table below, the table shows the regression of Behavioral Intention and also the P value from education level as our dummy variables. The P value shows that there is no significant difference between different education levels, all of our P values are more than 0.05.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.095 .191 21.401 .000

Secondary -.040 .072 -.050 -.561 .575 Undergraduate .029 .049 .064 .596 .551 Postgraduate .004 .043 .009 .103 .918 a. Dependent Variable: Behavioral Intention Table 88: Regression analysis about education level for Thailand and Vietnam: Behavioral Intention

Attitude

According to or table below, the regression coefficient shows the P value of greater than 0.05, this reflects that there is no significant difference of the Attitude between education levels.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.131 .158 26.179 .000

Secondary .045 .059 .067 .760 .448 Undergraduate .037 .041 .099 .916 .360 Postgraduate .006 .035 .015 .167 .868 a. Dependent Variable: Attitude Table 89: Regression analysis about education level for Thailand and Vietnam: Attitude

165

Perceived Usefulness

According to our table below, this shows that the Perceived Usefulness did not have any significant difference between education levels. All of the P values are more than 0.05.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.139 .161 25.685 .000

Secondary .031 .061 .045 .510 .610 Undergraduate .042 .042 .108 1.001 .317 Postgraduate .023 .036 .057 .632 .528 a. Dependent Variable: Perceived Usefulness Table 90: Regression analysis about education level for Thailand and Vietnam: Perceived Usefulness

Perceived Ease Of Use

According to our table below, the P value shows that there is no significant difference between education levels of Perceived Ease Of Use since the P value is greater than 0.05.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.206 .168 25.103 .000

Secondary -.018 .063 -.026 -.290 .772 Undergraduate -.005 .043 -.012 -.110 .912 Postgraduate -.031 .038 -.075 -.835 .404 a. Dependent Variable: Perceived Ease Of Use Table 91: Regression analysis about education level for Thailand and Vietnam: Perceived Ease Of Use

Subjective Norm

According to our table below, this show that P value between the education levels of the variable Subjective Norm. The P value of Secondary level is 0.10, which means the value is statistically significant. There is the mean difference when it comes to Secondary group with the Subjective Norm.

166

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 3.238 .240 13.513 .000

Secondary .232 .090 .227 2.578 .010 Undergraduate .105 .062 .182 1.702 .089 Postgraduate .029 .054 .048 .538 .591 a. Dependent Variable: Subjective Norm Table 92: Regression analysis about education level for Thailand and Vietnam: Subjective Norm

Perceived Risk

Refer to our table below, this shows that the three education levels have the P value greater than 0.05, making this value not statistically significant for the variable Perceived Risk

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 3.440 .219 15.705 .000

Secondary .098 .082 .105 1.187 .236 Undergraduate .074 .057 .142 1.311 .190 Postgraduate .042 .049 .076 .845 .399 a. Dependent Variable: Perceived Risk Table 93: Regression analysis about education level for Thailand and Vietnam: Perceived Risk

Covid-19

According to our table below, the P value between the education levels show that only the Undergraduate level is statistically significant, with the P value of 0.024 which is considered lower than 0.05. Therefore, there is mean difference from Undergraduate respondents toward the Covid19 issues. This group react towards the Covid19 issues.

Coefficientsa Standardized Model Unstandardized Coefficients Coefficients t Sig.

167

B Std. Error Beta 1 (Constant) 3.744 .199 18.816 .000

Secondary -.039 .075 -.045 -.516 .606 Undergraduate .117 .051 .239 2.271 .024 Postgraduate .075 .045 .147 1.672 .095 a. Dependent Variable: Covid19 Table 94: Regression analysis about education level for Thailand and Vietnam: Covid-19

Age group

Behavioral Intention

The table below shows that the P values between different age groups with our variable Behavioral Intention and it shows that the P value of all age group has no significant difference.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 3.750 .253 14.807 .000

Age18to24 .246 .132 .251 1.862 .063 Age25to34 .145 .088 .227 1.655 .099 Age35to44 .089 .067 .169 1.340 .181 Age45to54 .079 .055 .155 1.439 .151 Age55to64 .019 .054 .024 .340 .734 a. Dependent Variable: Behavioral Intention Table 95: Regression analysis about age group for Thailand and Vietnam: Behavioral Intention

Attitude

For the Attitude as our next variable, in the P value column, this shows that there is no P value of any age group that has exceeded the value below 0.05, making our assumption that there is no mean difference between age groups.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.500 .209 21.559 .000

168

Age18to24 -.108 .109 -.135 -.995 .320 Age25to34 -.105 .072 -.200 -1.451 .148 Age35to44 -.045 .055 -.104 -.823 .411 Age45to54 -.058 .045 -.138 -1.282 .201 Age55to64 -.058 .045 -.089 -1.289 .198 a. Dependent Variable: Attitude Table 96: Regression analysis about age group for Thailand and Vietnam: Attitude

Perceived Usefulness

According to our table below, the P value from different age groups with the variable Perceived Usefulness contain the value of greater than 0.05, therefore, there is no mean difference or no significant difference between age groups.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.181 .213 19.588 .000

Age18to24 .047 .111 .057 .421 .674 Age25to34 .016 .074 .030 .216 .829 Age35to44 .045 .056 .101 .800 .424 Age45to54 .018 .046 .041 .379 .705 Age55to64 .025 .046 .039 .554 .579 a. Dependent Variable: Perceived Usefulness Table 97: Regression analysis about age group for Thailand and Vietnam: Perceived Usefulness

Perceived Ease Of Use

Taking a look at our table below, we can see that the P values between all age groups are greater than 0.05, which means there is no significant difference between age groups when it comes to the variable Perceived Ease Of Use.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.306 .222 19.395 .000

Age18to24 -.062 .116 -.072 -.534 .593

169

Age25to34 -.073 .077 -.131 -.952 .342 Age35to44 -.021 .058 -.045 -.353 .725 Age45to54 -.031 .048 -.070 -.648 .517 Age55to64 -.023 .048 -.034 -.485 .628 a. Dependent Variable: Perceived Ease Of Use Table 98: Regression analysis about age group for Thailand and Vietnam: Perceived Ease Of Use

Subjective Norm

For the variable Subjective Norm, all the P values are more than 0.05. Therefore, we can assume that there is no mean difference and no is no significant difference between the age groups.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 3.792 .314 12.063 .000

Age18to24 -.068 .164 -.055 -.414 .679 Age25to34 -.158 .109 -.196 -1.451 .147 Age35to44 .035 .083 .053 .427 .669 Age45to54 -.015 .068 -.023 -.218 .827 Age55to64 .030 .068 .030 .445 .657 a. Dependent Variable: Subjective Norm Table 99: Regression analysis about age group for Thailand and Vietnam: Subjective Norm

Perceived Risk

Our next variable is Perceived Risk, as our P values are all greater than 0.05, we can assume that there is no significant difference between age groups.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 3.667 .289 12.678 .000

Age18to24 -.007 .151 -.006 -.047 .962 Age25to34 -.017 .100 -.023 -.169 .866 Age35to44 .052 .076 .087 .685 .494

170

Age45to54 .007 .063 .012 .116 .908 Age55to64 .051 .062 .057 .818 .414 a. Dependent Variable: Perceived Risk Table 100: Regression analysis about age group for Thailand and Vietnam: Perceived Risk

Covid-19

For the Covid-19, the significant level of all age groups except for Age between 55 – 64, are greater than 0.05, only the age group of 55 – 64 is at 0.05 which is less than 0.05. Therefore, we can assume that there is significant difference of the age group of 55 – 64 when it comes to Covid-19 issues. They react towards the issue of Covid-19 more than other age groups.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.208 .263 16.026 .000

Age18to24 -.028 .137 -.027 -.204 .838 Age25to34 -.028 .091 -.040 -.302 .762 Age35to44 .022 .069 .038 .312 .755 Age45to54 -.096 .057 -.177 -1.691 .092 Age55to64 -.160 .056 -.191 -2.827 .005 a. Dependent Variable: Covid19 Table 101: Regression analysis about age group for Thailand and Vietnam: Covid19

Vietnam Education level

Behavioral Intention

First of all, all of the P values among this item are larger than 0.05 so we can assume that there is no significant difference between educational level.

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 4.079 .210 19.455 .000 1 Secondary -.089 .083 -.114 -1.076 .283

171

Undergraduate -.062 .056 -.137 -1.111 .268 Postgraduate -.023 .048 -.052 -.468 .640 a. Dependent Variable: Behavioral Intention Table 102: Regression analysis about education level for Vietnam: Behavioral Intention

Attitude

Similarly, all of the P values among Attitude are over 0.05 so there is no significant difference between educational level in this item.

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 4.228 .181 23.322 .000

Secondary -.020 .071 -.030 -.287 .775 1 Undergraduate -.063 .049 -.161 -1.308 .192 Postgraduate -.039 .042 -.104 -.931 .352 a. Dependent Variable: Attitude Table 103: Regression analysis about education level for Vietnam: Attitude

Perceived Usefulness

For Perceived Usefulness, there is no P values smaller than 0.05 so we can conclude that there is no significant difference between educational level in this item.

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 4.155 .186 22.295 .000

Secondary .011 .073 .016 .149 .881 1 Undergraduate -.039 .050 -.097 -.787 .432 Postgraduate -.001 .043 -.003 -.029 .977 a. Dependent Variable: Perceived Usefulness Table 104: Regression analysis about education level for Vietnam: Perceived Usefulness

172

Perceived Ease Of Use

It’s the same of Perceived Ease Of Use, there is no significant difference between educational level because all of P values among this items are larger than 0.05 Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 4.175 .183 22.801 .000

Secondary -.028 .072 -.042 -.394 .694 1 Undergraduate -.087 .049 -.217 -1.777 .077 Postgraduate -.057 .042 -.149 -1.343 .180 a. Dependent Variable: Perceived Ease of Use Table 105: Regression analysis about education level for Vietnam: Perceived Ease of Use

Subjective Norm

In Subjective Norm, we can define that only “Secondary” has P value 0.004 which is less than 0.05 so we can assume that Secondary level is statistically significant and it reacts toward the Subjective Norm.

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 3.132 .257 12.189 .000

Secondary .293 .101 .291 2.895 .004 1 Undergraduate -.045 .069 -.077 -.661 .509 Postgraduate -.021 .059 -.038 -.359 .720 a. Dependent Variable: Subjective Norm Table 106: Regression analysis about education level for Vietnam: Subjective Norm

Perceived Risk

There is no significant difference between educational level in Perceived Risk variable because all of the P values in this variable are over 0.05

Coefficientsa

173

Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 3.513 .239 14.678 .000

Secondary .072 .094 .081 .764 .446 1 Undergraduate -.044 .064 -.083 -.679 .498 Postgraduate -.009 .055 -.019 -.171 .864 a. Dependent Variable: Perceived Risk Table 107: Regression analysis about education level for Vietnam: Perceived Risk

Covid19

There is no significant difference between educational level in Covid-19 variable because all of the P values in this variable are over 0.05

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 3.684 .237 15.543 .000

Secondary -.145 .093 -.162 -1.550 .122 1 Undergraduate .033 .063 .063 .519 .604 Postgraduate .036 .055 .072 .661 .509 a. Dependent Variable: Covid-19 Table 108: Regression analysis about education level for Vietnam: Covid-19

Age group

Behavioral Intention

The below table summarize all the p-value between different age group toward Behavioral Intention. We have P value for the age group as 18 – 24 is 0.364, 25 – 34 is 0.299, 35 – 44 is 0.445, 45 – 54 is 0.296, 55 – 64 is 0.309. As we can see all of the above P values are larger than 0.05 so we can assume that there is no significant different among age group in Behavioral Intention.

174

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 3.000 .914 3.282 .001

Age18to24 .419 .461 .395 .910 .364 Age25to34 .319 .306 .505 1.041 .299 1 Age35to44 .176 .231 .310 .765 .445 Age45to54 .194 .185 .367 1.048 .296 Age55to64 .161 .158 .241 1.019 .309 a. Dependent Variable: Behavioral Intention Table 109: Regression analysis about age group for Vietnam: Behavioral Intention

Attitude

As we can see from the below table, none of P value are smaller than 0.05 so we can assume that there is no significant difference between age group toward Attitude.

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 3.000 .794 3.781 .000

Age18to24 .513 .400 .558 1.284 .200 Age25to34 .332 .266 .606 1.249 .213 1 Age35to44 .278 .200 .563 1.387 .167 Age45to54 .211 .161 .461 1.314 .190 Age55to64 .202 .137 .351 1.478 .141 a. Dependent Variable: Attitude Table 110: Regression analysis about age group for Vietnam: Attitude

Perceived Usefulness

Similarly, we have no P value in Perceived Usefulness less than 0.05 so there is no significant difference between age group in this variable.

175

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 3.000 .814 3.686 .000

Age18to24 .524 .410 .555 1.278 .202 Age25to34 .361 .273 .643 1.325 .186 1 Age35to44 .265 .205 .523 1.288 .199 Age45to54 .214 .165 .454 1.294 .197 Age55to64 .229 .140 .387 1.632 .104 a. Dependent Variable: Perceived Usefulness Table 111: Regression analysis about age group for Vietnam: Perceived Usefulness

Perceived Ease Of Use

All of the P value of age group in Perceived Ease of Use are also less than 0.05 so there is no significant different between age group among this variable.

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 3.000 .802 3.740 .000

Age18to24 .441 .404 .472 1.091 .277 Age25to34 .284 .269 .510 1.056 .292 1 Age35to44 .234 .202 .467 1.154 .250 Age45to54 .202 .163 .434 1.241 .216 Age55to64 .214 .138 .366 1.549 .123 a. Dependent Variable: Perceived Ease of Use Table 112: Regression analysis about age group for Vietnam: Perceived Ease of Use

Subjective Norm

In Subjective Norm, we can define that there is no significant difference between age group because all of the P values are larger than 0.05.

176

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 3.000 1.099 2.731 .007

Age18to24 -.016 .554 -.012 -.029 .977 Age25to34 -.094 .368 -.115 -.256 .798 1 Age35to44 .174 .277 .236 .627 .531 Age45to54 .134 .223 .195 .603 .547 Age55to64 .179 .190 .206 .942 .347 a. Dependent Variable: Subjective Norm Table 113: Regression analysis about age group for Vietnam: Subjective Norm

Perceived Risk

For Perceived Risk, we can define that there is no significant difference between age group because all of the P values are larger than 0.05.

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 3.000 1.043 2.876 .004

Age18to24 .141 .526 .116 .268 .789 Age25to34 .123 .350 .170 .352 .725 1 Age35to44 .181 .263 .278 .689 .492 Age45to54 .097 .212 .160 .459 .647 Age55to64 .125 .180 .163 .695 .488 a. Dependent Variable: Perceived Risk Table 114: Regression analysis about age group for Vietnam: Perceived Risk

Covid-19

It’s the same for Covid-19 when we define that there is no significant difference between age group because all of the P values are larger than 0.05.

177

Coefficientsa Model Unstandardized Coefficients Standardized t Sig. Coefficients B Std. Error Beta

(Constant) 3.000 1.009 2.973 .003

Age18to24 .423 .509 .346 .832 .406 Age25to34 .272 .338 .373 .803 .423 1 Age35to44 .248 .255 .377 .972 .332 Age45to54 .024 .205 .040 .119 .906 Age55to64 -.012 .174 -.016 -.068 .946 a. Dependent Variable: Covid-19 Table 115: Regression analysis about age group for Vietnam: Covid-19

Thailand Education level

Behavioral Intention

For the variable Behavioral Intention, the P values between all education level is greater than 0.05, therefore, there is no significant difference.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.250 .524 8.104 .000

Secondary -.003 .181 -.004 -.016 .987 Undergraduate .054 .132 .120 .411 .681 Postgraduate .055 .110 .105 .498 .619 a. Dependent Variable: Behavioral Intention Table 116: Regression analysis about education level for Thailand: Behavioral Intention

Attitude

For our next variable, the Attitude, all of our P values between all the education level is greater than 0.05, therefore we can assume that there is no significant difference.

178

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.250 .454 9.356 .000

Secondary .060 .157 .091 .385 .700 Undergraduate .055 .114 .143 .486 .628 Postgraduate .055 .095 .122 .575 .566 a. Dependent Variable: Attitude Table 117: Regression analysis about education level for Thailand: Attitude

Perceived Usefulness

The table below shows that regression between education levels and the variable Perceived Usefulness, the P values are all greater than 0.05, therefore it is not statistically significant.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.333 .401 10.812 .000

Secondary .010 .138 .016 .069 .945 Undergraduate .046 .101 .132 .452 .652 Postgraduate .053 .084 .133 .633 .527 a. Dependent Variable: Perceived Usefulness Table 118: Regression analysis about education level for Thailand: Perceived Usefulness

Perceived Ease Of Use

In our table below shows that the P value between the education levels that there is no P values less than 0.05, therefore there is no significant difference.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.500 .479 9.389 .000

Secondary -.075 .165 -.107 -.452 .651

179

Undergraduate -.023 .120 -.056 -.192 .848 Postgraduate -.023 .100 -.048 -.227 .821 a. Dependent Variable: Perceived Ease Of Use Table 119: Regression analysis about education level for Thailand: Perceived Ease Of Use

Subjective Norm

According to our table below, this shows that all of our values are greater than 0.05 which means there is no significant difference between the education levels for the variable Subjective Norm.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.125 .518 7.967 .000

Secondary -.114 .178 -.149 -.636 .525 Undergraduate -.002 .130 -.005 -.018 .985 Postgraduate .032 .108 .061 .294 .769 a. Dependent Variable: Subjective Norm Table 120: Regression analysis about education level for Thailand: Subjective Norm

Perceived Risk

According to the table below, this shows that the P values of all education levels are greater than 0.05, which means that there is no significant difference between education levels for the variable Perceived Risk.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 2.750 .650 4.231 .000

Secondary .342 .224 .359 1.527 .128 Undergraduate .296 .163 .528 1.811 .071 Postgraduate .223 .136 .345 1.640 .102 a. Dependent Variable: Perceived Risk Table 121: Regression analysis about education level for Thailand: Perceived Risk

180

Covid-19

According to the table below, this shows the variable Covid-19 between the education levels. All of the education levels are containing the P values of less than 0.05. The secondary level has P value of 0.2 and the P value of Undergraduate level is 0.004 and lastly the Postgraduate level has P value of 0.007. Therefore, we can say that all of the education levels are statistically significant. There is mean difference and therefore, all education levels react about the Covid-19.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 3.125 .469 6.668 .000

Secondary .378 .162 .542 2.340 .020 Undergraduate .340 .118 .829 2.884 .004 Postgraduate .266 .098 .563 2.716 .007 a. Dependent Variable: Covid-19 Table 122: Regression analysis about education level for Thailand: Covid-19

Age group

Behavioral Intention

The table below shows the P values between different age groups with our variable Behavioral Intention and it shows that the P value of age group between 18 – 24 has the P value of 0.002, the age group of 25 – 34 has P value of 0.002 and the age group 35 – 44 has P value of 0.009 and the age group between 45 – 54 has P value of 0.044 which are below 0.05. Meaning that these age groups are statistically significant when it comes to the variable Behavioral Intention. Therefore, they would react to the statement relates to the intention to use mobile payment. Except only the age group of 55 – 64 where the P value is greater than 0.05 (0.45).

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 3.818 .217 17.566 .000

Age18to24 .369 .116 .465 3.184 .002 Age25to34 .241 .079 .417 3.059 .002

181

Age35to44 .156 .059 .364 2.644 .009 Age45to54 .101 .050 .235 2.021 .044 Age55to64 -.053 .070 -.054 -.756 .450 a. Dependent Variable: Behavioral Intention Table 123: Regression analysis about age group for Thailand: Behavioral Intention

Attitude

According to our table below, this shows that the age group between 55 – 64 has P value less than 0.05, the rest of age groups have the P value greater than 0.05. Therefore, the age group of 55 – 64 is considered statistically significant when it comes to the attitude to use mobile payment. This age is also reacting towards the statement about the attitude to use mobile payment.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.614 .192 24.087 .000

Age18to24 -.060 .102 -.088 -.587 .558 Age25to34 -.048 .070 -.096 -.687 .493 Age35to44 -.023 .052 -.063 -.448 .655 Age45to54 -.049 .044 -.133 -1.116 .265 Age55to64 -.154 .062 -.183 -2.497 .013 a. Dependent Variable: Attitude Table 124: Regression analysis about age group for Thailand: Attitude

Perceived Usefulness

Refer to table below, the P values between age groups are all greater than 0.05, therefore we can assume that there no significant difference between the age groups with the variable Perceived Usefulness.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.288 .170 25.175 .000

Age18to24 .097 .091 .159 1.062 .289

182

Age25to34 .066 .062 .149 1.067 .287 Age35to44 .084 .046 .255 1.811 .071 Age45to54 .039 .039 .120 1.008 .315 Age55to64 -.027 .055 -.036 -.494 .622 a. Dependent Variable: Perceived Usefulness Table 125: Regression analysis about age group for Thailand: Perceived Usefulness

Perceived Ease Of Use

Our next variable is the Perceived Ease Of Use where the P values are all greater than 0.05, therefore we could assume that there is no significant difference between age groups and variable Perceived Ease Of Use.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.364 .204 21.372 .000

Age18to24 .005 .109 .007 .045 .964 Age25to34 .031 .074 .059 .423 .673 Age35to44 .016 .055 .040 .282 .778 Age45to54 -.008 .047 -.020 -.171 .865 Age55to64 -.081 .066 -.091 -1.236 .218 a. Dependent Variable: Perceived Ease Of Use Table 126: Regression analysis about age group for Thailand: Perceived Ease Of Use

Subjective Norm

According to our table below, this shows that the P value of the age group between 18 – 24 is 0.022 and the age group 25 – 34 is 0.008 respectively, which means that between these two age groups, there is significant difference when it comes to the Subjective Norm. These age group is considered the main mobile payment users in Thailand and therefore, they are reacting towards to the Subjective Norm more than other age groups.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 3.659 .217 16.880 .000

183

Age18to24 .267 .116 .338 2.308 .022 Age25to34 .210 .079 .364 2.664 .008 Age35to44 .098 .059 .228 1.656 .099 Age45to54 .027 .050 .063 .541 .589 Age55to64 -.016 .070 -.017 -.230 .818 a. Dependent Variable: Subjective Norm Table 127: Regression analysis about age group for Thailand: Subjective Norm

Perceived Risk

According to our table below, this shows that the P values between the age groups for the variable Perceived Risk are all greater than 0.05 which means that there is no significant difference between the age groups and the variable Perceived Risk.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 3.636 .279 13.039 .000

Age18to24 .141 .149 .143 .945 .346 Age25to34 .113 .101 .157 1.113 .267 Age35to44 .060 .076 .113 .794 .428 Age45to54 .032 .064 .059 .492 .623 Age55to64 .123 .090 .101 1.368 .173 a. Dependent Variable: Perceived Risk Table 128: Regression analysis about age group for Thailand: Perceived Risk

Covid-19

According to the table below, the P values of all age groups are more than 0.05 for the variable Covid19 which means that there is no significant difference between age groups.

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) 4.386 .203 21.578 .000

Age18to24 .005 .108 .006 .042 .966 Age25to34 .053 .074 .102 .724 .470

184

Age35to44 .029 .055 .075 .533 .594 Age45to54 -.013 .047 -.032 -.269 .788 Age55to64 -.044 .066 -.049 -.664 .507 a. Dependent Variable: Covid-19 Table 129: Regression analysis about age group for Thailand: Covid-19

Summarization of regression analysis for control variables between all datasets

For education, the combination datasets have no special significant difference for almost every construct, except for the construct subjective norm and Covid-19 where in subjective norm, secondary level respondents seem to have reaction towards the society, family and friends with regards of using mobile payment. For covid19 and the intention to use mobile payment, undergraduates for the combination dataset react to the situation of pandemic. The result of the undergraduates that react to the pandemic situation is similar to Thailand dataset but not for the Vietnam dataset. For Thailand dataset, people who obtained all kinds of education react towards the covid19 situation. For Vietnam, there is no significant difference in any of the construct for the education level. For age group, the combination dataset shows that only the construct covid19 and the age group 55 – 64 react towards the agenda of the covid19 situation. For Vietnam dataset, the age group has nothing to do with the intention to use mobile payment at all. In Thailand, the age groups played significant roles to the intention to use mobile payment. When referring to our data for the behavioral intention, almost all age groups between 18 to 54 has reacted towards the intention to use mobile payment in Thailand. The attitude can only be found to have significance level of the age between 55 – 64 which is older people. In Thailand the younger generation, age between 18 to 34 has react towards the subjective norm which is the influence of the friends, family members and influencers.

Multiple Regression Analysis

Combination of two countries dataset

Our multiple regression analysis for the combination of the dataset between Thailand and Vietnam in the tables below indicates the result of our analysis, starting from the Model Summary table, we can see that our R value is at 0.720 which is considered high. It also means the model is a good fit. Next value is our R square, where the R square is .519, that can be interpreted as independent variables explain 51.9% of variability of the dependent variable and the rest 48.1%

185

(100% - 51.9 % = 48.1%), will explained by other factors about the variability of the dependent variable. Moreover, when R- square is more than 0.26, this shows that there is a large effect size in this combination dataset.

In the Model Summary table, we should also look at the Sig. F Change column to see if the Model is significant to use or not. As seen below, the value is below 0.001 which is considered significant at the level of 0.001. Next, In ANOVA table, significance level is also falling below 0.001. Therefore, it is significant at the level 0.001 which we can conclude that the regression model has a good fit for the data.

In table 132, the column called Standardized Coefficients, we look at our data, it is stated that if the absolute Beta value is higher, the stronger the effect of the independent variable to the dependent variable. For us, our dependent variable is Behavioral Intention. Our highest value of our Standardized Beta is the Attitude to the Behavioral Intention (Beta coefficient = .443), this reflects on the strong weights of Attitude to the Behavioral Intention. Next, Perceived Usefulness is also containing a high Beta value comparing to other independent variables, it contains the value of .252. Therefore, we can confirm that Attitude and Perceived Usefulness influence Behavioral Intention.

By considering the P values, both Attitude and Perceived Usefulness contain p-value below 0.001 (it is significant at level 0.001). Therefore, we can assume that we will reject the null hypotheses and accept the alternative hypotheses as we have assumed earlier. Therefore, we can summarize that Attitude and Perceived Usefulness influence towards the Behavioral Intention to use Mobile Payment in both countries. Consequently, the p-value of Perceived Ease Of Use (0.826), Subjective Norm (0.645), Perceived Risk (0.177), and Covid19 (0.089) are considered not statistically significant at the level of both 0.05 and 0.01, therefore, we will accept the null hypotheses. We support our analysis that these four variables have no influence on the Behavioral Intention to use Mobile Payment in both countries.

Model Summary Change Statistics R Adjusted R Std. Error of the R Square F Sig. F Model R Square Square Estimate Change Change df1 df2 Change 1 .720a .519 .513 .61325 .519 89.061 6 496 .000 a. Predictors: (Constant), Covid19, Perceived Risk, Subjective Norm, Perceived Usefulness, Perceived Ease Of Use, Attitude Table 130: Model Summary for dataset of Thailand and Vietnam

186

ANOVAa Model Sum of Squares df Mean Square F Sig. 1 Regression 200.964 6 33.494 89.061 .000b Residual 186.535 496 .376 Total 387.499 502

a. Dependent Variable: Behavioral Intention b. Predictors: (Constant), Covid19, Perceived Risk, Subjective Norm, Perceived Usefulness, Perceived Ease Of Use, Attitude Table 131: ANOVA test result for dataset of Thailand and Vietnam

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) .168 .189 .888 .375

Attitude .539 .076 .443 7.127 .000 Perceived Usefulness .301 .072 .252 4.164 .000 Perceived Ease Of Use .014 .062 .012 .220 .826 Subjective Norm -.014 .030 -.017 -.462 .645 Perceived Risk .039 .029 .044 1.352 .177 Covid19 .064 .037 .068 1.705 .089 a. Dependent Variable: Behavioral Intention Table 132: Coefficients result for dataset of Thailand and Vietnam

Vietnam

In this multiple regression analysis, we should first interpret the data in model summary table below. As we can see, we have our R value is 0.722 which is considered as moderately fit comparing to the standard. Next, we have R square is 0.521 which means that all the independent variables explain 52.1% the variability of dependent variable in this research model and the other factors will explain 47.9% the variability of dependent variable. However, our R square change is higher than 0.26 which illustrate that there is a large effect size in our model.

Model Summary Model R R Adjusted Std. Error of Change Statistics Square R Square the Estimate R Square Change F Change df1 df2 Sig. F Change 1 .722a .521 .509 .63901 .521 44.936 6 248 .000 a. Predictors: (Constant), C, PR, SN, PU, PE, AT

Table 133: Model Summary for dataset of Vietnam

187

Then, we will analyze the important indication in ANOVA table. As we can see that we have a good fit model when our significant level is less than 0.05.

ANOVAa Model Sum of Squares df Mean Square F Sig. Regression 110.093 6 18.349 44.936 .000b

1 Residual 101.266 248 .408

Total 211.359 254

a. Dependent Variable: BI b. Predictors: (Constant), C, PR, SN, PU, PE, AT Table 134: ANOVA test result for dataset of Vietnam

In this table, we should focus on the Beta values under Standardized Coefficients column. The higher value of this value, the stronger extent that independent variables impact on dependent variable. As we can see, Attitude (0.514) has the highest beta value comparing to other factors which mean that this factor has strongest impact on dependent variable. In addition, Perceived Usefulness (0.294) also strongly impacts on the Behavioral Intention when it has large beta value after Attitude. Besides, it’s necessary to consider p-value in our analysis. As we can see from the Coefficient table, Attitude is the only factor which has p-value less than 0.001 so we decide to reject to null hypothesis of the predictor Attitude and accept the alternative hypothesis as we assumed previously. Besides, the p-value of Perceived Usefulness is 0.004 which falls below 0.05 so we will also accept the null hypothesis of this predictor and accept the alternative hypothesis as we mentioned earlier.

On the other hand, the p-value of other factors such as Perceived Ease of Use (0.446), Subjective Norm (0.070), Perceived Risk (0.085), Covid-19 (0.645) are all higher than 0.001 which are considered as not statistically significant at the level of both 0.05 and 0.001. Therefore, we will accept the null hypotheses of these predictors which means these factors have no influence on the Behavioral Intention of Mobile Payment usage in Vietnam.

Coefficientsa Model Unstandardized Standardized t Sig. Coefficients Coefficients B Std. Error Beta

(Constant) .248 .257 .967 .335

Attitude .514 .098 .445 5.263 .000 1 Perceived Usefulness .294 .100 .262 2.944 .004 Perceived Ease of Use .064 .084 .057 .763 .446

188

Subjective Norm -.067 .037 -.087 -1.818 .070 Perceived Risk .067 .039 .078 1.727 .085 Covid-19 .021 .045 .023 .461 .645 a. Dependent Variable: BI Table 135: Coefficients result for dataset of Vietnam

Thailand

Our multiple regression analysis in the tables below shows the result of our analysis, starting from the Model Summary, we can see that our R value is at 0.626 which is considered not too low or too high which means the model is moderately fit. Next value is our R square, where the R square is .391, that can be interpreted that the independent variables explain 39.1% of variability of the dependent variable and the rest (100% - 39.1% = 60.9%), therefore the other factors will explain about 60.9% the variability of the dependent variable. However, we can say that there is a small positive linear association of the model. However, when R- square is more than 0.26, this means that there is a large effect size. In the Model Summary table, we should also look at the Sig. F Change column to see if the Model is significant to be used or not. We have found that the value is even below 0.001 which is considered significant.

Next step is when we look at the ANOVA table to see if the significance level is below 0.001, in this case our value is even smaller than 0.001, which we can summarize that the regression model has a good fit for the data.

In the table of 138, the column called Standardized Coefficients, we look at our data, it is stated that if the absolute Beta value is higher, the stronger the effect of the independent variable to the dependent variable, in this case is the Attitude. The highest value of our Beta is the Attitude to the Behavioral Intention (Beta coefficient = .310), this reflects on the strong effect of the predictor Attitude to the Behavioral Intention. Perceived Usefulness is also containing a high Beta coefficient comparing to other independent variables with the value of .305. We can assume that these independent variables including Attitude and Perceived Usefulness have an effect on the dependent variable, Behavioral Intention.

In addition to the same table, we considered p-value important to our analysis, The Attitude (p-value = 0.000) and Perceived Usefulness (0.000) contain p-value below 0.001 (it is significant at level 0.001). Therefore, we can conclude that we will reject the null hypotheses of the predictor Attitude and Perceived Usefulness and accept the alternative hypotheses as we have assumed

189

earlier. Therefore, we can conclude that Attitude and Perceived Usefulness have an influence towards the Behavioral Intention to use Mobile Payment in Thailand.

Lastly, as the p-value of Perceived Ease Of Use (0.627), Subjective Norm (0.293), Perceived Risk (0.849), and Covid19 (0.302) are considered not statistically significant at the level of both 0.05 and 0.01, therefore, we will accept the null hypotheses and analyze that these four variables have no influence on the Behavioral Intention to use Mobile Payment in Thailand

Model Summary Change Statistics R Adjusted R Std. Error of the R Square F Sig. F Model R Square Square Estimate Change Change df1 df2 Change 1 .626a .391 .376 .58542 .391 25.831 6 241 .000 a. Predictors: (Constant), Covid19, Perceived Risk, Subjective Norm, Perceived Usefulness, Perceived Ease Of Use, Attitude Table 136: Model Summary for dataset of Thailand

ANOVAa Model Sum of Squares df Mean Square F Sig. 1 Regression 53.115 6 8.853 25.831 .000b Residual 82.594 241 .343 Total 135.710 247

Coefficientsa Standardized Unstandardized Coefficients Coefficients Model B Std. Error Beta t Sig. 1 (Constant) .650 .351 1.853 .065

Attitude .360 .096 .310 3.768 .000 Perceived Usefulness .399 .106 .305 3.766 .000 Perceived Ease Of Use -.041 .085 -.038 -.486 .627 Subjective Norm .067 .064 .067 1.055 .293 Perceived Risk -.008 .042 -.010 -.190 .849 Covid19 .073 .071 .067 1.035 .302 a. Dependent Variable: Behavioral Intention Table 138: Coefficients result for dataset of Thailand 190

Summarization of multiple regression analysis between all datasets

According to our tables as seen above, we can see that all of the combination datasets has the same result. Only the attitude and perceived usefulness have significant difference across all respondents in both Vietnam and Thailand and other constructs are not. This give us a short conclusion that attitude, and perceived usefulness is the two factors that impact on the behavioral intention to use mobile payment in Vietnam and Thailand.

8. Discussions

Discussions based on the factors of research model

Discussions between countries

Figure 21: Evaluation model between two countries. Source: Self-edited

After all, we would like to summarize our findings between two countries the concise model as figure 21. As we can see from the above figure, all of consistent lines represent for supported results and inconsistent lines represent for not supported results relatively. Besides, the correlation between two variables are also reflected by Beta value which is taken from the multiple regression analysis. And the significant levels of *** or ** of Beta value are explained as *** reflects on the p-

191

value that has significant level of 0.001 or lower and ** means the significant level of 0.05 and below.

According to the above model, Attitude is the factor that has strongest impact on behavioral intention to use mobile payment in both Vietnam and Thailand. Besides, hypothesis 2 is also supported which means that Perceived Usefulness has influence on the behavioral intention of mobile payment usage in these countries. However, the hypothesis 3, 4, 6, 7 are not supported so we can conclude that the factors of Perceived Ease of Use, Subjective Norm, Perceived Risk, Covid-19 have no impact on behavioral intention of mobile payment usage in both Vietnam and Thailand. And the factor of Perceived Cost was removed during the CFA test therefore we will not process any further analysis with this factor afterward.

Discussions in the context of Vietnam

Figure 22: Evaluation model of Vietnam. Source: Self-edited

In the context of Vietnam, our findings are shortly summarized in the above model. Each of the lines as shown were described by the standardized Beta values taken from the multiple regression analysis and also the significant levels of *** or **, as *** reflects on the p-value that has significant level of 0.001 or lower and ** means the significant level of 0.05 and below.

Similar to the finding between two countries, we find that Attitude is the factor which have strongest impact on the intention to use mobile payment of Vietnamese. Furthermore, this model

192

also indicates that Perceived Usefulness also affects to the mobile payment usage of Vietnamese. On the other hand, the result illustrates that hypothesis 3, 4, 6, 7 are not supported so we can conclude that these factors of Perceived Ease of Use, Subjective Norm, Perceived Risk and Covid- 19 have no impact on the intention to use mobile payment of Vietnamese. Additionally, we will not process any further analysis with Perceived Cost because this factor was deleted during CFA test.

As we can see from the findings, the factor Attitude and Perceived Usefulness all impact on intention to use mobile payment in both Thailand and Vietnam.

Discussions in the context of Thailand

Figure 23: Evaluation model of Thailand. Source: Self-edited

According to the above figure, for the context of Thailand, the results were summarized into the evaluation of the model. Each of the lines as shown were described by the standardized Beta values taken from the multiple regression analysis and also the significant levels of *** or ** , as *** reflects on the p-value that has significant level of 0.001 or lower and ** means the significant level of 0.05 and below.

For our hypothesis 1, we accepted the hypothesis as the Attitude is a significant factor that has influenced the Behavioral Intention and we also have accepted the hypothesis number 2 where the factor Perceived Usefulness has influenced the Behavioral Intention to use mobile payment in Thailand. Furthermore, there are also four more hypotheses that we concluded that 193

the hypotheses were not supported by the results. The four hypotheses were the hypothesis number 3,4,6 and ,7. Therefore the factors Perceived Ease of Use, Subjective Norms, Perceived Risk and Covid-19 were found to be no influence towards the use of mobile payment in Thailand. Lastly, For the Perceived Cost factor, we have decided to delete it during the good fit model process of CFA therefore we did not proceed further with the regression analysis with control variables and multiple regression analysis.

By comparing to the multiple regression from the combined dataset and also from Vietnam, we have found that the similarity between both countries are very similar. We only accept the hypotheses 1 and 2 and also not support the rest of the hypotheses. This reflects that when people have a strong attitude and their perception of mobile payment as a useful tool. Then it influence the intention to use the mobile payment.

9. Conclusion

Theoretical implications

Theoretical implications in general

By comparing our results to the general findings from many countries about the mobile payment. The first filled gap is including the country context in our conceptual parts and covering region variety criteria in data collection which are barely found in the previous researches about this topic. We have found that an article from Zhang, Y. et al ,2018, Khan, H. et al. 2015, and Lin, X. et al., 2019 they have found that the Subjective Norm have influence over the Behavioral Intention in China, Qatar and South Korea, meanwhile in Thailand and Vietnam have the opposite findings. However, our results could be confirmed with the article from Tan, K. 2019, where the Chinese consumers in Malaysia has held strongest attitude towards the use of mobile payment. According to Teng, P. et al ,2018. The findings show that attitude is the main factor which influence customer to adopt to mobile payment service in China which is similar to our findings. According to an article from Zhang, Y. et al ,2018, the people in China has low perceived risk comparing to the people from the USA where Perceived Risk is still influencing the intention to use mobile payment.

194

that it is the Perceived Usefulness and Attitude were the two strongest predictors which is also different from India. We have confirmed the Perceived Usefulness construct with the result from Singapore, an article from Seetharaman, A. et al, 2017, they have found that perceived usefulness, and transaction security influenced strongly to intention to use mobile payment. However, our result of Perceived Cost has been deleted due to the CFA process, therefore we still couldn’t compare this construct to the result from Seetharaman, A. et al, 2017, where in Singapore, Perceived Cost has a very strong influence over to the intention to use mobile payment.

Theoretical implications for Vietnam

We will have a look back to chapter 5.3.2 where we illustrated all the research gaps that we found about behavioral intention to use mobile banking or mobile payment in Vietnam. Then, we will compare to these research gaps with our findings to illustrate the major points that we can cover in our research.

First of all, we have a short review back to the research of Liang (2016) to see what are the research gap that we can fulfil. The study of Liang (2016) only mentioned about mobile banking in general not mobile payment in specific. On the other hand, our research is more specific when we mainly focus on the mobile payment service. Furthermore, Liang’s research only tested on non-user because the research was taken even before the penetration of mobile payment into Vietnam. Our study could cover this research gap since we conducted the research conduct after 2 years that mobile payment services operating in Vietnam. Besides, we could also fulfil the gap of theories background when we already wrote up about country context, emerging market in general. About the testing variables, we had same result to this research when our study indicated that Subjective Norm had no influence on the consumer behavior of mobile payment usage.

Then, we will have a look at the research of Dao (2019). She also researched about mobile payment and applied UTAUT model where the testing factors were performance expectancy, effort expectancy, social influence (subjective norm) and trust. As Subjective Norm was the most discussing factor in previous studies about mobile payment or mobile banking. However, the research of Dao (2019) concluded that this factor had positive influence on the consumer behavior which was totally different with our findings. So, we could one again supported that the Subjective Norm factor had no influence on the intention to mobile payment services.

Next, in the research of Nguyen and Dao (2016), they used the TAM and TPB as their research model with six testing factors as perceived usefulness, perceived ease of use, perceived enjoyment, perceived trust, subjective norm, enjoyment and perceived behavioral control. One

195

again, we had the opposite finding with this research when it concluded that Subjective Norm significantly impacted on the consumer behavior of mobile payment usage. However, the positive influence of Perceived Usefulness was supported by our research since we also conclude that this factor had influence on the consumer behavior of mobile payment usage. Besides, we could also fill the gap of region variety which they didn’t proceed in their research. In our research, we distributed our surveys in three different regions of Thailand and Vietnam to ensure about the quality of our data. Moreover, the current user of mobile payment was tested in our research which Nguyen and Dao couldn’t do in their research in 2016.

There were many opposite conclusions about Subjective Norm in the previous researches about mobile payment or mobile banking in Vietnam. For example, in the research of Phan et al. (2020), they stated that this factor has positive influence on the consumer behavior to use mobile payment services. So, our research played an important role to support the contrast side that Subjective Norm was not important to predict the consumer behavior of mobile payment usage. Although the research of Phan el al., (2020) was most updated, their target sample was only Ha Noi (North). Our research could fill this gap since we distributed our data across the country from North to South.

Theoretical implications for Thailand

According to the research gap from the article reviews in chapter 5.3.3. By comparing to an article from Boonsiritomachai, W. & Pitchayadejanant K. (2017) we have found that we can fill the gap by confirming that the Subjective Norm factor has no influence over the mobile payment and also their article is focusing only on the mobile banking making this point of result similar. We have used the Perceived Ease of Use and also Perceived Usefulness as our independent variables whereas the article from Boonsiritomachai, W. & Pitchayadejanant K. (2017) have also used TAM as their ideal model but they did not use those two important factors from TAM. At the end, the predictor Perceived Usefulness is proved to be supporting our hypothesis and it has influence over the Behavioral Intention.

By comparing the article from Phonthanukitithaworn, C. et al., (2015), we have found that the Subjective Norm has no influence on the Behavioral Intention which is opposite to the early adopters group from their article. Moreover, in their article, the Perceived Usefulness have found to be no influence over the Behavioral Intention but we have found that the Perceived Usefulness is the strong predictor of using mobile payment for the current users. In the article, it has only focused on the early aopters on the mobile payment users, therefore we can fulfill the research gap for the current users. We have also confirmed about the same factor Perceived Usefulness

196

to the Behavioral Intention to use mobile payment, according to the article from Chansaenroj, P. & Techakittiroj, R. (2015), their results were conflicted to our results. By comparing their results, the Perceived Usefulness, Perceived Ease of Use and also Perceived Risk were found to have influence over the Behavioral Intention. Perceived Cost was found to have no influence over the Behaviroal Intention. Therefore, by comparing our findings and this article, we can conclude that we can only confirm the similar result with only the factor Perceived Usefulness and also we cannot really assume the Perceived Cost since we have decided to delete this construct during the CFA stage to make a model fit. Furthermore, the article from Chansaenroj, P. & Techakittiroj, R. (2015) and Phonthanukitithaworn, C. et al., (2015) were focused only the Bangkok areas and not all regions around Thailand, therefore our research can also say that we have collected the Thai mobile payment users around the country from all the regions. Their articles did not use the moderators such as education level and also age group as our research. We can assumed that we fulfill research gaps by doing the regression analysis with the control variables.

By comparing our results to the general findings from many countries about the mobile payment. We have found that an article from Zhang, Y. et al ,2018, Khan, H. et al. 2015 , and Lin, X. et al., 2019 they have found that the Subjective Norm have influence over the Behavioral Intention in China, Qatar and South Korea, meanwhile in Thailand it is the opposite. However, our results could be confirmed with the article from Tan, K. 2019, where the Chinese consumers in Malaysia has held strongest attitude towards the use of mobile payment. According to Teng, P. et al ,2018. The findings show that attitude is the main factor which influence customer to adopt to mobile payment service in China which is similar to our findings. According to an article from Zhang, Y. et al ,2018, the people in China has low perceived risk comparing to the people from the USA where Perceived Risk is still influencing the intention to use mobile payment.

Our results could be confirmed with China that people trust in the system and has no issues with the risk of using mobile payment applications. In India, perceived ease of use is considered the strongest predictor for behavior intention (Shankar, A. & Datta, B., 2018). By comparing to our results , it has confirmed that it is the Perceived Usefulness and Attitude were the two strongest predictors which is also different from India. We have confirmed the Perceived Usefulness construct with the result from Singapore, an article from Seetharaman, A. et al, 2017, they have found that perceived usefulness, and transaction security influenced strongly to intention to use mobile payment. However, our result of Perceived Cost has been deleted due to the CFA process, therefore we still couldn’t compare this construct to the result from Seetharaman, A. et al, 2017, where in Singapore, Perceived Cost has a very strong influence over to the intention to use mobile payment.

197

Practical implications

From results of Vietnam, it’s indicated that the secondary level of education reacts to the Subjective Norm. The age range of this group lies under 18 years old. It’s easy to understand why this group reacts to the Subjective Norm because this age group is easily impacted by their influencers. And this group is also a potential target customer since they usually catch up with the latest technologies or innovations. Hence, the marketers should involve relevant key opinion leaders (KOLs), influencers, talents, stars, etc. when they do the marketing campaigns for their mobile payment services. This will be an effective way to strongly persuade this group to become your users.

As we can see from our findings that Attitude is the strongest predictor towards intention to use mobile payment services in Vietnam. So, we can quickly conclude that most of people have the positive opinion about mobile payment services and they tend to use this service when they are able to reach it. Therefore, in the management level, we should develop the marketing strategies in two directions. The first direction aims to current users, we should strongly focus on retention programs for the loyal customers. We should develop more specific incentives, promotions for the current customers to give them more interesting experiences when using your services. On the other hand, we should also focus on quality improvement and customer service to ensure the customer satisfaction. The second direction targets to non-users who never use any mobile payment service or have no awareness about this service. This market is quite potential since the current user portion is still low comparing to the whole population. The management roles should focus more on above the line marketing activities to raise customer awareness about this new service. And marketing message should focus on the usefulness, convenience of this service towards the users since the Perceived Usefulness is also an important predictor to use mobile payment in Vietnam. The application design should be simple, easy to use which can ensure that all of the age group will be able to use the application easily. Besides, the manager can try to reach as many merchants as possible to increase the convenience of this service. And it’s better if the company can cooperate with many different business sectors from convenient stores, retail stores, supermarkets, public transportation, hospital, airline, hospitality, etc. to enhance the convenience of service and approach many more users.

In our research result from the regression analysis with the control variable, we have found out that all education levels in Thailand were reacted toward the Covid19 situation. This reflects on the possibility that we can also introduced the marketing campaigns and advertisements related to the situation of Covid19 such as the encouragement to use mobile payment applications with the buying behavior during the pandemic. Encourage people to use less cash and also use more of the applications and also promote the products that were related to the pandemic situation so 198

they can get more bonus points in the mobile payment system such as masks, alcohol gels. Furthermore, many hospital and the Covid-19 test centers should be able to link the POS payment system with the mobile payment applications, again, hospital should also encourage patients to use the m-payment methods.

By looking at our result from the age group, in Thailand we have found out almost all age groups between 18 to 54 has reacted towards the intention to use mobile payment in Thailand. This also can be interpreted in many things, most of our respondents were all the current users therefore the intention to use m-payment is always relevant to everyone, not just only divided by the age group. Moreover, the current users in Thailand has a strong intention to pursue the use of m-payment, this means they are already loyal customers of m-payment applications. Therefore, as a manager, it is strongly suggested that the loyalty programs should be used intensively to the current users. For example, the current users will also receive news about new promotion and bonuses very often so they can be more active and engage to use the m-payment. Not only the new promotions, but also building the circle of payment where a person who use m-payment application will gain more benefits when transferring the money to others who also use the same m-payment applications.

When making marketing strategies, our results from the multiple regression analysis has shown that Attitude and the Perceived Usefulness were the two strongest predictors in the Thai market. Therefore, the suggestions would be that managers should always do advertisements and campaigns about how useful it is to use m-payment. May be the m-payment companies should also considered link their system to the transportation payment systems such as the express way for cars, BTS sky train, MRT, Airlines where Thai people will use the m-payment applications to pay for these everyday lives and gain additional bonuses such as tickets free every 10 to 20 purchases (just for an example). This will allow them to think how useful it is to use the m-payment and it will eventually involve in their everyday life; this will create a strong attitude towards the use of m-payment.

10. Limitations

There were some subjective and objective reasons which caused a few of limitations in our research. First of all, the unexpected Covid-19 crisis made some changes in our research plan and somehow restricted our research. For example, we were not able to reach more non-user because we couldn’t conduct paper-based survey as the initial plan. It’s known that most of the non-users are over 65 years old and they barely interacted with internet or web services and mobile devices so it’s very difficult to collect the e-survey from people over 65 years. As the initial

199

plan, we would like to test for both non-user and user but we were not able to reach enough target of non-users so we decide to test only with the current user to see their behavioral intention.

One more limitation that we have in this research was the time constraints, we only had around one months to distribute and collect the data so we can only obtain the target of around 200 respondents in each country. If the timeline was more redundant, we could approach more respondents and collect more data. Furthermore, we also planned to apply Qualitative method as Focus group in this research. However, the time limitation didn‘t allowed us to process all the things as plan.

11. Direction for future study

Even though we tried to cover as much as what have not been done in previous research about mobile payment, there are still many limitations. And this will be a potential key for the researcher to develop their studies in the future. First, the behavioral intention cannot completely reflect the action in reality. Therefore, it is essential for the future study to indicate the proportion of people who are willing to use mobile payment in reality comparing to the assumed behavioral intention. Second, our research is able to compare only among Southeast Asia countries. It is very useful if the future research can broaden the scope to compare Southeast Asia results with Europe or US results. Besides, the future studies are highly recommended to include the Qualitative research method into their studies so they can measure the respondent answers not only in the numbers but also in words. The Qualitative Method with the specific Focus Group can help the researcher deeply understand the insight of customers. Furthermore, the research samples can broaden to many different generations such as Gen X, Gen Y, Baby Boomer, Millennials so researchers are able to compare between the generation in their intention towards mobile payment usage. Additionally, the future studies can consider having the professional interviews with mobile payment companies or providers so it allows the research to understand on the management perspective.

200

12. References

Abrahão, R. d. S., Moriguchi, S. N., & Andrade, D. F. (2016). Intention of adoption of mobile payment: An analysis in the light of the Unified Theory of Acceptance and Use of Technology (UTAUT). RAI Revista De Administração E Inovação, 13(3), 221–230. https://doi.org/10.1016/j.rai.2016.06.003

Accenture.com. (2020). Driving the future of payments, 10 megatrends. Retrieved from https://www.accenture.com/_acnmedia/pdf-62/accenture-driving-the-future-of-payments- 10-mega-trends.pdf.

Ajibade, P. (2018). Technology Acceptance Model Limitations and Criticisms: Exploring the Practical Applications and Use in Technology-related Studies, Mixed-method, and Qualitative Researches, pp. 1–13. Retrieved from http://digitalcommons.unl.edu/libphilprac/1941. Ajzen, I. (1991). The Theory of Planned Behavior. In P. van Lange, A. Kruglanski, & E. Higgins (Eds.), Handbook of Theories of Social Psychology: Volume 1 (pp. 438–459). SAGE Publications Ltd. https://doi.org/10.4135/9781446249215.n22 Arneson, K. (2020). ‘Cash is king’ has long been the motto of German consumers and small business owners – but Covid-19 is bringing rapid change. Retrieved from https://www.bbc.com/worklife/article/20200520-will-coronavirus-change-germans-love-of- cash Auer, R. & Cornelli, G. & Frost, J. (2020). Covid-19, cash, and the future of payments. Bank for International Settlements 2020, No. 3. Bloomenthal, A. (2020). On the Move: Mobile Commerce. Investopedia. Retrieved 20 September 2020, from https://www.investopedia.com/terms/m/mobile-commerce.asp. Bader, J. (2019) Payment Methods Report 2019. E-wallets, Mobile wallets, and P2P payments. The Paypers. Bank of Thailand (2018). Payment Systems Roadmap. No. 4. Retrieved from https://www.bot.or.th/English/PaymentSystems/PolicyPS/Documents/PaymentRoadmap_ 2021.pdf Beldad, A.D & Hegner, S.M. (2018). Expanding the Technology Acceptance Model with the Inclusion of Trust, Social Influence, and Health Valuation to Determine the Predictors of German Users’ Willingness to Continue using a Fitness App: A Structural Equation Modeling Approach. International Journal of Human–Computer Interaction, 34(9), p. 882- 893, DOI:10.1080/10447318.2017.1403220 Bergthaler, M. (2019). Payment Methods Report 2019. E-wallets, Mobile wallets, and P2P payments. The Paypers.

201

Betuel, E. (2020). Covid-19 may change one common American behavior forever. Retrieved from https://www.inverse.com/mind-body/will-the-coronavirus-kill-cash Bezhovski, Z. (2016). The Future of the Mobile Payment as Electronic Payment System. European Journal of Business and Management 8(8). 2222-2839

Bhat, A. (2020). Non-Probability Sampling: Definition, Methods and Examples | QuestionPro. Retrieved 14 May 2020, from https://www.questionpro.com/blog/non-probability-sampling/ Black, W., & Babin, B. J. (2019). Multivariate data analysis: Its approach, evolution, and impact. In The Great Facilitator (pp. 121-130). Springer, Cham. Bob. 2018. The Rise of Thailand’s Cashless Economy. Retrieved form https://appsynth.net/blog/thailand-cashless-economy/ Bonett, D. & Wright, T. (2014). Cronbach's alpha reliability: Interval estimation, hypothesis testing, and sample size planning. Journal of Organizational Behavior. 36. 10.1002/job.1960. Boonsiritomachai, W., & Pitchayadejanant, K. (2017). Determinants affecting mobile banking adoption by generation Y based on the Unified Theory of Acceptance and Use of Technology Model modified by the Technology Acceptance Model concept. Kasetsart Journal of Social Sciences. Advance online publication. https://doi.org/10.1016/j.kjss.2017.10.005 Bosamia, M., & Patel, D. (2019). Wallet Payments Recent Potential Threats and Vulnerabilities with its possible security Measures. International Journal of Computer Sciences and Engineering, 7(1), 810–817. https://doi.org/10.26438/ijcse/v7i1.810817

Bentler, P.M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246. Capgemini.com. (2020). Retrieved 15 September 2020, from https://www.capgemini.com/fr-fr/wp- content/uploads/sites/2/2017/07/world_payments_report_2014.pdf. Cheung, M. (2019). China Mobile Payment Users 2019. eMarketer. Retrieved 8 October 2020, from https://www.emarketer.com/content/china-mobile-payment-users-2019. Clement, J. (2020). Topic : Mobile payments worldwide. Statista. Retrieved 21 November 2019, from https://www.statista.com/topics/4872/mobile-payments-worldwide/. Cao, T. K., Dang, P. L., & Nguyen, H. A. (2016). Predicting consumer intention to use mobile payment services: Empirical evidence from Vietnam. International Journal of Marketing Studies, 8(1), 117-124. Chansaenroj, P. & Techakittijoj, R. 2015: Factors Influencing the Intention to Use Mobile Banking Services in Bangkok, Thailand (ISBN: 978-93-85465-78-9), pp. 54 - 58.

Chawla, A. (2020). Covid-19 as a Catalyst in Deeper Acceptance of Digital Payment Platforms. Retrieved from https://www.counterpointresearch.com/covid-19-catalyst-digital-payment/ Check J &, Schutt, RK. (2011). Research methods in education

202

Chhonker, M. S., Verma, D., & Kar, A. K. (2017). Review of Technology Adoption frameworks in Mobile Commerce. Procedia Computer Science, 122, 888–895. https://doi.org/10.1016/j.procs.2017.11.451

Child, D. (2006). The essentials of factor analysis. (3rd ed.). New York, NY: Continuum International Publishing Group. Citypopulation.de (2019). Thailand division: Planning regions and provinces. Retrieved from https://www.citypopulation.de/en/thailand/prov/admin/ Cobanoglu, C.& Yang, W. & Shatskikh, A. & and Agarwal, A. (2015) "Are Consumers Ready for Mobile Payment? An Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences. New York: Routledge, https://doi.org/10.4324/9780203771587 Coronavirus. (2020). Retrieved 11 June 2020, from https://www.who.int/emergencies/diseases/novel-coronavirus- 2019?gclid=Cj0KCQjwrIf3BRD1ARIsAMuugNvjW8g6_FU_krnCAST049fmId_hDfphzlTre- LMFc7ydc8MCipE6TYaAv8NEALw_wcB Cropley, A. J. (2019). 2nd updated, revised, and enlarged edition. Qualitative research methods: A practice-oriented introduction for students of psychology and education. Riga, Latvia: Zinātne. (open access – doi: 10.13140/RG.2.1.3095.6888) Daoud, J.I (2017). Multicollinearity and Regression Analysis. Journal of Physics: Conference Series, 949, 12009.

Dahlberg, T., Guo, J., & Ondrus, J. (2015). A critical review of mobile payment research. Electronic Commerce Research and Applications, 14(5), 265-284. Dahlberg, T., Mallat, N., Ondrus, J., & Bachfischer, A. (2006). Mobile payment market and research-Past present and future. Helsinki mobility roundtable. Dahlberg, T. & Mallat, N. & Öörni, A. (2003). Trust enhanced technology acceptance modelconsumer acceptance of mobile payment solutions: Tentative evidence. Retrieved from https://www.researchgate.net/publication/267934058_Trust_enhanced_technology_acce ptance_modelconsumer_acceptance_of_mobile_payment_solutions_Tentative_evidence Dahlberg, T., Mallat, N., Ondrus, J., & Zmijewska, A. (2008). Past, present and future of mobile payments research: A literature review. Electronic Commerce Research and Applications, 7(2), 165–181. https://doi.org/10.1016/j.elerap.2007.02.001

Dash, M. & Bhusan R.B. & Samal, S. (2014). Determinants of Customers’ Adoption of Mobile Banking: An Empirical Study by Integrating Diffusion of Innovation with Attitude. Journal of Internet Banking and Commerce 19(3). P.2-21

Davis, F. (1989): Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology. In MIS Quarterly 13 (3), p. 319. DOI: 10.2307/249008.

203

De Sena Abrahão, R., Moriguchi, S. N., & Andrade, D. F. (2016). Intention of adoption of mobile payment: An analysis in the light of the Unified Theory of Acceptance and Use of Technology (UTAUT). RAI Revista de Administração e Inovação, 13(3), 221-230. Demirguc-Kunt, A., Klapper, L., Singer, D., & Van Oudheusden, P. (2015). The global findex database 2014: Measuring financial inclusion around the world. The World Bank. Demography - Elderly population - OECD Data. (2020). Retrieved 18 May 2020, from https://data.oecd.org/pop/elderly-population.htm#indicator-chart Dennehy, D., & Sammon, D. (2015). Trends in mobile payments research: A literature review. Journal of Innovation Management, 3. https://doi.org/10.24840/2183-0606_003.001_0006 Dhakal, C. (2018). Interpreting the Basic Outputs (SPSS) of Multiple Linear Regression. International Journal of Science and Research (IJSR). 10.21275/4061901. Digital payments nearly double in March amid virus outbreak. (2020). Retrieved 11 June 2020, from https://www.bangkokpost.com/business/1891110/digital-payments-nearly-double- in-march-amid-virus-outbreak

Diop E.B. & Zhao S.& Duy, T.V. (2019). An extension of the technology acceptance model for understanding travelers’ adoption of variable message signs. PLoS ONE 14(4). https://doi.org/10.1371/journal.pone.0216007 Doan, E. (2020). Smartphone users in Vietnam 2015-2022 | Statista. Retrieved 13 May 2020, from https://www.statista.com/statistics/467739/forecast-of-smartphone-users-in-vietnam/ Doing business in Vietnam. (2019). Retrieved from https://www.pwc.com/vn/en/publications/2019/pwc-vietnam-dbg-2019.pdf Dougn, D. (2020). Vietnam is experiencing a boom in mobile payments - Vietnam Insider. Retrieved 20 May 2020, from https://vietnaminsider.vn/vietnam-is-experiencing-a-boom- in-mobile-payments/ Du, S., & Li, H. (2019). The Knowledge Mapping of Mobile Commerce Research: A Visual Analysis Based on I-Model. Sustainability, 11(6), 1580. https://doi.org/10.3390/su11061580

Enberg, J. (2019). Global Mobile Payment Users 2019. eMarketer. Retrieved 8 October 2020, from https://www.emarketer.com/content/global-mobile-payment-users-2019.

Examination of Consumer Acceptance of Mobile Payment Technology in Restaurant Industry," Hospitality Review: Vol. 31 (4) Featherman, M. S., & Pavlou, P. A. (2003). Predicting e-services adoption: a perceived risk facets perspective. International journal of human-computer studies, 59(4), 451-474. Fichman, R.G. (1992). Information Technology Diffusion: A Review of Empirical Research. International Conference on Information Systems (ICIS). Field, A. (2000). Discovering Statistics using SPSS for Windows. London – Thousand Oaks – New Delhi: Sage publications 204

Faber, E., & Bouwman, H. (2003). Designing business models for mobile payment services. Proceedings of ICEB, Singapore. Fatonah, S., Yulandari, A., & Wibowo, F. W. (2018). A Review of E-Payment System in E- Commerce. In Journal of Physics: Conference Series (Vol. 1140, No. 1, p. 012033). Fiallos, F., & Wu, L. (2005). Digital Money: Future Trends and Impact on Banking. Financial Institutions, and eBusiness. Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18, 39-50. Fornes, G., & Mendez, A. (2014). China-latin america axis. Palgrave Macmillan. Foster, K. & Greene, C., & Stavins J. (2018). The 2018 Survey of Consumer Payment Choice: Summary Results. Frost, J. (2017). How to interpret R square in regression analysis. Retrieved from http://statisticsbyjim.com/regression/interpret-squared-regression/ Gbongli, K., Xu, Y., & Amedjonekou, K. M. (2019). Extended Technology Acceptance Model to Predict Mobile-Based Money Acceptance and Sustainability: A Multi-Analytical Structural Equation Modeling and Neural Network Approach. Sustainability, 11(13), 3639. https://doi.org/10.3390/su11133639

Gallagher M.W., Brown T.A. (2013) Introduction to Confirmatory Factor Analysis and Structural Equation Modeling. In: Teo T. (eds) Handbook of Quantitative Methods for Educational Research. SensePublishers, Rotterdam. https://doi.org/10.1007/978-94-6209-404- 8_14 Ghasemi, A. & Zahediasl, S. (2012). Normality Tests for Statistical Analysis: A Guide for Non- Statisticians. International journal of endocrinology and metabolism. 10. 486-489. 10.5812/ijem.3505. Girish, V., & Manu, M. Y. (2020). Impact of COVID-19 on Mobile Banking Services. Studies in Indian Place Names, 40(70), 2790-2797.

Goodhue, D., & Thompson, R. (1995). Task-Technology Fit and Individual Performance. MIS Quarterly, 19(2), 213-236. doi:10.2307/249689 Gupta, S., Nguyen, T. D., Rojas, F. L., Raman, S., Lee, B., Bento, A., ... & Wing, C. (2020). Tracking public and private response to the covid-19 epidemic: Evidence from state and local government actions (No. w27027). National Bureau of Economic Research. Hair, J. F., Ringle, C. M., & Sarstedt, M. (2013). Partial least squares structural equation modelling: Rigorous applications, better results and higher acceptance. Long Range Planning, 46, 1– 12. Hu, L.T. & Bentler, P.M. (1999). Cut-off criteria for fit indexes in covariance structure analysis. Con-ventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55.

205

Hubley A.M. (2014) Discriminant Validity. In: Michalos A.C. (eds) Encyclopedia of Quality of Life and Well-Being Research. Springer, Dordrecht. https://doi.org/10.1007/978-94- 007-0753-5_751 Hord, J. (2005). How electronic payment works. Online]. HowStuffWorks. Available: http://money. howstuffworks. com/personal-finance/onlinebanking/electronic-payment1. htm [Accessed: 23-February-2017]. Ha, C. (2020). COVID-19 offers new growth opportunity for mobile money in Vietnam - Vietnam Insider. Retrieved 11 June 2020, from https://vietnaminsider.vn/covid-19-offers-new- growth-opportunity-for-mobile-money-in-vietnam/ Hadi, N. & Abdullah, N. & Ilham, S. (2016). An Easy Approach to Exploratory Factor Analysis: Marketing Perspective Noor Ul Hadi. Journal of Educational and Social Research. 6. 215- 223. 10.5901/jesr.2016.v6n1p215. Ho Chi Minh City, Vietnam Population (2020) - Population Stat. (2020). Retrieved 13 May 2020, from https://populationstat.com/vietnam/ho-chi-minh-city Imtiaz, M. A., & Maarop, N. (2014). A Review of Technology Acceptance Studies in the Field of Education. Jurnal Teknologi, 69(2). https://doi.org/10.11113/jt.v69.3101

Intention to Use Mobile Payment System by Ethnicity: A Partial Least Squares Multi-group Approach. Asian Journal of Business Research Vol 9 (1). DOI:10.14707/ajbr.19005 J.P. Morgan Global Payment Trends (2019). E-Commerce Payments Trends: Thailand. Thailand E-Commerce Insights. Retrieved from https://www.jpmorgan.com/merchant- services/insights/reports/thailand Jabri I.M & Sohail, M.S (2012). Mobile Banking Adoption: Application of Diffusion of Innovation Theory. Journal of Electronic Commerce Research 13(4). P.319 - 391 Jones, S. & Gu, J. (2012). Vietnam’s Regions and Key Economic Zones. Retrieved from https://www.vietnam-briefing.com/news/vietnams-regions-key-economic-zones.html/

Joshi, A., Kale, S., Chandel, S., & Pal, D. (2015). Likert Scale: Explored and Explained. Current Journal of Applied Science and Technology, 7(4), 396–403. https://doi.org/10.9734/BJAST/2015/14975

Kang, J. (2018). Mobile payment in Fintech environment: trends, security challenges, and services. Human-centric Computing and Information Sciences, 8(1), 1-16

Kim, C., Tao, W., Shin, N., & Kim, K. S. (2010). An empirical study of customers’ perceptions of security and trust in e-payment systems. Electronic commerce research and applications, 9(1), 84-95.

Kim, C., Mirusmonov, M., & Lee, I. (2010). An empirical examination of factors influencing the intention to use mobile payment. Computers in Human Behavior, 26(3), 310-322.

206

Kaiser, H. F. (1960). The application of electronic computers to factor analysis. Educational and Psychological Measurement, 20, pp. 141- 151. https://doi.org/10.1177/001316446002000116

Kaiser, H. F. (1974). An index of factorial simplicity. Psychometrika, 39, pp. 31–36 Katona, G. (1968). Consumer Behavior: Theory and Findings on Expectations and Aspirations. Kemp, S. (2020). Digital 2020: Vietnam — DataReportal – Global Digital Insights. Retrieved 18 May 2020, from https://datareportal.com/reports/digital-2020-vietnam Kemp, S. 2020. Digital 2020: Thailand. Retrieved from https://datareportal.com/reports/digital- 2020-thailand Khan, H. & Alshare, K.: Factors Influence Consumers' Adoption of Mobile Payment Devices in Qatar 13 (6). DOI: 10.1504/IJMC.2015.072100.

Kim, S.H. & Yoo B.K. (2020). Factors Influencing the Reuse of Mobile Payment Services in Retail. Journal of Distribution Science 18-3 (2020) pp. 53-65 http://dx.doi.org/10.15722/jds.18.3.202003.53 King, W. R., & He, J. (2006). A meta-analysis of the technology acceptance model. Information & Management, 43(6), 740–755. https://doi.org/10.1016/j.im.2006.05.003

King, K. M., & Jackson, K. M. (2019). Improving the implementation of quantitative methods in addiction research: Introduction to the special issue. Addictive Behaviors, 94, 1–3. https://doi.org/10.1016/j.addbeh.2019.05.003 Kohl, S. & Eydgahi, A. (2017). A systemativ review of technology adoption frameworks and their applications. Journal of Technological Management Innovation 12(4). Korean Users. Sustainability 2019, 11, 5532. doi:10.3390/su11195532 Lai, P. C. (2017). THE LITERATURE REVIEW OF TECHNOLOGY ADOPTION MODELS AND THEORIES FOR THE NOVELTY TECHNOLOGY. Journal of Information Systems and Technology Management, 14(1). https://doi.org/10.4301/S1807-17752017000100002

Lewis-Beck, M. S., Bryman, A., & Futing Liao, T. (2004). The SAGE encyclopedia of social science research methods (Vols. 1-0). Thousand Oaks, CA: Sage Publications, Inc. doi: 10.4135/9781412950589 Lal & Rajiv & Sachdev (n.d.). "Mobile Money Services—Design and Development for Financial Inclusion." Harvard Business School Working Paper, No. 15-083. Lance, C. E., Butts, M. M., & Michels, L. C. (2006). The sources of four commonly reported cutoff criteria: What did they really say? Organizational Research Methods, 9(2), 202–220. Landreneau, K. J., & Creek, W. (2009). Sampling strategies. Available on: http://www. natco1. org. Leducq, D., & Scarwell, H. (2018). The new Hanoi: Opportunities and challenges for future urban development. Cities, 72, 70-81. doi: 10.1016/j.cities.2017.08.003

207

Li, J. & Wang, J. & Wangh, S., & Zhou, Y. (2019). Mobile Payment With Alipay: An Application of Extended Technology Acceptance Model. IEEE Access, 7, 50380–50387. https://doi.org/10.1109/ACCESS.2019.2902905

Li, L (2010). A Critical Review of Technology Acceptance Theory. Retrieved from http://swdsi.org/swdsi2010/SW2010_Preceedings/papers/PA104.pdf.

Liang, C. C. (2016, January). Subjective norms and customer adoption of mobile banking: Taiwan and vietnam. In 2016 49th Hawaii International Conference on System Sciences (HICSS) (pp. 1577-1585). IEEE. Lin, X. & Wu, R. & Lim, Y. & Han, J. & Chen, S. (2019). Understanding the Sustainable Usage Intention of Mobile Payment Technology in Korea: Cross-Countries Comparison of Chinese and Liu, G.-S., & Tai, P. T. (2016). A Study of Factors Affecting the Intention to Use Mobile Payment Services in Vietnam. Economics World, 4(6). https://doi.org/10.17265/2328- 7144/2016.06.001 Longyara, T. & Van H.T. (2015). Diffusion of Innovation in Asian: A Study of Mobile NFC (Near Field Communication) Payment in Korea and Thailand. International Journal of Managerial Studies and Research 3(10). P.36-42 Lu, L. (2019). Mobile Payments—Why They Are So Successful? Open Journal of Business and Management, 07(03), 1131–1143. https://doi.org/10.4236/ojbm.2019.73078

Luna, I., Liébana-Cabanillas, F., Sánchez-Fernández, J., & Muñoz-Leiva, F. (2019). Mobile payment is not all the same: The adoption of mobile payment systems depending on the technology applied. Technological Forecasting and Social Change, 146, 931–944. https://doi.org/10.1016/j.techfore.2018.09.018

Lwoga E.T. & Lwoga, N.B. (2017). User Acceptance of Mobile Payment: The Effects of User- Centric Security, System Characteristics and Gender. The Electronic Journal of Information Systems in Developing Countries 81(3). P. 1-24 Mahmood, A. & Thaw, Y & Dominic, P.D. (2009). A Study on a Factors That Influence the Consumers’ Trust on E-commerce Adoption. International Journal of Computer Science and Information Security 4(1). Mallat, N., & Dahlberg, T. (2005). Consumer and Merchant Adoption of Mobile Payment Solutions. In T. Saarinen, M. Tinnilä & A. Tseng (Eds.), Managing Business In A MultiChannel World: Success Factors For E-Business. Hershey, PA 17033, USA: Idea Group Publishing. Masihuddin, M., Khan, B. U. I., Mattoo, M. M. U. I., & Olanrewaju, R. F. (2017). A survey on e- payment systems: elements, adoption, architecture, challenges and security concepts. Indian Journal of Science and Technology, 10(20), 1-19.

208

McKibbin, W. J., & Fernando, R. (2020). The global macroeconomic impacts of COVID-19: Seven scenarios. McMillan, J and Schumacher, S (2010). Research in Education. Evidence base Inquiry. 7th Edition, International Edition Boston: Pearson Education Inc Meadows, K. A. (2003). So, you want to do research? 4: An introduction to quantitative methods. British Journal of Community Nursing, 8(11), 519-526. Mercury-processing.com, (2020). Covid-19 (corona visus) impact on payments and fintech. Retrieved from https://mercury-processing.com/industry-news/covid-19-coronavirus- impact-on-payments-and-fintech/ Mertler, C. & Reinhart, R. (2016). Advanced and multivariate statistical methods: Practical application and interpretation. Sage Publications Inc. Mobilepaymentstoday, (2019). Thailand, Vietnam embrace mobile payments amid cashless push. Retrieved from https://www.mobilepaymentstoday.com/news/thailand-vietnam-embrace- mobile-payments-amid-cashless-push/ Momani, A., & Jamous, M. (2017). The Evolution of Technology Acceptance Theories. International Journal of Contemporary Computer Research (IJCCR), 1, 50–58. Mondego, D. (2018). Understanding the influence of trust on mobile payment systems adoption: A literature review, 8, 1–5. Morse, J. (1991). Qualitative nursing research. United States. Muijs., D. (2004). Doing Quantitative Research in Education (p. 1). London : SAGE Publications Ltd. Nga, V. T. H., Cuong, T. H., & Nga, B. T (2020). Factors affecting the use of mobile payment service at sales points: A case study in Long Bien district, Hanoi, Vietnam. Nguyen, P. Y., Lee, S. T., Kang, W. M., Kim, J. S., & Gim, G. Y. (2015). An Empirical Study on Factors Affecting Customer Intention to Use Mobile Payment System in Vietnam. Journal of the Korea Society of IT Services, 14(4), 171–184. https://doi.org/10.9716/KITS.2015.14.4.171 Nguyen, T. N., Cao, T. K., Dang, P. L., & Nguyen, H. A. (2016). Predicting Consumer Intention to Use Mobile Payment Services: Empirical Evidence from Vietnam. International Journal of Marketing Studies, 8(1), 117. https://doi.org/10.5539/ijms.v8n1p117 Nunnally, J.C. (1978) Psychometric theory. 2nd Edition, McGraw-Hill, New York. Norman, G. (2010). Likert scales, levels of measurement and the “laws” of statistics. Adv in Health Sci Educ 15, 625–632. https://doi.org/10.1007/s10459-010-9222-y Odoyo Collins Otieno, Samuel Liyala, Benson Charles Odongo, Silvance Abeka (2016). Theory of reasoned action as an underpinning to technological innovation adoption studies. Oh, J.S. & Park, C.U. & Lee S. B. (2014). NFC-Based Mobile payment Service Adoption and Diffusion. Journal of Convergence 5(2). P. 8 -14

209

Ondrus, J., & Pigneur, Y. (2005). A Disruption Analysis in the Mobile Payment Market.Paper presented at the 38th Hawaii International Conference on System Sciences, Hawaii, USA, January 3-6 Ondrus, J., & Pigneur, Y. (2006b). A Systematic Approach to Explain the Delayed Deployment of Mobile Payments in Switzerland. Paper presented at the 5th International Conference on Mobile Business, Copenhagen, Denmark, June 26-27 Okazaki, S. (2005). New Perspectives on M-Commerce Research. Journal of Electronic Commerce Research 6(3). Oliveira, T and Martins, M, F. (2011). Literature Review of Information Technology Adoption Models at Firm Level. The Electronic Journal Information Systems Evaluation Volume 14 (1), pp110- 121 O'nions, C. (2020). Ipsos.com. Retrieved 3 April 2020, from https://www.ipsos.com/sites/default/files/2016-11/The-millennial-influence-27-10-16.pdf. Ono, Y. (2017, June 1). Mobile payments are catching on strong in Southeast Asia. Nikkei Asian Review. https://asia.nikkei.com/magazine/20170601/Business/Mobile-payments-are- catching-on-strong-in-Southeast-Asia Porcellana, S. (2016). Integrate Mobile Payment & Marketing Strategy - Marketo. Marketo Marketing Blog - Best Practices and Thought Leadership. Retrieved 9 October 2020, from https://blog.marketo.com/2016/07/6-ways-to-integrate-mobile-payment-into-your- consumer-marketing-strategy.html Prudon, P. (2015). CFA as a tool in research using questionnaires: a critique. Comprehensive Psychology. 4. Article 10. 10.2466/03. CP.4.10. Pathirana, P.A., & Azam, S. M. (2017). Factors influencing the use of mobile payments — A conceptual model. https://doi.org/10.1109/NITC.2017.8285656 Patil, P.P. & Dwivedi, Y.& Rana, N. (2017). Digital Payments Adoption: An Analysis of Literature. 16th Conference on e-Business, e-Services and e-Society, pp.61-70, 10.1007/978-3-319- 68557-1_7. hal-01768529 Phonthanukitithaworn, & C. Sellitto, C. & Fong, M. (2015). User Intentions to Adopt Mobile Payment Services: A Study of Early Adopters in Thailand. Journal of Internet Banking and Commerce, 20(1). https://www.researchgate.net/profile/Chanchai_Phonthanukitithaworn/publication/282741 683_User_Intentions_to_Adopt_Mobile_Payment_Services_A_Study_of_Early_Adopters _in_Thailand/links/57dbc8b108ae4e6f18455bb4/User-Intentions-to-Adopt-Mobile- Payment-Services-A-Study-of-Early-Adopters-in-Thailand.pdf Pousttchi, K. & Wiedemann, D. (2007): What Influences Consumers' Intention to Use Mobile Payments? Retrieved from https://pdfs.semanticscholar.org/a4dd/ce87edaa6d2bf14d0bd362e1c9beb24b030b.pdf.

210

PWC, (2020). Mobile payments in Vietnam fastest growing globally, Thailand emerges second in Southeast Asia. Retrieved from https://www.pwc.com/th/en/press-room/press- release/2019/press-release-30-04-19-en.html. Razali, M. N. & Bee, Y. (2011). Power Comparisons of Shapiro-Wilk, Kolmogorov-Smirnov, Lilliefors and Anderson-Darling Tests. J. Stat. Model. Analytics. 2. R.O. Mueller, G.R. Hancock, in International Encyclopedia of the Social & Behavioral Sciences, 2001 Rao, S.V. (2015). E-Wallet A ‘Pay’volution. Retrieved from https://www.researchgate.net/publication/302586756_E-_WALLET_- _A_%27PAY%27VOLUTION

Rodgers, G. (2019). The 7 best place to visit in Northern Thailand. Retrieved from https://www.tripsavvy.com/the-best-places-to-visit-in-northern-thailand-4147334 Rogers, E. M. (1983). Diffusion of innovations (3rd ed.). New York, London: Free Press; Collier Macmillan.

Rogers, E.M. (1995) Diffusion of Innovations. 4th Edition, the Free Press, New York. Roscoe, A. M., Lang, D., & Sheth, J. N. (1975). Follow-up Methods, Questionnaire Length, and Market Differences in Mail Surveys: In this experimental test, a telephone reminder produced the best response rate and questionnaire length had no effect on rate of return. Journal of Marketing, 39(2), 20-27. Schober, P & Boer, C. & Schwarte, L. (2018). Correlation Coefficients: Appropriate Use and Interpretation. Anesthesia & Analgesia. 126. 1. 10.1213/ANE.0000000000002864. Sedgwick, P. (2014). Understanding P values. BMJ (online). Pp. 349. DOI: 10.1136/bmj. g4550. Seetharaman, A., Kumar, K. N., Palaniappan, S., & Weber, G. (2017). Factors Influencing Behavioural Intention to Use the Mobile Wallet in Singapore. Journal of Applied Economics & Business Research, 7(2), 116–136. Shon, T. H., & Swatman, P. M. (1998). Identifying effectiveness criteria for Internet payment systems. Internet Research. Senaviratna, N.& Cooray, T. (2019). Diagnosing Multicollinearity of Logistic Regression Model. Asian Journal of Probability and Statistics. 1-9. 10.9734/ajpas/2019/v5i230132. Shaakumeni, S. N., & Csapó, B. (2019). Exploring the Factorial Validity of the Beliefs about Nature of Science Questionnaire. Science Education International, 30(1), 38-44 Singh, P. R. I. Y. A. N. K. A., Supriya, N., & Joshna, M. S. (2013). Issues and challenges of electronic payment systems. International Journal for Research in Management and Pharmacy, 2(9), 25-30. Smutkupt, P., Krairit, D., & Esichaikul, V. (2010). Mobile marketing: Implications for marketing strategies. International Journal of Mobile Marketing, 5(2).

211

Shankar, A., & Datta, B. (2018). Factors Affecting Mobile Payment Adoption Intention: An Indian Perspective. Global Business Review, 19(3_suppl), S72-S89. https://doi.org/10.1177/0972150918757870 Sharma, S. K., Mangla, S. K., Luthra, S., & Al-Salti, Z. (2018). Mobile wallet inhibitors: Developing a comprehensive theory using an integrated model. Journal of Retailing and Consumer Services, 45, 52–63. https://doi.org/10.1016/j.jretconser.2018.08.008 Sharma,R. & Mishra, R. (2014). A Review of Evolution of Theories and Models of Technology Adoption. IMJ 6(2). Retrieved from https://www.researchgate.net/publication/295461133_A_Review_of_Evolution_of_Theori es_and_Models_of_Technology_Adoption Smartphone users in Thailand 2013-2022 | Statista. (2019). Retrieved 13 May 2020, from https://www.statista.com/statistics/467191/forecast-of-smartphone-users-in-thailand/ Syahadiyanti. L, & Subriadi A.P. (2018). International Journal of Economics and Financial Issues 8(3). P.219-226 Sykes, L M, Gani, F, & Vally, Z. (2016). Statistical terms Part 1: The meaning of the MEAN, and other statistical terms commonly used in medical research. South African Dental Journal , 71(6), 274-278. Retrieved from http://www.scielo.org.za/scielo.php?script=sci_arttext&pid=S0011- 85162016000600009&lng=en&tlng=en. Tabachnick, B. G., & Fidell, L. S. (1991). Software for advanced ANOVA courses: A survey. Behavior Research Methods, Instruments, & Computers, 23(2), 208-211. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston, MA: Allyn & Bacon. Taguchi, N. (2018). Description and explanation of pragmatic development: Quantitative, qualitative, and mixed methods research. System, 75, 23–32. https://doi.org/10.1016/j.system.2018.03.010 Tan, K. & Memon, M. & Sim, P. & Leong, C. & Soetrisno, F & Hussain, K. (2019). Taylor, S., & Todd, P. (1995a). Understanding information technology usage: A test of competing models. Information Systems Research, 6, 144–176. doi:10.1287/isre.6.2.144 Taylor, S., & Todd, P. (1995b). Assessing IT Usage: The Role of Prior Experience. MIS Quarterly, 19(4), 561-570. doi:10.2307/249633 Taylor,L. & Churchill, B. (2013). Descriptive Statistics with SPSS. Chapter 9: Using SPSS for Descriptive Statistical Analysis. Oxford University Press. Third Edition. Pp. 195 - 224 Teo, E., Fraunholz, B., & Unnithan, C. (2005). Inhibitors and facilitators for mobile payment adoption in Australia: A preliminary study. Paper presented at the 4th International Conference on Mobile Business, Sydney, Australia, July 11-13.

212

Teng, P. K., Ling, T. J., & Seng, K.W. K. (2018). Understanding Customer Intention to Use Mobile Payment Services in Nanjing, China. International Journal of Community Development & Management Studies, 2, 49-60, Retrieved from: http://ijcdms.org/Volume02/v2p049- 060Teng4446.pdf

Tran, C. (2019). Ho Chi Minh City: How Vietnam's Emerging Megacity Will Develop. Retrieved 13 May 2020, from https://www.vietnam-briefing.com/news/ho-chi-minh-city-how-vietnams- emerging-megacity-will-develop.html/ Trochim, W. (2020). Research Methods Knowledge Base. Retrieved 14 May 2020, from https://conjointly.com/kb/sampling-in-research/ Trütsch, T. (2016). The impact of mobile payment on payment choice. Financial Markets and Portfolio Management, 30(3), 299–336. https://doi.org/10.1007/s11408-016-0272-x U.S. Library of Congress. (2020). Regions, retrieved from http://countrystudies.us/thailand/37.htm Varshney, U., & Vetter, R. (2001). A framework for the emerging mobile commerce applications. In Proceedings of the 34th Annual Hawaii International Conference on System Sciences. https://doi.org/10.1109/HICSS.2001.927208

Valcourt, E., Robert, J. M., & Beaulieu, F. (2005, August). Investigating mobile payment: supporting technologies, methods, and use. In WiMob'2005), IEEE International Conference on Wireless and Mobile Computing, Networking and Communications, 2005. (Vol. 4, pp. 29-36). IEEE. Venkatesh, V. & Bala, H. (2008). Technology Acceptance Model 3 and a Research Agenda on Interventions. Decision Sciences - DECISION SCI. 39. 273-315. 10.1111/j.1540- 5915.2008.00192. x. Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information technology: Toward a unified view. MIS quarterly, 425-478. Venkatesh, V., Thong, J., & Xu, X. (2012). Consumer Acceptance and Use of Information Technology: Extending the Unified Theory of Acceptance and Use of Technology. MIS Quarterly, 36(1), 157-178. Retrieved from http://www.jstor.org/stable/41410412 Venkatesh,V. & Davis,F.D. (2000). A Theoretical Extension of the Technology Acceptance Model: Four Longitudinal Field Studies. Management Science 46(2):186-204. http://dx.doi.org/10.1287/mnsc.46.2.186.11926 Vietnam GDP | 1985-2019 Data | 2020-2022 Forecast | Historical | Chart | News. (2020). Retrieved 18 May 2020, from https://tradingeconomics.com/vietnam/gdp Vietnam, F. (2020). Overview of Vietnam’s Major E-Wallet and Mobile Payment Players - Fintech Singapore. Retrieved 18 May 2020, from https://fintechnews.sg/32843/vietnam/overview- of-vietnams-major-e-wallet-and-mobile-payment- players/https://fintechnews.sg/32843/vietnam/overview-of-vietnams-major-e-wallet-and- mobile-payment-players/

213

Wang, Y., Hahn, C., & Sutrave, K. (2016). Mobile payment security, threats, and challenges. 2016 Second International Conference on Mobile and Secure Services (MobiSecServ), 1-5.

Williams, L.J. & Holahan, P.J. (1994). Parsimony-Based Fit Indices for Multiple-Indicator Models: Do They Work? Structural Equation Modeling 1(2). p. 161-189. World Payments Report 2016. Capgemini Worldwide. (2020). Retrieved 15 September 2020, from https://www.capgemini.com/resources/world-payments-report- 2016/#:~:text=across%20the%20globe.-,Capgemini%2C%20in%20a%20new%20collab oration%20with%20BNP%20Paribas%20brings%20the,(KRIIs)%20that%20govern%20t hem.

Wurmser, Y. (2019). US Mobile Payment Users 2019. eMarketer. Retrieved 8 October 2020, from https://www.emarketer.com/content/us-mobile-payment-users-2019.

Wani, T. (2015). Innovation Diffusion Theory Review & Scope in the Study of Adoption of Smartphones in India. Journal of General Management Research vol3. Issue2. Pp.101 - 118. Williams, M. D., Rana, N. P., & Dwivedi, Y. K. (2015). The unified theory of acceptance and use of technology (UTAUT): a literature review. Journal of Enterprise Information Management, 28(3), 443–488. https://doi.org/10.1108/JEIM-09-2014-0088 Wollenberg, A. (2014). Consumer Behaviour in the Smartphone Market in Vietnam. International Journal of Innovation, Management and Technology, 5(6). https://doi.org/10.7763/IJIMT.2014.V5.550 Wong, W. H., & Mo, W. Y. (2019). A Study of Consumer Intention of Mobile Payment in Hong Kong, Based on Perceived Risk, Perceived Trust, Perceived Security and Technological Acceptance Model. Journal of Advanced Management Science, p. 33–38. https://doi.org/10.18178/joams.7.2.33-38

World Payments Report 2019. World Payments Report. (2020). Retrieved 2 April 2020, from https://worldpaymentsreport.com/. Worthington, R. L., & Whittaker, T. A. (2006). Scale Development Research: A Content Analysis and Recommendations for Best Practices. The Counseling Psychologist, 34(6), pp. 806– 838. https://doi.org/10.1177/0011000006288127 Wu, J., Liu, L., & Huang, L. (2017). Consumer acceptance of mobile payment across time. Industrial Management & Data Systems, 117(8), 1761–1776. https://doi.org/10.1108/IMDS-08-2016-0312

Xu, C., Mongo, P., & Ganiyu, S. (2020). Model construction and empirical study on mobile commerce user satisfaction. Current Psychology. Advance online publication. https://doi.org/10.1007/s12144-020-00877-z

214

Yeh, H. (2020). Factors in the Ecosystem of Mobile Payment Affecting its Use: From the Customers' Perspective in Taiwan. Journal of Theoretical and Applied Electronic Commerce Research, 15(1), 0. https://doi.org/10.4067/S0718-18762020000100103 Yong, A. & Pearce, S. (2013). A Beginner’s Guide to Factor Analysis: Focusing on Exploratory Factor Analysis. Tutorials in Quantitative Methods for Psychology 2013, Vol. 9(2), pp. 79- 94. Yu, H. C., Hsi, K. H., & Kuo, P. J. (2002). Electronic payment systems: an analysis and comparison of types. Technology in Society, 24(3), 331-347.

Yousafzai, S. Y., Foxall, G. R., & Pallister, J. G. (2007a). Technology acceptance: a meta‐ analysis of the TAM: Part 1. Journal of Modelling in Management, 2(3), 251–280. https://doi.org/10.1108/17465660710834453

Yousafzai, S. Y., Foxall, G. R., & Pallister, J. G. (2007b). Technology acceptance: a meta‐ analysis of the TAM: Part 2. Journal of Modelling in Management, 2(3), 281–304. https://doi.org/10.1108/17465660710834462

Yuvaraj S. (2016). The factors influencing the buying behavior of the consumers with digital payment modes. Journal of management (JOM). Vol. 5 Issue 1. pp.74-80 Zhang, Y., Sun, J., Yang, Z., & Wang, Y. (2018). What Makes People Actually Embrace or Shun Mobile Payment: A Cross-Culture Study. Mobile Information Systems, 2018, 1–13. https://doi.org/10.1155/2018/7497545

215

13. Appendix: Questionaire

We are Thuy Tran and Pailin Kunnawat and we are in our last semester of the Master. Management in Johannes Kepler University Linz. Currently, we are doing our Master thesis about “Factors influencing the consumers’ behavior intention to use mobile payment in Vietnam and Thailand. Therefore, we really appreciate if you can spend a few minutes to conduct this survey to support for our research.

Our questionnaires have divided into three different parts. The first part will be about the usage of mobile payment, current or non-users, the service they have used and the frequencies. The second part will be about the questions related to our hypotheses; these questions are using the rating scale (likert scale between 1-5). The last part is the demographic data, genders, age groups, education levels, careers and regions of each country.

Part I: General information

1. Are you currently using any mobile payment method?

□ Yes □ No 2. If yes, what are the services that you chose?

(Vietnam)(VietnamNetBridge, 2019) □ ZaloPay □ MoMo □ ViettelPay □ Moca □ Airpay □ GrabPay □ Others (Thailand) (Bob, 2018) □ True Money □ AIS □ mPay □ PaySocial □ Airpay □ Rabbit LINE pay □ Prompt pay □ Others (7-11 App wallet) 3. How often do you use mobile payment method for your transactions?

216

□ Almost everyday □ At least once a week □ A few times a month □ A few times a year □ Less than once a year □ Never (Williams, D.R. 2016)

Part II: Personal attitude towards the usage of mobile payment method

In which extent that you can agree or disagree with each of these statements while (1): strongly disagree, (2) disagree, (3) Neither agree nor disagree, (4) agree, (5) strongly agree. Please kindly mark an "X" mark in the box of your answer. (Adapted from Venkatesh, V. 2003; De Sena Abrahão et al., 2016; Zhang et al., 2018)

Behavioral Intention 1 2 3 4 5 I will always use mobile payment in the future I will always try to use mobile payment in my daily life Once I had access to mobile payment services, I would really use them I will recommend mobile payment to my friends, relatives, colleagues I plan to use mobile payment in the next 6-12 months

(Adapted from Taylor, S. et al., 1995) Attitude 1 2 3 4 5 The use of mobile payment is a good idea The use of mobile payment is convenient The use of mobile payment is beneficial The use of mobile payment is interesting

(Adapted from Davis, 1989; Kim et al., 2010) Perceived Usefulness 1 2 3 4 5 Using mobile payment would enable me to pay quickly

217

Using mobile payment would make it easier for me to conduct transactions Using mobile payment would be advantageous I would find mobile payment a useful tool for paying

(Adapted from Davis, 1989) Perceived Ease of Use 1 2 3 4 5 I believe that when I use mobile payment, the process will be clear and understandable I believe that it is easy for me to become skillful at using mobile payment I believe that mobile payment is easy to use

(Adapted from Venkatesh, V. 2003) Subjective Norm 1 2 3 4 5 People who are important to me (friends/family members) think that I should use mobile payment People who influence my behavior (influencers) think that I should use mobile payment People whose opinions that I value (boss, famous people) prefer that I use mobile payment

(Adapted from De Sena Abrahão et al., 2016; Luarn et al, 2005; Wei et al., 2009) Perceived Cost 1 2 3 4 5 I believe the mobile payment services would be very expensive I would have financial barriers (e.g. purchase of telephone and communication time expenses) in order to use the mobile payment services I believe I would have to do a lot of effort to obtain the information that would make me feel comfortable in adopting mobile payment. I believe that the transaction fees for using mobile payment will be high

218

(Adapted from Featherman et al., 2003) Perceived Risk 1 2 3 4 5 Compared to traditional payment method, I believe that using mobile payment is riskier I believe that there will be high potential for loss associated with using mobile payment (for instance, loss of my financial details to thieves) I believe that there will be too much uncertainty associated with using mobile payment (for instance, money does not get through to the receiver due to a network problem)

(Adapted from Girish et al., 2020) The impact of Covid-19 on mobile payment 1 2 3 4 5 Covid-19 has increased your use of mobile payment services

(to understand whether the Covid-19 is the reason to increase the use of mobile payment) Mobile payment services are more convenient during Covid-19 lock down

(to understand if the Covid-19 is a factor to motivate people to use mobile payment during lockdown) Mobile payment service helped to maintain social distance while making payment or receipt during Covid-19 lock down

(to understand the attitude of people when using mobile payment during lockdown, whether it is because the social distance or not) COVID-19 has not spread through mobile payment as it doesn’t involve touching of currencies

219

(to understand the attitude of people when using mobile payment during lockdown, whether it is because touching of cash or not) I will always use mobile payment applications during and even after COVID-19 lock down

(to understand whether Covid-19 is the only reason to use mobile payment or it is another motivation for people to continue using it)

Part III: Personal Information

1. Gender

□ Female □ Male 2. Age (New) □ < 25 □ 25 - 30 □ 31 - 35 □ 36 - 40 □ 41 – 45 □ 46 - 50 □ > 50 (Zhang, Y., Sun, J., Yang, Z., & Wang, Y. 2018) 3. Monthly Income (currencies and range will be according to each country) • (Vietnam) (VND) (Nguyen, T. N. et al. 2016)

□ Under 5000000 □ 5000000 - 10000000 □ 10000001 - 15000000 □ 15000001 - 20000000 □ Over 20000000 • (Thailand) Thai Baht

□ 0 – 24999 □ 25000 - 49999 □ 50000 - 74999 □ 75000 - 99999 □ 100000 - 150000 □ > 150000 (Driediger, F., & Bhatiasevi, V. 2019) 4. Educational Level

□ Highschool □ Associate □ Bachelor’s degree □ Master Degree □ Ph.D or higher level (Zhang, Y., Sun, J., Yang, Z., & Wang, Y. 2018) 5. Career

(Thailand and Vietnam) □ Student/ College Student

220

□ Government Officer □ Employee □ Entrepreneur □ Freelancer □ Others □ No job (Boonsiritomachai, W. & Pitchayadejanant, K. 2017)

6. Regions

(Vietnam) (Jones, S. & Gu, J. 2012) □ Northern Zone □ Central Zone □ Southern Zone (Thailand) (Phompradit et al. 2011) □ North □ Central □ South □ Northeast □ East □ West

221