Báo cáo nghiên cứu khoa học: Examining the role of AI – enabled voice assistants in affecting consumer motivations for online shopping: the mediating factors role awe experience, price value, sales promotion, and E–WOM

RESEARCH REPORT EXAMINING THE ROLE OF AI—- ENABLED VOICE ASSISTANTS IN AFFECTING CONSUMER MOTIVATIONS FOR ONLINE SHOPPING: THE MEDIATINGFACTORS ROLE AWE EXPERIENCE, PRICE VALUE, SALES PR

INTERNATIONAL SCHOOL, VIETNAM NATIONAL UNIVERSITY, HA NOI

FACULTY OF ECONOMICS AND MANAGEMENT

a ŒD

RESEARCH REPORT

EXAMINING THE ROLE OF AI—- ENABLED VOICE ASSISTANTS IN AFFECTING

CONSUMER MOTIVATIONS FOR ONLINE SHOPPING: THE MEDIATINGFACTORS ROLE AWE EXPERIENCE, PRICE VALUE, SALES PROMOTION, AND E

— WOM

Instructor : Dr Bui My Trinh

Students : Nguyen Quang Duy — IB2021B

Nguyen Ngoc Lan — IB2021BNguyen Thi Tuyet Vi — AC2021CDuong Thuy Anh — AC2021B

TEAM LEADER INFORMATION

- Program: International Business

- Address: Thuan Thanh — Bac Ninh

- Phone no /Email: 0985527696/ quangduyvjk9597 @ gmail.com

II Academic Results (from the first year to now)

Academic year Overall score Academic rating

Ist semester, 3rd year | 3.70 | Quite good

III Other achievements:

- Language: English and †⁄4 3: (Chinese)

- Google Digital Marketing & E-commerce Certificate (2023)

- Business Writing Certificate of University of Colorado Boulder (2023)

- 3 semesters consecutive scholarships of International School — Vietnam National University

- Scholarship from Lazada corporation for excellent students (2022)

- Scholarship from HTI corporation for students who have great academic results and

scientific research activities (2022-2023 academic year)

- Project Leader of the Scientific Research Competition of International School-VNU and

gained the Third Prize at University Level.

- Second Prize for Video Contest in the position of Leader

- Participate in the Semi-final Eureka Scientific Research Competition organized by VietnamNational University, HCM for students nationwide.

Hanoi, 15th April 2024Advisor Team Leader

(Sign and write fullname) (Sign and write fullname)

44 Mig Trinh Piguet Gog Deg

This research consumed a huge amount of work, research, and dedication Still, implementationwould not have been possible if we did not have the support of Dr Bui My Trinh, teammates,and individuals Therefore we would like to extend our sincere gratitude to all of them

We are deeply grateful to our advisor - Ph.D Bui My Trinh for her guidance, encouragement,and intellectual contributions throughout this research endeavor Her wisdom, expertise, andunwavering support have been instrumental in shaping the direction and quality of this work

We are thankful to the International School - Hanoi National University for their financial andlogistical support and for providing necessary guidance concerning research implementation

We would also like to express our appreciation to the participants who willingly volunteeredtheir time and provided valuable data for this study Their involvement was crucial in gatheringthe data that formed the basis of our research findings

Finally, we would like to thank our teammates for their collaboration and fruitful discussionsduring this research Their contributions to data analysis, experimental design, and manuscriptpreparation were greatly appreciated

The contributions of the aforementioned Dr Bui My Trinh, teammates, and individuals havebeen instrumental in the completion of this research project, and we are truly grateful for their

support.

Student

Nguyen Quang Duy

"hố ca na 18PIN] á0(00 0i): 1 192.7 E— WOM 05 19

3 RESEARCH HYPOTHESIS DEVELOPMENT ng HH, 20

3.1 MCT and 00ìi6i1-30ì0i/900500 1777 d 203.2 MCI and awe €X€TI€TICC - G0 1119919112101 TH ng HH Hệ 233.3 Awe experience, Price value, sales promotion, E-WOM, and purchase 1ntentions 25

4 METHODOLOGY HH HH TH TH HH HH HH TT TT TT 28

V0 - 314.3 PLS-SEM mẽ 314.4 Fuzzy Set Qualitative Comparative Analysis (fs(QCA) - - LH ngư, 32

5 DATA V0 904 33

5.1 Measurement model analysis - - - << + E3 1E E981 931 91 1v nh ng trưy 335.2 Structural model analysis 8 385.3 [soi 000i 1, 17777 455.4 Results of f8QCA analysis - - G1 HH TH TH TH HH Hệ 46

6 DICUSSION 0134 49

7 THEORETICAL CONTRIBUTIONS - Án HH HH HH HH Hiệp 33

8 MANAGERIAL IMPLICA TIONS - ng HH ng 54

9 CONCLUSION AND LIMITATIONS 000 cccccccccccsccsecsecseceeceecseeseeeeeceeeeeeeaceseeseeseeseeaeeaes 57

9.1 CONCLUSION eee 'dd ' 57

9.2 Limitations and future research 1333333311111 11 1 1E 555555

APPENDIX

LIST OF FIGURESFigure 1: Research MOdeÌ - - c1 39113113318 111 11 11191111 11T HH TH ng rệt 20

Figure 2: Results of the structural model (*p<0.05, **p<0.01, ***p<0.001) 43

LIST OF TABLESTable 1: Sample characteristics (M=300) cecceesceseseeeseeeseceeeesecesecceesecesecseeeseceseceeeeeeeseceeeeatenes 29

Table 2: Factor loadings Of Measures eee cece cee cee 1111 1 E1 111 1 1 H1 TT HH HH HH 33

Table 3: Assessment of reliability and convergent validity (n=300) -<+<<<sx+s2 35

Table 4: Discriminant validity - Fornell-Lacker criterion (N= 300) 555 «<< ++ssxssess 37

Table 5: Discriminant validity - Heterotrait-monotrait ratio of correlations (HTMT) 38

Table 2346) 00 i17 38

Table 7: VIF values in the structural mO(€Ì, - < +6 %6 1E 91833 E93 E8 99 2v vn riết Al

Table 8: f2 effect SizeS c.c.c.ccccccscsessssssssesesesesesssssssscscscsessescscssscscsvssessscssscscsesesesscssssscscscsessecessseeees 41

Table 9: Bootstrapping result of direct paths - - - c6 S1 2311321113911 9 1 1H 1n ng kiệt 43

Table 10: Results of hypotheses f€StITE - Án TH TH nh HH HH nh ng nh nh 44

Table 11: Bootstrapping results of mediating ©ŸÍ€C(S - <6 5 11+ S9 vn ng 46

Table 12: Main configurations for high consumer purchase intentions (fsQCA) - 48

Table 13: Demographics Questionnaires (English V€TSIOT)) 5 5 + +++£*+£+vsesesesersre 76

Table 14: Research Questionnaires (English V€TSIOT)) - c5 2 S3 rrre 77

Table 15: Demographics questionnaires (Vietnamese V€TSIOT)) - - 5 55 5< + sskeserserrers 81

Table 16: Research Questionnaires (VietnaMese VErsiON) :cccccsscccesssecessseceesseecesssecessseeesssees 82

PROJECT NAME

EXAMINING THE ROLE OF AI —- ENABLED VOICE ASSISTANTS IN AFFECTING

CONSUMER MOTIVATIONS FOR ONLINE SHOPPING: THE MEDIATINGFACTORS ROLE AWE EXPERIENCE, PRICE VALUE, SALES PROMOTION, AND E

AI-we have a limited grasp of their effects on enhancing the purchase intentions of onlineshoppers This study employs a novel theoretical model grounded in consumer innovativeness,broaden-and-built theory, and stimulus-organism-response model to investigate the impact ofmotivated consumer innovativeness to use Al-enabled voice assistants on online shoppers’purchase intentions and awe experience The model was evaluated with survey data from 300digital voice assistant customers The data was examined using partial least squares structuralequation modeling (PLS-SEM) and fuzzy-set qualitative comparative analysis (fsQCA) PLS-SEM revealed that awe experience, price value, sales promotion, and eWOM mediate therelationship between the role of AI-enabled voice assistants (Functional, Hedonic, Social, andCognitive MCI) and voice shoppers’ perceptions of purchase intentions The results fromfsQCA results suggest that multiple, distinct, and equally effective combinations of functionalMCI, hedonic MCI, social MCI, cognitive MCI, awe experience, price value, sales promotion,

and E-WOM exist to achieve high intention to purchase Seven solutions are presented thatlead to high intention to purchase The study complements to existing literature on consumerinnovativeness, Al-based voice assistants, and online buying These findings can helpbusinesses enhance their usage of voice assistants for online consumers.

Keywords (3 — 5 words)

Keywords: Al-enabled services Voice assistants, Motivated consumer innovativeness (MCI),

PLS-SEM, fsQCA, Awe experience, Purchase intentions

SUMMARY REPORT IN STUDENT RESEARCH,

2023-2024 ACADEMIC YEARPurposes/aim: Intending to investigate the role and functions of Al-enabled voice assistants inimpacting consumer motivations for online shopping Besides, we also applied some mediatingfactors like awe experience, price value, sales promotion, and E-WOM to evaluate the effects ofAl-enabled voice assistants on consumers’ purchase intention

Design/Methodology/Approach: 300 responses were obtained from Vietnamese consumers.Data was analyzed with combination of both PLS-SEM and fsQCA by using SMARTPLS andfsQCA software

Findings: PLS-SEM results revealed that awe experience, price value, sales promotion, andeWOM mediate the relationship between the role of Al-enabled voice assistants (Functional,Hedonic, Social, and Cognitive MCI) and voice shoppers’ perceptions of purchase intentions

fsQCA: suggests that multiple, distinct, and equally effective combinations of functional MCI,hedonic MCI, social MCI, cognitive MCI, awe experience, price value, sales promotion, E—WOM exist to achieve high intention to purchase

Managerial Implications: Marketing managers had better aim at the user-friendliness, coolness,customization, and individually tailored characteristics of Al-enabled services Refreshing andaccelerating the innovation in their performance by employing AI-enabled VA services are given.Marketing managers are suggested to give priority to sales promotion in mediating aweexperience and purchase intention Carrying out promotion campaigns collaborated with thetrendiness and fastest update in product information of AlI-enabled VA services helps e-storesstand out The development of an interface that enables the distribution of the VA app wouldbenefit e-retailers

Limitations: there is a limited degree of generalizability in the results because the data wereonly collected from Vietnam Future research might examine how the suggested theoreticalmodel applies to other locations and locales, which could improve its generalizability

Thanks to Alexa voice assistants which were introduced by Amazon in 2014, theaccessibility of smart speakers has been getting wider (Hu et al., 2019) Smart speakers, after that,became the most rapidly expanding consumer tools since the smartphone (Simms, 2019), andonline shoppers have a greater tendency to utilize artificial intelligence (AI) enabled e-tailservices such as voice assistants (VA), augmented reality (AR) or virtual reality (VR) to makepurchases (Balakrishnan et al., 2021; Brannon Barhorst et al., 2021; Grewal et al., 2017; Rabassa

et al., 2022) Forecasts suggest that by 2024, the number of digital voice assistants being used indevices around the world will reach 8.4 billion units — a number higher than the world’spopulation (Statista, 2024) Voice assistants have been extensively installed by Siri, Alexa, andGoogle, which give customers intriguing experiences when purchasing online (Aw et al., 2022;Whang, C., 2018)

Voice assistants or voice agents run on smartphones or specialized devices, engaging users

in conversations to aid them with multiple tasks or resolving user requests on a broad range oftopics in real-time using natural language processing (Hoy, 2018; Kautish et al., 2023; Vedula etal., 2023) Many brands have had voice assistants to assist their online selling (Klaus &Zaichkowsky, 2020) For instance, Starbucks recently launched a new feature in its mobile appcalled My Starbucks Barista, that lets consumers order and pay for food and drinks by speaking.Further, the firm developed an Amazon Alexa skill, that makes it possible for users to repurchasewhatever they like by communicating to their Echo speaker or any Alexa-powered device (Perez,2017) Plenty of brands resort to AI-based shopping assistants to claim the flourishing essence ofsuch technologies for online shopping (Kristen Stephens, 2022; Szeja, R., 2023)

An intuitive interface, effortless installation, learning, and well-informed capabilities

through voice control of AI-enabled technology have made voice assistants a user-friendlychoice for online shoppers (Kautish et al., 2023) The speech-recognition system enables voiceassistants to swiftly respond to their users orally, getting rid of the need for manual control (Lee,1988) and, as a result, making an impression on online shoppers The AI voice assistant analyzesuser queries using online data and user interactions before delivering the most appropriateresponse based on relevant facts (Whang, C., 2018)

Scholars have assumed that the VA technology research is well-timed and deservesscholarly attention (McLean et al., 2021) and initially successfully identify determinants

concerning voice assistants such as the adoption of the new technology: AI-based Vas (Kessler

& Martin, 2017), the impact of attitude and trust on the behavioral intention (Ashrafi & Easmin,2023), the investigation of the role of customer perceptions of VAs empathy and risk on VAsadoption (Coker & Thakur, 2023) Nevertheless, when it comes to the online shopping scenario,these previous articles are not explicit As officially recognized by scholars (Klaus &Zaichkowsky, 2022; Mari, 2019), the motivations of VAs technology usage for online shoppingare still ambiguous

Consumers’ adoption of Voice VA assistants for purchasing has been slow and restricted(Vitezié & Perié, 2021), which is why it is critical to perceive the motivations for the use of VAsthat prompt online shopping purchases Previous research has stressed the relationship betweeninnovativeness and its adoption by individuals, but the insight into this link was quite unclear(van Rijnsoever & Donders, 2009) Some researchers suggested investigating the drivingobjectives behind the adoption of an innovation (QUINLAN, 2008) presenting it as a moreargumentative justification of the adoption decisions (Kautish & Sharma, 2018) A few scholarshave researched the implications of functional, hedonic, social, and cognitive motivations on theadoption of innovative products (Saeed et al., 2014; Vandecasteele & Geuens, 2010), especially,the motivated consumer innovativeness (MCI) scale developed to evaluate consumers’motivation to employ innovative products or services (Vandecasteele & Geuens, 2010) Onlyseveral studies have justified the motivated consumer innovativeness regarding innovativetechnology use (Seyed Esfahani & Reynolds, 2021a) However, its influence on the purchaseintention of online shoppers is still unavailable As a whole, notwithstanding the vitality ofexamining the true motivations for using AI-enabled technology for buying, little is known aboutthat association, particularly in the VA context, postulating a research gap

In addition, consumer satisfaction in technology-related fields, especially, consumeremotional experiences cannot be pretermitted (Tai et al., 2021) The experience of awe,sometimes known as the 'wow effect,’ is an advanced emotional state that can motivate people toutilize new technologies like virtual reality (Hinsch et al., 2020a) Well-perceiving positiveemotions, like awe, is imperative for marketing research (Guo et al., 2018), for example,companies like Apple make an effort to inspire awe in their products, hence this concept retainssignificant relevance among marketers Marketers could take advantage of the awe feeling to

customers’ purchase intention (Alzahrani, 2020) Yet, the significance of awe in VA adoption forshopping in online platforms has to be studied.

Highlighting the impact of electronic word of mouth (eWOM) on customers’ purchaseintentions has been pointed out by scholars (Sa’ait et al., 2016) Consumers are increasinglysearching and using online reviews before purchasing a range of items; eWOM features likereview volume and valence have a substantial influence on new product sales (Chung et al., 2020)making it very relevant from a marketing point of view For this reason, we decided to includeeWOM in our conceptual framework as a paramount determinant influencing new technologyadoption

Scholars have also emphasized the relevance of sales promotion and purchase behaviours(Suryani & Syafarudin, 2021) It contributes to imparting product messages to clients, drivingdemand, establishing market positioning, and influencing purchasing decisions (Suresh et al.,2015) Consumers virtually pay attention to sales promotion at the first approach on anyshopping platform and as a consequence, this is a primary factor included in our research model

in association with technology innovativeness adoption

Finally, many studies have asserted that customers are willing to make purchases at any cost

in exchange for convenience or sacrifice convenience (Pham et al., 2018) With the growingprominence of a customer-driven approach, research indicates that price value plays a criticalrole throughout exchange operations (Wu et al., 2014) By making a comparison between theprosperity and trade-offs in online purchasing scenarios, customers can make decisions based ontheir rationale (Rajaguru, 2016) Marketers have been endeavouring to create an optimal pricevalue for their products (Razak et al., 2016) which can foster customers’ purchase intention.Therefore, we also added this element to our research model

As previously stated, the motivations for making purchases using AI-enabled VAtechnology have now been identified, indicating a crucial research gap The current study, which

is based on the consumer innovativeness concept, broaden-and-build theory (B&BT), andstimulus-organism response (SOR) model, investigates the influence of motivated consumerinnovativeness (MCI) on the purchase of products via VA technologies, taking into account themediation of eWOM, awe experience, price value and sales promotion The model assesses theimpact of the four (functional, hedonic, social, and cognitive) MCI components (stimulus) on theawe experience (organism) via eWOM, price value, sale promotion, and buying intention Awe

experience, in turn, is claimed to affect eWOM, pricing value, sales promotion, and customerpurchasing intention The proposed research questions are as follows:

RQI: How do functional, hedonic, social, and cognitive MCI for using VAs affectpurchasing intention via VAs and awe experience?

RQ2: How do awe experience, price value, sales promotion, and eWOM affect onlineshoppers’ intention to purchase through VAs?

The research offers significant theoretical developments to the literature Our research isamong the first studies that give insight into customer motivations for using VAs to purchasegoods A model based on consumer innovativeness and broaden-and-build theory sheds light onthe motivated consumer innovativeness for AI-enabled VA use as well as its effect on purchaseintention, awe experience, sales promotion, and price value among online shoppers

Thereby, we put forward the tenuous literature on VA utilization in online shopping (S Lee

et al., 2023) This knowledge reinforces consumer behaviour theory and allows practitioners tomake well-informed decisions about product design and communication strategies (Hinsch et al.,2020a; Septianto et al., 2020) Emphasizing the essential part of consumer innovativeness intechnology adoption (Li et al., 2011; Vandecasteele & Geuens, 2010), we extend the notion tothe VA and online shopping contexts, providing valuable theoretical insights Marketers haverecognized the importance of awe experiences in their strategies Nonetheless, the relationshipbetween awe and consumer behaviour has been scarcely proven in studies (Guo et al., 2018).Taking into consideration the significance of emotional experiences in technology-related areas

of study, incorporating the concept of awe experience into the VA and online shopping literatureoffers novel theoretical perceptivities Besides, the relevance of eWOM behaviors as asubstantial theoretical concept for innovative technology cannot be overlooked (Cheung et al.,2021) Hence, this study brings an emphasis on eWOM behaviors in the advanced AJ-enabled

VA technology for online shopping The introduction of price value and sales promotion inonline shopping using Al-enabled VAs is a new aspect that we offer in this search, still, theyhave been studying for the genuine effect on purchase intention potentially being studied in thecontext of technology innovation Marketers may leverage the clues to deliver a pleasant userexperience using VAs based on consumer motivations that can elicit purchase intention, aweexperiences, sales promotion, eWOM behaviors and price value among online consumers

2 THEORETICAL BACKGROUND

Al is disrupting corporate models and driving digital transformation Smart speakers are themost rapidly expanding AI-based consumer devices since the smartphone (Simms, 2019).Therefore, Business executives are currently considering how they may be used to boost salesand enhance customer shopping experiences Samantha is commonly referred to as Alexa or Siri

in the real world The voice-assistant software operates in continuous listening mode, waiting for

a keyword to activate (McLean & Osei-Frimpong, 2019) When the keyword is heard, the systemrecords the user's voice and sends it to the main server for processing with natural languageprocessing and machine learning algorithms The server answers the user's command and sendsthe response to the voice assistant, which is subsequently played back to the user (Hoy, 2018).This communication mechanism allows users to have natural conversations with their voiceassistants (Schweitzer et al., 2019)

Voice-assistants are progressively substituting traditional search engines due to their ability

to answer sophisticated inquiries from customers (Smith, 2020) For instance, consumers can askthe voice assistants to search for “the best UV protection mask under 50 thousand VND” or

“which is the most prevalent jeans in the market” Similarly, an increasing number of studiesindicate that consumers are using voice assistants to place online orders for their shopping needs(Klaus & Zaichkowsky, 2020)

The use of voice assistants for Internet shopping represents a paradigm shift (Klaus &

Zaichkowsky, 2020) Three major phenomenological observations are: (a) consumers are

increasingly considering product qualities or product benefits as a search criterion (O’Boyle &Picaro, 2013) (b) Instead of looking for brands directly, consumers are now interacting withthem by asking their voice assistants questions about them The algorithms of the correspondingvoice assistants retrieve this data, which in turn affects the purchasing decisions made byconsumers; (c) customer preferences are changing from visual stimuli to auditory cues from theirvoice assistants A few studies have been conducted in the literature to investigate thepsychological aspect of consumers’ decision-making in the context of artificial intelligence (AI)and how disruptive AI-based technologies, like voice assistants, are affecting consumerpreferences (Hoy, 2018; Smith, 2020) Nevertheless, it is currently unclear what the role of voiceassistants when using AI-enabled voice assistants to carry out various tasks, such as findinginformation, and giving correct desired information to consumers Hence, Voice assistants will

help marketers innovate and remodel consumer interactions, allowing them to provide a bettervalue proposition to their current and prospective clients The current study intends to fill thisgap by utilizing the MCI framework as an overarching theoretical framework in the area of voiceassistants.

Source: Apple's Siri Is The Most Popular Virtual Assistant In The World: Report (Mathur, 2019)

2.1 Motivated consumer innovativeness (MCI)

The term “motivated customer innovativeness” condenses the idea that people’s desire to trynew products or experiences is fueled by various motivations (Midgley & Dowling, 1978).Innovative customers are frequently drawn to new technology that exhibits innovative behaviorsbecause they crave for uniqueness (Hwang et al., 2019) According to the technology acceptancemodel (hereafter TAM) (Davis, 1989), there are two key factors shape people’s opinions theninfluence people willingness to try new things and end up using them These are perceivedusefulness and perceived ease of use However, there are believes that this theory is insufficient

to account for the uptake of new technology, numerous academics have expanded on the TAMtheory to the MCI concept (Hwang et al., 2019), MCI is the term for the internal and externalvariables that influence consumers’ innovative purchasing behavior This has four theoreticaldimensions: functional, hedonic, social, and cognitive motivations (Hwang et al., 2024) Each of

(Vandecasteele & Geuens, 2010) AI-enable Voice Assistants (VA) is a cutting-edge technologythat makes voice-based product shopping easier This VA provides a distinctive viewpoint socustomers may be persuaded to utilize them for shopping or other purchase-related tasks fordifferent reasons.

2.2 Broaden - and - build theory

The broaden-and-build theory,(hereafter B&BT), is a foundational and well-known modelfor more accurately capturing the special characteristics of positive emotions including joy,interest, contentment and love (Fredrickson, 2001) These positive emotions broaden anindividual’s momentary thought-action repertoire, (whether through play, exploration or similaractivities—positive emotions promote the discovery of novel and creative actions, ideas andsocial bonds), and build that individual’s resources; ranging from physical and intellectualresources, to social and psychological resources (Huppert et al., 2004) Importantly, theseresources function as reserves that can be drawn on later to improve the odds of successfulcoping and survival

Artificial intelligence, voice assistants, virtual reality, mixed reality, and other forms oftechnology-driven automation are all thought to be infused with the idea of a real mind that iscapable of imposing wisdom and emotions Al-enabled systems were categorized into twocategories by Russell and Norvig (2016) Human performance-rationality (a system that thinksand behaves rationally) and human reasoning-behavior (a system that thinks and acts likehumans) Through the subtle blending of the two dimensions, AlI-enabled services are usingemotions to reshape and transform consumer experiences (Huang et al., 2019)

2.3 Stimulus-organism-response model

According to Mehrabian and Russell (1974), environment stimulus(S) generates anemotional reaction (O) that leads to behavioral responses (R) of consumers So, the S-ORproposes that when a person gets exposed to external stimuli, the ‘inner organism changes’which leads to behavioral responses The S-O-R model has been widely implemented innumerous online environments (Sharma et al., 2021; Sharma, Fadahunsi, et al., 2022) According

to the S-O-R paradigm, environmental cues function as external stimuli and foreshadow theorganism as a mediator between the stimulus and the response, emphasizing the influence on the

psychological processes of the consumers (Mehrabian & Russell, 1974) Emotions, expectations,and thought processes are all influenced by motivated consumer innovativeness The S-O-Rmodel was utilized in the study to investigate how consumers engage with the features of onlinefashion portals (stimuli) and how this contributes to the creation of an amazing experience(organism), eWOM, and buy intentions (response).

2.4 Awe experience

The sensation of amazement and wonder is a pleasant emotion that is exhibited by the awe

or wow effect (Kautish & Khare, 2022) Emotions are crucial in understanding why different enabled products, virtual reality (VR) or augmented reality (AR), are accepted (Chirico et al.,2018; Quesnel et al., 2018; Yaden et al., 2019) There are two characteristics of awe: theperceived vastness and a need for accommodation (Keltner & Haidt, 2003) The strong force of

AI-an emotional stimulus, which cAI-an break or subdue people's willpower AI-and cause them to feellittle, helpless, afraid, and humble, is referred to as the perceived vastness When someoneexperiences something that is difficult to grasp or goes beyond what they have previouslyexperienced, they may feel confused and surprised In these situations, they may needaccommodations Awe inspires people to play and stretch their creative boundaries, which may

be shown in inventive and intellectually stimulating conduct in addition to social and physicalactivity Through awe emotions, Al-enabled services are changing and redefining consumerexperiences (Huang et al., 2019)

2.5 Price value

A core premise in consumer behavioral research 1s that value maximization (Zeithaml et al.,2020) is an important underlying rationale for people's choices Value is defined by prospecttheory (Kahneman & Tversky, 1979) as the perceived gain or loss relative to a baseline It assertsthat individuals choose to engage in behaviors that result in the highest payments Price value forproducts is a compromise between "give" and "get" components Zeithaml (1988) definedperceived value as a consumer's assessment of a product's usefulness based on their perceptions

of what they receive and provide From the view of consumer choice, consumers weigh theadvantages and costs of a product while making a decision (Kahneman & Tversky, 1979;Zeithaml, 1988) In this study, price value refers to consumer’s perception of the trade-off

value by delivering tailored information and targeted recommendations based on data alreadyprovided by the user (e.g., current/past decisions, purchases, requirements) via algorithms anddata processing analytics tools (Huang, 2018) Additionally, the hands-free nature of voiceassistants allows users to make faster judgments and save time and effort (Rhee & Choi, 2020).

2.6 Sales promotion

Sales promotion is a crucial strategy for marketers to meet sales targets and boostprofitability It involves using short-term stimuli and motivational strategies to improvepurchasing behavior and induce consumers to switch from competing brands (Kotler & Keller,1994) Sales promotion is a key component of the marketing mix that attracts customers andboosts short-term sales volume (Keyan & Natarajan, 2019) The consumer market adapts tochanging lifestyles and offers a variety of promotional strategies, including presents for sportsshoes, cash returns, discounts and coupons, and cash or gifts for returned certificates (Chang,2017) The effectiveness of sales promotion instruments is linked to consumer perception, whichperceives sales promotions as opportunities to benefit (Sinha & Verma, 2018) In this study,sales promotion also refers to consumers’ perceptions of the benefits they will get frompurchasing products As a result, by utilizing Al-enabled voice assistants, consumers can readilylocate product promotion information, allowing them to stimulate and arouse their desire to buy

2.7 E—- WOM

Electronic word of mouth, or eWOM, is the current term for this new type of online WOMcommunication (Yang, 2017) Compared to traditional media, consumers view eWOM as a farmore trustworthy medium (Cheung & Thadani, 2012) eWOM is an anonymous source ofinformation that allows consumers to share their experiences, expertise, and opinions via onlineplatforms such as discussion forums, blogs, and review websites (Daugherty & Hoffman, 2014).EWOM communication is well known for being a trustworthy, non-commercial source ofinformation that significantly affects the attitudes and actions of consumers when makingpurchases (Arif, 2019) With the help of e-WOM, information may be quickly communicatedamong current and prospective buyers According to scholars, e- WOM influences the image of aspecific brand Thus, e-WOM plays an important role in influencing product purchases bychanging its look based on information shared online by current customers (AI Halbusi, H., &Tehseen, S et al., 2018) Therefore, adopting Al-enabled voice assistants in this study can give

customers with reputable sources of information, such as reviews, based on the data provided byusers, increasing their buy intent.

3 RESEARCH HYPOTHESIS DEVELOPMENT

Hedonic MCI

Social MCI

E-WOM

Figure 1: Research Model

3.1 MCI and purchase intentions

Purchase intentions are a sign that a consumer may be considering or willing to make apurchase in the near future (Kang et al., 2020) There is enough data to support the idea thatpurchasing intentions and MCI aspects are related However, the VA context is unaware of thisrelationship Stated differently, there is a lack of explanation in the literature about the reasonsbehind the use of VAs that leads to purchase intentions Researchers have talked about howimportant it is to look at psychological reasons when examining technology adoption (Lin &Filieri, 2015) For example, Lin and Filieri (2015) found that consumer innovation encouragesthe intention to continue using technology Purchase intentions are directly impacted byconsumers' inventiveness for new technology-based services, claims Cao (2021) Comparable

findings about how MCI facets—functional, hedonic, social, and cognitive—affect purchaseintentions can be found in the body of existing research (Seyed Esfahani & Reynolds, 2021a;Vandecasteele & Geuens, 2010).

This study examines how functional, hedonic, social, and cognitive incentives influenceonline buyers' purchase intentions when using AI-enabled VA technology

3.1.1 Functional MCI and Purchase IntentionsThe functional MCI alludes to “consumer innovativeness motivated by the functionalperformance of innovations and focuses on task management and accomplishment improvement”(Vandecasteele & Geuens, 2010) The functional MCI depicts the functional or utilitarianbenefits such as enhanced efficiency, greater production, and risk (Vandecasteele & Geuens,2010) conceived by the customer that provokes the use of the innovative product (Venkatraman

& Price, 1990) VA functions in conjunction with machine learning, allowing the completion oftasks without explicit instructions, thereby delivering functional advantages to users (McLean &Osei-Frimpong, 2019) Customers purchase things online because of the functional benefitsprovided by the VAs (Moriuchi, 2019) For example, it is swifter when it comes to speed,compatibility, and accessibility, and it spares customers time while purchasing This is because

VA technology allows customers to perform multiple tasks with minimal effort and provides therelief of job completion, such as product-making purchases, without a need to type or hold thedevice (McLean & Osei-Frimpong, 2019) Many researchers noticed that functional MCIimpacts purchase intentions (Chopra, 2019; Seyed Esfahani & Reynolds, 2021b) Suchfunctional advantages given by the VA could boost the purchasing intentions of online buyers.Based on the literature, we propose the following hypothesis:

Hypothesis la: Functional MCI to use Al-enabled VA has a positive impact on ConsumerPurchase intentions

3.1.2 Hedonic MCI and Purchase IntentionsHypothesis 1b: Hedonic MCI to use AT-enabled VA has a positive impact on ConsumerPurchase intentions

The hedonic MCI is defined as "consumer innovativeness motivated by affective or sensorystimulation and satisfaction" (Vandecasteele & Geuens, 2010) The hedonic motivationsstimulate consumers to experiment with innovative items for excitement or pleasure(Vandecasteele & Geuens, 2010) It is linked with sensory stimulation among customers and

represents feelings such as pleasure (Hwang et al., 2019) These customers utilize the productowing that it's strange to others (Hwang et al., 2019) Previous research has described the usage

of modern technologies such as VA for hedonic characteristics (McLean & Osei-Frimpong,2019) Earlier studies have explored the importance of hedonic MCI in predicting purchaseintentions (Hinsch et al., 2020a) Sharma, Dwivedi, et al., (2022) showed that the sense offulfillment in using sophisticated devices has an enormous effect on behavioral intentions Ifcustomers enjoy using VAs, they will be more likely to use them for a variety of activities,including shopping On the basis of this justification, we claim that hedonic motivations have thepotential to raise purchase intention for goods because buyers would feel excitement, joy, orpleasure Therefore, we propose that:

3.1.3 Social MCI and Purchase IntentionsSocial MCI is defined as "consumer innovativeness motivated by the self-assertive socialdesire for differentiation" (Vandecasteele & Geuens, 2010) Brown and Venkatesh (2005) definesocially motivated consumer innovativeness (SMCD as the usage of innovative items to improveone's social standing or to attract others According to the literature, customers' adoption of AI-enabled novel products is influenced by their sense of belonging Several consumers believe thatutilizing a certain technology, such as the VA, elevates their social standing (McLean & Osei-Frimpong, 2019) Researchers have pointed out the symbolic component in socially drivenconsumer innovativeness illustrated in the creation of social identity through the adoption ofinnovative items (Roehrich, 2004), such as VA for shopping activities These customers wouldlike to achieve their social connection goals (Vandecasteele & Geuens, 2010) while appearingtrendy and smart to get social values Customers, for example, buy smartwatches to show offtheir social recognition, and this desire stimulates their purchase intentions (Patel et al., 2023).People's behavior is determined by what motivates them (Hwang et al., 2019), hence we assertthat socially driven consumer innovativeness is likely to increase the purchase intentions of

to reach the limit of their cognitive bounds (Vandecasteele & Geuens, 2010) Consumers makeuse of innovative technologies to intrigue their minds and achieve cognitive objectives includingexploration, perception, and mental creativity (Hwang et al., 2019) For instance, people mustselect suitable words and sentences to complete their task According to Sharma et al (2021),tasks should be controlled and executed at specific times These examples highlight the use ofcognitive abilities in the operation of VAs.

Hypothesis 1d: Cognitive MCI to use Al-enabled VA has a positive impact on Consumers'Purchase Intentions

3.2 MCI and awe experience

By the previously indicated conceptual backdrop, this study predicted how MCTI affectedcustomers’ feelings of astonishment In recent marketing research (Hinsch et al., 2020a; Kim etal., 2021) as well as consumer psychology literature (Kautish & Khare, 2022; Rudd et al., 2018),awe has been recognized as a conceptually significant construct Consumers' sense of wondercan be increased by cutting-edge technologies (Hinsch et al., 2020) Exposure to moderntechnology can elicit awe since it is a complicated emotion that develops when one is provoked

by something enormous that surpasses past information schema (Kim et al., 2021; Shiota et al.,2007) According to the study, buyers' sensations of self-transcendent and potentiallytransformative awe may be triggered by Al-enabled VAs Customers may be in awe of theunderlying incentives (functional, hedonistic, social, and cognitive) to employ interactive AI-enabled VA technology (Mishra et al., 2022)

3.2.1 Functional MCI and awe experienceFunctional motivation to use innovative products can be explained by the customer's desirefor convenience, time-saving, and accuracy (Hwang et al., 2019) These customers are drawn tothe goods because of their practicality and functionality Researchers have demonstrated awe inthe requirement for accommodations and proposed that customers experience awe as a result ofvirtual technology (Quesnel & Riecke, 2018) In addition, amazement is a powerful feeling thatsomeone has when they are confronted with a lot of stimuli (Hinsch et al., 2020) Customers arelikely to be amazed by the functional gains that Al-enabled VA technology offers (McLean et al.,2021) Therefore, it is suggested that:

Hypothesis 2a: Functional MCI to use Al-enabled VA has a positive impact on the Aweexperience

3.2.2 Hedonic MCI and awe experiencePeople who seek enjoyment or excitement experiment with novel and innovativetechnologies (Mishra et al., 2022) According to Hwang et al (2019), hedonism was classified as

a psychological incentive Previous research has noted that the awe is triggered by thepsychological drive or schema's astounding impact (Hinsch et al., 2020) Shoppers may alsoenjoy using AlI-enabled technology to explore the fashion portals (Kautish & Khare, 2022).Exposure to new technologies can cause amazement (Guo et al., 2018) as well as excitement ordelight (McLean & Osei-Frimpong, 2019) For instance, customers can utilize voice search ontheir cellphones to find the newest things with ASOS's Enki app Customers who use AT-enabled

VA for purchasing are likely to create awe in the eyes of other customers and enjoy orexperience excitement (McLean & Osei-Frimpong, 2019) Drawing from this conversation, wecontend that online consumers are likely to feel in amazement when they utilize VA technologyfor their amusement Therefore, it is suggested:

Hypothesis 2b: Hedonic MCI to use Al-enabled VA has a positive impact on the Aweexperience

3.2.3 Social MCI and awe experienceAccording to Quesnel and Riecke (2018), the feeling of awe falls into the range of self-transcendent encounters with wellness tools and an affective state of social interconnectedness.Previous research has established the awe sensation that is generated by social motives.According to the author, social cues were a major factor in the awe experience that people hadfor commercial objects Since fashion products are commercial, we incorporate this idea into ourresearch The social impact's immensity can be explained by awe (Hinsch et al., 2020).Furthermore, as an emotion that delves into the underlying psychology of a person's socialpresence or status, awe can be grounded in the social component A greater social status isconferred upon users of Al-enabled VA, as it is regarded as prestigious (Mishra et al., 2022).Therefore, we contend that the employment of VA by socially driven online consumers maycause astonishment or the "wow" effect in them Drawing from the existing literature'sdiscussions, we propose that:

Hypothesis 2c: Social MCI to use Al-enabled VA has a positive impact on the Aweexperience

Awe demonstrates the need for adjustment that Keltner and Haidt's (2003) cognitiveparadigm elicits Through the cognitive constitution of events and structures, people make sense

of the world (Guo et al., 2018) An individual's intellectual schemas may be called upon byexposure to unusual vastness, resulting in a transposition that necessitates a conceptualization(Guo et al., 2018) When people come upon something that defies their preexisting knowledgeand mental models, they experience awe and feel compelled to revise their psychological schema.Awe sensations arise when certain items are accommodated, requiring mental schemas to beadjusted For instance, consumers of fashion may apply mental schemas to search results whenthey utilize VA technology to look for products To cite, Uniqlo provides text chats and voice-based personalized recommendations Consumers experience astonishment and mentaladaptations when exposed to new and inventive technologies (Hinsch et al., 2020) Consequently,

we argue that online consumers who are driven by cognitive aspects could be in awe whenutilizing VA technology to look for products Based on these arguments from the literature, wepropose the following hypothesis:

Hypothesis 2d: Cognitive MCI to use Al-enabled VA has a positive impact on the Aweexperience

3.3 Awe experience, Price value, sales promotion, E-WOM, and purchase intentions

3.3.1 Awe experience and purchase intentionsAwe is recognized as a separate positive emotion related to awe evoked by different objects,distinct from happiness, joy, and pride (Kim et al., 2021; Septianto et al., 2020) Customers mayexperience awe when they are exposed to exquisite items, such as products (Septianto et al.,2020) Exposure to items that challenge customers’ existing mental schemas to make sense ofnovel experiences might result in the awe or wow effect (Hinsch et al., 2020b), which in turnevokes buying intentions for products (Guo et al., 2018) According to Kautish and Khare (2022),voice assistants create awe among shoppers Al-enabled technologies demand users to changetheir mental representations, leading to awe and behavioral outcomes (Hinsch et al., 2020) likepurchase intents (Guo et al., 2018) Previous studies have confirmed the impact of the aweexperience on customers’ decision-making (Guo et al., 2018) Customers’ awe increases theirdesire to test a product and piques their curiosity or learning (Septianto et al., 2020) Previousresearch has shown that awe experiences can influence purchasing intention (Guo et al., 2018).Thus, we infer the following hypothesis from the discussion of the literature:

Hypothesis 3: Awe experience has a positive impact on Purchase intentions among onlineshoppers.

3.3.2 Awe experience and price valuePsychologists believe that cognitive and emotive variables influence human behaviorconcurrently Piaget initially postulated that emotion and cognition are intertwined (Piaget, 1981).Piaget believed that, while cognitive and emotional ways of processing information are distinct,there are some interactions between them, stating that "cognition is like the motor while emotion

is like the gasoline; the gasoline enables the motor to run well but does not change the structure

of the car" Individual evaluation of value is a cognitive process, but individual emotions canplay a role as affective factors and decision-making behavior is influenced by the interaction ofcognitive and emotional processes (Lemerise & Arsenio, 2000) Wirtz and Bateson, amongothers, believed that incorporating emotional elements into the pure cognitive method ofdetermining consumer pleasure would improve the intention with which it could be anticipated(Wirtz & Bateson, 1999) Yuksel and Yuksel have shown that positive emotions canconsiderably increase customer satisfaction (Yiiksel & Yiiksel, 2007) Awe is a pleasant feelingthat may improve consumer satisfaction Therefore, utilizing AI enabled VAs leading consumers

to awe experience and cognition of value Deploying this attestation from the existing literature,the following hypothesis is proposed:

Hypothesis 4: Awe experience has a positive effect on Price value among online shoppers.3.3.3 Awe experience and sales promotion

Research has demonstrated that awe experiences can have a major impact on consumerbehavior, particularly during sales campaigns These experiences can influence evaluative andselection processes in ambiguous decision-making circumstances, thus leading to a beneficialeffect on sales promotion efficacy (Ahmmad et al., 2024) Furthermore, awe, a commonexperience-based emotion, can readily arise among consumers in online brand communities,

altering their involvement and buying decisions, further revealing a potential favorable

association between awe experience and sales promotions (Zhao et al., 2022) Marketers canexploit the awe experience to design effective promotions and induce behavioral intention infashion shopping by using digital voice assistants (Kautish et al., 2023) Thus, based on thisevidence it is proposed:

Hypothesis 5: Awe experience has a positive effect on Sales promotion among onlineshoppers.

3.3.4 Awe Experience and E-WOMInitiating emotions such as awe experience prompts the sharing of information with othersand leads to beneficial eWOM behaviors (Guo et al., 2018) Consumers are more likely to usenovel technology, such as AI-enabled voice assistants when they perceive it positively.Customers who experience awe are more likely to share their enthusiasm for the goods withothers (Guo et al., 2018) Awe experience is a feeling outcome that encourages consumers toshare information with others and results in positive eWOM behaviors (Guo et al., 2018).Positive emotions like awe can increase customers’ propensity to utilize AlI-enabled services,engage with others, and recommend/eWOM (M.-H Huang & Rust, 2018; Perez-Vega et al.,2021) Based on these arguments, the subsequent hypothesis is suggested:

Hypothesis 6: Awe experience has a positive effect on E-WOM among online shoppers

3.3.5 Price value and purchase intentionsHollebeek et al (2014) argued that consumer behaviors are a result of perceived price value.For example, when customers believe they have received a high level of value for their moneyfrom a consumption experience, they are more likely to engage in experience-related activitiessuch as reviews, recommendations, and referrals In mobile service research, it was shown thatthe more that users valued utilizing the mobile service, the more likely they were to sharepositive eWOM (Cheshin et al., 2018) As a result, when customers believe they are obtainingexcellent value from their VAs, they are more inclined to refer them Furthermore, it has beendiscovered that the perceived pricing value of mobile applications exerts a direct and positiveinfluence on users' intentions to use the information they supply (Cheshin et al., 2018) Asperceived price value rises, we anticipate that consumers will be more likely to follow VA advice.Following research by Kusumawati and others (Kusumawati et al., 2014), the sole factorinfluencing consumers’ intentions to purchase music items is price value When a consumerviews the value they have received from a product positively, they believe that the benefitsoutweigh the costs, making it worthwhile to continue using the product In the VA context, whenconsumers think that they gain value from using the technology, they develop a positive attitudetoward it, leading them to continue using it Based on the preceding points, we offer thefollowing hypothesis:

Hypothesis 7: Price value has a positive impact on Purchase Intentions among onlineshoppers.

3.3.6 Sales promotion and purchase intentions

In the public market, promotions are a crucial factor in consumer purchases due to theabundance of products and high product uniformity, leading producers to prioritize promotions(Dehkordi et al 2012) Lee and Olafsson (2009) revealed that Sales Promotion has a significantbeneficial influence on Purchase Intention The strongest perception of promotion in customerpurchase intention was found by Li et al (2011), with notable impacts Sun (2010) examined thenoteworthy impacts of promotions on the intention to purchase Besides, using digital voiceassistants, marketers may create advertisements that effectively encourage consumers to makefashion-related behavioral intentions (Kautish et al., 2023) Thus, it is proposed:

Hypothesis 8: Sales promotion has a positive impact on Purchase Intentions among onlineshoppers

3.3.7 E-WOM and purchase intentionseWOM has long been discussed as a successful online marketing tool for consumers toexchange information (Filieri, 2015) There are lots of online platforms offered by the Internetlike product/service reviews, social media, etc (Kautish et al., 2021; Wang et al., 2018) Beforepurchasing goods or services, other buyers might feel more at ease thanks to the information left

by past clients Moreover, favorable or negative service evaluations determine how positiveeWOM is channeled and play a crucial role in influencing purchase intentions (Patel et al., 2023).Buyers are more likely to purchase after reading a favorable voice shopping review (Chung et al.,2020; Filieri, 2015; Wang et al., 2018) We contend that when customers' eWOM effect onvoice-based online purchasing is considerable, it is likely to boost their purchase intention based

on the findings in the literature

Hypothesis 9: E-WOM has a positive impact on Purchase Intentions among onlineshoppers

4, METHODOLOGY

4.1 Data collection

Data were collected during the period between February and March 2024 A Google Form

social media platforms (Facebook, Zalo, and Messenger) Vietnam represents an appropriateenvironment for this study because Vietnam is the country with the fastest Internet growth rate inthe region and among the countries with the highest growth rate in the world according toVECOM (VECOM, 2021) Moreover, in 2022, the number of e-commerce users in Vietnamrecorded around 57 million and the market value in e-commerce reached approximately 16.4billion U.S dollars In 2023, this grew 25% in 2023, reaching 20.5 billion U.S dollars (StatistaResearch Development, 2024) Additionally, large businesses are interested in investing ininformation technology infrastructure e-commerce is much higher than that of small and mediumbusinesses, these businesses rate sales on social networks and e-commerce platforms as moreeffective The rate of websites integrating online interaction features with customers has reached78%, and of the two websites that interact online with customers, one uses automatic chatbots,

applying AI in their work (VECOM, 2023) To ensure the eligibility of respondents, a link that

provides a practical trial of AI voice assistant was added to a survey A convenience sample wasadopted and it is a type of nonprobability sampling This sampling approach also known as grab,accidental, or opportunity sampling, involves selecting a sample from the population that iseasily accessible, convenient, or close at hand Stated differently, convenience sampling consists

of the participants who are most easily available to the researcher This technique makes theprocess of collecting data quick, easy, and inexpensive (Isaac, 2023) A total of 300 responseswere collected during the period and no disqualified responses after eliminated in our research.Table | presents the demographic information of the respondents

How often do you use Al-enabled

VAs in online shopping?

et al., 1991), (Jarvenpaa & Todd, 1996), and (Sirdeshmukh et al., 2002) Sales promotion (4items), E-WOM (5 items), and Purchase intentions (6 items) were adapted from (Kotler et al.,2013), (Chopra, 2019), and (Harrigan, 2022) respectively A seven-point Likert scale [Stronglydisagree (1) — Strongly agree (7)] served as the anchor for these items The questionnaire wasinitially developed in English and then translated into Vietnamese using a translation method toensure the clarification of the questionnaire for Vietnamese customers to fulfill the survey.

4.3 PLS-SEM

In this study, data was analyzed using partial least squares structural equation modeling(PLS-SEM) (Hair et al., 2014) A well-established methodology in IS and marketing researchenables researchers to statistically and reliably examine causal correlations between severalindependent and dependent variables simultaneously (Chen et al., 2021; Gu et al., 2019) Thisapproach entails reviewing the measurement model to ensure the reliability and validity of themodel's measurement constructs, as well as assessing the structural model to test the model'shypotheses This data research was conducted with Smartpls software version 4

The values of item loadings, Cronbach's alpha, composite reliability, average varianceextracted (AVE), and the Fornel-Larker criterion were used to evaluate the measurement model(Hair et al., 2017) The ultimate goal was to confirm the reliability of our scale Factor loadings,

Cronbach's alpha, and composite reliability values were expected to be greater than the 0.7thresholds to support item and construct reliability; AVE values had to be greater than the 0.5threshold to support convergent validity; and to validate discriminant validity, HTMT has to bebelow 0.85 and the square root of the AVE for each construct should be greater than thecorrelation between the constructs (Fornell-Larcker criterion).

The path coefficients and significance levels were used to examine the structural model Thesignificance of the pathways was evaluated using p-values obtained by doing a bootstrap analysiswith 5000 subsamples to confirm the stability of the acquired results (Hair et al., 2017) To

determine the predictability of our model, we employed the R-squared (R?) value R? values of

0.75, 0.50, or 0.25 are commonly referred to as substantial, moderate, or weak in informationsystems and marketing studies However, consumer behavior research focusing on customer

satisfaction issues prefer to view R? values of 0.20 as significant (Hair et al., 2017).

4.4 Fuzzy Set Qualitative Comparative Analysis (fsQCA)

Qualitative comparative analysis (QCA) is defined as a set-theoretic analytical approach thatutilizes Boolean algebra and fuzzy set theory fsQCA, a novel form of QCA, employs a morerigorous consistency evaluation based on set theory, allowing the inclusion of continuous andinterval-scale variables in relation to causal circumstances and outcomes Essentially, fsQCAdiffers from traditional regression, with its presuppositions of asymmetrical interactions (i.e.,causes leading to a particular outcome may be different from causes leading to the non-existence

of an identical outcome), equilibrium (i.e., multiple paths or configurations can lead to the sameoutcome), and causal complexities (1.e., distinct combinations of causal precursors can lead tothe same outcome) (Ragin, 2009; Woodside, 2013)

As opposed to traditional regression, fsQCA is based on the assumption that theconsequence of interest is given by one or more combinations of various causal factors, ratherthan a single cause and its net effect (Woodside, 2013) This approach provides a more profoundempirical and theoretical assessment of the most advantageous combinations of variables forparticular results (Hughes et al., 2019) Researchers have urged the adoption of asymmetricconfigural analysis in interpreting complicated incidents, especially when involving humanbehavior that is often unlikely to adhere to a symmetry stance (Schmitt et al., 2017)

For this reason, we applied fsQCA to analyze complex models involving the interplaybetween MCIs and purchase intentions (i.e., functional, hedonic, social and cognitive MCIsversus purchase intentions); the relationship between awe experience, price value and salepromotions and purchase intention in verifying the importance of each factor in customers’online purchases when using AI-based VAs services.

Two major measures are used in fSQCA to assess the parameters of the fit including consistencyand coverage (Ragin, 2009) Consistency is the degree to which causal combinations lead to acomparable consequence, demonstrating the strength of subgroup connections (Ragin, 2006) Incontrast, the coverage index quantifies the percentage of variance explained and is analogous to

the R2 in traditional regression analysis More particularly, raw coverage represents the

proportion that a causal combination encompasses the outcome, while unique coverage partitionsthe raw coverage and indicates the proportion that the result is covered solely by a causalcombination but not concurrently by other combinations (Ragin, 2009)

5 DATA ANALYSIS

5.1 Measurement model analysis

All of the reliability and validity measurements fell within the acceptable threshold.

Cronbach's alpha and composite reliability ratings above the threshold of 0.70 All item loadingsabove the 0.7 criteria, AVE results were likewise higher than the criterion of 0.50 Thus, thereliability and validity of the items and constructs were confirmed

Table 2: Factor loadings of measures

AW CMCI | EWOM | FMCI | HMCI | PI PV SMCI | SPAWI 0.901

Table 3: Assessment of reliability and convergent validity (n=300)

Constructs Items OL CR AVE Cronbach’s

HŠ MCI 0.890

H6 MCI 0.907

Social MCI S1 MCI 0.905 0.939 0.720 0.922

S2 MCI 0.874 S3 MCI 0.760 S4 MCI 0.822 S5 MCI 0.827

S6 MCI 0.895

Cognitive MCI C1 MCI 0.911 0.937 0.750 0.917

C2 MCI 0.831

C3 MCI 0.866 C4 MCI 0.839 C5 MCI 0.880 Awe experience AWI 0.901 0.932 0.697 0.913

Table 4 shows that the square root of each construct's AVE exceeds its maximum correlationwith other constructs, indicating discriminant validity of the latent variables according to theFornell-Larcker criterion and the HTMT pairs of latent variables are all below the threshold of0.85 (Table 5).

ES a ee

cmcr_ jose joss | | | | | | | |

rev [oss fos |osm foam losw Joos lama] |_|

SMCI 0.520 10.500 | 0.449 0.541 | 0.645 | 0.554 | 0.498 | 0.849

sp | 0.561 0.414 | 0.558 0.401 | 0.575 | 0.583 |0.559 | 0.475 | 0.879

Note: FMCI: Functional MCI, HMCI: Hedonic MCI, SMCI: Social MCI, CMCI: Cognitive MCI,AW: Awe experience, PV: Price value, SP: Sales promotion, E-WOM: Electronic Word of Mouth,PI: Purchase Intentions

Table 5: Discriminant validity - Heterotrait-monotrait ratio of correlations (HTMT)

Construct AW CMCI |EWOM | FMCI | HMCI) PI PV | SMCI SP

Table 6: Cross-loadings

AW CMCI EWOM EFMCI HMCI PI PV SMCI SP

AWTI 0.901 0.460 0556 0.503 0.568 0.580 0.573 0.467 0.555

AW2 0.820 0.391 0.448 0.442 0.456 0.449 0.448 0.372 0.463AW3 0.808 0.371 0.476 0.414 0.502 0.502 0477 0.433 0.424

AW4 0.830 0.356 0.489 0.394 0.485 0.499 0.491 0.425 0.464

AWS5 0.765 0.335 0.407 0.410 0.446 0417 0.393 0.437 0.358

AW6 0.879 0.422 0.490 0.426 0.520 0.548 0.541 0.469 0.520CIMCI 0.472 0.911 0.397 0.458 0.539 0.483 0.499 0.488 0.394

C2MCILI 0.345 0.831 0.304 0.364 0.418 0.340 0.381 0373 0.311 C3MCI 0.435 0.866 0.332 0.423 0522 0.491 0.452 0.445 0.367

F3 MCI 0415 0.387 0.434 0.795 0.476 0.425 0.342 0.447 0.330

F4 MCI 0.419 0.382 0.440 0.817 0496 0.432 0.340 0.438 0.290FSMCI 0.413 0.346 0.470 0.836 0494 0443 0.321 0.433 0.295F6 MCI 0.426 0.409 0.451 0.868 0.504 0.456 0.370 0.438 0.341F7 MCI 0.424 0.424 0.433 0.886 0.570 0.458 0.360 0.471 0.364HIMCI 0.610 0.547 0.542 0.599 0.910 0.588 0.551 0.620 0.548H2MCI 0.564 0.501 0.524 0.570 0.915 0.588 0.516 0.611 0.533H3MCI 0470 0.491 04/5 0.488 0.856 0.499 0.406 0.527 0.464

H4MCI 0472 0.449 0.437 0.514 0.850 0.501 0.443 0523 0.459H5MCI 0.516 0.506 0.467 0.504 0.890 0.520 0.443 0553 0.524

H6MCI 0532 0.510 0.475 0.539 0.907 0.567 0.489 0.591 0.526

PH 0.563 0.459 0.570 0.540 0620 0.921 0.593 0.535 0.5/3PI2 0525 0.431 0.504 0.446 0482 0.860 0554 0.504 0.509PI3 0.496 0.416 0.452 0.406 0.487 0.828 0.494 0.451 0.440PI4 0.547 0.444 0.547 = 0.524 0.571 0.885 0.571 0.505 0.509PIS 0.511 0.420 0.518 0.457 0.540 0.869 0.568 0.449 0.535PI6 0.531 0.439 0.567 0.491 0.528 0.906 0.548 0.473 0.500PVI 0.586 0.498 0.543 0.404 0.501 0598 0.928 0.467 0.544PV2 0.528 0.440 0.526 0.364 0.470 0.551 0.922 0443 0.487

PV3 0.505 0.483 0.520 0.411 0.518 0.596 0.920 0.465 0.523 PV4 0.553 0.462 0.542 0.380 0.499 0593 0.928 0.466 0.510

SIMCI 0.521 0.486 0.453 0.547 0.620 0.553 0.483 0.905 0.487S2MCI 0.452 0.448 0.346 0.448 0559 0.484 0.439 0.874 0.433

S3MCI 0.383 0.378 0.329 0.414 0452 04399 0.372 0.760 0.315

S4MCI 0.401 0.377 0.358 0.420 0.526 0441 0.422 0.822 0.371SSMCI 0398 0.357 0.373 0.404 0512 0.420 0.371 0.827 0.328

S6MCI 0.471 0.479 0.413 0.500 0.595 0.501 0.436 0.895 0.454

SP1 0.559 0.363 0.518 0.376 0532 0.531 0.567 0.458 0.928SP2 0.535 0.367 0.524 0.368 0517 0.541 0.519 0.441 0.910SP3 0.502 0.385 0.504 0.348 0523 0.549 0.458 0.446 0.913SP4 0.450 0.405 0.496 0.375 0533 0.512 0.499 0.389 0.909

5.2 Structural model analysis

5.2.1 Evaluate collinearity issues

Hair et al (2017) suggests assessing VIF, the value of all combinations between endogenousand exogenous constructs, to identify collinearity issues in the structural model In other words,this represents the values of all predictor constructs in the model The researchers found that VIF