Acceptance of the metaverse: a laboratory experiment on augmented and virtual reality shopping

Purpose – In recent years, there has been significant interest in adopting XR (extended reality) technologies such as VR (virtual reality) and AR (augmented reality), particularly in retail. However, extending activities through reality-mediation is still mostly believed to offer an inferior experience due to their shortcomings in usability, wearability, graphical fidelity, etc. This study aims to address the research gap by experimentally examining the acceptance of metaverse shopping. Design/methodology/approach – Thisstudyconductsa2(VR:withvs.without) 3 2(AR:withvs.without) between-subjectslaboratoryexperimentinvolving157participantsinsimulateddailyshoppingenvironments. This study builds a physical brick-and-mortar store at the campus and stocked it with approximately 600


Introduction
With the aid of the Internet, consumers can more efficiently browse, search and acquire information about products and services (Grewal et al., 2004).At the same time, shopping became a more cognitively-oriented and information-heavy task, losing some of the multidimensional experientiality of Brick-and-Mortar (B-a-M) shopping (Grewal et al., 2004).However, with the recent progress of XR (extended reality) technologies such as VR (virtual reality) and AR (augmented reality), it is now believed that the experientiality of shopping might be making a comeback as XR may be able to combine the advantages and benefits of digital and physical shopping (Bonetti et al., 2018;Xi and Hamari, 2021), so offering a more optimal mix of experientiality and utility.Such a futuristic-sounding shopping experience can be represented by the concept of "Metaverse Shopping." It has been felt that these recent technological developments regarding the metaverse may open the door to a new shopping experience (Pizzi et al., 2019;Greenwood et al., 2016).For example, large international retail giants such as IKEA (VR kitchen), Amazon (AR/VR applications) and Walmart (AR scanner) have started to move toward technology-based innovative business, and the global market for VR and AR in retail has been expected to reach USD 1.6 billion by 2025 (Joshi, 2019).However, these "hype" technologies for creating the metaverse have also raised concerns among retailers and consumers that the additional XR modalities might reduce the shopping experience in terms of comfort, information overload, usability, richness, responsiveness and even introduce a change in behavioral norms.Currently, consumers' attitude and willingness to adopt metaverse shopping are still ambiguous, and many retailers have gradually lost confidence and interest in investing in new technologies (Boardman et al., 2020).In academia, there is a growing interest in exploring user experiences within the metaverse, as evidenced by recent literature reviews and conceptual studies (see Barrera and Shah, 2023;Dwivedi et al., 2023;Hadi et al., 2023).This surge in interest is particularly notable in the realm of consumer experiences within the metaverse, specifically within the context of activities such as shopping (see Table 1 in Section 2).However, it is noteworthy that in general, there is only a limited amount of comprehensive empirical studies that focus on the factors and perceptions of the metaverse (Damar, 2021;Shen et al., 2021), which has hindered the development of a deep understanding of the fundamental value and function of the metaverse.First, regarding the concept of metaverse, there is no consensus of prior studies on what the metaverse is and what it actually consists of.Therefore, the representative XR technologies, including AR and VR, have rarely been investigated together and compared.Second, it should be noticed that the majority of extant literature has examined the adoption of existing and available XR applications in the market, based on surveying the users in general.In the fields of Information Systems (IS) and Information Technology (IT), legacy technology acceptance

Acceptance of the metaverse
research has started to face resistance as it often provides only iterative additions to the literature and especially disregards methodological ambition.Consequently, it is especially important to investigate its adoption through experimental settings where users have the opportunity to acquire a first-hand experience of the technology.Furthermore, synthesizing findings from previous studies is quite challenging due to the fact that prior research has either focused on the effects of single-or multiple-specific attributes of AR/VR on adoption or has examined different shopping contexts involving various interactions.Therefore, it is still unknown whether consumers are generally willing to adopt XR in general for shopping (rather than any specific features) and what factors would influence the adoption.
To address the research gap, we experimentally examine the acceptance of metaverse shopping using the technology acceptance model, aiming at answering the research questions of whether and how AR and VR technology differently affect consumers' perceived usefulness and perceived enjoyment, perceived ease of use, attitude, as well as their future intention to reexperience XR shopping.We conducted a 2 (VR: with vs. without) 3 2 (AR: with vs. without) between-subjects laboratory experiment with 157 participants in real-life daily shopping environments, combined with a psychometric questionnaire.The cutting-edge research question, important findings and experiment-based empirical approach offer considerable empirical, theoretical and methodological research contributions.In addition, they provide valuable practical implications for practitioners who seek to utilize virtual technologies for retail transformation, as well as for developers and designers of metaverse platforms.
This study is structured as follows.Section 2 describes the theoretical background and hypothesis development.The experimental method and implementation are detailed in Section 3, including laboratory experiment design, participants, materials, measures and procedure.Section 4 analyzes the structural equation model and Section 5 provides a detailed discussion of the results, theoretical contributions and practical implications of the work.Section 6 presents the conclusions, and the Section 7 points out future research directions based on the limitations that are acknowledged.

Background and hypothesis
2.1 Theoretical background 2.1.1Metaverse: AR and VR.Metaverse broadly represents artificially generated, modified, diverse and extended realities that are built on the convergence of virtual technologies and multimodalities (Dwivedi et al., 2022;Pamucar et al., 2022;Rauschnabel et al., 2022).The term "Metaverse" was first coined by Neal Stephenson in his 1992 science-fiction novel Snow Crash and refers to the dystopian virtual world.Nowadays, this concept is no longer limited to the imagination; it is becoming a futuristic reality that provides full-scale, persistent and interactive experiences (Kozinets, 2022).Even though there is a lack of an accurate definition as to what Metaverse is, it has been considered as interchangeable with the concept of XR (extended reality), which forms an umbrella term for AR (augmented reality), VR (virtual reality) and MR/AV (mixed reality/augmented virtuality).XR or Metaverse is a generic expression that covers all digitalized, replaced, duplicated, alternative, augmented and new realities (Ç€ oltekin et al., 2020;Rauschnabel et al., 2022), and these reshaped realities could be fully or partially virtual (Hwang and Chien, 2022).Figure 1 describes different realities in the metaverse.
VR is one of the core technologies employed in the metaverse.The understanding of VR has deepened over the past 30 years, from being loosely used to refer to things such as the Internet as a whole (Climo, 2001), digital worlds (Lee and Chung, 2008) and virtual games (Zyda, 2005), to describing computer-based 3D virtual environments (Brooks, 1999).This also applies in relation to VR shopping literature where VR has described the web 3D technology used for producing an online store (e.g.web-based applications: see Wang and Datta, 2010; Second Life retail store: see Vrechopoulos et al., 2009).However, these descriptions usually refer to the provided digital content, while today VR refers to human-computer interfaces and most commonly to stereoscopic head-mounted displays (HMD).It should be acknowledged that the creation of a VR environment usually relies on what is known as presence-inducing or immersive computing technology (Berg and Vance, 2016) that more or less fully covers the field of view of the user through the use of large screens or HMD (Zikic, 2007).Thus, VR is touted to block out the real world (Manis and Choi, 2019) and to digitally duplicate or substitute the "real reality" (Dincelli and Yayla, 2022;Xi and Hamari, 2021;Xi et al., 2023;Yim et al., 2017).VR creates a highly "realistic" digital world for users and a sensation of "being there" (Kipper, 2013;Hardiess et al., 2015).Research and discussions in the field have traditionally focused on HMD-based virtual reality, which provides a higher level of immersive experience (Dincelli and Yayla, 2022).
For a long time, AR has been misleadingly considered as a specific subform of VR (Wedel et al., 2020) and limitedly anchored to the use of computer-generated images superimposed on a view of a "real-world" physical environment (Javornik, 2016b;Olsson et al., 2013;Pantano et al., 2017;Voogt and Fisser, 2015).However, AR provides access to additional "augmenting" information (Pantano and Servidio, 2012;Rese et al., 2014) that can take a variety of forms and provide sensory stimulation using graphics, text, video, sound or other virtual elements (Van Krevelen and Poelman, 2010) such as scent (Lu and Smith, 2007) and tactile stimulation (Azuma, 1997).AR is also agnostic as to the "reality" it is augmenting (Rovaglio et al., 2012).Virtual reality can also be augmented by superimposing other content onto it.This has separately been referred to as augmented virtuality (AV) (Albert et al., 2014;Riar et al., 2023) and in the present study corresponds to one of the treatments of the 2 3 2 experiment (where AR elements are superimposed onto the view of virtual reality in Group 4).In retail, with the help of recognition and tracking techniques (e.g.location-based and image-based AR), relevantly digitalized content can be triggered and displayed on screens in different forms according to specific elements (e.g.images, objects, barcodes, QR codes and location) in the immediate surroundings (Aggarwal and Singhal, 2019).
2.1.2Acceptance of metaverse shopping.The Technology Acceptance Model (TAM) (Davis, 1989;Van der Heijden, 2004) has been the workhorse theory used to guide research models of the past 30 years that investigate why people adopt and continue using different technologies, ranging from the machines of yesteryear (e.g.mainframes : Panko, 1988) to contemporary technologies such as gamification (Hamari and Koivisto, 2015), VR (Disztinger et al., 2017) and robotics (Conti et al., 2015).According to the seminal vision of TAM, the attitude and behavioral intention of using a specific technology are determined by two factors: perceived usefulness and perceived ease of use (Davis et al., 1989).However, when IS were more pervasively adopted into leisure use and consequently became a source of intrinsic need satisfaction (Davis et al., 1992;Deci and Ryan, 2000), TAM was expanded to incorporate the hedonic factor of enjoyment ( Van der Heijden, 2004).Later iterations conceptualizing the Acceptance of the metaverse nature of IS in relation to hedonic and instrumental aspects have developed the realization that most systems serve a so-called dual purpose (Wu and Lu, 2013), where the enjoyment of the system can further lead to utility, due to the higher motivational benefits involved (see, e.g.Koivisto and Hamari, 2019;K€ ose et al., 2019).Thus, to better understand the intrinsic motivation of technology usage, perceived enjoyment (the extent to which the activity of using the technology is perceived to be enjoyable in its own right: Davis et al., 1992) was extended into the TAM model, which aims at providing self-fulfillment to the technology users ( Van der Heijden, 2004).More recently, TAM has been the predominant theory used to examine the acceptance and continued use of innovative technology (Wu et al., 2011;Ayeh et al., 2013;Bailey et al., 2017).This theory has been applied and empirically supported in retailing to predict the adoption of various technologies such as self-service retail technology (Kaushik and Rahman, 2015), online shopping (Wei et al., 2018), mobile shopping (Fuentes and Svingstedt, 2017) and quick response (QR) codes (Kim and Woo, 2016).
Metaverse has been believed to blur the boundary between the physical and virtual worlds (Soliman et al., 2017;Suh and Prophet, 2018) as well as the boundary between utility and hedonism, thus making metaverse technologies akin to dual-purposed ISs (Hamari and Koivisto, 2015;Lee and Kim, 2022;K€ ose et al., 2019;Koivisto and Hamari, 2019;Kool and Agrawal, 2016;Wu and Lu, 2013).Since user acceptance is an important facet of a successfully implemented new technology (Davis, 1993;Dillon and Morris, 1996), a considerable amount of research has focused on the acceptance of the metaverse, and the TAM model is the most commonly used theoretical framework (see, e.g.Aburbeian et al., 2022;Almarzouqi et al., 2022;Akour et al., 2022).Ahead of research covering various fields such as healthcare (Almarzouqi et al., 2022), sports and recreation (Akour et al., 2022;Huang et al., 2023) and education (Papakostas et al., 2023), prior literature has also empirically investigated the acceptance of metaverse (especially VR and AR) in different shopping contexts based on the TAM model (see Table 1) [1].In terms of research theme and methodology, most studies have examined the acceptance-related aspects of a single AR or VR shopping application based on a user evaluation-based questionnaire (see, e.g.Altarteer and Charissis, 2019;Pantano and Servidio, 2012;Pantano et al., 2017), and the effects of VR and AR on shopping experiences have rarely been investigated and compared together.As for the technologies that have featured in prior studies, most AR retail-related studies have adopted an existing AR application such as the IKEA furniture app (Javornik, 2016a;Rese et al., 2014Rese et al., , 2017) ) and virtual try-on technology (e.g.sunglasses: see Pantano et al., 2017; clothing: see Huang and Liao, 2015).Accordingly, the selection of AR devices has also been limited to handheld devices (e.g.tablets: Spreer and Kallweit, 2014;mobile phones: Rese et al., 2014; PC with a web camera: Pantano et al., 2017).Due to the limitations of AR devices and programs, the results and conclusions of existing AR studies are basically only applicable to online shopping based on computer or mobile screens and are hence hard to disseminate to a wider range of products.Furthermore, there is also insufficient research on the application of AR (especially wearable AR technology) in other shopping settings such as B-a-M stores.
In contrast, most VR-related studies have been conducted based on self-developed programs for the corresponding devices, such as a HMD (e.g.Peukert et al., 2019) and projector (e.g.Altarteer and Charissis, 2019).However, for the experiment environment used in studies, most VR programs or prototypes have simply created places for shopping activities to take place (e.g.VR shelf: see Sikstr€ om et al., 2016; product display: see Altarteer and Charissis, 2019).These experiment settings have limitations for constructing a realistic store that supports the consumer's shopping journey from product search to purchase decision-making.Thus, it is also hard to generalize the empirical results of studies regarding the adoption and acceptance of VR retail.More importantly, in either VR or AR retail studies, XR technologies have mostly been investigated as the context of the research rather than as part of the research model in the form of variables (see, e.g.Rese et al., 2017;Huang and Liao, INTR 34,7 2015).Thus, while these studies relate to XR technology, the research models employed do not allow conclusions to be drawn about how XR has affected the shopping experience (e.g. compared with a real-world scenario).In addition, it should also be noted that most VR-related studies have not systematically discussed the acceptance of VR in shopping (concentrating mainly on individual variables), nor have they been based on relevant theory (e.g.Ohta et al., 2015) or had a clear research framework (e.g.Verhulst et al., 2016).Therefore, the empirical evidence lacks knowledge on the acceptance of retail AR and VR technologies in a realistic setting, built on a suitable theoretical basis.

Hypothesis development 2.2.1 Acceptance of AR retail.
According to the extant literature related to AR, vividness, novelty and interactivity have been considered as the three main attributes of AR (Azuma, 1997;McLean and Wilson, 2019).The vividness of AR is seen as the ability to produce a sensory-rich mediated environment (Steuer, 1992) and is usually related to the aesthetic aspects and display quality of information (Flavi an et al., 2019;McLean and Wilson, 2019).The novelty of AR refers to the unique and personalized content experienced through the AR display (McLean and Wilson, 2019).The interactivity of AR technology can allow sensory information to be digitally overlaid in either the physical or virtual environment according to user preference (Rovaglio et al., 2012).Overall, it can be seen that the digitized display of information in more vivid, novel and interactive ways can lead to high perceived enjoyment.Thus, we propose the following hypothesis: H1(a).AR has a significant positive effect on perceived enjoyment.
However, the interactivity of AR usually requires mental and physical user effort (Heeter, 1989;Nam, 2015) and therefore has the potential to reduce the perceived ease of use.Compared with non-AR mediated shopping (e.g.traditional physical shopping), consumers need to make more physical and mental effort when using devices, operating AR systems and processing information from multiple channels.For example, there are limitations in information processing for current AR technology in terms of response time speed and physical resource consumption.More importantly, using AR in shopping activities requires consumers to mentally process and integrate various forms of information from multiple channels simultaneously (split-attention effect, see, e.g.Sweller, 2005), which might lead to high cognitive load and mental effort.As these aspects can potentially decrease the subjective evaluation of the perceived ease of use of AR for shopping, we propose the following hypothesis: H1(b).AR has a significant negative effect on perceived ease of use.
It should be noted that in the shopping process, the main role of AR is to digitally display controllable information in the environment (McLean and Wilson, 2019;Yim et al., 2017).Thus, AR alone may not influence shopping performance and efficiency in terms of, e.g.information-seeking and decision-making.In other words, the use of AR may not influence consumers' evaluation of how useful the information is.The argument offered in most previous studies that AR can increase shopping value is mainly due to the "additional" content/information provided rather than AR technology as a human-computer interface itself (see, e.g.Rese et al., 2017;Huang and Liu, 2014;Yim et al., 2017).Accordingly, there should be an assumption that no differences in the shopping experience in terms of perceived usefulness would be seen between AR and non-AR environments when the total amount and content of information are controlled to be the same and when the AR technology itself does not significantly hamper the experience.In this study, we propose the following null hypothesis and its corresponding alternative hypothesis: H1(c) 0 .AR has no significant effect on perceived usefulness.H1(c) a .AR has a significant effect on perceived usefulness.
A null hypothesis is often used in scientific research to test whether there is a significant relationship between two or more variables and is acceptable in the IS field (see, e.g.Johnston et al., 2015;Dong and Wu, 2015;Jiang and Benbasat, 2007).Here, the null hypothesis is put forward to examine whether there is no effect of AR on perceived usefulness.In this study, we do not debate null hypothesis significance testing (NHST) within the hypothetico-deductive tradition (see Mertens and Recker, 2020), but we believe that research contributions should not only depend on significant results.Thus, even a supported null hypothesis can still bring meaningful insights, especially to such a new and promising research area.
2.2.2 Acceptance of VR retail.The core feature of VR is to situate the user in an immersive computer-mediated environment, i.e. by creating the environment in a three-dimensional model and allowing the user to experience it through display technology that appears to insert the user into that environment (Siegrist et al., 2019).With the aid of interactive modalities (Schnack et al., 2019), VR can create similar perceptions and experiences in the virtual world as in the "real" world and thus achieve the goal of diminishing the relevance of the difference between the real and the virtual (Lee et al., 2013).However, in some studies, VR might have the characteristics of the so-called "magic circle" of the game world (Castronova, 2008; also see "non-consequentiality": Xi and Hamari, 2021), which has provided opportunities for a variety of activities that are typically either not done well or safely in real life without concerns for the consequences involved (Schultheis and Rizzo, 2001;Reid, 2004).From this point, VR may give rise to curiosity and playfulness in the user, as the actions they undertake within VR may not have similar consequences to their real-life equivalents.For example, in a real shopping context, the consumer must be careful not to damage goods on display, whereas, in VR, consumers can be allowed to freely use and interact with virtual products, even in an extreme, playful and risky way (Xi and Hamari, 2021).Thus, we propose the following hypothesis: H2(a).VR has a significant positive effect on perceived enjoyment.However, as representative wearable technology, VR usually requires a higher amount of physical resources, which leads to high demand for comfort, wearability and functionality.Many critics of HMD systems have reported restricted movement, motion sickness, limited field of view and discomfort (McComas et al., 1998;Reid, 2004), together with complex manipulations and difficulty using controllers, all of which indicate that VR requires a high degree of effort and has a high potential to lead to a low ease of use perception toward shopping.We have to admit that such limitations might be overcome with the development of VR technology, but this is still hard to achieve.Therefore, our assumption of perceived ease of use in the current study is based on the potential of using existing consumer-grade VR devices for providing shopping experiences and can be put forward in the following hypothesis: H2(b).VR has a significant negative effect on perceived ease of use.
Similar to AR, we can reasonably assume that there should be no distinct difference in perceived usefulness between VR and non-VR shopping environments when the VR content is purposefully replicated to correspond with current reality.As technology for substituting reality (Xi and Hamari, 2021), VR would not change or affect a shopping decision-making process that involves duplicated information and the environment.Accordingly, there should be an assumption that no differences in the shopping experience in terms of perceived usefulness would be seen between the VR and non-VR environments.Therefore, we propose the following null hypothesis and a corresponding alternative hypothesis: H2(c) 0 .VR has no significant effect on perceived usefulness.H2(c) a .VR has a significant effect on perceived usefulness.
2.2.3 TAM in XR retail.TAM proposes that perceived usefulness and perceived ease of use determine a person's attitude toward using a technology, which in turn determines their intention to use it (Ha and Stoel, 2009).In the study of Van der Heijden (2004), perceived enjoyment was added to the extended TAM model as the additional motivational determinant of acceptance.This has been empirically examined in a number of technology-related studies in business and retail sciences (Ha and Stoel, 2009;Van der Heijden and Verhagen, 2004;Lee et al., 2006;Lee and Chang, 2011).Thus, we hypothesize that in XR shopping, the relationships between different constructs, such as perceived usefulness, perceived ease of use, perceived enjoyment, attitude and intent to use, would be consistent with the theoretical foundation of TAM (Ajzen and Fishbein, 1975;Davis et al., 1989Davis et al., , 1992;;Van der Heijden, 2004)-see H3 to H6. Figure 2 further depicts the research model and the hypotheses featured in this study.
H3. Perceived ease of use has a significant positive effect on (a) perceived enjoyment, (b) perceived usefulness and (c) attitude.
H4. Perceived enjoyment has a significant positive effect on attitude.
H5. Perceived usefulness has a significant positive effect on attitude.
H6. Attitude has a significant positive effect on the intention to visit.

Research method 3.1 Design
This study adopted a 2 (VR: with vs. without) 3 2 (AR: with vs. without) between-subjects experiment design (see Plate 1).The independent variables were therefore composed of the dummy-coded variables of AR (with 5 1, without 5 0) and VR (with 5 1, without 5 0).For the measurement of perceived ease of use, we employed the NASA Task Load Index (NASA-TLX) Metaverse retail

Note(s):
The dotted lines between variables indicate the potential path relationships.Since there might be some unexpected and effects between AR / VR, attitude and intention to visit, we also tested the direct effects of AR/VR on attitude and intention to visit in this study Source(s): Author's own creation/work

Research framework in this study
Acceptance of the metaverse (Tang et al., 2003).The remaining variables were composed of the psychometric constructs dictated by the technology acceptance model, including perceived usefulness, perceived enjoyment, attitude and behavioral intention (Davis, 1989;Van der Heijden, 2004;Vijayasarathy, 2004).

Participants
The data for this study was collected from September to November 2019.A total of 165 participants with diverse nationalities were recruited from students at a Finnish university to participate in a shopping-related experiment on the university premises (see Section 3.5).Three disqualified participants were identified, and five outliers were omitted during the datacleaning stage, leaving a sample of 157 for analysis.Among the 157 participants, 54.1% identified as male, and the ages of 122 participants ranged from 20 to 29 years old.Eighty-nine (56.7%) participants were undergraduate students, and 89 (56.7%) had a monthly income of less than V499.The recruited participants came from different countries, including Finland (33.1%), Germany (8.3%), Vietnam (8.3%),China (7.6%), Spain (5.7%), Russia (4.5%), India (3.2%), France (3.2%), Italy (2.5%) and elsewhere.The F-test results in Appendix 1 indicate that there was no significant difference between samples across the four groups (all p-values >0.05).
We considered the following ways to minimize non-response bias during data collection.First, we offered incentives to motivate participants.Second, we ensured the respondents' anonymity.Third, we sent reminders and assisted with rescheduling.Finally, we encouraged and ensured that all participants completed all tasks in the experiment study.In addition, the results of T-tests for comparisons between 50 early respondents and 50 late respondents showed that there were no significant differences in demographic and technology use characteristics, which suggests that there was no non-response bias.
As part of the experimental procedure, each participant was given a gift card (virtual value V10) to buy products in the second-hand record shop during the experiment.The participants were not otherwise compensated; however, each participant got to keep the records they selected in the shop [2].The purpose of this type of compensation scheme was to garner realistic decisions that are made in the shop environment, as the compensation depends upon the rational choices made by the participant.Upon arrival at the laboratory, each participant was randomly assigned to one of four groups (Group 1, n 5 40; Group 2, n 5 41; Group 3, n 5 40; Group 4, n 5 36).The study design and procedure adhered to the ethical guidelines set by the Finnish National Board on Research Integrity TENK guidelines (2019).
3.3 Materials 3.3.1 Shops.A physical LP record shop (4.24 meters 3 5.09 meters 5 21.58 square meters) was built on the university campus.The shop functioned as the experimental setup for each condition.In the control group (shop using neither AR nor VR), the shop functioned as a common B-a-M shop.In Group 2 (shop using only AR), the shop functioned similarly, but with the exception of product information being displayed through an AR headset display (Microsoft HoloLens: see Section 3.3.4)rather than through printed product information.In Group 3 (shop using only VR), the same room was used.However, the "control" condition was fully replicated in virtual reality (e.g.textures, lighting conditions and geometry) by using laser scanning and Unity 3D for modeling (LiDAR [3] technology (direct time-of-flight) provided by FARO (brand name) Laser Scanner Focus 3D: see Section 3.3.4).The shop in Group 4, utilizing both VR and AR, combined the conditions from Group 3 with the superimposed product information as employed in the conditions of Group 2. In Group 4, the head position of the participant was tracked to predict which record the participant was looking at (similar to the shop used for Group 2), but without using image recognition.The shop floor plan, size, decoration and layout were identical in each condition (see Plate 1).Each wall was equipped with three layers of three-row shelves to achieve a uniform distribution of the products across the room (in total 27 shelves and two products were displayed on each shelf).The video source for each shopping condition is provided as an open-access supplementary file at https://cutt.ly/XR-shopping.
3.3.2Products.In order to improve the internal and external validity of the research, productrelated factors were considered such as providing the same interactivity in all conditions and facilitating information searching and processing for the shopping experience was easy to be modeled in the computer-based environment for VR-mediated conditions.No gender or cultural bias was present in the products, and a large number of real products (approximately 600) were used to simulate a realistic shopping process.Eventually, second-hand English LP Records [4] (31.4 cm 3 31.4cm) were selected as the experiment materials for this study since participant familiarity with the specific artists/bands on display among the records in the shop could be well controlled.As with the general shopping process in daily life, participants searched for and processed relatively simple product information, interacted with products, made a purchase decision with the given V10 gift card and took their chosen products back home.
3.3.3Extra product information.In regular shopping, consumers usually seek extra information to make purchase decisions.In this study, AR was operationalized as the way of information presentation (the information page was either attached to the shelf or displayed as being superimposed on the environment).The extra product information was gathered from the Discogs website [5], including general product information (e.g.label, format, country, release year, genre and style), track list, social statistics, companies and credit information (see Plate 2).
3.3.4Apparatus.Microsoft HoloLens.Microsoft HoloLens version 1 AR glasses were used in the shopping condition with only AR (Group 2) in order to superimpose a computergenerated image with the LP record information on the user's view of the real world.The Microsoft HoloLens is equipped with a 1,280 3 720 display resolution (per eye), producing a total of 2.3 million light points and weighs 579 grams.
FARO Laser Scanner Focus 3D.The Faro laser scanner focus 3D is a precise and powerful high-speed 3D scanner for various applications.It has a scanning rate of one million points per second and a scanning range of up to 130 meters, a resolution of up to 70 megapixels (color), with a field-of-view coverage of 3608(horizontal) by 3008(vertical).
Valve Index.In shopping conditions with VR technology (Group 3 and 4), the participants used the Valve Index headset (tethered with a free-movement enabling 3.5 meters cable and scanned by two base stations) and its controllers.The Valve Index headset is a fullyimmersive headset weighing 809 grams and is equipped with dual 1,440 3 1,600 RGB LCDs, runs at 120 Hz and covers about a 130-degree field-of-view.It is equipped with two controllers (one for each hand) that enable a haptic interface for grabbing records naturally via 87 sensors to track hand and finger positions.
Plate 3 presents the main devices used for constructing and simulating the different shopping environments.

Measurements
In this study, the main constructs included perceived ease of use, perceived usefulness, perceived enjoyment, attitude and intention to visit and were adapted from existing literature.In terms of perceived usefulness, three items, including efficient, useful and productive, were developed in this study, referring to previous literature (Davis, 1989;San Mart ın and Herrero, 2012).In addition, perceived enjoyment was measured with four items taken from Van der Heijden (2004); attitude was measured with two items; and intention to visit was measured with three items adapted from INTR 34,7 Vijayasarathy (2004).A seven-point Likert scale was used to measure perceived usefulness, perceived enjoyment, attitude and intention to visit, ranging from 1 (representing strongly disagree/improbably) to 7 (strongly agree/very probably).From a technology perspective, perceived ease of use can be seen as the extent to which individuals view the use of technology as being effortless, resulting in enhanced productivity, performance, efficiency and control (Davis, 1989;McLean and Wilson, 2019).To measure perceived ease of use, we employed the NASA-TLX (Tang et al., 2003;Hart and Staveland, 1988), where scoring is achieved by rating 6 items of workload from 0 to 100 in increments of 5, including mental demand, physical demand, effort, frustration, performance and temporal demand.The items of the NASA-TLX were reversed and further converted into a seven-point Likert scale.See Table 2 for all measures.The loadings of EUS 2 and EUS 4 were lower than 0.6 (Wong, 2013); therefore, they were omitted.

Acceptance of the metaverse
Following Podsakoff et al. (2003), potential common method bias in the current study was controlled through the design of the study's procedures and statistical controls.Regarding procedural remedies, the measures of investigated variables were collected from different sources to eliminate the effects of, e.g.consistency motifs, social desirability tendencies and any tendencies on the part of the rater to acquiesce or respond in a lenient manner.Existing scales of previous studies were used to reduce measurement errors, and the order of all measurements was randomized via the advanced features of the SurveyGizmo survey platform (now Alchemer).In addition, the scale items were tested and improved through a pilot test with 20 participants before conducting the formal experiment.In this study, we did not collect any identifiable information and anonymized all of the participants' answers.Regarding statistical controls, Harman's one-factor test was used to determine the presence of common method variance bias among the self-report variables.Only 33.67% of the variance in the variables could be explained by one factor.Thus, no general factor was apparent.
We conducted a confirmatory factor analysis to test the reliability, convergent validity and discriminant validity of the measurement model by using SmartPLS 3.0 software.Specifically, we used out-loadings of the items, internal consistency reliability (Cronbach's α), composite reliability of the latent variable (CR) and average variance extracted (AVE) to reflect the scale's reliability (see Table 2 and Table 3).The values of Cronbach's ɑ and CR for all variables exceeded 0.7 (Hair et al., 2006), which indicated that the level of reliability of the scales is satisfied.In addition, all of the value loadings in this study exceeded 0.6 (Chin et al., 1997;Hair et al., 2006) and ranged from 0.629 to 0.927 at a significance level of 0.05 or less.The average variance extracted (AVE) for all constructs was higher than 0.5 (Bagozzi and Yi, 1988).Therefore, the measurement instrument used in this study had a satisfactory convergent validity.
Discriminant validity was assessed using two approaches.First, discriminant validity was assessed using the method proposed by Fornell and Larcker (1981) by comparing the square root of each AVE with the correlation coefficients for each construct.According to Table 3, the square roots of the AVEs are larger than those of the inter-construct correlation coefficients.Overall, the discriminant validity for this measurement model can be accepted, and supports the distinctions found between constructs.Second, the heterotrait-monotrait (HTMT) ratio of correlation was employed and the results indicated no discriminant validity problems, given that all HTMT values were significantly less than 1 (Henseler et al., 2016) 3.5 Procedure 3.5.1 Recruitment.To keep participant reactions as realistic as possible, we did not disclose the key research questions in the recruitment advertisements, but participants were informed of the entire research purpose after each experiment had been concluded.We described the research purpose as "a study of shopping experience" without disclosing any of the experiment details.We used the same content in both online (using the intranet system used by the university community) and offline advertisements (tri-fold flyers) for recruiting university student volunteers.A total of 265 students entered our recruitment system, and a final total of 165 students finally participated in the experiment [6].
3.5.2Pre-survey.When the participant arrived at the lab during a designated time slot, the researchers introduced the entire experiment process.They instructed participants to read the consent form of the study and to fill out the pre-survey.
3.5.3Tutorial.Once the pre-survey was completed, researchers guided the participant to the shop room and introduced the experiment procedure step by step, according to the instruction page (see Appendix 2).All participants were randomized to join one of the four groups.For the three XR-mediated shopping conditions, the researchers introduced the devices and guided participants on how to wear and use the headset and controllers.For both Groups 3 and 4, two tutorial shopping programs were developed without revealing any details of the shops.
3.5.4Experiment.Participants were asked to spend 10 min in the shop and make their purchase decisions independently and in accordance with their own preferences.When the shopping time ended, researchers asked participants to pay with the gift card they had received.Each participant was told that if he/she needed any help or felt uncomfortable during shopping, a short break during the experiment could be offered, and they would be welcome to request additional breaks.When the shopping was completed, participants were asked to fill out a post-survey.
3.5.5Pilot study.Before the actual experiment, a pilot study (N 5 20) was conducted for all conditions to test the measurement items contained in the pre-survey and post-survey, the experimental procedure, instructions, apparatus and methods.Especially in the shopping conditions with VR, instructions for the use and guidance of the HMD headsets and corresponding programs were key aspects of the pilot studies.

Results
The analysis of structural equation modeling was conducted using Smart-PLS 3.0 software, with parameters set as follows: bootstrapping (2000 iterations), a two-tailed test and a significance level of 0.05.With regard to the hypotheses related to the effect of AR technology: AR led to poorer usability (perceived ease of use) but did not have an effect on perceived usefulness or perceived enjoyment (see Table 4 for the full results).The results show that AR had a significant negative association with perceived ease of use (β 5 À0.229, p 5 0.004).Thus, H1(b) was supported.In addition, there was no significant relationship between AR and either perceived enjoyment (β 5 0.107, p 5 0.165) or perceived usefulness (β 5 0.108, p 5 0.208), confirming our hypothesis that AR itself would not lead to better or inferior perceived usefulness and rejecting our expectation that AR would lead to higher perceived enjoyment.Thus, H1(a) was rejected; the null hypothesis H1(c) 0 was supported; and the alternative hypothesis H1(c) a was rejected.Accordingly, when compared with non-AR shopping, AR technology offers neither inferior nor better hedonic value in terms of enjoyable and pleasant experiences incurred during shopping, nor does it change the utilitarian value of information in items of perceived usefulness.The results also empirically confirm the proposed hypothesis that AR technology decreases the perceived ease of shopping.However, this reduction did not lead to a reduction in perceived enjoyment or perceived usefulness.
VR had no significant association with the two constructs of technology acceptance: perceived ease of use (β 5 À0.011, p 5 0.894) and perceived usefulness (β 5 À0.062, p 5 0.465) toward shopping, while it had a positive relationship with perceived enjoyment (β 5 0.173, p 5 0.045).Accordingly, H2(a) and the null hypothesis H2(c) 0 were both supported, while H2(b) and the alternative hypothesis H2(c) a were rejected.Therefore, contrary to our hypothesis and previous studies (regarding, e.g.discomfort: Mart ınez-Navarro et al., 2019; cybersickness: Liu and Uang, 2011), the VR implementation did not negatively affect the ease of the shopping task.
Unlike what is commonly found in technology acceptance research, there was no significant effect between ease of shopping and perceived enjoyment and/or perceived usefulness.Perceived ease of use was not significantly associated with either perceived usefulness (β 5 0.117, p 5 0.326) or perceived enjoyment (β 5 0.051, p 5 0.702), which deviates from the bulk of technology acceptance literature (discussed later in the manuscript).Thus, H3(a) and H3(b) were rejected.Since perceived ease of use was negatively related to attitude (β 5 À0.250, p 5 0.001), hypothesis H3(c) was rejected.This interesting observation may be a proxy indication that difficulty of use may co-occur with, for example, curiosity, interest to try again and/or general novelty that could be a powerful predictor of a (re)visit intention.However, this was unaccounted for by the TAM model (i.e. in the construct of perceived enjoyment).Consistent with a long vein of TAM research, perceived enjoyment and perceived usefulness were positively associated with attitude (β 5 0.364, p < 0.001 and β 5 0.207, p 5 0.004, respectively), and attitude was further positively associated with intention to visit (β 5 0.814, p < 0.001), so hypotheses H4, H5 and H6 were all supported.The direct effects between AR/VR and attitude and between AR/VR and intention to visit were all insignificant (all p-values >0.05), indicating that overall, AR and VR did not directly decrease (or increase) the attitude and future (re)visit likelihood of customers.INTR 34,7

Main findings
In this study, we conducted a 2 (VR: with vs. without) 3 2 (AR: with vs. without) betweensubjects experiment on the effects of VR and AR on the technology acceptance of XR (shopping) and its effect on customers' attitude and visit intention.In summary, we found that AR had a significant negative effect on perceived ease of use, with no significant effects on either perceived enjoyment or perceived usefulness (Finding 1).Meanwhile, VR had a significant positive effect on perceived enjoyment, with no significant effects on either perceived ease of use or perceived usefulness (Finding 2).We also found that in the metaverse shopping context, perceived enjoyment and perceived usefulness significantly and positively influenced consumers' attitude and consequently, their intention to (re)visit metaverse shops (Finding 3).However, previous findings regarding the positive effects of perceived ease of use on perceived enjoyment, perceived usefulness and attitude were not confirmed in this study.Surprisingly, worse perceived ease of use led to a positive attitude toward shopping in the metaverse (Finding 4).Those results altogether indicate that although AR would have a significant negative effect on perceived ease of use, such an effect may not negatively influence consumers' adoption of AR shopping.Below, we present a comprehensive discussion of the main findings combined with our research contributions regarding empirical, theoretical and methodological aspects, along with practical implications for retailers and designers of metaverse shopping.

Research contributions
This study enriches the metaverse and retail literature by providing important and valuable evidence based on rigorous experiment design on consumers' acceptance and adoption of metaverse shopping.As presented in Table 1, investigations on the acceptance of metaverse in previous literature have mainly been based on the evaluation of specific AR or VR applications (e.g.Javornik, 2016a;McLean and Wilson, 2019;Oyman et al., 2022), and there are concerns about the robustness of findings in the extant literature and their generalizability.More importantly, due to an inconsistent understanding of metaverse-related concepts, AR and VR have not been discussed and are rarely compared in one study.To the best of our knowledge, this study is the first to conceptually discuss and empirically examine different virtual technologies, including AR and VR, which significantly extend the current literature.At the outset of this study, we assumed that wearing HMD and using ways of interacting outside the real world would introduce limitations to the ease of the studied task (shopping).
The results indicate that while AR did have a significantly negative effect on ease of shopping, a lower level of ease of shopping was also positively associated with attitude.Such results indicate that the overall shopping experience might not suffer tremendously beyond the issues caused by usability, as long as the XR implementation is suitable and of high quality-i.e.comparable to the physical experience.These research results provide empirical evidence and inspiration for researchers involved in the interdisciplinary study of IS, humancomputer interaction and marketing science to explore the role and value of XR technology in retail management.
Furthermore, this study makes a considerable theoretical contribution to technology acceptance by examining the TAM framework in the metaverse context.Compared with previous yet segmented discussions on the advantages and drawbacks of XR shopping which influence its adoption (Hawkins, 2022;Kliestik et al., 2022), this study has extended the TAM framework (see, e.g.Aburbeian et al., 2022;Almarzouqi et al., 2022;Akour et al., 2022) to explore the utilitarian and hedonic value of XR shopping systems.Considering the technology acceptance of XR adds possible interesting findings to the theory of technology adoption as a whole and challenges the established understanding that IS are regarded as more attractive when they are easier to use.On the one hand, the level of perceived ease of use regarding XR technology would not seem to significantly influence the sense of enjoyment and usefulness.Such results differ from the findings of previous literature that perceived ease of use has a positive association with perceived enjoyment (e.g.Kim and Forsythe, 2008;Pantano and Servidio, 2012;Rese et al., 2017) and perceived usefulness (Huang and Liao, 2015;Kim and Forsythe, 2008) and indicates the optimistic aspect of XR that a difficulty in using technology may not always be detrimental.On the other hand, a low level of usability (i.e.low perceived ease of use) leads to a positive attitude toward using XR.This interesting observation may be examined in future studies by exploring the psychological mechanisms of unique and positive features of the systems involved.The difficulty of use may co-occur with issues such as coolness (Kim et al., 2015), newness and awkwardness (Bunz et al., 2021), challenge (Altarteer et al., 2013) and consumer curiosity toward XR technology (Manis and Choi, 2019).Therefore, an interest to try again and/or the perception of general novelty could be a powerful predictor of attitude and behavioral intention.However, the TAM model was unable to account for this.
In terms of methodological contribution, this study provides an important reference point for future studies on undertaking a randomized controlled study on the effects of technology on adoption.While this study tested the difference of experience stemming from VR or AR technologies applied to a shopping context, the design can be replicated in other contexts where the goal is to isolate the effect of the core essence of the technology.As mentioned earlier, the majority of previous studies have failed to isolate and compare the effects of AR and VR.Due to a lack of a conceptually accurate understanding of AR and VR, it could have been challenging for previous studies to carry out comparative research in identical settings.Moreover, particularly in the design of experiments, researchers have often overlooked confounding factors (such as adding information for consumers or attempting to make the scenario completely different from the B-a-M).This gap in the extant literature has made it close to impossible to realistically infer just what the technology itself implies to considerations of XR acceptance and adoption.An important aspect to recall about the present study is that it attempted to create XR-mediated shopping experiences as close as possible to the original B-a-M scenario, with other external factors (e.g.interactive features, amounts of information, prior product knowledge, visual perceptions and body movement range) being strictly controlled.Hands-free wearable AR glasses were used rather than handheld devices such as a tablet or mobile phone, together with a high-resolution VR HMD with controllers that enabled natural interactions rather than a trigger, joystick, or mouse and keyboard.These features facilitated free body movement rather than sitting in front of a computer (e.g.Mart ınez-Navarro et al., 2019) or moving in a limited space created by a big projector (e.g.Verhulst et al., 2016) or a CAVE [7] system (e.g.Bign e et al., 2016).AR or VR technology did not provide any extra information, features or other aids for the shopping task beyond the core features of VR and AR headsets that would not have been present in the B-a-M scenario.This study maximized the attributes and nature of XR technology to reduce the limitations caused by technological immaturity, as well as to control almost all of the possible factors that might affect the results of the study.Therefore, one of the main contributions of the present study is that it is able to show that XR technologies are able to replicate the B-a-M experience without a clear negative impact on consumer/user experience, which has been seen as a major hindrance to XR adoption by both companies and consumers.

Practical implications
The findings from this study serve to strengthen the confidence of retailers in adopting metaverse-oriented retail and business strategies.For a long time, retailers have been applying digital retail technology to supplement B-a-M shops in order to improve INTR 34,7 accessibility and convenience (Childers et al., 2001).While the Internet revolutionized how we acquire information and make decisions (such as in the context of shopping), it greatly reduced the experientiality and multisensory nature of cognitive and emotional processes.Now, with the advent of XR, it is believed that this experientiality of decision-making might be coming back across domains.However, warranted concerns over the usability and naturalness of XR technologies have loomed over their adoption by firms and their customers.Overall, this study has been able to alleviate the negative expectations toward the use of XR in different domains, particularly in the context of retailing, which was the focus of our study.In this study, we showed that a well-implemented replication of physical activities through VR and AR technologies was overall on par with the physical counterpart when it comes to adoption and acceptance.In particular, VR even led to a higher level of perceived enjoyment and even though the workload appeared to be higher when using AR, it did not prove to be bothersome with respect to perceived usefulness, perceived enjoyment, attitude and willingness to use it in the future.
Retailers are encouraged to make use of the different advantages of VR and AR to provide a virtual shopping experience.In our study, the three digital realities (VR, AR and AV) were compared with B-a-M shopping (between-subjects study design) and for experimental purposes, all participants were required to participate in the experiments in the same physical space.This comparison might not be completely fair for XR as a lot of the value of XR rests in the same aspects as can be seen in the value of the Internet and in particular, it can be used from home.In this study, B-a-M and XR showed the same levels of value, which is a hugely positive result as it shows that shopping from home through XR technology may offer an equal experience, saving valuable time and money for the customer while reducing the costs of maintaining B-a-M premises.Retailers should also notice the potential of combining VR and AR together in the retail environment (similar to the AV shopping conditions featured in Group 4).Based on the results of this study, the additional AR technology did not decrease the perceived ease of use toward shopping in the virtual reality environment.This indicates that AR can provide more value when used in virtual reality.Other interactive technologies such as recommendation systems (Ahn et al., 2015;Guo and Elgendi, 2013), robotics and artificial intelligence (Cruz et al., 2019;Tao and Zhang, 2017) can also be integrated into virtual reality using interactive AR modalities.
When creating an XR-mediated shopping environment, the product type should be taken into consideration by both retailers and designers.In this study, it seems that the virtual visual experience brought by 3D programming technologies and HMD devices can create the same or even a better shopping experience (e.g.enjoyment) as in physical reality.However, it is undeniable that products that rely heavily on other dimensions of experientiality, such as food (taste), perfume (smell), clothing (touch and trial experience) and speaker (audio), require sensory modalities such as auditory, olfactory, artificial flavor and haptic feedback to be included in virtual reality.That said, this study at least provides empirical evidence for shopping experiences that mainly involve visual information such as product appearance and packaging.
Moreover, it can be seen that XR plays an important role in influencing shopping intention, even with a relatively low ease of use.Mobile-based XR technologies such as mobile AR applications and VR cardboard offering a 360-degree view are favored by most retailers, and these devices have relatively high economic value and low development costs.However, they do not offer many advantages in creating high-experientiality shopping due to the limitations of hand, head and body movement, interaction function and a continued need to be hand-held.In this study, consumers had a high acceptance of using the wearable HMD device for shopping.Thus, retailers could consider applying these research results obtained in the laboratory setting to the practical retail environment, by choosing suitable wearable XR devices and developing corresponding interactive programs for providing optimal shopping experiences.

Conclusions
While the adoption and hype around XR technologies have been volatile and oscillating, the current consensus has been somewhat set on the pessimistic view that XR technologies would not be able to provide an experience that would be on par with the real experience (Yung and Khoo-Lattimore, 2019).Therefore, while there have been a lot of warranted and unwarranted expectations to transform many tasks of the real world into fully digitalized ones, these fears have hindered the adoption of XR.In this study, we set out to build a realistic and high-quality implementation that transformed a complicated task of shopping in real life into a digitalized experience through VR and AR technologies.We constructed a physical B-a-M shop and stocked it with approximately 600 products with accompanying product information and pricing.We then created three XR-mediated shopping experiences employing accurate laser scanning to model the shop in a 1-on-1 scale as a 3D environment: an AR shop (where the product information was superimposed onto reality), a VR shop (where the experience took place completely in VR) and an AV shop (where the experience completely took place in VR and the product information was superimposed onto reality).Based on the results from 157 participants, this study indicates that AR and VR can indeed offer an experience that is on par with and even better than the normal B-a-M experience.In support of this, the findings show that AR and VR did not significantly and negatively influence users' perceived usefulness and that VR significantly increases perceived enjoyment, while AR had a significantly negative effect on perceived ease of shopping.Interestingly, we found that poorer perceived ease of shopping was significantly and positively associated with attitude while having no significant effect on perceived enjoyment and perceived usefulness.

Limitations and future directions
This study has some limitations that should be discussed.While not a limitation per se, it is worth repeating for future comparison and reproducibility of the results that the perceived ease of use was measured through the NASA-TLX instrument of workload, as it is better suited for measuring the overall ease of a task or an activity (or the lack thereof).Therefore, our measure was not only limited to the perceived ease of use of the technology but reached onto the entire task performed by using the technology.We recommend that future studies on technology acceptance employ more comprehensive measures of the workload involved in order to further increase the granularity of what kinds of dimensions of perceived ease of use/ workload (e.g.rating and weight) may be differentially affected by XR and how they further affect other psychological and behavioral outcomes.
Regarding the employed metaverse technologies featured in this study, one of the limitations is that the investigated user experience was mainly based on HMD-based visual experience.The adoption and acceptance of other sensory modalities for haptic, olfactory and auditory experience and full-body movement can be researched in future studies.Additionally, this study measured user experience through the participants' subjective self-reports, which might possibly be biased or inaccurate and can only reflect their past or delayed experiences rather than their immediate or real-time experiences.To mitigate these limitations, future researchers might use a combination of research methods, such as observational studies, indepth interviews and objective performance metrics (see, e.g.neural recordings: Saffari et al., 2023;gaze analysis: Chen et al., 2023;immersive netnography: Kozinets, 2022), in order to gain a more comprehensive understanding of consumer experiences.
Even though a possible limitation of this study lies in TAM's simplicity, TAM is still a very useful tool through which to gauge the acceptability of new technologies and is applicable as a theoretical foundation for investigating the acceptance of metaverse shopping.However, it is acknowledged that TAM has a limitation in regard to its INTR 34,7 comprehensive coverage of factors leading to the adoption and acceptance of IS and IT.Therefore, to enrich and develop the extant technology acceptance theories, future studies can examine different models and theories of individual acceptance, such as the Motivational Model, the Theory of Planned Behavior, the Model of PC Utilization and the Innovation Diffusion Theory in different metaverse contexts.
Possibly stemming from our substitution of perceived ease of use with workload measured by NASA-TLX, our results indicated that usability was conversely correlated with user attitude.Again, while not necessarily a limitation of the present study, our research points toward a possible limitation in the technology adoption model and its operationalization of measurement.Novel technologies (such as VR and AR) might pose confounding factors for technology adoption that are not only undetectable by mere perceived values, including of use, enjoyment and usefulness but may also shift path coefficients between extant variables of the model.In the present study, an unexpected finding where poorer usability predicted positive attitude hints toward a co-occurring hidden variable that correlated with poor ease of use but which is simultaneously able to create a positive experience that is not explained either by enjoyment or usefulness.Similar odd findings regarding perceived ease of use have been detected in existing studies; however, research and in-depth investigation into such concealed phenomena are lacking.
The music product category selected in this study has sufficient product information to facilitate information-seeking and decision-making behaviors during the shopping process and has the technical advantage of being suited for creating 3D virtual objects.On reflection, it seems to have been the best choice for the XR-mediated shopping experiment, especially based on student samples.However, the results of this study might be limited to the mediarelated consumption context and products such as books, postcards and posters may also meet experiment design requirements, which can be investigated in future studies to examine the robustness of the results yielded in the current study.In addition, for other types of products, such as large-size products (e.g.furniture, houses or cars), clothing, luxury products and entertainment products, consumers are usually eager to get more experiential value and trial possibilities when shopping.To address this, other product types can be investigated in future experimental studies and in order to conduct a more generalizable and international study, large-size international surveys can also be considered to improve the external validity of the research results.
It should also be mentioned that this study only investigated the shopping experiences of single consumers and did not research interactions with others such as other consumers and shopping assistants.Social aspects have been considered as an essential factor in retail research (e.g. in relation to arousing positive emotions: McGrath and Otnes, 1995; technology acceptance and attitude: Hassanein and Head, 2007;buying intention: Chen et al., 2022).In particular, VR has been considered as an "empathy machine" (Buji c et al., 2020), which is intuitively correlated with pro-social behaviors in the shopping context.However, due to the limitations of current XR devices, technologies and experiment design, there is still a research gap in regard to whether and how social factors influence XR shopping experiences.Thus, future XR shopping research can bring social factors into consideration, such as shopping in the presence of others, different types of social interactions and group shopping.
Notes 1.Studies that used multiple screens/monitors were included given that they took the expansion of the FOV (field of view) into consideration, so as to create a more immersive experience based on developing technologies and devices (see Xi and Hamari, 2021).
product condition were evenly and randomly distributed to record pools per treatment group to fill the store shelves.For each participant, 54 products were displayed on three layers of three-row shelves (2 products per shelf) in the shop.After completing the experiment, the participant was given the records he/she had "bought."Any records "bought" by previous participants were restocked with new records in the same price category.
4. An LP (derived from "long playing" or "long play") is an analogue sound storage medium-a phonograph record format characterized by a speed of 33 1⁄3 rpm and 12-or 10-inch (30-or 25-cm) diameter.
5. Discogs is a website and crowdsourced database of information about audio recordings, including commercial releases, promotional releases and bootleg or off-label releases (https://www.discogs.com/).
6.As indicated in section 3.2 Participants, three disqualified participants were identified, and five outliers were omitted during the data cleaning stage, leaving a remaining sample of N 5 157 for analysis.
Figure 1.Metaverse The horizontal axis represents AR technology, value ranges from 0 (without) to 1 (with); the vertical axis represents VR technology, value ranges from 0 (without) to 1 (with) Source(s): Author's own creation/work Focus 3D (middle) and the Valve Index VR headset and controllers (right)

Table 2 .
Measurement scales Note(s):The loadings of items EUS 2 and EUS 4 were lower than 0.6; therefore, they were omitted in the analysis Source(s): Author's own creation/work . The diagonal (italic values) means the square root of AVE Source(s): Author's own creation/work Standard Regression Coefficient, CI 5 Confidence Interval; the significance level was set at 0.05 Source(s): Author's own creation/work