Complete Research Paper from attached feedback
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4, pp. 673-697/December 2005 673
RESEARCH ARTICLE
THE EFFECTS OF VIRTUAL REALITY ON CONSUMER LEARNING: AN EMPIRICAL INVESTIGATION1
By: Kil-Soo Suh School of Business Yonsei University 134 Shinchon-dong, Seodaemoon-ku Seoul 120-749 KOREA kssuh@base.yonsei.ac.kr
Young Eun Lee Sauder School of Business University of British Columbia 2053 Main Mall Vancouver, British Columbia V6T 1Z2 CANADA lee@sauder.ubc.ca
Abstract
As competition in business-to-consumer e- commerce becomes fiercer, Web-based stores are attempting to attract consumers’ attention by ex- ploiting state-of-the-art technologies. Virtual reality (VR) on the Internet has been gaining prominence recently because it enables consumers to experi- ence products realistically over the Internet, there-
1Ron Weber was the accepting senior editor for this paper. Vivek Choudhury was the associate editor. Dennis Galletta, Lorne Olfman, and Jason Thatcher served as reviewers.
by mitigating the problems associated with con- sumers’ lack of physical contact with products. However, while the employment of VR has in- creased in B2C e-commerce, its impact has not been explored extensively by research in the IS field.
This study investigates whether and under what circumstances VR enhances consumer learning about products. In general, VR enables consu- mers to learn about products thoroughly by pro- viding high-quality three-dimensional images of products, interactivity with the products, and in- creased telepresence. In addition, congruent with the theory of cognitive fit, the effects of VR are more pronounced when it exhibits products whose salient attributes are completely apparent through visual and auditory cues (because most VR on desktop computers uses only those two sensory modalities to deliver information). Based on these attributes, we distinguish between two types of products—namely, virtually high experiential (VHE) and virtually low experiential (VLE) pro- ducts—in terms of the sensory modalities that are used and required for product inspection. Hypoth- eses arising from the distinctions expressed by these terms were tested via a laboratory experi- ment. The results support the predictions that VR interfaces increase overall consumer learning about products and that these effects extend to VHE products more significantly than to VLE products.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
674 MIS Quarterly Vol. 29 No. 4/December 2005
Keywords: Virtual reality, consumer learning, interface design in e-commerce, cognitive fit
Introduction
As competition in business-to-consumer e- commerce becomes fiercer, Web-based stores are adopting and using state-of-the-art technologies in their attempts to attract the fickle and selective attention of consumers. Toward this end, virtual reality (VR) has emerged as a technology that provides users with realistic, interactive computer environments (Li et al. 2001). VR interfaces pro- vide high-quality three-dimensional images of products, interactivity with the products, and increased telepresence (Klein 2003; Steuer 1992). With the assistance of VR, users can experience products virtually by examining and manipulating the visual images, functions, and features of pro- ducts in a variety of ways. VR brings verisimilitude to Web-based stores, partially alleviating the major constraints caused by the lack of contact between consumers and products online (Klein 2003).
Previous marketing research into three-dimen- sional (3D) advertising (which is made possible on a two-dimensional (2D) screen using VR tech- nology) has demonstrated that consumer learning is enhanced by such interfaces (Li et al. 2001, 2002, 2003). Compared to products presented in 2D modes, consumers tend to understand products better, prefer them to other products, and are more inclined to buy products when they are presented with 3D advertising. However, while the employment of VR has increased in B2C e- commerce, its impact has not been explored extensively by research in the IS field (Walsh and Pawlowski 2002).
Thus, the first goal of this study is to examine whether the use of VR in Web-based storefronts positively influences consumer learning, including consumer intentions to purchase. While positing a positive overall effect from the use of VR, we assert that its effectiveness is greatest when the qualities of VR correspond to the salient attributes of products, specifically, the attributes consumers
consider most important in their purchase deci- sions. This argument is based on the theory of cognitive fit, which identifies a contingent effect of technologies: better performance results when interfaces correspond to the nature of particular tasks to be accomplished (Goodhue and Thomp- son 1995; Vessey 1991). In the context of B2C e- commerce, the nature of any particular product is an important influence on consumer tasks, parti- cularly in processes of searching for and acquiring information and making decisions about purchases (Levin et al. 2003; McCabe and Nowlis 2001). Consequently, the second objective in this study is to investigate whether and how the impact of VR on consumer learning may be contingent upon product type.
We investigate the moderating effect of product type on VR effectiveness in terms of the sensory stimuli VR is able to convey. In particular, current VR can reasonably generate and transmit only visual and auditory stimuli through monitors and (sometimes) speakers, given that the kinds of VR adopted by most Web-based stores do not employ more sophisticated tools, such as force-reflecting devices. Hence, the effects of VR are manifested only when it represents products that require vision and hearing for inspection. We use the term virtually high experiential (VHE) to characterize such products and to distinguish them from virtually low experiential (VLE) products, whose evaluation is best accomplished by senses other than vision or hearing.
In the sections below, we first review previous research on VR and consumer learning in order to derive our hypotheses on the overall impact of VR on consumer learning. Next, we present a review of existing literature on the theory of cognitive fit and product attributes, which provides a founda- tion for our theories on the relations between different product types and VR. We then explicate product types and predict the moderating effects of the different product types on consumer learning. Our research method is explained, including the operationalization of independent and dependent variables and the experimental design and procedures. Data analysis and results are pre- sented in the subsequent section. Finally, we
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 675
conclude with a summary of the results of the experiment, the implications for theory and practice, and suggestions for future research.
Theoretical Background and Hypotheses
Figure 1 presents the research model developed and applied in this study. The figure illustrates that VR, with high media richness, interactivity, and telepresence, enhances consumer learning about products. In addition, VR can best replicate con- sumer inspections of products that require vision and hearing, because VR on desktop computers generally uses only these two sensory modalities to deliver information. Hence, the effects of VR are more pronounced when it exhibits VHE products whose salient attributes are completely apparent through visual and auditory cues, while the effects are limited in regards to VLE products whose salient attributes are best described by other sensory cues or from secondary sources. These assertions and the corresponding hypoth- eses are developed below.
Virtual Reality
VR is a computer-generated, interactive, 3D environment in which people become immersed (Wexelblat 1993). Depending on the extent of this immersion, VR applications can be classified into two categories: immersive VR and non-immersive VR (Mills and Noyes 1999). In the former, users wearing head-mounted displays are totally sur- rounded by enclosed virtual environments. Non- immersive VR, on the other hand, is conveyed most commonly by desktop or laptop computers. Thus, users’ VR experiences are limited to what they see on their display monitors and what they hear from their speakers. The present study focuses on non-immersive VR interfaces, because most Web-based stores have implemented non- immersive VR, generally due to the high expense
and cumbersome equipment required for immersive VR.
VR provides high media richness (i.e., high levels of representational quality and volume of content in a mediated environment). The degree of media richness is determined by the sensory depth and breadth of an interface (Steuer 1992). Depth refers to the quality of information within each channel. Breadth, on the other hand, refers to the number of sensory dimensions simultaneously presented. VR increases sensory depth, espe- cially in the visual sense, as it can transmit more detailed 3D images than 2D static images, particularly through zoom and rotation functions (Klein 2003). Simultaneously, VR has the capa- bility to increase the breadth of a sensory interface inasmuch as it often stimulates multiple sensory channels, although, in the context of B2C e- commerce, most VR targets only two senses: vision and hearing.
VR also provides high interactivity (i.e., the degree to which users can manipulate the form and content of a mediated environment in real time; Steuer 1992). Interactivity is achieved when users are provided with immediate feedback through their perceptions that a mediated environment is modified based on their input (Klein 2003). This relates to the prominent features of VR. It offers a high level of control over computer-mediated environments, both in terms of user abilities to adjust the information according to their individual interests and concerns, and, in general, their ability to be active, rather than passive, in their engagement with the information (Pimentel and Teixeira 1994, pp. 20-21).
Through high media richness and interactivity, VR can generate compelling feelings of telepresence (Biocca 1997; Klein 2003). Telepresence is a sense of “being there” in an environment by means of a communication medium (Reeves and Nass 1996; Steuer 1992). Based upon sensory stimuli conveyed by a VR interface, human beings can create a perceptual illusion of being present and highly engaged in a mediated environment, while they are in reality physically present in another place (Biocca 1997).
Suh & Lee/Effects of Virtual Reality on Consumer Learning
676 MIS Quarterly Vol. 29 No. 4/December 2005
Media Richness Knowledge
Attitude
Purchase Intentions
Interactivity
Telepresence
Strengths of VR Interfaces
H1a, H1b, H2, H3
Virtually Low Experiential Products
Virtually High Experiential Products
Vision and Hearing
H4a, H4b, H5, H6
Fit (Sensory modalities used by VR and required for product descriptions)
Consumer Learning
Product Types
Media Richness Knowledge
Attitude
Purchase Intentions
Interactivity
Telepresence
Strengths of VR Interfaces
H1a, H1b, H2, H3
Virtually Low Experiential Products
Virtually High Experiential Products
Vision and Hearing
H4a, H4b, H5, H6
Fit (Sensory modalities used by VR and required for product descriptions)
Consumer Learning
Product Types
Figure 1. Research Model
Consumer Learning
Previous research has indicated that rich, inter- active, and engaging presentations of information enhance consumer learning (Kim and Biocca 1997; Li et al. 2003). Consumer learning refers to any process that changes a consumer’s memory and behavior as a result of information processing (Arnould et al. 2001). Learning is open to outside influences, given that it is not always a process of achieving absolute truth; furthermore, it remains vulnerable to external factors such as consumers’ familiarity with particular products, their motiva- tions, and the ambiguity of various information
environments (Hoch and Deighton 1989). VR reduces ambiguity by providing rich information, and it motivates consumers by enabling them to interact with products (Kempf and Smith 1998). Thus, by providing VR, Web retailers can positively influence consumer learning about products.
Lavidge and Steiner (1961) first established a research tradition that has investigated consumer learning from three dimensions (cognitive, affec- tive, and conative), and that has accumulated a rich history of research. For instance, it has been asserted that,
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 677
traditionally, effective consumer learning is assumed to be a critical mediator of consumption and ascertained from cogni- tive, affective, and conative dimensions. Likewise, numerous techniques for mea- suring effectiveness are intended to examine components from the same domains (Li et al. 2003, p. 398; emphasis added).
The sequence along which the three dimensions occur has yet to be fully explicated, however. Hierarchical models of effects have posited that the process is sequential, beginning with cognition, proceeding through affects, and concluding with conation. In practice, consumers first comprehend cognitively a given message and form positive, neutral, or negative attitudes toward products. They then develop their intentions for action. However, for less-complicated products that can be sampled relatively easily, researchers have proposed an alternative path from cognition to conation and affect (see Smith and Swinyard 1982). Confronted with such disagreements regarding the actual sequence, MacInnis and Jaworski (1989) have provided an integrative framework that includes antecedents to learning, information processing, and consequences of this process. Notably, their framework separates infor- mation processing from its consequences. Fol- lowed by exposure to stimuli, information pro- cessing occurs. As a result, cognitive, affective, and conative responses are invoked.
In the present study, we examine VR effects on consumer learning in terms of cognitive, affective, and conative dimensions, based on observations made by MacInnis and Jaworski (1989) and Li et al. (2003).2 The cognitive dimension determines the extent to which information about products enhances consumer comprehension. It can be measured based on either actual or perceived
knowledge (Bettman and Park 1980). The affec- tive dimension, on the other hand, identifies whether or not consumer attitudes are influenced by particular stimuli (McKenzie et al. 1989). Finally, conative measurements investigate behav- ioral responses to various stimuli, such as pur- chase intentions that may be invoked by the stimuli (see Li et al. 2003).
In terms of cognitive processes, Kempf and Smith (1998) have claimed that product trials enhance consumer comprehension about products. The richness of VR allows consumers to examine realistic 3D images of products from various angles and distances. The interactive capabilities of VR permit them to sample various product func- tions and features. Simultaneously, consumers experience strong telepresence, which engages them in learning processes and hence increases their comprehension of the objects (Klein 2003; Li et al. 2003). On the other hand, consumers feel less telepresence in static interfaces consisting of still pictures, inasmuch as pictures are relatively lean media compared to VR. Moreover, they are rarely interactive. Therefore, we expect that VR interfaces can enable consumers to comprehend products better than static interfaces.
H1a. Compared to static interfaces, VR interfaces increase consumers’ actual product knowledge.
H1b. Compared to static interfaces, VR interfaces increase consumers’ per- ceived product knowledge.
Next, from the perspective of the affective and conative dimensions, consumers form either posi- tive, neutral, or negative attitudes and purchase intentions toward products as they examine the products. VR, with its higher media richness, in- creased interactivity, and higher telepresence, affects consumers’ attitudes and purchase inten- tions regarding products more significantly than static pictures. However, more-thorough examina- tions of products do not always result in positive consumer attitudes and strong purchase inten- tions. If consumers realize the drawbacks of a product, the influence on their attitudes and
2The sequence along which the three dimensions occur is not considered in this paper because we do not intend to partake in the debate in the marketing field, which remains unresolved. Benbasat and Zmud (2003) have warned IS researchers of the danger of including or over- investigating issues that are best left to scholars in other disciplines.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
678 MIS Quarterly Vol. 29 No. 4/December 2005
purchase intentions is negative. Regardless of whether the influence is positive or negative, the enhanced examination of products possible with VR can change consumers’ attitudes and pur- chase intentions.
H2. Consumers’ attitudes toward products presented with VR interfaces differ from their attitudes toward products presented with static interfaces.
H3. Consumers’ purchase intentions toward products presented with VR interfaces differ from their purchase intentions toward products presented with static interfaces.
Theory of Cognitive Fit and Product Types
While VR generally influences consumer learning, the degree to which VR affects consumer learning varies contingent upon the particular product type being examined. The theory of cognitive fit posits that a match between IT applications and users’ tasks is important for the realization of positive results from IT (Vessey 1991). Goodhue and Thompson (1995) have found that, when users perceive that the characteristics of a technology match the characteristics of their tasks, they be- lieve the technology exerts a more positive impact on their effectiveness and productivity in achieving their goals. Furthermore, prior marketing research has demonstrated that an important influence on the consumer task is the nature of the product (McCabe and Nowlis 2001). That is, the attributes of products affect consumers’ tasks of searching for and acquiring information and making decisions about their purchases (Levin et al. 2003). Given the influence of product type on consumers’ tasks, we claim that the advantages of VR are aug- mented only in relation to products whose critical attributes can be assessed adequately by the characteristics of VR.
Consumers can experience products in three ways: directly, indirectly, and virtually (Li et al.
2003). Specifically, consumers experience pro- ducts with physical or actual trials (i.e., direct experience), through secondhand source informa- tion such as advertising or labels (i.e., indirect experience), or with virtual representations of the products, such as by using VR (i.e., virtual experience). The primary difference between the three lies in the human senses that are involved. All five senses of active organisms (i.e., orien- tation, hearing, touch, taste-smell, and vision) can be used in direct experience (Klein 2003; Schiffman 1990). Conversely, none of these senses are used directly in indirect experience, because no physical contact with products is involved. Two senses (i.e., vision and hearing) are used in virtual experiences, because the VR adopted by most Web-based stores uses only monitors and speakers.3
The distinction between virtual and direct experi- ences parallels distinctions that can be made between different kinds of product attributes because experiencing different attributes requires the use of distinct senses. Specifically, direct experiences are most suitable for experience attributes (e.g., the taste of food; see Table 1) because the use of one or more of the five senses is often required (Nelson 1975). Indirect experi- ences, on the other hand, are sufficient for search attributes (e.g., calories of food) that do not involve the direct use of these senses (Wright and Lynch 1995). However, attributes that are best supported by virtual experience have not been fully explicated by previous research.
3Web-based VR does not support receivers’ touch, taste- smell, or orientation because it does not employ the devices required for replicating such sensations, including head-mounted displays and force-reflecting devices. On the other hand, vision is considered to be the most important sense in the context of generating virtual experiences through realistic 3D displays (Gigante 1993; Pimentel and Teixeira 1994, p. 95). VR can also generate auditory sensations. However, auditory stimuli play a secondary role in the generation of virtual experiences when accompanied by visual stimuli (Pimentel and Teixeira 1994, pp. 95-98). Thus, in the present study, products are characterized primarily by their visible qualities.
Table 1. Summary of Product Experience and Attributes
Product Experience Product Attributes
Product Experience Definition Media Used
Sensory Modalities Used
Product Attributes Examined Definition Examples
Direct Experience
Consumers learn about products through physical contact using full sensory modalities
N/A (Physical contact)
Orientation, vision, hearing, haptic, taste-smell
Experiential Attributes
Attributes that one can assess only through direct experience
The appearance of clothing, the taste of food, the smell of perfume
Indirect Experience
Consumers learn about products from secondary sources without physical contact
Secondary sources
N/A (No direct physical contact involved)
Search Attributes
Attributes that can be fully described by secondary sources
Components of vitamins, calories of food
Virtual Experience
Consumers learn about products in a mediated environment using visual/ auditory modalities
VR interface* Vision, hearing Virtually Experiential Attributes
Attributes that can be described only through visual or auditory sensory channels
The shape of a wristwatch, the movement of a sofa- bed frame, color choice, and the click of a camera shutter button
*VR interface refers to the VR operated on a desktop platform, given the current state of technology.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
680 MIS Quarterly Vol. 29 No. 4/December 2005
In this light, Li et al. (2003) have discussed types of products that are suited for virtual experience by investigating the senses used for conveying the products. They have classified products as geo- metric, material, or mechanical. Geometric pro- ducts are those whose attributes consumers can understand fully by visual cues, whereas material products are those whose attributes require haptic cues (i.e., physical touching) in addition to visual inspection (McCabe and Nowlis 2001). Mechani- cal products, a third category corresponding to consumer desires to interact with a product when examining it, are those whose attributes can be evaluated by behavioral interaction in addition to merely touching the product (Li et al. 2003).
Despite the contribution by Li and his colleagues to the classification of product types based on virtual experience, this classification system is not exhaustive. They deal only with visual and haptic senses with little explanation of other senses. Geometric products are related primarily to vision, material products are associated with vision and touch, and mechanical products involve vision and touch (because behavioral interaction is achieved through a visual sense in VR). Thus, products whose attributes require senses other than vision and touch do not fall into any of these categories. Consider a bag of potato chips and a candy bar, the examples of geometric products proposed by Li et al. (2003). These products require con- sumers’ taste and smell for satisfactory information to be obtained, not just vision, the sense that characterizes geometric products. Thus, this type of product is not exactly a geometric product, but neither is it explicitly a material nor a mechanical product.
Furthermore, because this classification does not effectively distinguish the senses that are most suitable for virtual experiences from those that are not, it is difficult to predict the impact of virtual experience consistently. For example, to assess material products (e.g., a sweater), consumers must see and touch them. Thus, both visual and haptic senses are important. However, because VR cannot fully support haptic senses, consumers can assess only those attributes requiring visual cues. Thus, the effects of virtual experience on
this type of product are not anticipated easily. Given the absence of a proper product type associated with the effect of VR that emphasizes the use of vision and hearing, we provide a new classification of product types. Specifically, we refer to the attributes that can be experienced with visual and auditory senses as virtually experiential attributes. This category includes (1) the shape and appearance of a product (e.g., the shape of wristwatches and the appearance of furniture), (2) possible changes in the form of a product (e.g., the assembly of a toy, the movement of a sofa-bed frame) and possible changes in its content (e.g., color choice), and (3) the sound of a product (e.g., the click of a camera shutter button).
In contrast, product attributes that can only be assessed through senses other than vision and hearing are not assessed ideally through virtual experience in the current state of technology. Furthermore, the attributes that are better assessed through indirect experience generally cannot be examined by virtual experiences in satisfactory ways. In other words, certain attri- butes are not virtually experiential at this time. For example, the bouquet of wine, which requires consumers to use another sensory channel, or the reliability of a PDA, which can be examined satisfactorily through secondary sources, are not virtually experiential attributes.4
We define salient attributes of products as those attributes that are most prominent and important when consumers make decisions about pur- chasing the products. For example, with clothing and similar products, consumers emphasize the ability to examine the designs of the products, while with products like compact disks they like to listen to the products. We refer to products whose salient attributes are mainly virtually experiential as virtually high experiential (VHE) products. We refer to products whose salient attributes are not
4We are positing that the effects of virtual experiences are enhanced insofar as VR supports the sensory stimuli required for consumers to assess particular attributes faithfully and realistically. Hence, no matter how crea- tively a VR interface is designed, some attributes can be best appreciated with senses other than vision and hearing. Therefore, for these attributes, the additional knowledge gained by VR viewers will be limited.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 681
primarily virtually experiential as virtually low experiential (VLE) products.
These types of products moderate the degree to which VR affects consumer learning, since the strengths of VR (media richness, interactivity, and telepresence) are enhanced by visual and auditory cues. That is, visual and auditory cues intensify media richness of VR, establish interactivity,5 and build perceptual illusions that generate tele- presence. Therefore, when VR is applied to VHE products, consumers are able to visually examine and interact with products by sampling features of the products. However, when VR is used for VLE products, the functionality of VR may not be as useful as it is for VHE products because the salient attributes cannot be represented effectively through vision and hearing. Thus, VR for VLE products does not contribute as much to con- sumers’ knowledge, attitudes, and purchase intentions as it does for VHE products.
H4a. Increases in consumers’ actual know- ledge, effected by VR interfaces, are more significant for VHE products than VLE products.
H4b. Increases in consumers’ perceived knowledge, effected by VR interfaces, are more significant for VHE products than VLE products.
H5. The impacts of VR interfaces on con- sumer attitudes toward products are more significant for VHE products than VLE products.
H6. The impacts of VR interfaces on con- sumer purchase intentions toward products are more significant for VHE products than VLE products.
Research Method
A laboratory experiment was employed to empi- rically test the effects of VR on consumer learning and the moderating effect of product types. The experiment allowed close control over indepen- dent, moderating, dependent, and possibly con- founding variables to achieve a high degree of internal validity (Singleton and Straits 1999, p. 183). To enhance mundane realism, the similarity of experimental events to real experiences (Single- ton and Straits 1999, p. 194), and the generali- zability of the findings, we selected products sold in real Web-based stores and interfaces developed by a commercial VR application provider.
Experimental Design
A 2 × 2 factorial design with a within-subject factor and a between-subject factor was used. The within-subject factor, interface design, had two levels: VR and static. The between-subject factor, product type, had two levels: VHE and VLE. Besides economizing on the number of partici- pants (Gravetter and Wallnau 2000), the use of a within-subject design for the interface design enabled control over individual differences in comprehension and memory abilities, which otherwise could have significantly influenced the dependent variables (Sternthal and Craig 1982).
Because interface design was a within-subject factor, different products were employed for each interface design. This prevented the learning effect that could occur through repeated searches for information about the same products. How- ever, if the differences between products were too dramatic, a fair comparison of independent vari- ables could not be drawn. Consequently, the inter- face effect would not be detected. If consumers preferred certain products much more strongly than others, for example, it would be unlikely that interface design would change their attitudes. Therefore, controlling for consumers’ previous attitudes toward products was essential. Con- ducting a pretest and counterbalancing products for each interface accomplished this outcome. First, the pretest was conducted with a pilot group
5The interactivity of VR is represented in a series of visual states portraying the forms or content of products. Consider the operation of a camera shutter as an example: the shutter is closed in the initial stage; the user provides an input (e.g., a mouse click); this input provokes the shutter button to be pressed and released; the shutter opens and closes accordingly. The sequence of visually changing states of the camera manifests the user’s interaction with the camera.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
682 MIS Quarterly Vol. 29 No. 4/December 2005
VR Static
H1 H2 H3 H4
H1 H2 H3 H4
L1 L2 L3 L4
L1 L2 L3 L4
VHE Product
VLE Product
Hi: VHE product Li: VLE product i: 1, 2, 3, 4 (the number
of products)
VR StaticVR Static
H1 H2 H3 H4
H1 H2 H3 H4
L1 L2 L3 L4
L1 L2 L3 L4
VHE Product
VLE Product
Hi: VHE product Li: VLE product i: 1, 2, 3, 4 (the number
of products)
Figure 2. Counterbalancing Products
that was demographically similar to the experiment participants. Purchase intentions for 10 VHE and 10 VLE products were measured. Among them, four products ranking fourth to seventh in each product type were selected, because the inclusion of the most or least preferred products could hinder realization of the interface effect. In addi- tion, as shown in Figure 2, we counterbalanced products for each interface. Under the VHE (and VLE) conditions, half the participants were pre- sented the first and third products in VR and the second and fourth products in static pictures. The other half was presented the interfaces in the reverse order. To eliminate any potential extra- neous effects in the experiment participants were assigned randomly to each condition and asked to navigate freely around demonstrations of the four products.
Participants
For the main experiment, 85 participants were drawn from a pool that comprised undergraduate
students taking a financial management course in a large university. Participation was voluntary; all participants were offered $10 gift certificates to encourage their participation in the experiment. Of the 85 participants, 57 were male and 28 were female. On average, they were 23.3 years old, and 85 percent were business students. They surfed the Internet for 12.63 hours per week, and 71 percent had previous experience with VR (87 percent had previously purchased products in Web-based stores). A Chi-square analysis re- vealed no significant differences in gender, area of study, previous experience with VR, or online purchase experience among the groups. A one- way ANOVA further revealed no significant differences between the groups in terms of age, number of years in university, or average time spent surfing the Internet. Participants were also asked if they possessed sufficient product knowledge to make suitable purchase decisions on a seven-point Likert scale (1 for “strongly disagree” and 7 for “strongly agree”). Their prior knowledge of each class of products was not statistically different (VHE = 3.7, VLE = 4.1, t = 1.269, p = .208).
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 683
Web-Based Store Design
Interface Design
We constructed four Web-based stores, each offering four products. Two used a VR interface; two used a static interface. With the exception of counterbalancing the interfaces, the two stores for each product type were identical. They contained the same products, information, and design, there- by ensuring information symmetry.
Hoffman and Novak (1996) have categorized information technologies into the groupings of dynamic and static. Dynamic technology includes audio, full-motion video, and animation, while static technology contains text, images, and graphics. Accordingly, VR interfaces, as a kind of dynamic technology, employ 3D representations of pro- ducts, while static interfaces are built around still pictures of products.6
The Web-based stores were developed using HTML, ASP (Active Server Page), JavaScript, and Flash, all of which are commonly used by online stores. Each Web-based store consisted of an introduction page, a page with product lists, and four detailed-information-about-product pages. Product-display windows were attached to each page. From the product-list pages, participants were able to proceed to pages detailing informa- tion about each product by selecting the product names. The detailed-information-about-product pages contained both specific information about the products (such as functions, features, sizes, and benefits), and “Show Me This Product” buttons. Selecting these buttons led to product display windows, in either a VR or a static interface, which allowed participants to examine the products. Appendix A provides examples of the VR and static interfaces for a sample product.
The VR interfaces were developed in collaboration with a commercial VR application provider, 3DIGM Co. Ltd. The same programmer, who was not
informed of the research objective, built VR inter- faces for all eight products to ensure that the quality and functions of the VR were as similar as possible for VHE and for VLE products. 3DIGM’s virtual-reality engine, NOVA, extending VRML 97, was used to implement the VR interface. Participants viewed this VR interface using 3DIGM’s VR plug-in, NOVA Viewer. The Web- based stores were stored on a commercial Web server that participants accessed through a LAN with a T1 connection. IBM-compatible 1.4 GHz Pentium4 PCs with 256 megabytes DDR SDRAM, 17-inch color monitors, and GeForce, a VR sup- porting graphics card, were used for the experi- ment. Participants browsed the Internet using Microsoft Internet Explorer (Version 5.5). With this configuration, retrieval of information, including VR representation, was almost instantaneous.
Choice of Products
We chose a computer table as a VHE product because the salient attributes center on external appearance and functionality that can be fully represented by visual stimuli. Consumers could examine the external appearance, such as the table design, by rotating the image and by magni- fying or diminishing it. In addition, they could examine alterations to the table, for example, by adjusting the height or angle of the table, opening and closing drawers, and pulling out keyboard trays. We selected a desktop computer for our VLE product because most of its salient features cannot be effectively described by vision and hearing. Rather, they can only be described sufficiently by secondary sources (e.g., CPU type, memory size, and hard-disk storage capacity). The components for which appearance (i.e., visual stimuli) would generally be important, including peripherals (e.g., a monitor, speakers, keyboard, and mouse), were excluded from the description (hence, from the product display), and only the main processor unit was included in the experi- ment. Recall, four products in each product type were selected to prevent learning effects and to increase the generalizability of the results.
For a manipulation check, a pretest was conducted with 35 participants who were demographically
6A pilot study indicated that the quantity of pictures on a Web site (single or multiple) causes no significant impact on dependent variables for VHE products; hence, a single picture was used for the static format.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
684 MIS Quarterly Vol. 29 No. 4/December 2005
similar to the main experiment participants. The salient attributes for the two types of products were ascertained using Fishbein and Ajzen’s (1975) free-elicitation method. Participants were asked to write down the attributes that would be important to them when buying computer tables and desktop computers. They were directed to report the attri- butes related to product quality per se, putting aside other attributes such as price and brand that might influence their perceptions of product quality. For the computer table, 94 percent of the respon- dents listed appearance as the most salient attri- bute, followed by convenience of use (56 percent), durability (53 percent), functionality (50 percent), and storage space (35 percent). For the desktop computer, the most salient attribute was perfor- mance (71 percent), followed by features (66 percent), appearance (54 percent), capacity for future expansion (43 percent), warranty (43 per- cent), and reliability (23 percent). This result is consistent with our definitions of VHE and VLE, thereby ensuring the manipulation was successful.
To ensure that this manipulation would also be successful in the main experiment, on the post- experimental questionnaire, participants were asked to rate the degree to which each attribute was experiential by VR on a seven-point Likert scale (1 for “VR did not enable me to experience this attribute at all” and 7 for “VR fully enabled me to experience this attribute”). This method was adopted from a study by Kempf and Smith (1998), which parallels our research purpose and context. For the computer table, all attributes with the exception of durability had relatively high experi- entiality ratings (above 4) on the seven-point scale. In contrast, only one of the six salient attributes (i.e., appearance) was classified as relatively high for the desktop computer. The weighted average experientiality ratings of all of the salient attributes for the computer table and the desktop computer were 5.10 and 3.16 respectively.
The participants were also asked to classify the given products into VHE or VLE categories after the experiment, and the result again confirmed the success of the manipulation. Over 90 percent (90.7 percent) of the participants presented with computer tables categorized the products as VHE, while 78.6 percent of the participants viewing
desktop computers classified the products as VLE (P2 = 41.47, p <.0001). Hence, we believe our identification of a computer table as a VHE product and a desktop computer as a VLE product is valid.
Dependent Variables
Recall, the effects of VR on consumer learning can be measured in cognitive, affective, and conative dimensions (Hutchinson and Alba 1991; Li et al. 2002, 2003; Lutz 1975). We adopted actual and perceived knowledge as indicators of cognitive learning (Bettman and Park 1980). Actual knowl- edge was measured by a comprehension test. Such tests have been employed previously by a number of studies on the effects of information- presentation formats (Agarwal et al. 1999; Jarven- paa and Machesky 1989). This type of test exam- ines elements in a presentation with the intention of measuring whether viewers identify the ele- ments correctly (Gemino 1999). Six items con- taining correct or incorrect information about the products in the experiment were developed. The content validity of all items was assessed carefully by conducting a pilot test employing participants demographically similar to the experimental group (to ensure that the questions were not biased toward specific interfaces or products). The parti- cipants answered 45 out of a total of 48 questions correctly, demonstrating no differences in the rate of correct answers across interfaces and products. Given the high rate of correct answers, only minor revisions were made to the questionnaire, which was used for the experiment.
To assess perceived product knowledge, three existing Likert-scale items were adopted (Li et al. 2002; Smith and Park 1992). Participants’ atti- tudes toward the products were measured by adopting an established scale using seven-point semantic differential items (Bruner 1998; Li et al. 2003). Participants’ purchase intentions were assessed using an existing seven-point semantic differential scale (Bearden et al. 1984; Li et al. 2003). The participants answered these questions for each of the four products separately.
We measured the relative preference between the products using a pair-wise comparison method for
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 685
further analysis. This method has a foundation in analytical hierarchy process (AHP), which allows comparisons of many alternatives, yielding a rela- tive preference for each alternative (Saaty 1980). Participants were presented with a pair of products and were asked to distribute 100 points depending on the degree to which they preferred each individual product.
Experimental Procedure
Prior to undertaking the experiment, participants were assigned randomly to one of the four treat- ment conditions and trained for 10 minutes with each of the VR and the static interfaces. To ensure identical training, the same instructor taught each participant. After the participants fully understood how to manipulate the interfaces, they were asked to navigate freely around the Web- based store for as long as they wanted. The parti- cipants spent about 1 minute reviewing detailed information for each product. No significant time difference existed between the products presented in the VR interface (66 seconds) and in the static interface (59 seconds). Nonetheless, participants used more time to view displays on products in the VR interface (148 seconds) than the static inter- face (45 seconds). Examining products through VR usually takes more time because interactive VR product visualization allows participants to move, rotate, and magnify images. After navi- gating around the store, participants were asked to minimize their browser windows and to complete the actual knowledge test. Next, participants continued with the remainder of the experimental session while referring to the sites. Each stage of the experiment began with the distribution of ques- tionnaires and ended with their removal, because questionnaires from the prior stage might influence responses in later stages.
Results and Analysis
Analyses of all dependent measures were con- ducted using SPSS for Windows Version 10.0.
Cronbach’s Alpha was used to assess reliability for perceived knowledge, product attitudes, and pur- chase intentions. All constructs displayed accep- table levels (.79, .95, and .96, respectively). Furthermore, the scales were tested for internal consistency and a specified factor structure using confirmatory factor analysis, which indicated they were unidimensional (Table 2). Composite mea- sures for perceived product knowledge, product attitudes, and purchase intentions were then constructed by aggregating the multiple items. These measures were used in the subsequent analyses.
Prior research suggests that experience (Taylor and Todd 1995) and gender (Gefen and Straub 1997) influence individuals’ perceptions and use of IT. We first tested whether demographic differ- ences explained variance in the dependent variables. Multivariate regression analyses found no significant covariate effects of gender and experience (F(16, 150) = 1.05, p = .41 in Wilks’ test). Thus, based on the principle of parsimony (Box et al. 1994, p. 16), we did not include these variables in the subsequent analysis.
Table 3 shows the means and standard deviations of the dependent variables. The Pearson’s pro- duct-moment correlation coefficients indicate that the dependent variables were moderately corre- lated with each other. Thus, the MANOVA model was applied to the data to test for interface-design and the product-type effects on these measures. The interface-design main effect (F(4, 80) = 34.226, p < .001) and the interaction effect between interface design and product type were significant (F(4, 80) = 17.515, p < .001). The pro- duct-type main effect, however, was not significant (F(4, 80) = 2.159, p > .05).
Because the MANOVA results were significant, these results were further analyzed using individ- ual ANOVAs to examine the effects of the inde- pendent variables on each dependent variable. We applied the Bonferroni family adjustment to the level of significance. We set our value at 0.05/4 or 0.0125 to control for an inflated Type I error that arises from multiple tests.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
686 MIS Quarterly Vol. 29 No. 4/December 2005
Table 2. Reliability and Factor Loadings
Construct Reliability Item Factor 1 Factor 2 Factor 3
Perceived Product Knowledge .79 PK1 0.31790 0.11963 0.66948
PK2 0.07611 0.11104 0.73991
PK3 0.07219 0.03296 0.71537
Product Attitudes .95 PA1 0.79164 0.37548 0.18622
PA2 0.76585 0.47864 0.17914
PA3 0.69065 0.29558 0.22531
PA4 0.81416 0.39890 0.14449
PA5 0.73632 0.27342 0.17117
PA6 0.79008 0.47727 0.08890
Purchase Intentions .96 PI1 0.45566 0.81167 0.09823
PI2 0.40863 0.85672 0.10525
PI3 0.38981 0.85732 0.15433
PI4 0.38775 0.83181 0.11813
Table 3. Descriptive Statistics
Product/ Interface
Measurement
VHE Product VLE Product
VR Static TotalVR Static Sub- total VR Static
Sub- total
Actual Product Knowledge
Mean 5.35 4.76 5.05 5.08 5.00 5.04 5.22 4.88 5.05
(S.D.) (.72) (.80) (.56) (.69) (.79) (.59) (.71) (.80) (.78)
Percentile Score*
89.2% 79.3% 84.2% 84.7% 83.3% 84.0% 87.0% 81.3% 84.2%
Perceived Product Knowledge
Mean 5.17 3.23 4.20 4.37 4.06 4.21 4.77 3.64 4.21
(S.D.) (.64) (1.09) (.69) (.73) (1.00) (.80) (.80) (1.12) (1.12)
Product Attitudes
Mean 4.93 3.81 4.37 4.50 4.05 4.27 4.71 3.93 4.32
(S.D.) (.78) (.86) (.37) (.70) (.65) (.53) (.77) (.77) (.86)
Purchase Intentions
Mean 4.37 3.09 3.73 3.58 3.15 3.37 3.98 3.12 3.55
(S.D.) (.91) (.99) (.44) (.85) (1.00) (.73) (.96) (.99) (1.06)
*There are 6 test items, thus percentile score can be calculated by dividing mean score by 6.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 687
Hypotheses Testing
A repeated-measure ANOVA was run to analyze the effects of interface design and product types on each dependent variable. Table 4 shows the results. The interface design significantly affected all dependent variables. Participants reported significantly higher scores for the actual product knowledge test in the VR treatment (mean = 5.22/6.00 or 87.0 percent) than the static treatment (mean = 4.88/6.00 or 81.3 percent). Thus, H1a, which predicts that VR interfaces have more signi- ficant effects than static interfaces on consumers’ actual product knowledge, is supported. Per- ceived product knowledge in the VR treatment (mean = 4.77) was also higher than the static treatment (mean = 3.64). Moreover, the partici- pants reported more positive attitudes toward products in the VR treatment (mean = 4.71) compared to those in the static treatment (mean = 3.93). Likewise, purchase intentions for products in the VR treatment (mean = 3.98) were more pronounced than those in the static treatment (mean = 3.12). These results support hypotheses H1b, H2, and H3, which predict the superior ef- fects of VR interfaces on perceived product knowl- edge, product attitudes, and purchase intentions.
Product type exhibited a significant moderating effect in the interface designs for all of the dependent variables except actual product knowl- edge. For the VHE products, the mean scores of perceived product knowledge, product attitudes, and purchase intentions increased 60 percent, 29 percent, and 41 percent respectively when they were represented by the VR interface. On the other hand, for the VLE products, the mean scores increased only 8 percent, 11 percent, and 14 percent respectively (see Figure 3). Therefore H4b, H5, and H6, which state that the impact of VR interfaces on perceived product knowledge, product attitudes, and purchase intentions will be greater for VHE products than VLE products, are supported. However, H4a, regarding actual pro- duct knowledge, is not supported at the conser- vative significance level of .0125, although the enhancement of actual product knowledge by the VR treatment for VHE products (12.4 percent) was greater than that for VLE products (1.6 percent).
To investigate the nature of the interaction effects, separate t-tests were conducted comparing the mean differences between the VR and static treatments of the dependent variables for each product type. In the case of VHE products, the values of all three variables were significantly different between the VR and the static treatments (t = 122.86, p < .0001 for perceived product knowl- edge; t = 25.06, p < .0001 for product attitudes;, and t = 24.99, p < .0001 for purchase intentions). For the VLE products, however, the difference in purchase intentions between the VR and the static treatments was not significant (t = 5.85, p = .02), whereas the other two variables exhibited signi- ficant differences (t = 7.98, p = .0073 for perceived product knowledge; t = 12.35, p = .0011 for pro- duct attitudes), although the t-values for all three dependent variables for VHE were greater than those for VLE. In summary, the results demon- strate that the VR interface increases consumer learning about VHE products more than it does for VLE products, but it does not promote purchase intentions for VLE products.
Additional Analyses: Relative Preference
Given that the participants’ attitudes toward pro- ducts differed significantly even when VLE products were represented by VR, we analyzed relative preference to investigate further whether a real difference existed in preference across dif- ferent product types. Our measurement of relative preferences (“Which product do you prefer when you compare the two?”) is based on the AHP method (Saaty 1980). Participants in our experi- ment were given four products. Thus, they com- pared a total of six pairs of products. The analysis was completed using Expert Choice 2000 (2nd
Edition for Groups). Table 5 shows the results of the relative preference scores, along with the inconsistency ratios.
The VHE product H1 received the highest pre- ference score (0.392) when presented by the VR interface, whereas this preference score dropped to third place (0.267) when the same product was
Suh & Lee/Effects of Virtual Reality on Consumer Learning
688 MIS Quarterly Vol. 29 No. 4/December 2005
Table 4. ANOVA Results
Actual Product Knowledge (R2 = .616)
Perceived Product Knowledge (R2 = .824)
DF MS F p Power DF MS F p Power
Within-Subjects
Interface Design 1 4.860 10.331 .002** .888 1 53.65 118.32 .000** 1.000
Interface Design × Product Type
1 2.760 5.867 .018 .668 1 28.72 63.329 .000** 1.000
Error (Interface Design)
83 .470 83 .453
Between-Subjects
Product Type 1 .0048 .007 .932 .051 1 .0069 .006 .938 .051
Error (Product Type)
83 .661 83 1.118
Product Attitudes (R2 = .527)
Purchase Intentions (R2 = .548)
DF MS F p Power DF MS F p Power
Within-Subjects
Interface Design 1 26.20 36.642 .000** 1.000 1 31.08 29.847 .000** 1.000
Interface Design × Product Type
1 4.765 6.664 .012* .723 1 7.737 7.429 .008** .769
Error (Interface Design)
83 .715 83 1.041
Between-Subjects
Product Type 1 .403 .967 .328 .163 1 5.662 7.820 .006** .789
Error (Product Type)
83 .417 83 .724
*p < .0125; **p < .01
displayed by the static interface. Meanwhile, participants’ preferences for product H2 rose from the lowest (0.135) to the highest (0.325) when its interface design was changed from static to VR. These results suggest that consumer preferences toward VHE products can be altered significantly by presenting them with VR interfaces. On the other hand, participants who were provided with
VLE products did not profess dramatic changes in their preferences, regardless of the type of inter- face design that was used. Although products L1 and L4 were interchanged in their positions from third to fourth, the experimental consumers’ pre- ferences did not vary significantly whether the products were represented in VR or in static interfaces.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 689
3.23
4.06
5.17
4.37
0
1
2
3
4
5
6
VHE VLE
Static VR
3.81 4.05
4.93 4.5
0
1
2
3
4
5
6
VHE VLE
Static VR
Perceived Product Knowledge
Product Attitudes
Figure 3. Differences Between VHE and VLE
Suh & Lee/Effects of Virtual Reality on Consumer Learning
690 MIS Quarterly Vol. 29 No. 4/December 2005
3.09 3.15
4.37
3.58
0
1
2
3
4
5
VHE VLE
Static VR
H1 (0.392) H3 (0.311)
H2 (0.325) H4 (0.289)
L3 (0.372)
L4 (0.207)
Purchase Intentions
Figure 3. Differences Between VHE and VLE (Continued)
Table 5. Preference Listing Across Groups
Preference 1st 2nd 3rd 4th Inconsistency
Ratio
VHE Product
Condition 1 H4 (0.161) H2 (0.135) 0.00
Condition 2 H1 (0.267) H3 (0.119) 0.00
VLE Product
Condition 3 L2 (0.270) L1 (0.239) L4 (0.119) 0.00
Condition 4 L3 (0.353) L2 (0.246) L1 (0.194) 0.01
= the VR interface static = the static interfaceVR Hi = VHE product Li = VLE product i = 1, 2, 3, 4 (the number of products)
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 691
Discussion and Conclusions
VR on the Internet has been gaining prominence in recent years because it enables consumers to experience products virtually over the Internet, alleviating consumers’ lack of physical contact with products. Furthermore, compared to a static inter- face, the results of laboratory experiments demon- strate that participants exhibit significantly higher levels of actual and perceived product knowledge, product attitude, and purchase intentions with a VR interface. The results also demonstrate that the type of a product (either VHE or VLE) moder- ates the effects of VR interfaces on consumer learning. Unexpectedly, the moderating effect of product type on consumers’ actual product knowl- edge, effected by VR interfaces, was not sup- ported at the conservative significance level of .0125. Nonetheless, the moderating effect seemed to occur because the enhancement of actual product knowledge achieved by the VR interface for VHE products (12.4 percent) was much greater than for VLE products (1.6 percent). This suggests a potential limitation to our findings, which is discussed further below. The results of separate t-tests contradicted our expectation that VR does not improve participants’ actual and per- ceived product knowledge and product attitudes for VLE products. These results can be explained by increases in flow, which highly engages and motivates users in computer-mediated environ- ments, thereby enhancing learning and positive experiences with products when a technology is vivid and interactive (Hoffman and Novak 1996).
Several limitations should be considered when interpreting the results of the present study. First, the instrument for assessing actual product knowl- edge does not seem sensitive enough to detect the apparent moderation effect of product types. The fact that we paid careful attention to the development of the instrument to ensure it did not favor any type of products or interfaces might result in less sensitivity in detecting differences in results.
Next, we selected products whose salient attri- butes are perceived through their visual aspects rather than their auditory aspects and hence the
experiment did not contemplate all sensory stimuli that VR can reasonably present. However, visual stimuli are the main sensory cues in producing virtual experiences (Pimentel and Teixeira 1994, p. 146). Participants reported more positive results for the VHE products, even with the absence of auditory cues that might have expanded their learning about the products. These points suggest that the focus on visual cues in the product choice did not significantly inhibit the moderation effects of product types from occurring.
Another limitation arises due to the use of student participants. However, because the present study addressed an individual decision-making situation in B2C e-commerce with which 87 percent of the participants had previous experiences and em- ployed tasks that were familiar to students, the use of student participants does not present a signi- ficant threat to validity (McKnight et al. 2002). Finally, this study did not demonstrate tangible benefits of VR in relation to its costs. Increases in purchase intentions do not always result in corre- sponding increases in actual purchases. It is difficult to estimate what increase in purchase intentions will be sufficient to compensate busi- nesses for the costs involved in providing VR interfaces.
This study has contributed to both theory and practice. First, by conducting a laboratory experi- ment with control over potentially extraneous variables, it empirically tested the impact of VR on consumer learning. Although VR has been avail- able since the 1970s in a number of areas, such as B2C e-commerce, architecture, education, medicine, and computer-supported collaborative work, the impact of VR use has seldom been explored in the IS field (Walsh and Pawlowski 2002). Second, we have proposed two types of online products that incorporate considerations of salient product attributes into VR technology. These may be useful in future VR research re- garding Web-based stores. Simultaneously, this research provides empirical support for the theory of cognitive fit inasmuch as the impact of IT is limited depending on whether or not a particular IT application, such as VR, can support the require- ments necessary to complete a given task (Agarwal et al. 1999, Agarwal et al. 1996).
Suh & Lee/Effects of Virtual Reality on Consumer Learning
692 MIS Quarterly Vol. 29 No. 4/December 2005
The current study provides useful guidelines for implementing VR interfaces. If Web-based stores want to enhance consumer learning with less con- cern for the costs necessary to achieve this goal, they can achieve it by adopting VR for all products. However, because implementing VR is still expen- sive and more labor-intensive than other tech- nologies (e.g., still pictures and 2D animation), it might not be worthwhile for e-commerce sites dealing in VLE products to introduce VR because the costs may exceed the benefits. Two sugges- tions might assist Web retailers seeking to enjoy the benefits of VR interfaces while minimizing the costs of development. First, a vendor can use VR to represent only those VHE products that they want to highlight in a Web-based store. By con- trasting products a retailer particularly wants to sell against other products, the retailers may induce the sales of the product as intended. Alternatively, a vendor might display selected attributes (i.e., those that are virtually experiential) only through VR, while displaying other qualities in static modes.
Several promising avenues for future research arise. First, researchers might consider investi- gating the effects of VR with regard to products with salient auditory aspects along with visual aspects because these two elements reflect VHE and VLE products more adequately than the focus on visual stimuli in the current study.
Furthermore, longitudinal research into potential increases in the endurance of consumer learning through VR interfaces would be another interesting avenue to explore. We conducted additional tests to assess the persistence of the interface effect and found interesting results that may guide future research. We asked participants which product they “remember most” (cognitive dimension) and which product they “want to buy most” (conative dimension) a week after the experiment. Approxi- mately 89 percent of the previous participants appeared for the data collection. For VHE pro- ducts, 76 percent of returning participants remembered the product represented in the VR interface most clearly, while only 24 percent of
participants remembered the product in the static interface more clearly. In addition, 71 percent of participants most wanted to buy a product that had been represented in VR, while only 29 percent preferred the products in the static interface. A sample test revealed that these differences of 52 percent for cognition (Z = 3.24, p = .0012) and 42 percent for conation (Z = 2.60, p = .0094) were statistically significant. However, the differences were relatively small (31 percent for cognition and 21 percent for conation) for VLE products and statistically insignificant at the .05 level. Thus, the effects of VR interfaces seem to last even after a week, while the strength of persistence is less for VLE products.
As e-commerce becomes more pervasive, advanced Web technologies, including VR inter- faces, will be more widely adopted for Web-based stores. We need to understand whether and under what conditions VR has positive effects on con- sumers learning about products. The present study implies that acceptance of top-of-the-line Web technologies would not be a panacea for all circumstances in e-commerce contexts; rather, the acceptance of a suitable Web technology that supports the salient characteristics of products is crucial.
Acknowledgements
The authors would like to thank Dr. Hyun Suk Kim and Mr. Jeahong Ahn at 3DIGM Co. Ltd. for imple- menting a virtual reality interface, Dr. Byoung Seon Choi and Ms. Eun Yi Chung at Yonsei University for their statistics advice, Ms. Sunhye Chang and Mr. Jung Wook Han for their assistance with the experiment, and YISRI at Yonsei Business Research Institute for providing research facilities. Special thanks to Dr. Izak Benbasat, and to the senior editor, the associate editor, and three anonymous reviewers for their valuable comments. This work was supported by a Korea Research Foundation Grant (KRF-2002-041-B00232).
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 693
References
Agarwal, R., Prabuddha, D., and Sinha, A. P. “Comprehending Object and Process Models: An Empirical Study,” IEEE Transactions of Software Engineering (25:4), July/August 1999, pp. 541-556.
Agarwal, R., Sinha, A., and Tanniru, M. “Cognitive Fit in Requirements Engineering: A Study of Object and Process Model,” Journal of Manage- ment Information Systems (13:2), 1996, pp. 137-162.
Arnould, E. J., Price, L., and Zinkhan, G. Con- sumers, McGraw-Hill, New York, 2001.
Bearden, W. O., Lichtenstein, D. R, and Teel, J. E. “Comparison Price, Coupon, and Brand Effects on Consumer Reactions to Retail Newspaper Advertisements,” Journal of Retailing (60:2), 1984, pp. 11-34.
Benbasat, I., and Zmud, R. W. “The Identity Crisis Within the IS Discipline: Defining and Com- municating the Discipline’s Core Properties,” MIS Quarterly (27:2), June 2003, pp. 183-194.
Bettman, J. A., and Park, W. C. “Effects of Prior Knowledge and Experience and Phase of the Choice Process on Consumer Decision Making Processes: A Protocol Analysis,” Journal of Consumer Research (7), December 1980, pp. 234-248.
Biocca, F. “Cyborg's Dilemma: Progressive Embodiment in Virtual Environments,” Journal of Computer Mediated-Communication (3:2), 1997(available online at http://www.ascusc.org/ jcmc/vol3/issue2/biocca2.html).
Box, G. E. P., Jenkins G. M., and Reinsel, G. C. Time Series Analysis: Forecasting and Control (3rd ed.), Prentice Hall, Englewood Cliffs, NJ, 1994.
Bruner, G. C. “Standardization and Justification: Do Ad Scales Measure Up?,” Journal of Current Issues and Research in Advertising (20:1), 1998, pp. 1-18.
Fishbein, M., and Ajzen, I. Belief, Attitude, Inten- tion and Behavior: An Introduction to Theory and Research, Addison-Wesley, Reading, MA, 1975.
Gefen, D., and Straub, D. W. “Gender Differences in the Perception and Use of E-Mail: An Exten- sion to the Technology Acceptance Model,” MIS Quarterly (21:4), December 1997, pp. 389-400.
Gemino, A. Empirical Comparison of System Analysis Modeling Techniques, Unpublished Ph.D. Dissertation, The University of British Columbia, Vancouver, Canada, 1999.
Gigante, M. A. “Virtual Reality: Enabling Tech- nologies,” in Virtual Reality Systems, R. A. Earneshaw, M. A. Gigante, and H. Jones (Eds.), Academy Press, London, 1993, pp. 15- 22.
Goodhue, D. L., and Thompson, R. L. “Task- Technology Fit and Individual Performance,” MIS Quarterly (19:2), June 1995, pp. 213-236.
Gravetter F. J., and Wallnau, L. B. Statistics for the Behavioral Sciences: A First Course for Students of Psychology and Education (5th ed.), Wadsworth, Belmont, CA, 2000.
Hoch, S. J., and Deighton, J. “Managing What Consumers Learn from Experience,” Journal of Marketing (53), April 1989, pp. 1-20.
Hoffman, D. L., and Novak, T. P. “Marketing in Hypermedia Computer-Mediated Environments: Conceptual Foundations,” Journal of Marketing (60), July 1996, pp. 50-68.
Hutchinson, J. W., and Alba, J. W. “Ignoring Irrelevant Information: Situational Determinants of Consumer Learning,” Journal of Consumer Research (18), December 1991, pp. 325-345.
Jarvenpaa, S., and Machesky, J. “Data Analysis and Learning: An Experimental Study of Data Modeling Tools,” International Journal of Man- Machine Studies (31), 1989, pp. 369-391.
Kempf, D. S., and Smith, R. E. “Consumer Pro- cessing of Product Trial and Influence of Prior Advertising: A Structural Modeling Approach,” Journal of Marketing Research (35:3), August 1998, pp. 325-338.
Kim, T., and Biocca, F. “Telepresence via Tele- vision: Two Dimensions of Telepresence May Have Different Connections to Memory and Persuasion,” Journal of Computer-Mediated Communications (3:2), September 1997 (avail- able online at http://www.ascusc.org/jcmc/vol3/ issue2/kim.html).
Klein, L. R. “Creating Virtual Product Experiences: The Role of Telepresence,” Journal of Inter- active Marketing (7:1), 2003, pp. 41-55.
Lavidge, R. J., and Steiner, G. A. “A Model for Predictive Measurements of Advertising Effec- tiveness,” Journal of Marketing (25), October 1961, pp. 59-62.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
694 MIS Quarterly Vol. 29 No. 4/December 2005
Levin, A. M., Levin, I. P., and Heath, C. E. “Pro- duct Category Dependent Consumer Prefer- ences for Online and Offline Shopping Features and Their Influence on Multi-Channel Retail Alliances,” Journal of Electronic Commerce Research (4:3), 2003, pp. 85-93.
Li, H., Daugherty, T., and Biocca, F. “Charac- teristics of Virtual Experience in Electronic Commerce: A Protocol Analysis,” Journal of Interactive Marketing (15:3), Summer 2001, pp. 13-30.
Li, H., Daugherty, T., and Biocca, F. “Impact of 3- D Advertising on Product Knowledge, Brand Attitude, and Purchase Intention: The Medi- ating Role of Presence,” Journal of Advertising (31:3), Fall 2002, pp. 43-58.
Li, H., Daugherty, T., and Biocca, F. “The Role of Virtual Experience in Consumer Learning,” Journal of Consumer Psychology (13:4), 2003, pp. 395-408.
Lutz, R. J. “Changing Brand Attitudes through Modification of Cognitive Structure,” Journal of Consumer Research (1), March 1975, pp. 49- 59.
McCabe, D. B., and Nowlis, S. M. “Information Integration across Online and Offline Shopping Environments and Its Effects on Consumers’ Purchase Decisions,” in Proceedings of the Experiential E-Commerce Conference, Frank Biocca (Ed.), East Lansing, MI, 2001.
MacInnis, D. J., and Jaworski, B. J. “Information Processing from Advertisements: Toward an Integrative Framework,” Journal of Marketing (53), October 1989, pp. 1-23.
McKenzie, S. B., Scott, B., and Lutz, R. J. “An Empirical Examination of the Structural Ante- cedents of Attitude toward the Ad in an Adver- tising Pretesting Context,” Journal of Marketing (53), April 1989, pp. 48-65.
McKnight, D. H., Choudhury, V., and Kacmar, C. “Developing and Validating Trust Measures for e-Commerce: An Integrative Typology,” Infor- mation Systems Research (13:3), September 2002, pp. 334-359.
Mills, S., and Noyes, J. “Virtual Reality: An Over- view of User-Related Design Issues, Revised Paper for Special Issue on Virtual Reality: User Issues in Interacting with Computers,” Inter- acting with Computers (11), 1999, pp. 375-386.
Nelson, P. “The Economic Consequences of Advertising,” Journal of Business (48:2), 1975, pp. 213-241.
Palmer, J. W., and Griffith, D. A. “An Emerging Model of Web Site Design for Marketing,” Com- munications of the ACM (41:3), 1998, pp. 44- 51.
Pimentel, K., and Teixeira, K. Virtual Reality: Through the New Looking Glass (2nd ed.), Intel/McGraw-Hill, New York, 1994.
Reeves, B., and Nass, C. The Media Equation: How People Treat Computers, Televisions, and New Media Like Real People and Places, Cambridge University Press, Cambridge, England, 1996.
Saaty, L. The Analytical hierarchy Process, McGraw-Hill, New York, 1980.
Schiffman, H. R. Sensation and Perception: An Integrated Approach (3rd ed.), John Wiley & Sons, New York, 1990.
Singleton Jr., R. A., and Straits, B. C. Approaches to Social Research (3rd ed.), Oxford University Press, New York, 1999.
Smith and Park 1992 Smith, R. E., and Swinyard, W. R. “Information
Response Models: An Integrated Approach,” Journal of Marketing (26), Winter 1982, pp. 81- 93.
Smith, R. E., and Swinyard, W. R. “Cognitive Response to Advertising and Trial: Belief Strength, Belief Confidence and Product Curiosity,” Journal of Advertising (17:3), 1988, pp. 3-14.
Sternthal, B., and Craig, C. S., Consumer Behavior: An Information Processing Perspec- tive, Prentice-Hall, Englewood Cliffs, NJ, 1982.
Steuer, J. “Defining Virtual Reality: Dimensions Determining Telepresence,” Journal of Com- munication (42:4), 1992, pp. 73-93.
Taylor, S., and Todd, P. “Assessing IT Usage: The Role of Prior Experience,” MIS Quarterly (19:4) 1995, pp 561-570.
Vessey, I. “Cognitive Fit: A Theory-Based Analy- sis of the Graphs Versus Tables Literature,” Decision Sciences (22), 1991, pp. 219-240.
Walsh, K. R., and Pawlowski, S. D. “Virtual Reality: A Technology in Need of IS Research,” Communications of the Association for Infor- mation Systems (8), 2002, pp. 297- 313.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 695
Wexelblat, A. Virtual Reality Application and Exploration, Academic Press Professional, Boston, 1993, pp. 3-44.
Wright, A. A., and Lynch, J. “Communication Effects of Advertising Versus Direct Experience When Both Search and Experience Attributes Are Present,” Journal of Consumer Research (21), 1995, pp. 708-718.
About the Authors
Kil-Soo Suh is a professor of Information Systems at Yonsei University, Seoul, Korea. He received his Ph.D. in Management Information Systems from Indiana University. His research interests are in the areas of interface design for electronic commerce, communication media, conceptual modeling, and implementation of information. His
work has been published in Decision Support Systems, Electronic Commerce Research and Applications, Information & Management, Informa- tion Systems Research, Journal of Global Informa- tion Management, and many Korean journals. He is a member of the editorial board for Information & Management.
Young Eun Lee is a Ph.D. candidate in Manage- ment Information Systems at the Sauder School of Business, University of British Columbia. She received her MBA and B.A. in Psychology at Yonsei University, Seoul, Korea. Her research focuses on human-computer interaction, stationary and mobile electronic commerce, and intelligent product recommendation agents. Her work has been published in International Journal of Elec- tronic Commerce and Communications of the ACM.
Suh & Lee/Effects of Virtual Reality on Consumer Learning
696 MIS Quarterly Vol. 29 No. 4/December 2005
Appendix A
A Sample Interface Design
Static Interface VR Interface
A Web-Based Store for VHE Products
Suh & Lee/Effects of Virtual Reality on Consumer Learning
MIS Quarterly Vol. 29 No. 4/December 2005 697
Static Interface VR Interface
A Web-Based Store for VLE Products