User Behaviour In Virtual Worlds

Published Date: 02 Nov 2017

As VWs constitute a promising new environment for e-business, it is important to better understand the profile of users that visit these worlds.

According to the results presented in Table 3.8, 83.3% of the "Social Communication" users (group 1) visit VWs at least once a week. The corresponding percentage is greater (92.5%) for "E-Commerce" users and for the third group (76.5%).

The following Table (Table 3.9) highlights the social aspect of VWs. It is noteworthy that the percentage of users of group #1 (13.3%) and #3 (5.9%) that do not visit other social web sites is greater than that of "E-Commerce" users (2.5%).

As part of our study of VW user profiles, we also investigated how the users first learned about the existence of VWs (Table 3.10). For the first group, most of the users (86.8%) learned about VWs from friends (offline and online) and through e-mails. The same applies to 75% of the respondents of the second group and 70.6% of the third group. It is notable that only 5.9% of the respondents of the third group were informed through scientific articles and journals, while 20% of "E-Commerce" users, randomly.

The majority of users, especially those of the first two groups, seem to embrace the idea that VWs are becoming an emerging alternative retail channel. Nevertheless, approximately one in four (26.5%) of the users of the third group do not (Table 3.11).

Looking further into the perception of VWs as an e-business outlet, it was considered important to investigate what types of stores or business users visit in VWs. As users had the ability of choosing more than one option, Table 3.12 depicts the percentages of users that chose only one option and the percentage of users that chose more than one option (combination of answers). The findings show that 33.3%, 15% and 20.1% of the users within each group respectively (i.e., for groups 1,2 and 3), visit apparel stores only. However, the frequency that the second "E-Commerce" users group visit a combination of the stores, is greater (77.5% ) to that of the first "Social Communication" users group. It should be noted that no respondent (in any of the groups) visits grocery stores in VWs, either as a sole option, or in combination with other stores.

The amount of money that people spend in transactions in VWs is depicted in the following Table (Table 3.13). The first and the third group have been excluded as users that belong to these groups do not make any economic transactions in VW.

Approximately, half of the users/shoppers (47.5%) spend up to thirty Euros per month, while (17.5%) spend more than a hundred.

Having reviewed the characteristics of the research sample which provides a better understanding of VWs’ use and preferences, the next section proceeds to discuss the factors characterizing store atmosphere and driving store selection in VWs.

3.6. EXPLORATORY STUDY: FACTOR ANALYSIS OF STORE ATMOSPHERE DETERMINANTS & STORE SELECTION CRITERIA

An exploratory factor analysis research was considered as the most appropriate approach to address the aforementioned goals outlined in the research framework. (i.e., section 3.2.). The electronic questionnaire that was developed served as the data collection instrument of this study.

3.6.1. Store Atmosphere Determinants: Layout as a Distinct Dimension

Table 3.14 displays the results of factor analysis for store atmosphere determinants. A minimum of five subjects per variable is required for factor analysis (Malhotra and Birks 2000). This requirement is fully met in the case of this research that involves 9 variables and 104 subjects. Tests of normality (Kolmogorov-Smirnov and Shapiro-Wilk) and linearity support the appropriateness of the factor analytical model. Furthermore, the several sizable correlations resulted from the correlation matrix, imply that the matrix is appropriate for factor analysis (Hair et al. 2006). Also, multicollinearity and singularity were conducted to check if any of the squared multiple correlations are near or equal to one. Finally, Bartlett’s test of sphericity (Approx. Chi- Square 138.716, df 36.000, Sig 0.000) and Kaiser-Mayer-Olkin measure (0.643) were conducted in order to prove the appropriateness of the model (Coakes et al. 2009).

Table 3.14 displays the three factors that were extracted. Storefront, store theatrics, colors, music and graphics were grouped in one factor (Factor #1). This factor is labeled Store Appeal because all these attributes are related to the "artistic" part of a store (e.g., the store as a theater), the way the aesthetics of the store are perceived by customers. Crowding, product display techniques and innovative store atmosphere services were grouped in a second factor (Factor #2). This factor is labeled Innovative Atmosphere; these elements are directly related to the innovative aspect offered by VREs in the sense that 3D technology provides such capabilities for displaying products, providing services and manipulating crowding that are new to the world of retailing. Also, innovative product display techniques (this is actually a core retail service) guide avatars’ navigational behavior within the store and, therefore, affect the crowding dimension. Finally, Store Layout constitutes the only attribute included in Factor #3. This finding highlights the importance of this graphical user interface dimension as a major consumer influencing factor in V-Commerce, in the sense that consumers perceive it as a selection criterion that is not related to others. Therefore, this factor should be investigated on its own, similarly to the relevant research practice. This finding confirms the available knowledge on the topic of online store layout effects on consumer behaviour in the context of multichannel retailing (Baker et al. 1994; Burke 2002; Grewal and Baker 1994; Griffith 2005; Lohse and Spiller 1999; Merrilees and Miller 2001; Simonson 1999).

Furthermore, it should be noted that all factor scores indicate that consumers attach significant importance to them when they select a V-Commerce store to conduct purchases (Average scores: Factor 1: 3.42, Factor 2: 3.88, Factor 3: 3.84). This finding is consistent with an earlier study in Web retailing by Vrechopoulos et al. (2001). Specifically, that study found that consumers attach high importance to store atmosphere variables when they select a Web based retail store to conduct their purchases. It also reported that the score consumers attached to the importance of store selection criteria is higher for potential shoppers compared to the current ones. This finding was attributed to the various concerns (e.g., security, effectiveness, etc.) that a shopper has when he/she uses a new retail channel to conduct purchases. Similarly, since the percentage of current V-Commerce shoppers is lower that the potential ones it is expected to obtain such high average scores for the store selection criteria. In other words, consumers that plan to adopt a new shopping channel, compared to the current ones, usually attach higher importance to the majority of the potential criteria in order to select a particular store (Vrechopoulos et al. 2001).

Finally, it should be underlined that the resulted factors’ content (i.e., variables) differ from earlier research on both conventional and traditional web retailing, implying that VWs’ visitors perceive them differently. Thus, factor analysis results do not confirm established knowledge; this finding, along with the implications of all findings of this empirical research are discussed extensively later on in this chapter.

3.6.2. Store Selection Criteria

The store selection criteria (i.e., store attributes) identified through the literature and discussed in chapter 2, were complemented by the responses of experts in the preliminary qualitative study, leading to the list of variables presented in the first column of Table 3.15. This constitutes a concise, rather than an exhaustive, list of store selection criteria in VWs. The selection of a concise list of criteria is a deliberate choice in the research design, because shopping through VWs is an emerging phenomenon.

Therefore, current or potential consumers may not be experienced enough to provide reliable answers when evaluating complicated and advanced VW store features. For example, the ability of flying with the avatar instead of walking in a 3D online environment is likely to be a feature that users are not familiar with. Factor analysis was used to examine the structure of interrelationships among variables, leading to a smaller set of underlying factors (Hair et al. 1995). The variables of Table 3.15 were grouped into four underlying factors for store selection in VWs.

The appropriateness of the model for factor analysis was thoroughly tested. First, the sample of 104 respondents exceeds the requirement of a minimum of five subjects per variable for factor analysis (Malhotra 2000). Furthermore, several variables were sufficiently correlated with each other. Also, multicollinearity and singularity were conducted to check if any of the squared multiple correlations are near or equal to one. Additionally, Bartlett’s test of sphericity (Approx. Chi- Square 215.389, df 66.000, Sig 0.000) and Kaiser-Mayer-Olkin measure (0.636) suggest that the data structure was adequate for factor analysis (Coakes et al. 2009). Principal components analysis and principal axis factoring are among the most commonly used methods for factor analysis, leading in most cases to the same results (Coakes et al. 2009). Principal axis factoring was adopted in the present study and the factors that extracted are based on the eigenvalue criterion (eigenvalues greater than 1 should be included in the model). After retrieving the number of factors, the varimax rotation procedure was adopted, that is an orthogonal procedure enabling the enhanced interpretability of the factors (Malhotra 2000). The results of the factor analysis are presented in Table 3.15.

Factor #1 had positive loadings on Variety of the Products, Quick Access and Easy Walking through the Virtual Store, Prices of the Products and Store Atmosphere. This factor is labeled "Core Store Features". Specifically, Variety of the Products enables "one-stop-shopping" and is preferred by consumers both offline and online, mainly due to time constraints. Quick Access and Easy Walking through the Virtual Store also constitutes a core store feature since it is related to ease of use and convenience. Prices of the Products constitutes a critical success factor for e-tailing due to the tremendous information search and evaluation of alternatives capabilities offered to the online users today. The importance of price is also strengthened by the current global economic climate. In sum, these three variables constitute core store features of a retail store in VWs, as they do in offline and online retail stores (Vrechopoulos 2005).

Conversely, Store Atmosphere, which constitutes a distinct factor in offline and 2D online retail stores (Vrechopoulos 2005) (i.e., usually, it is perceived differently by consumers), is identified in this research as one of the core store features for VW stores. This finding could be explained by the advanced graphic capabilities (i.e., 3D) that may be available in a VW retail store. This implies that VWs consumers hold high expectations for Store Atmosphere (perhaps due to their familiarity with online gaming 3D interfaces), similar to their expectations for reasonable prices, convenience and "one-stop-shop" capabilities. Therefore, consumers that select the virtual retailing shopping channel to conduct their purchases perceive Store Atmosphere in a similar fashion with the other three variables (Variety of the Products, Quick Access and Easy Walking through the virtual store and Prices of the Products). This finding is also supported by the fact that the average score of this factor was the highest one observed (the average score of responses was 4, in the five-point Likert scale used) compared to the remaining three factor scores.

Factor #2 has positive loadings on Quality of the Products, Store Reputation and Value Added Services and Customer Support. This factor is labeled "Peripheral Store Features". Specifically, while Quality of the Products is usually (in studies similar to the present one) grouped with product variety and price attributes, this was not the case in the present study. Also the average score of responses for this factor (3.44 in the five-point Likert scale) is lower than the corresponding scores of factors #1 and #3 in the total ranking. Probably, consumers believe that nowadays most of the products have reasonable quality and, therefore, price and variety are more important than quality. Also, the experience of shopping through Web 1.0 (i.e., the early stage of web stores with static pages where interactive or social features were lacking) contributed towards confronting any concerns regarding quality of products and services bought "from distance" (i.e., not through the physical store where consumers have more options than online for testing product quality). The same may stand for Store Reputation. Finally, Value Added Services and Customer Support could be also characterized as "peripheral" services because consumers are aware that such type of services may be offered online due to the combination of technological capabilities with low cost.

Factor #3 has positive loadings on Security and Privacy Protection, thus it has been labeled "Security and Privacy". This label highlights users’ concerns about issues such as security in transactions and privacy, as these first arose with the advent of the internet. Thus, this grouping was expected. However, while these attributes usually obtain the highest scores (see Vrechopoulos et al. 2009), in the case of the present study they were ranked as the second most important store dimension (i.e., average score of responses was 3.9 in the five-point Likert scale). This finding could be probably explained by the fact that VWs users are usually experienced Internet users and, therefore, are not as concerned about security issues or privacy protection as early shoppers in the Web 1.0 environment were.

Factor #4 has positive loadings on My Friends Visit the particular store, Quality of Advertising and Exhibitions and Entertaining Activities within the store. This factor, labeled "Social and Promotional Impulsion", characterizes people (or avatars!) that are extrovert and motivated by their friends or are looking for amusement and entertainment. Specifically, advertising, exhibitions and entertaining activities within the store constitute elements of the promotional mix. Also, the effects of friends constitute a promotional tool in the sense that these friends may operate as reference groups (e.g., opinion leaders) and, thus, companies invest in formulating their opinions and use them as promoters of their VW stores. This grouping implies that consumers perceive their friends’ influence (e.g., through online "word-of-mouth/mouse") as comparable to promotional effects. In other words, it appears that consumers perceive any type of promotional effect similarly. However, this factor obtained the lowest score compared to the other factors (the average score of responses was 3.37 in the five-point Likert scale). This is consistent with the findings of earlier studies exploring the influence of advertising and promotion on online or offline store selection criteria (see Vrechopoulos 2005).

3.7. CONCLUSIVE RESEARCH APPROACH

The second phase (Phase B) of the initial study follows a conclusive research design to investigate:

How do these store selection criteria influence the choice of users-consumers for visiting a virtual retail store?

Which are the differences per type of user-consumer (i.e., shoppers vs. non-shoppers) in terms of these store selection criteria?

How are the specific capabilities provided by VWs’ platforms perceived by users?

Which factors seem to influence sales of virtual retail stores?

3.7.1. Research Hypotheses Formulation of the Initial Study

There are several studies in the context of brick-and-mortar and web retailing addressing the different characteristics and behavioral patterns of shoppers (multichannel or not) and non-shoppers. They all seek to investigate various motivations for brick-and-mortar as well as web activity. Indicatively, in the context of web retailing, Vijayasarathy (2004) reported that users’ general acceptance of the internet affected their shopping behavior accordingly. Similarly, Farag et al. (2006) and Sorce et al. (2005) showed that demographics play an important role in the shopping adoption process, while Vrechopoulos et al. (2009) found significant differences between VWs’ retail store selection criteria in terms of the importance consumers attach to them.

Thus, it is assumed on the one hand that the store selection factors derived through the exploratory study in the previous section significantly differ in terms of the importance both shoppers and non-shoppers (i.e., the total sample of the study) attach to them. On the other hand, it is assumed that VW shoppers perceive the importance of store selection factors differently to VW users who are non-shoppers (hereafter called VW non-shoppers). In other words, it is important to investigate whether the differences observed between the four factors’ average scores have any statistical significant difference, as well as to investigate whether such potential differences (and/or ranking of importance) apply to both shoppers and non-shoppers. Thus:

Hypothesis 1(a): There are statistically significant differences in the importance that all VW users (shoppers and non-shoppers) attach to store selection factors (i.e., Factors 1, 2, 3 and 4).

Hypothesis 1(b): There are statistically significant differences in the importance that VW shoppers attach to store selection factors (i.e., Factors 1, 2, 3 and 4).

Hypothesis 1(c): There are statistically significant differences in the importance that VW non-shoppers attach to store selection factors (i.e., Factors 1, 2, 3 and 4).

In the same vein, it is important to compare shoppers and non-shoppers in terms of the importance they attach to each of the four factors separately, in order to investigate whether and why these groups exhibit different behavioral patterns and attitudes towards each of these factors. The results of such comparison can contribute to the design of targeted promotional and communication campaigns in the sense that a company could approach shoppers and non-shoppers differently, according to the importance they attach to different store selection criteria. Melancon (2011), based on the study of Yee (2006) who investigated the typology of users’ motivations in virtual environments, argues that information on the motivations of different groups of users is valuable for marketers wishing to enhance users’ experiences through branded policies. Jin (2009) argues that the majority of consumers are "inexperienced" shoppers in the context of VWs. Thus, investigating their attitudes towards VWs store selection criteria is important, especially because non-shoppers may visit VWs stores, search for and evaluate information, decide online and buy offline (or even buy online but through Web 1.0 online retail stores). Based on this discussion, the following research hypotheses are formulated in order to investigate the perceptions of different types of users (shoppers vs. non-shoppers):

Hypothesis 2: There are statistically significant differences in each store selection factor between VW shoppers and non-shoppers:

Hypothesis 2.1: There are statistical significant differences in Core Store Features between VW shoppers and non-shoppers

Hypothesis 2.2: There are statistical significant differences in Peripheral Store Features between VW shoppers and non-shoppers

Hypothesis 2.3: There are statistical significant differences in Security and Privacy between VW shoppers and non-shoppers

Hypothesis 2.4: There are statistical significant differences in Social and Promotional Impulsion between VW shoppers and non-shoppers

Focusing on the novel features of VWs, it is interesting to investigate whether these account for differences in perception between shoppers and non-shoppers. According to Hackbarth et al. (2003), shoppers are more likely to adopt and use a system than non-shoppers, as they spend more time exploring its capabilities. Also, computer anxiety is likely to create negative feelings in the direction of use (Venkatesh 2000). Computer anxiety is the notion or even the worry of an individual as far as the use of computers is concerned (Shen and Eder 2009). Webster et al. (1993) claim that if a computer task is too endeavoring, it will probably cause a negative effect on anxiety. In the same vein, Hoffman and Novak (1996) state that in a very demanding environment (e.g., with many options and buttons) users will consider that their capabilities are not enough to face environment requirements. Based on this claim, Shen and Eder (2009) investigated the factors that influence users to visit VWs for business purposes and concluded that computer anxiety does not influence the users’ perceived ease of use (PEOU) of the Second Life VW. However the results of their study imply that the difficulty or ease an individual faces with technology use, influence the use of Second Life respectively (Shen and Eder 2009). Specifically, in Second Life, the process of creating an avatar may not be a one-step process. Inexperienced users have to face issues such as creating (or buying) skin, clothes, body, face and shoes. Also, the directional buttons that can be used to direct an avatar in a virtual place can be time consuming, for users not familiar with teleporting and flying capabilities, when visiting a virtual mall. In sum, creating an avatar and navigating through VWs are considered as difficult in-world activities (Kaplan and Haenlein 2009). Therefore, it is anticipated that non-shoppers consider the processes of creating an avatar and walking around and visiting places in a virtual reality world more difficult than shoppers do. Thus:

Hypothesis 3: There are statistically significant differences between VW shoppers and non-shoppers in terms of their perceived difficulty in:

Hypothesis 3.1: Creating an avatar

Hypothesis 3.2: Walking around and visiting places in a virtual world

Users that visit VWs frequently are expected to be more experienced than those who are not frequent visitors and, in accordance with the findings of computer anxiety studies aforementioned, more likely to be engaged in shopping activities. At the same time, there are some economic, political, virtual experiences and regulatory issues in VWs that are similar to the physical world ones (Messinger et al. 2009). Indicatively, naturalness of virtual in-world activities may generate a familiar environment for visitors, creating or strengthening consumption of virtual or real products (Vrechopoulos et al. 2009). Herrington and Capella (1995) state that store design decisions relate to the time that customers spend shopping. Indeed, Eroglu et al. (2001) found that virtual store design influences the time that customers spend within a Web site. Similarly, van der Heijden (2000) and Li et al. 1999) state that Web site characteristics determine the duration of a Website visit. Moreover, the time spent shopping in a virtual store has proved to be an important factor that positively affects the amount of money spent in virtual environments (Shih 1998). In sum, several studies in the past (both offline and online) attempted to predict shopping behavior employing "sales" (or money spent) as the dependent variable in any given research design.

Bellman et al. (1999) note that "the most important information for predicting online shopping habits are measures of past behavior." Furthermore, they state that "looking for product information on the Internet is the most important predictor of online buying behavior" (p.35-38). As far as the context of VWs is concerned, Vrechopoulos et al. (2009) attempted to measure the predicting power of online activity related determinants (e.g., perceived usefulness, perceived ease of use and entertainment, time spent within the store, promotional sales and impulse purchases) on the overall evaluation of a virtual reality store layout, but did not find any significant relationships. However, according to Huang (2008) perceived ease of use is the strongest predictor of e-consumer attitudes followed by perceived usefulness, irritation and entertainment. In light of this earlier work and in order to investigate the determinants of shopping behavior in the Virtual Reality Retailing (VRR) environment further, the following hypothesis is posited:

Hypothesis 4: The amount of money spent in to the VRR environment is predicted by the:

Hypothesis 4.1: Frequency of visiting Virtual Worlds

Hypothesis 4.2: Perceived difficulty (vs. ease of use) of creating an avatar

Hypothesis 4.3: Perceived difficulty (vs. ease of use) of walking around and visiting places in a virtual world

Hypothesis 4.4: Perceived similarity between virtual and physical worlds

Hypothesis 4.5: The time spent in virtual worlds

Hypothesis 4.6: The average time spent in a store

It should be noted that all the research hypotheses articulated in this research are primarily supported by relevant literature regarding electronic retailing channels other than virtual reality retailing. This is justified by the great similarities that exist among these channels, but is mainly due to the fact that relevant literature in the context of VWs is scarce. Besides, supporting research hypotheses referring to new retail channels (e.g., VWs) by employing literature from other channels (e.g., Web 1.0) constitutes a common research practice, especially when research on the new channel is in its infancy. For example, Web 1.0 electronic retailing initial research attempts employed literature from conventional retailing to support the investigated research hypotheses.

3.7.2. Results of the Conclusive Research of the Initial Study

Hypotheses H1(a), H1(b) and H1(c) were tested through ANOVA with Post-Hoc comparisons. The necessary assumptions of population normality and homogeneity of variance are met in all three cases. Results are presented in Tables 3.16, 3.17, and 3.18, respectively.

The results of the importance that shoppers and non-shoppers of VWs (i.e., the total sample) attach to store selection criteria, as depicted in Table 3.16, indicate that there are statistically significant differences observed among the factor means (marked by the asterisk in the fourth column of the Table). The means of each factor depict that the whole sample attaches more importance to factor#1 ("Core Store Features", mean=4), and secondly to factor#3 ("Security and Privacy", mean=3.9), without, however, any significant difference observed between them. Also, both factors significantly differ from the other two (factor #2: "Peripheral Store Features", mean = 3.44, and factor #4: "Social and Promotional Impulsion", mean = 3.37), whereas no significant differences are observed between factors #2 and #4. These results imply that all respondents perceive "Core Store Features" and "Security and Privacy" as the most important selection criteria, of equal importance, when selecting a store within this virtual world. Thus, Hypothesis 1(a) is confirmed. However, it should be noted that all factors scored greater than 3.3, indicating that all factors are perceived as important by VW users.

Similarly, ANOVA was used to test whether there are statistically significant differences between VRR store selection criteria (i.e., Factors 1, 2, 3 and 4) in terms of the importance that VRR shoppers attach to them (Hypothesis 1(b) – see Table 3.17). The significant differences derived among factors are the same as in the whole sample (Hypothesis 1(a)) with slightly different scores observed in factors’ means (Factor#1= 3.9, Factor#3= 3.87, Factor#2= 3.54, Factor#4= 3.39). However, the ranking remains the same. Thus, hypothesis 1(b) is also confirmed.

The testing of Hypotheses 1(a) and 1(b) confirm the available knowledge as thoroughly discussed earlier in this chapter, especially as far as price, product variety, ease of use, security and privacy issues are concerned.

Testing Hypothesis 1(c) showed that results for VW non-shoppers are slightly different from the ones discussed above referring to VW shoppers (i.e., those of Hypotheses 1(a) and 1(b)). Specifically (Table 3.18), non-shoppers attach the highest importance to "Security and Privacy" (mean = 3.95) and then to "Core Store Features" (mean = 3.72). However, also in this case, there was no statistically significant difference observed between these factors, implying that these two factors are perceived similarly for non-shoppers and shoppers. The fact, however, that "Security and Privacy" scored slightly higher than "Core Store Features" could be explained by the fact that non-shoppers do not shop online (at least until now) because they may have some reservations about the security standards employed online and about the fair use of their data. One could also claim that non-shoppers are not experienced in buying through VWs (as shoppers are), and, therefore, they are more concerned about something that they have not done before. Furthermore, it should be noted that "Core Store Features" (i.e., prices, product variety, ease of use, etc.) do not significantly differ from "Social and Promotional Impulsion" (Factor #4 mean = 3.3). This finding could be explained by the fact that VWs’ non-shoppers spent their time in this world not for shopping, but mainly for social communication, entertainment and similar purposes. Therefore, they are used to enjoying such services and attach significance to them when selecting a virtual reality store. Finally, "Peripheral Store Features" (mean = 3.24) was found to be the least important factor. In sum, Hypothesis 1(c) is confirmed.

As discussed earlier, non-shoppers visit virtual reality stores in VWs in order to search for information, evaluate the alternatives, use customer services, etc. So, part of the non-shoppers’ decision making process may be conducted online while they purchase products and services offline (this is a common consumer behavioural practice applied in Web 1.0 retailing since the appearance of web based retail stores). Thus, non-shoppers are aware of VWs’ store features and use these to select which VWs’ stores to visit. This decision making process of non-shoppers makes them relevant for our research and this is why they are included in our sample and compared to VWs shoppers.

In order to test whether there are statistical significant differences between VRR shoppers and non-shoppers in terms of the importance they attach to each factor separately (i.e., Hypothesis 2) and in terms of their perceived difficulty regarding the processes of creating an avatar (Table 3.19) and walking around and visiting places in a virtual world (Table 3.20), t-Tests were conducted (Hypothesis 3). The output indicates that there are no significant differences observed among the mean values (p> 0.05) in both cases. Therefore, hypotheses 2 and 3 are not confirmed.

Specifically, the findings concerning hypothesis 2 imply that shoppers and non-shoppers attach the same importance to each factor, confirming indirectly the results of hypotheses 1(b) and 1(c) testing. As far as the results of hypothesis 3 testing are concerned, it is observed that non-shoppers are experienced enough to use VWs’ tools and services. Therefore, the fact that they do not shop online cannot be attributed to the difficulties they face in using and navigating through a VW.

Finally, in order to test H4, stepwise regression was adopted (Table 3.21). In this stepwise approach, each predictor variable enters or is excluded from the regression equation at a time (McIntyre et al. 1983). This procedure is used where the variables that explain most of the variation of the dependent variable need to be drawn from a large set of predictor variables (Malhotra 2000). In this initial research, this procedure was considered appropriate due to the exploratory nature of the independent variables that frame H4 (whereby not all variables are likely to be significant – see Malhotra 2000). The amount of money spent in a VW was used as the dependent variable, and the independent variables inserted in the model were: frequency of visiting VWs, perceived difficulty of creating an avatar, perceived difficulty of walking around and visiting places in a virtual world, perceived similarity between virtual and physical worlds, the time spent in virtual worlds and the average time spent in a store.

We can notice from the outcome (Table 3.21) that only the average time spent in the store and the frequency of visiting VWs, have been entered into the regression equation. These two variables explain 34.7% (R2) of the variability in the money spent in the store (F (2.68) = 18.042, p<0.5). The other variables failed to meet the selection criteria.

While the average time spent within the store is positively related to sales, it is surprising, and conflicting to earlier findings (as discussed in the formulation of the research hypotheses section), that frequency of visits is negatively related to money spent in VWs. This finding could be probably explained by the fact that shoppers that spend high amounts of money in VWs are mainly goal-oriented and, therefore, do not visit virtual reality stores often, but only when they want to accomplish a specific objective (i.e., find and directly buy a product or service). So, when they visit VWs’ stores they spend a considerable amount of money and time within the store in order to browse, compare, evaluate and buy their desired products or services. On the other hand, those shoppers that visit VW stores frequently seem to do that mainly for market research or entertaining purposes rather than shopping. However, this finding should be interpreted with caution since it is not in line with established knowledge (e.g., the Customer Relationship Management "Recency-Frequency-Monetary" metric) in the sense that frequency of visits, usually, is positively related to cross and up-sell (Strauss and Frost 2009). Thus, future research on this topic is highly recommended.

Therefore, Hypothesis 4 is confirmed as far as 4.1 (i.e., Frequency of visiting Virtual Worlds) and 4.6 (i.e., The average time spent in a store) are concerned. Hypotheses 4.2 (i.e., Perceived difficulty (vs. ease of use) of creating an avatar) and 4.3 (i.e., Perceived difficulty (vs. ease of use) of walking around and visiting places in a virtual world) were rejected, which indirectly confirm the findings of hypothesis 3 testing results, since difficulties of using and navigating through a VW did not appear to affect sales. Finally, as far as hypothesis 4.4 is concerned, it is clear that the perceived similarity between virtual and physical worlds does not have predicting power on sales since both shoppers and non-shoppers are experienced online users and, therefore, the amount of money they spend online is not determined by whether they perceive a VW as similar to the real world, but by other factors.

3.7. DISCUSSION OF THE INITIAL STUDY’S RESULTS

Virtual worlds constitute a relatively recent e-business context. Understanding it better would make a significant contribution to the study of contemporary e-business models. By exploring the profile and behavior of users in VWs, this initial research has contributed to this line of research.

One of the most important findings of the present initial study is the great amount of users that conduct e-commerce transactions in the "traditional" Web but don’t buy products over the internet in the context of VWs. While this merits further exploration, it can probably be explained either because users treat VWs as an entertaining or gaming oriented environment and not as a retailing channel, or because they are considered light users of VWs and are reluctant to commit to transactions in an environment that is deemed unstable.

This initial study explored store selection criteria in VWs stores. The empirical research led to the identification of four factors influencing store selection: Core Store Features, Peripheral Store Features, Security and Privacy, Social and Promotional Impulsion. The study outlined the importance of these factors for all respondents, and subsequently looked separately into the responses of shoppers in VWs and users of VWs who are non-shoppers.

The empirical results show that the "Core Store Features" factor, comprising of Variety of the Products, Quick Access and Easy Walking through the store, Prices of the Products and Store Atmosphere plays a major role for both shoppers and non-shoppers selecting to visit a virtual reality retail store. Other studies, as discussed in earlier sections, have shown that all these attributes are considered important in determining behavior both in traditional and web retailing. This study confirmed the presence and influence of these attributes in determining store selection criteria in the virtual reality retailing channel as well.

Also, the study demonstrated that users that have never conducted purchases through virtual reality environments are mostly concerned about security and privacy issues. This conclusion is in line with the extant related literature. A probable explanation is that VWs non-shoppers do not shop in online retailing (i.e., in Web 1.0) either. These users are, therefore, likely to exhibit a similar behavior in VWs, as most of security and privacy issues are similar to those of the Web 1.0 environment.

This study also showed that both the frequency of visiting VWs and the average time spent in a store directly predict the amount of money spent in virtual environments. Surprisingly however, as discussed in the previous section, the frequency of visits is negatively associated with sales. Another interesting finding is that perceived similarity between virtual and physical worlds does not seem to predict the amount of money spent; this is in line with the belief that traditional and virtual retailing channels are perceived quite differently by consumers despite the analogies between the two worlds.