An experimental investigation of the influence of virtual community characteristics on consumers' evaluations of an online sto

1

The effect of an integrated virtual community on the evaluation of an online store: Findings from an Internet experiment

We examined the impact of three characteristics of a virtual community within an online store on consumer evaluations of that store. In particular, we focused on the exertion of retailer influence, the quality of the virtual community and the degree of sociability. An Internet experiment with 477 participants using a professionally designed shopping Web site was conducted, confirming the influence of all three characteristics.

The effect of an integrated virtual community on the evaluation of an online store: Findings from an Internet experiment

Forrester Research estimates that 74% of Europe’s population will be online in 2011, and more than 176 million people will shop online (Dellner 2007). At the same time, the use of the web is changing dramatically. Applications that are summarized under the umbrella term “Web 2.0” used to refer to key concepts such as user empowerment and participation are changing the way consumers use the Web (O´Reily 2005). Implementing new strategies with regard to Web 2.0 will allow retailers to create competitive advantages. For example, a study by Brown, Tilton and Woodside (2002) showed that customers who participated in a virtual community (VC) provided by a Web site generated two-thirds of the sales of that site.

While research on the general concept of VCs is broad (Farquhar and Rowley 2006), there is still a lack of studies investigating its influence on consumer perceptions of the transactional and relational qualities of shopping Web sites. Furthermore, the effects of different VC characteristics are widely unknown. We therefore identified three VC characteristics as influencing factors of consumer evaluations based on prior literature and practical evidence, including the exertion of influence by the operator (i.e., the retailer), the quality of the VC and the degree of sociability. We investigated the effects of these determinants on consumer perceptions of the transactional and relational qualities of a shopping Web site by means of an online experiment.

Effects of virtual communities: Conceptual foundation and hypotheses

We conceptualize consumer evaluations using six constructs. The transactional dimension in our study includes Perceived Ease of Use, Perceived Usefulness (Davis 1989) and Perceived Risk (Mitchell 1999). To capture the relational perspective of consumer evaluations, we include three dimensions. Perceived Social Presence is defined as to extent to which a medium is perceived to be warm if it conveys a feeling of human contact and sensitivity to the user (Hassanein and Head 2004). Perceived Enjoyment refers to the extent to which the activity of using an information technology is perceived to be enjoyable (Davis, Bagozzi, and Warshaw 1992). Following McMillan and Hwang (2002), we conceptualize Perceived Interactivity as the extent to which users perceive the online store to enable (many-to-many) communication, to provide control (in terms of speed, navigation and content), and to respond to user requests.

The investigated determinants of consumer evaluations (i.e., VC characteristics) include the exertion of influence by the operator, the quality of the VC and the degree of sociability.

We hypothesize that the effect of the exertion of influence is nonlinear; that is, we expect a more positive evaluation of a moderate influence than that of either a high or a low exertion of influence (H1).

As mentioned by Preece (2001), the volume, quality and timeliness of the information provided in a VC are factors that affect member engagement in the community, implying that those factors have a strong impact on consumer evaluations of the VC. Referring to VC quality, we therefore hypothesize that consumers evaluate the online store more positively if an integrated VC is of high quality (H2).

The impersonality of online retailing is still one of the major problems facing this distribution channel, which may be resolved through the usage of Web 2.0 instruments. We thus focus on the concept of sociability, which refers to the extent of the support of social interaction online (Preece 2001), and we investigate the effect of introducing social cues to a VC. We assume that increasing sociability of the VC will lead to a more positive evaluation by visitors (H3).

Experimental Design

To test our hypotheses, a laboratory 3x2x2 (i.e., exertion of influence: low/moderate/high; VC quality: low/high; sociability: low/high) between-subject experiment with 477 participants was conducted. The VC included a discussion forum and a user profiles section. Manipulation of the exertion of influence was achieved by varying the frequency of administrator contributions. To manipulate VC quality, the number of contributions to the discussion forum, the comprehensiveness of content and the timeliness of contributions varied between groups. To manipulate the sociability of the VC, the user profiles section was either present or absent within the VC. To avoid participant awareness bias, we chose a non-forced exposure design and developed a cover story.

Major Results

The results of the 3x2x2 MANCOVA (with product involvement and internet experience as covariates) showed that the exertion of influence mainly affects transactional factors, especially the risk perceived by respondents. We showed that consumers tend to react positively to an increased exertion of influence by the retailer only to a certain point, beyond which an increasing influence leads to a negative effect. Moreover, we found a significant interaction between influence and sociability. Apparently, the presence of the user profiles section has some alleviative effect on the negative impact of the high exertion of influence.

The effect of VC quality was highly significant for the transactional constructs of perceived ease-of-use and perceived usefulness, indicating that comprehensiveness and timeliness of content are effective components of consumer transactional evaluations. With regard to relational components, the results showed that the use of a comprehensive and lively VC makes the task of shopping a more enjoyable experience. Further analysis confirmed our hypotheses of the effects of high versus low sociability of the VC on perceived risk, perceived social presence, perceived enjoyment and perceived interactivity. We found no support for the effect of sociability on perceived usefulness of the online store. Apparently, consumers in an online environment tend to disregard the importance of tie strength (i.e., in terms of similarity, expertise and accessibility) as posited by traditional word-of-mouth theory. With regard to relational factors increasing, the sociability of an integrated VC seems to be adequate to convey a feeling of human contact and warmth, thus increasing perceived social presence. Moreover, changing the nature of the online store by adding the user profiles section and increasing sociability was shown to have a positive effect on perceived enjoyment and perceived interactivity.

The effect of a virtual community on the evaluation of an online store: Findings from an Internet experiment

Forrester Research estimates that 74% of Europe’s population will be online in 2011, and more than 176 million people will shop online (Dellner 2007). In addition, the use of the Web is changing dramatically. Applications that are summarized under the umbrella term “Web 2.0” used to refer to key concepts such as user empowerment and participation are changing the way consumers use the Web (Cheung, Lee and Rabjohn 2008). Implementing new strategies with regard to Web 2.0 will allow retailers to create competitive advantages. For example, a study by Brown, Tilton and Woodside (2002) showed that customers who participated in a virtual community (VC) provided by a Web site only accounted for about one-third of the visitors but generated two-thirds of the sales. As such, assessing the potential market opportunities in utilizing a VC becomes critical (Kim and Jin 2006).

As the topic of VCs is still at the early stages of research, there is little consensus regarding the definition of a VC (Kim and Jin 2006). Preece (2001) and Preece and Maloney-Krichmar (2005) identified a participatory design and employed the concepts of usability and sociability as the most important elements of a VC. A comprehensive literature review by Lee, Vogel and Limayem (2003) resulted in the working definition that VCs are cyberspaces supported by computer-based information technology centered upon communication and interaction of participants to generate member-driven content, resulting in relationships. The implementation of a VC can combine different aspects with various focuses, such as communities of transaction or communities of interest as described by Armstrong and Hagel III (1996); therefore, VCs can fulfill various types of human needs.

While research on the general concept of VCs is broad (Farquhar and Rowley 2006) and the implementation of a VC is acknowledged as an influential feature of business-to-consumer (B2C) Web sites, there is still a lack of studies investigating the influence of VCs on consumer perceptions of the transactional and relational qualities of a shopping Web site. Only a few exploratory studies have reported on the usefulness of a VC in an e-commerce context (Gupta and Kim 2004). Furthermore, the effects of different VC characteristics on consumer evaluations are widely unknown. We therefore identified three VC characteristics as influencing factors of consumer evaluations from prior literature and practical evidence. These include the exertion of influence by the operator (i.e., the retailer), the quality of the VC and the degree of sociability. We investigate the effects of these determinants on consumer perceptions of the transactional and relational qualities of a shopping Web site by means of an online experiment.

Effects of virtual communities ON CONSUMER EVALUATIONS

We expect that the integration of a VC within the experimental shopping Web site to have a positive effect on consumer evaluations by improving their shopping experience and emotional reactions when visiting the online store. Past research has examined facets of consumer evaluation and acceptance of an information technology from two different perspectives (Dahui, Browne, and Wetherbe 2006). The first perspective, often called the transactional perspective, views a Web site as an information technology and thus focuses on cognitive factors such as perceived usefulness or perceived ease of use. A second perspective refers to the affective or relational components of a Web site such as perceived enjoyment. Recent literature has emphasized the importance of investigating the driving factors of Web site usage beyond cognitive factors, especially because of the changing nature of the Web (Loiacono, Watson, and Goodhue 2007).

Following the research that integrates the transactional and relational perspectives (Gefen and Straub 2004; Moon and Kim 2001), we conceptualize consumer evaluations using six constructs. The transactional perspective in our study includes the key antecedents of the intention to use a technology identified by the Technology Acceptance Model (TAM) (Davis 1989), namely, Perceived Ease of Use (PEOU) and Perceived Usefulness (PU). As originally defined by Davis (1989, 320), PEOU refers to “the degree to which the prospective user expects the target system to be free of effort”. With regard to e-commerce, PU of a Web site is “the extent to which an individual perceives a web-site to be useful in performing shopping tasks” (Kumar and Benbasat 2006, 428). We add Perceived Risk (PRisk) following Mitchell (1999) to comprise two components, namely, the perceived probability of a loss and the subjective evaluation of potential unfavorable consequences.

To capture the relational perspective of consumer evaluations, we consider three dimensions. Perceived Social Presence (PSP) is defined as to extent to which a medium is perceived to be warm if it conveys a feeling of human contact and sensitivity to the user (Hassanein and Head 2004). Perceived Enjoyment (PEn) refers to the extent to which the activity of using an information technology is perceived to be enjoyable, aside from performance consequences (Davis, Bagozzi, and Warshaw 1992). Following McMillan and Hwang (2002), this study defines Perceived Interactivity (PInt) as the extent to which users perceive the online store to enable (many-to-many) communication, to provide control (i.e., in terms of speed, navigation and content), and to respond to user requests.

HYPOTHESES

The investigated determinants of consumer evaluations (i.e., VC characteristics) include the exertion of influence by the operator, the quality of the VC and the degree of sociability.

Concerning the exertion of influence on the VC by the operator, this can, on the one hand, be valued positively by members. For example, an operator can establish codes of behavior (i.e., netiquette or guidelines) to contain potential conflict (Leimeister, Sidiras, and Krcmar 2006). Andrews (2002) argued that operators must facilitate the formation of subgroups as well as reward and acknowledge member contributions. In these cases, a certain level of exertion of influence and contribution to the VC by the operator is needed to establish such structural patterns. The positive effect of a higher exertion of influence is supported by the literature on trust, which proposes that an organization must demonstrate its honesty, benevolence and competence to its members (Gefen 2000). In the context of a VC, the operator can show benevolence by being concerned for the welfare of the members of the VC; it can also demonstrate its competence if its contributions to the VC are significant and generate value (Flavián and Guinalíu 2006).

On the other hand, a very strong influence by the retailer will make consumers feel that the retailer tries to act in an opportunistic manner and exercises too much control (Flavián and Guinalíu 2006). The negative effect can be explained by the elaboration likelihood model (ELM) (Bhattacherjee and Sanford 2006) as a theoretical model of information adoption in computer-mediated communication (CMC) contexts. ELM posits that a message can influence people’s attitudes and behaviors with source credibility acting as the peripheral influence (Cheung, Lee, and Rabjohn 2008). Source credibility refers to a message recipient’s perception of the credibility of a message source; information provided by highly credible sources is perceived to be useful and reliable (Ko, Kirsch, and King 2005). In the context of CMC, user-generated content has found to be more credible than marketer-generated information (Bickart and Schindler 2001), which means that increasing the exertion of influence by the operator beyond a certain level will lead consumers to evaluate the credibility of information more negatively. Furthermore, reactance theory (Brehm 1966) implies that when people feel that their freedom to choose an action is threatened, they become motivated to reestablish the threatened freedom, which is the case in which content generated by the operator dominates the VC and consumers feel that the retailer acts in an opportunistic manner. Reactance theory also indicates that reactance must reach a certain intensity and thus surpass a threshold point for the perception of threat to take effect. These arguments indicate that a higher exertion of influence by the operator does not in every case lead to a more positive evaluation by the consumers and vice versa. Hence, we assume the effect of the exertion of influence to be nonlinear. This means that given a moderate exertion of influence and either a high or a low exertion, we expect a more positive evaluation from moderate exertion.

We propose that shifting intensity of the exertion of influence by the operator of the VC (i.e., the retailer) will have an effect on consumer PRisk because information from an independent third-party (e.g., other users) is perceived as more credible than information that comes directly from the retailer (Weathers, Sharma, and Wood 2007). We also propose an impact on consumer PU, since information provided by commercial sources is perceived as less useful, as well as on relational factors, that is, PEn and PInt. We expect the latter impact because too much influence by the operator hampers the sense of community among users and their perceived control over their community. Thus, we hypothesize:

H1.1: Compared to a low exertion of influence, a moderate exertion of influence on the VC by the retailer will lead to (a) higher PU, (b) lower PRisk, (c) higher PEn and (d) higher PInt.

H1.2: Compared to a high exertion of influence, a moderate exertion of influence on the VC by the retailer will lead to (a) higher PU, (b) lower PRisk, (c) higher PEn and (d) higher PInt.

As one of the most relevant factors of a VC, the literature highlights the depth, timeliness and quality of content (Farquhar and Rowley 2006). In a recent study by Leimeister, Sidiras and Krcmar (2006), content-related factors were ranked high in terms of relevance by VC members. As mentioned by Preece (2001), the volume, quality and timeliness of the information provided in a VC are factors that affect member engagement in the community, implying that those factors have a huge impact on the positive or negative evaluation by consumers of the VC. We therefore hypothesize a positive effect of high quality information on the perceived usefulness and ease of use. We also expect consumers to perceive the online store as more enjoyable if the depth, timeliness and quality of content are at high levels. As described by Koufaris (2002), product information searches can be fun-seeking experiences. We expect that the use of a comprehensive and lively virtual community can make the shopping experience more fulfilling and enjoyable. Based on our conceptualization of perceived interactivity (i.e., in terms of communication, control and response), we assume that the quality of the VC will affect the perceived possibility of communication as well as the perceived control over navigation and content. Referring to VC quality, we therefore hypothesize:

H2: If the integrated VC is of high quality, consumers evaluate the online store more positively as compared to an online store in which the integrated VC is of low quality. In particular, they rate the online store more positively with respect to (a) PEOU, (b) PU, (c) PEn and (d) PInt.

The impersonality of online retailing is still one of the major problems facing this distribution channel, which may be solved by the usage of Web 2.0 instruments. For example, a VC may help substitute for the personal treatment a consumer is given in bricks-and-mortar stores (Flavián and Guinalíu 2005). The term sociability relates to the development of software, policies and practices to support social interaction online (Preece 2001). One important determinant of success concerning sociability is the provision of information about community members, including, for example, what roles they are taking or how experienced they are (Preece 2001). We therefore assume that increasing sociability by adding more personal information on users will lead to a higher PSP.

The traditional concept of word-of-mouth as relating to personal sources of information takes the relational closeness between a decision maker and a recommendation source as its basis (Brown and Reingen 1987). With the advent of Web 2.0 and the integration of personal information on users within user-generated content, there is the possibility to implement such strong ties in the form of electronic word-of-mouth (eWOM). This can help to create a higher PU of online information and a lower PRisk in the buying decision.

The theory behind the effect of sociability on perceived enjoyment is based on literature that classifies information systems as hedonic (Heijden 2004). We act on the assumption that adding more social cues to the VC will move the nature of the information system toward a hedonic information system, which will be perceived as more enjoyable. Finally, based on the conceptualization of perceived interactivity, we expect a strong impact of sociability on the perceived possibilities to communicate with others:

H3: If the integrated VC provides a higher degree of sociability, then consumers evaluate the online store more positively as compared to an online store in which the integrated VC provides a lower degree of sociability. In particular, they rate the online store more positively with respect to (a) PU, (b) PRisk, (c) PSP, (d) PEn and (e) PInt.

METHod

Experimental Design

To test our hypotheses, we used a professionally designed fictitious online-shop. We chose to integrate two different areas into the VC. First, there was an area with a discussion forum, which has been identified as a popular feature of VC Web sites (Pitta and Fowler 2005). Second, in order to address the aspects of a community of relationships and to manipulate sociability in the experiment, a user profiles section was added. Some screenshots illustrate the Web site and the VC (figure 1). All content within the online store and the VC has been sampled and adapted from real stores, discussion forums and virtual communities on the Web.

Figure 1

Sample screenshots

Product page with hyperlink to the VCStart page of the VC

Example from the user profile sectionDiscussion forum overview page

A laboratory 3x2x2 between-subject experiment was conducted. Given the need to control for shopping Web site and VC features, this design allowed us to isolate the effects of the different VC aspects on consumer responses.

Exertion of influence. Manipulation of factor 1 (exertion of influence: low/moderate/high) was achieved by varying the number of forum posts by an administrator. Note that the name and picture were chosen in a way that the association between the administrator and the retailer was obvious. In conditions with a low exertion of influence, the retailer did not appear in any forum post, whereas the discussion forum contained posts by the retailer in 50% of threads for the moderate condition, and posts appeared in every thread for the high condition.

VC quality. The quality of information is evaluated in terms of information content, accuracy, format and timeliness (Cheung, Lee, and Rabjohn 2008). In our study, we focused on comprehensiveness as well as timeliness as grouped under the umbrella term “virtual community quality”. To manipulate the virtual community quality (low/high), we varied the number of contributions to the discussion forum, the comprehensiveness of the content and the timeliness of contributions from group to group. The low quality setting contained very few threads with few posts within the threads; all entries were at least eight months old. We also manipulated the displayed number of views on the forum overview page.

Sociability. To manipulate the sociability (low/high) of the VC, a user profiles section was either present or absent within the VC. Depending on the experimental condition, the community start page contained a section to enter only the section with the discussion forum (conditions 1-6) or the discussion forum and the user profiles (conditions 7-12) or. Additionally, respondents who were assigned to conditions 7-12 also had the ability to follow hyperlinks on user aliases in the discussion forum. In the user profiles section, respondents had different ways of accessing the individual profiles (e.g., only the newest profiles or the full list of profiles could be viewed) and information on the profiled person, including profile picture, name, age, interests and his or her personal guestbook. All profiles were checked for plausibility (e.g., fit between age, gender and interest) and conformity with the cover story (e.g., language usage) in a pretest.

Table 1

Experimental conditions

Condition

Treatments

Factor 1: exertion of influence (3)

Factor 2: VC quality (2)

Factor 3: sociability (2)

1

Low

Low

Low

2

Moderate

Low

Low

3

High

Low

Low

4

Low

High

Low

5

Moderate

High

Low

6

High

High

Low

7

Low

Low

High

8

Moderate

Low

High

9

High

Low

High

10

Low

High

High

11

Moderate

High

High

12

High

High

High

Procedure

To avoid participant awareness bias, we chose a non-forced exposure design and developed a cover story to avoid respondents guessing the actual purpose of the study and to induce the impression that they would be examining a portion of a real online store. After a short briefing on the experimental task (i.e., respondents were told to search for information about products in the online store and to choose one product), participants were forwarded to the overview page, which displayed the first three products, namely, mp3 audio players. From this page, participants were able to access the other overview pages as well as the individual product pages containing detailed information about the products. A button with a hyperlink to the VC (labeled “Find out more in our community”) was present on the page header as well as at the end of every product page. Respondents were free to visit the VC. Participants who followed this link were lead to the VC. On every page of the VC, a link was available to return to the online store and the product pages. Respondents who did not enter the VC were assigned to the control group. After putting the selected product in the shopping cart and confirming their decision to purchase, respondents were forwarded to the online questionnaire. We first surveyed the dependent variables and then the variables for manipulation checks in order to avoid respondents recognizing the actual purpose of the study before answering the questions on the dependent variables.

Measurement

The dependent variables in this study were captured by an online questionnaire assessing constructs on a multiple-item seven-point scale (1=strongly disagree, 7=strongly agree). All measurement scales were pretested. To assess PEOU, we chose five reflective items based on Davis (1989) and Gefen, Karahanna and Straub (2003). The reflective five-item-scale for measuring PU was adopted from past applications of the TAM (Agarwal and Karahanna 2000; Davis 1989; Gefen, Karahanna, and Straub 2003). Both measurement models showed a high level of scale reliability; Cronbach´s Alpha was .91 for PEOU and .93 for PU. We chose a reflective measurement approach for the dimensions of PSP (five indicators, Cronbach´s Alpha = .97), all of which were adapted from validated constructs developed by Cyr et al. (2007). To capture PEn, we followed the conceptualization of Hassanein and Head (2005), Moon and Kim (2001) and van der Heijden (2004) (four indicators, Cronbach´s Alpha = .92). Measures of sampling adequacy values (MSA) and factor loadings for the items of all scales exceeded .50; item-to-total correlations surpassed .40. As AVE for all reflective constructs mentioned above is greater than the squared correlation between that construct and any other construct, discriminant validity is confirmed.

PRisk was operationalized in a formative way to capture functional risk, financial risk as well as the risk of mispurchase (Cunningham 1967). None of the indicators revealed multicollinearity problems (all VIFs were well below 10; condition indices were all below 30). We also assessed nomological validity by analyzing bivariate correlations between the formative construct and a single item of trust, a construct that is closely related to PRisk and has been included as additional construct in the questionnaire. Bivariate correlations were strong and significant (r = .57, p < .001). The formative measure of PInt consists of a second-order formative factor composed of three components. Factor analysis of six items selected from past research on interactivity (Jee and Lee 2002) revealed three reflective components, namely, control, response and communication. Based on the theoretical work of McMillan and Hwang (2002), a formative construct of PInt was computed from these three dimensions. None of the indicators revealed multicollinearity problems. Nomological validity was assessed by measuring bivariate correlations between the formative construct of PInt and a single item of interactivity, which also was included in the questionnaire. Bivariate correlations were highly significant (r = .73, p < .001).

Sample and Manipulation Checks

The final dataset consists of 477 participants, of which 264 participants visited the VC and 213 did not. Respondents range in age from 16 to 65 years, with a mean age of 32.3 years. 50.3% of participants are female.

Manipulation checks were performed to control for the effectiveness of the treatments. To test for the perceived exertion of retailer influence on the VC, which was measured by a single item, an ANOVA analysis indicated that the three groups were significantly different (F(2, 263) = 45.516, p < .001). The results of post-hoc Scheffé tests confirmed that statistically significant differences existed between all groups. Manipulation of VC quality was checked with a variable that was characterized by three indicators, namely, the number of contributions to the message board (threads), the comprehensiveness of content (posts) and the timeliness of contributions. As intended, participants in the low quality sample perceived the VC as of lower quality (M = 2.64) than those in the high quality sample (M = 4.69, F(1, 263) = 221.235, p < .001). To ensure the successful manipulation of sociability, we removed those respondents from the dataset that were assigned to conditions including the presence of the user profiles section (high sociability) but reported that they did not see that particular part of the VC. Moreover, entering the VC at any point was observed by collecting participant identifications in a SQL database; respondents who actually visited the VC but reported to have not visited the VC were removed from the data set.

RESULTS

The data was analyzed using a 3x2x2 MANCOVA due to the orthogonal between-subject design. Because past research showed a strong influence of product involvement and Internet experience on consumer behavior in an e-commerce setting, we included both as covariates. We first tested the overall null hypothesis as a screening analysis. The Wilks Lambda multivariate tests of overall group differences were all statistically significant for the main effects (p < .001). Regarding interaction effects, exertion of influence x sociability was found to be significant (Wilks´ Lambda: Multivariate F = 1.899, p < .05). Referring to factor 1, univariate between-subject tests showed that the exertion of influence was significantly related to PU, PRisk and PInt.

Table 2

Univariate tests of hypotheses

Independent variables

Dependent variables

Sum of Squares

df

Mean Square

F

Sig.

Partial η2

Observed Power a

Factor 1: Exertion of influence

PU

21.353

2

10.677

10.136

.000***

.075

.985

PRisk

15.713

2

7.857

10.548

.000***

.078

.988

PEn

2.950

2

1.475

1.666

.191NS

.013

.349

PInt

3.933

2

1.966

5.626

.004**

.043

.857

Factor 2: VC quality

PEOU

26.382

1

26.382

21.745

.000***

.080

.996

PU

17.594

1

17.594

16.703

.000***

.063

.983

PEn

8.126

1

8.126

9.175

.003**

.035

.855

PInt

1.332

1

1.332

3.811

.052NS

.015

.494

Factor 3: Sociability

PU

3.755

1

3.755

3.564

.060NS

.014

.468

PRisk

16.559

1

16.559

22.232

.000***

.082

.997

PSP

62.814

1

62.814

65.585

.000***

.208

1.000

PEn

25.072

1

25.072

28.309

.000***

.102

1.000

PInt

5.692

1

5.692

16.285

.000***

.061

.980

Factor 1 * Factor 3

PRisk

4.755

2

2.377

3.192

.043*

.025

.607

a computed using alpha = .05; *** p < .001, ** p < .01, * p < .05, NS not significant

Contrast analysis reveals that statistically significant differences existed among all groups for PU (contrast estimate (c.e.) low vs. moderate = -.333 with p < .05; c.e. moderate vs. high = .700 with p < .001), PRisk (c.e. low vs. moderate = .297 with p < .05; c.e. moderate vs. high = -.601 with p < .001) and PInt (c.e. low vs. moderate = -.216 with p < .05; c.e. moderate vs. high = .289 with p < .01). All comparisons are in the expected directions. We therefore find support for hypotheses H1.1a, H1.1b, H1.1d, H1.2a, H1.2b and H1.2d. The results show that consumers tend to positively evaluate a moderate exertion of influence by the retailer, whereas a high influence leads to a negative effect on PU, PRisk and PInt. Moreover, respondents in the group with low exertion of influence also rated the online store to be less useful for shopping tasks and perceive a higher risk than respondents in the group with moderate exertion of influence. Furthermore, respondents evaluated the online store to be less interactive. As table 1 reveals, the exertion of influence does not have an effect on PEn, and therefore, hypotheses H1.1c and H1.2c must be rejected. Obviously, the exertion of influence mainly affects transactional factors, especially the risk perceived by respondents.

Because a significant interaction was found between the factors exertion of influence and sociability by multivariate testing, the main effects were qualified by this interaction. To further examine this interaction, we performed univariate tests and found the interaction to be significant only for PRisk. Even though the interaction effect size was small, apparently the presence of the user profiles section and the linking of aliases to profiles in the discussion forum have some alleviative effect on the negative impact of the high exertion of influence (figure 2).

Figure 2

Joint effect of exertion of influence and sociability on PRisk

-0,75

-0,5

-0,25

0

0,25

0,5

0,75

Absence of User Profiles section

(low sociability)

Presence of User Profiles section

(high sociability)

low

moderate

high

Perceived Risk

low exertion of influence

moderate exertion of influence

high exertion of influence

Whereas respondents perceive a higher risk in purchasing from the online store in the group with high exertion of influence when the user profiles section is absent and sociability is low, this negative effect nearly disappeared when sociability was high and user profiles were present.

To test hypotheses H2, univariate F-tests showed that the high quality group (M PEOU = .528; M PU = .369; M PEn = .334) scored significantly higher than the low quality group (M PEOU = -.105; M PU = -.149; M PEn = -.018) with respect to PEOU, PU and PEn. There was no statistically significant difference between the two groups with respect to PInt. These findings supported hypotheses H2a, H2b and H2c, whereas H2d is rejected. The effect of VC quality was highly significant for the transactional constructs of PEOU and PU, indicating that comprehensiveness and timeliness of content are effective components of consumer transactional evaluations. With regard to relational components, the results showed that the use of a comprehensive and lively VC makes the shopping task a more enjoyable experience. Further analyses supported hypotheses H3b, H3c, H3d and H3e. The comparison of the effects of high versus low sociability of the VC on PRisk, PSP, PEn and PInt were significant with large effect sizes for PSP and PEn. We found no support for hypothesis H3a (F(1, 263) = 3.56, p > .05); that is, the degree of sociability does not affect the perceived usefulness of the online store. Apparently consumers in an online environment tend to disregard the importance of tie strength (i.e., in terms of similarity, expertise and accessibility) as posited by traditional WOM theory, because the presence of user profiles does not lead to a higher PU. With regard to relational factors, the presence of user profiles (i.e., indicating high sociability) seems to be adequate to convey a feeling of human contact and warmth, thus increasing PSP. Moreover, the effect of changing the nature of the online store by adding the user profiles section is clear for PEn and PInt.

IMPLICATIONS

This study investigated how to design an integrated VC within an online store with regard to different characteristics; thus, this study has various implications for marketers. First, retailers should pay attention to the exertion of influence on the VC. We showed that consumers tend to positively react to an increased exertion of influence by the retailer only to a certain point, beyond which an increasing exertion of influence leads to a negative effect on perceived usefulness, risk and interactivity. This effect seems to take the shape of an inverted U-form. However, online marketers may be able to compensate for the negative effect of a high exertion of influence by integrating user generated content areas and social cues within the shopping Web site, as there is a moderating effect of sociability on the effect of a higher exertion of influence with respect to perceived risk. Hence, e-retailers can, for example, boost sales with specific messages regarding new products or sales without consumer reactance if they simultaneously encourage a lively VC with a substantial number of social cues exchanged between members.

As indicated by past research, our study confirms that VC quality is also important in the e-commerce environment with regard to transactional factors like perceived ease of use or usefulness as well as relational factors like perceived enjoyment. Only if consumers find the information they need and have the feeling that they are part of a vital community the positive effect of the VC appears. Finally, our investigation shows that the existence of the user profiles section of the VC has an enormous impact on relational factors of consumer evaluations of the online store such as perceived social presence or enjoyment, which is demonstrated by the large effect sizes. Referring to the rapid growth of this community type and its impact on the everyday lives of consumers, retailers can be recommended to provide these profiles in their VCs.

Our study provides evidence on re-embedding strategies that utilize social cues to infuse social presence through a Web interface, which can help to make the virtual interaction more similar to traditional shopping environments and thus lead to an enhanced online experience (Cyr et al. 2007). However, there was no effect of adding such social cues on perceived usefulness. This reveals an important difference between offline WOM and eWOM. Research on offline word-of-mouth emphasizes the importance of the closeness of relationship between the decision maker and the recommendation source. This closeness is determined by how the decision maker is able to evaluate similarity between the decision maker and the recommender as well as the expertise and credibility of the recommender. In our study, consumers tend to focus on the benefit of collective intelligence provided by user-generated content in a VC rather than on detailed information about particular recommendation sources.

REFERENCES

Agarwal, Ritu and Elena Karahanna (2000), "Time Flies When You´re Having Fun: Cognitive Absorption and Beliefs About Information Technology Usage," MIS Quarterly, 24 (4), 665-94.

Andrews, Dorine C. (2002), "Audience-specific online community design," Communications of the ACM, 45 (4), 64-68.

Armstrong, Arthur and John Hagel III (1996), "The Real Value of ON-LINE Communities," Harvard Business Review, 74 (3), 134-41.

Bhattacherjee, Anol and Clive Sanford (2006), "Influence Processes for Information Technology Acceptance: An Elaboration Likelihood Model," MIS Quarterly, 30 (4), 805-25.

Bickart, Barbara and Robert M. Schindler (2001), "Internet Forums as influential sources of consumer information," Journal of Interactive Marketing, 15 (3), 31-40.

Brehm, Jack W. (1966), A theory of psychological reactance, Oxford: Academic Press.

Brown, Jacqueline Johnson and Peter H. Reingen (1987), "Social Ties and Word-of-Mouth Referral Behavior," Journal of Consumer Research, 14 (3), 350-62.

Brown, Shona L., Andrew Tilton, and Dennis Woodside (2002), "Online communities pay," McKinsey Quarterly, 19 (1), 17.

Cheung, Christy M. K., Matthew K. O. Lee, and Neil Rabjohn (2008), "The impact of electronic word-of-mouth: The adoption of online opinions in online customer communities," Internet Research, 18 (3), 229-47.

Cunningham, Scott M. (1967), "The Major Dimensions of Perceived Risk," in Risk Taking and Information Handling in Consumer Behavior, ed. Donald F. Cox, Boston: Boston University Press, 82-108.

Cyr, Dianne, Khaled Hassanein, Milena Head, and Alex Ivanov (2007), "The role of social presence in establishing loyalty in e-Service environments," Interacting with Computers, 19 (1), 43-56.

Dahui, Li, Glenn J. Browne, and James C. Wetherbe (2006), "Why Do Internet Users Stick with a Specific Web Site? A Relationship Perspective," International Journal of Electronic Commerce, 10 (4), 105-41.

Davis, Fred D. (1989), "Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology," MIS Quarterly, 13 (3), 319-40.

Davis, Fred D., Richard P. Bagozzi, and Paul R. Warshaw (1992), "Extrinsic and Intrinsic Motivation to Use Computers in the Workplace," Journal of Applied Social Psychology, 22 (14), 1111-32.

Dellner, Tom (2007), "European e-commerce," electronic Retailer magazine, 4 (6), 46-53.

Farquhar, Jillian and Jennifer Rowley (2006), "Relationships and online consumer communities," Business Process Management Journal, 12 (2), 162-77.

Flavián, Carlos and Miguel Guinalíu (2005), "The influence of virtual communities on distribution strategies in the internet," International Journal of Retail & Distribution Management, 33 (6), 405-25.

_____, (2006), "Virtual Communities and E-Business Management," in Encyclopedia of E-commerce, E-government, and Mobile Commerce, ed. Mehdi Khosrow-Pour, Hershey: IGI Publishing, 1163-1168.

Gefen, David (2000), "E-commerce: the role of familiarity and trust," Omega, 28 (6), 725-37.

Gefen, David, Elena Karahanna, and Detmar W. Straub (2003), "Trust and TAM in Online Shopping: An Integrated Model," MIS Quarterly, 27 (1), 51-90.

Gefen, David and Detmar W. Straub (2004), "Consumer trust in B2C e-Commerce and the importance of social presence: experiments in e-Products and e-Services," Omega, 32 (6), 407-24.

Gupta, Sumeet and Hee-Wong Kim (2004) "Enhancing the Commitment to Virtual Community: A Belief and Feeling Based Approach," in Proceedings of the AIS/ICIS 25th International Conference on Information Systems, Washington DC, 99-111.

Hassanein, Khaled and Milena Head (2005), "The Impact of Infusing Social Presence in the Web Interface: An Investigation Across Product Types," International Journal of Electronic Commerce, 10 (2), 31-55.

Hassanein, Khaled S. and Milena M. Head (2004) "Building Online Trust through Socially Rich Web Interfaces," in Proceedings of the Second Annual Conference on Privacy, Security and Trust (PST 2004), New Brunswick, 15-22.

Jee, Joonhyung and Wei-Na Lee (2002), "Antecedents and Consequences of Perceived Interactivity: An Exploratory Study," Journal of Interactive Advertising, 3 (1).

Kim, Hye-Shin and Byoungho Jin (2006), "Exploratory study of virtual communities of apparel retailers," Journal of Fashion Marketing & Management, 10 (1), 41-55.

Ko, Dong-Gil, Laurie J. Kirsch, and William R. King (2005), "Antecedents of knowledge transfer from consultants to clients in enterprise system implementations," MIS Quarterly, 29 (1), 59-85.

Koufaris, Marios (2002), "Applying the Technology Acceptance Model and Flow Theory to Online Consumer Behavior," Information Systems Research, 13 (2), 205-23.

Kumar, Nanda and Izak Benbasat (2006), "The Influence of Recommendations and Consumer Reviews on Evaluations of Websites," Information Systems Research, 17 (4), 425-39.

Lee, Fion S. L., Douglas Vogel, and Moez Limayem (2003), "Virtual community informatics: A review and research agenda," Journal of Information Technology Theory and Application, 5 (1), 47-61.

Leimeister, Jan Marco, Pascal Sidiras, and Helmut Krcmar (2006), "Exploring Success Factors of Virtual Communities: The Perspectives of Members and Operators," Journal of Organizational Computing & Electronic Commerce, 16 (3/4), 279-300.

Loiacono, Eleanor T., Richard T. Watson, and Dale L. Goodhue (2007), "WebQual: An Instrument for Consumer Evaluation of Web Sites," International Journal of Electronic Commerce, 11 (3), 51-87.

McMillan, Sally J. and Jang Sun Hwang (2002), "Measures of Perceived Interactivity: An Exploration of the Role of Direction of Communication, User Control, and Time in Shaping Perceptions of Interactivity," Journal of Advertising, 31 (3), 29-42.

Mitchell, Vincent Wayne (1999), "Consumer perceived risk: Conceptualisation and models," European Journal of Marketing, 33 (1/2), 163-95.

Moon, Ji Won and Young Gul Kim (2001), "Extending the TAM for a World-Wide-Web context," Information & Management, 38 (4), 217-30.

Pitta, Dennis A. and Danielle Fowler (2005), "Internet community forums: an untapped resource for consumer marketers," Journal of Consumer Marketing, 22 (5), 265-74.

Preece, Jenny (2001), "Sociability and usability in online communities: determining and measuring success," Behaviour & Information Technology, 20 (5), 347-56.

Preece, Jenny and Diane Maloney-Krichmar (2005), "Online Communities: Design, Theory, and Practice," Journal of Computer-Mediated Communication, 10 (4).

Rheingold, Howard (1993), The virtual community: homesteading on the electronic frontier, Reading: Addison-Wesley.

van der Heijden, Hans (2004), "User Acceptance of Hedonic Information Systems," MIS Quarterly, 28 (4), 695-704.

Weathers, Danny, Subhash Sharma, and Stacy L. Wood (2007), "Effects of online communication practices on consumer perceptions of performance uncertainty for search and experience goods," Journal of Retailing, 83 (4), 393-401.