This paper provides a deeper understanding of the role of social influences on positive eWOM behaviour (PeWOM) of heavy-users of online communities. Drawing on Social Interaction Utility Framework, Group Marketing and Social Learning Theories, we develop and test a research model integrating the interactions between the social climate of a website and Interpersonal Influences in PeWOM. 262 Spanish heavy-users of online communities were selected and the data analysed using partial least squares equation modelling. Overall, the model explains 59% of the variance of PeWOM on online communities. Findings reveal that interaction with other members of the online community (Social Presence) is the main predictor of PeWOM. Social Identity is a mediator between Social Presence and PeWOM. Interpersonal Influence has an important role as a moderator variable; the greater the impact of Interpersonal Influence, the stronger the relationship between Social Presence and PeWOM.
The development of online communities where consumers can exchange comments on, and assessments of, trips and accommodation has revolutionised the tourism industry (Banerjee and Chua, 2016; Ruiz-Mafe et al., 2016; Filieri et al., 2015). Online comments have become a key component for customers’ choice of tourism services (Bigne et al., 2017; Hur et al., 2017). With the emergence of social media technologies available to Internet and smartphone users, online communities such as TripAdvisor, Booking.com and Venere have empowered consumers to engage in product-related electronic word of mouth and have emerged as promotional tools for marketing and eCommerce. Given the wide variety of opportunities for existing customers to interact (Matute-Vallejo et al., 2015), this study focuses on positive electronic word of mouth (PeWOM) which means any positive comment made by current or potential consumers available to many people and institutions through the Internet (Hennig-Thurau et al., 2004).
Recent research on social media and eWOM (Cheung and Thadani, 2012; Filieri et al., 2015; Godes and Silva, 2012) has mainly focused on two topics. First, it has analysed the characteristics of social networking sites and customer reviews, which increase their credibility and usefulness for users, including source credibility, length of comment, volume of reviews, valence, etc. (Babić et al., 2016; Cheung and Thadani, 2012; Yan et al., 2016; You et al., 2015). Second, academic research has analysed the motivations and characteristics of individuals, like altruism, self-enhancement and curiosity that encourage active participation on websites (Babić et al., 2016; Bigné et al., 2015; Hennig-Thurau et al., 2004; Munar and Jacobsen, 2014). However, the social dimension of consumption and the group-level antecedents of eWOM communications have been neglected in previous studies on social media.
Due to the intense competition in the tourism sector, incorporating consumers’ social identification processes in the formation of PeWOM communications is a significant issue (Harris and Goode, 2004; He et al., 2012). Obviously, the proliferation of online communities creates advantages for consumers because of the broad range of opportunities they provide. But this forces travel operators to ensure PeWOM communication on the part of the users to maximise their competitive position. eWOM communication is a dimension of loyalty (Zeithaml et al., 1996) and, therefore, is a key factor for profitability (Reichheld, 1993; Ehigie, 2006). Despite recent research (e.g., Chu and Sung, 2015; Oh et al., 2014; Sun et al., 2016) that postulates that social influences affect group members’ intentions and, ultimately, their behaviour, to the best of our knowledge there is a lack of research testing the influence of social climate on consumer's eWOM behaviour. Moreover, previous research on social media has mainly followed a normative approach to explain social influences on consumer decision-making (Zhou et al., 2011), neglecting the role of voluntary influences (friends, colleagues, etc.).
When booking tourism products on an online community, consumers are generally unable to make valued judgements prior to purchase because of the lack of information regarding product quality, so they rely heavily on external advice to make decisions. Experience products, such as accommodation and restaurant services, are habitually reviewed by professional critics on different websites (e.g., www.lonelyplanet.com) and hotels and restaurants are rated on tourist guides. For tourist products, expert reviews are a major source of reliable information (Bigne et al., 2017; Chossat and Gergaud, 2003; Ho and Dempsey, 2010; Zhang et al., 2016).
An understanding of the mechanisms that drive PeWOM behaviour when consumers interact with technology is of high importance for tourist companies that seek to increase their customer base. However, from a theoretical perspective, studies still highlight a need for more empirical research on how to increase consumer participation in the use of social media and the factors that impact on intention to share information among travel-related social media users (Ayeh et al., 2013; Bigné et al., 2015; Hur et al., 2017). This paper examines group level antecedents of PeWOM on the heavy-user segment. Heavy-users are the most attractive segment for online communities. Therefore, understanding how to encourage them to provide active recommendations to other members is important both from a retention point of view (spreading PeWOM is an important indicator of loyalty among these highly attractive members) as well as an acquisition point of view (it helps attract new members to the travel review site, and given that the recommendations are provided by heavy-users, it is likely that these consumers will attract other heavy-users). According to Media Dependency Theory, the intensity of the relationship between consumer and media predicts the likelihood of a media message impacting individual's attitudes and behaviour (Ball-Rokeach, 1985). Consumers who use online reviews more often will be more willing to spread eWOM as a result of their dependency on media information resources (Ruiz-Mafe et al., 2014). Therefore, heavy-users are more likely to forward online information to other consumers than light users (Andreu et al., 2017; Ho and Dempsey, 2010).
The aim of this study is to provide a deeper understanding of the role of social influences on PeWOM behaviour of heavy-users of online communities. Understanding the role of social influences on PeWOM communication is strategically important for tourism companies and responds to recent calls for studies that go deeper into the antecedents of word-of-mouth communication (Filieri et al., 2015; Sun et al., 2016). This study intends to make three specific contributions to the literature. Previous research carried out on online communities highlights the role of Social Identity and affective commitment towards the community on online community participation (Arenas-Gaitan et al., 2013; Casaló et al., 2010). This paper analyses the direct and mediating effects of the components of social climate (Social Identity and Social Presence) of the online community on heavy-users’ PeWOM behaviour. Following Group Marketing Theory (Harmeling et al., 2017), we argue that Social Identity mediates the relationship between Social Presence and PeWOM disseminated by heavy-users of an online community. Second, the conceptual model proposed integrates the effect of social climate with the moderating role of informational influences on the relationship between social climate and consumer recommendation to use online travel communities to make purchases (PeWOM). In this respect, some authors (e.g., Wiertz and de Ruyter, 2007; Casaló et al., 2011) suggest that the analysis of the direct effects of Interpersonal Influence may only be restating the obvious and, therefore, we investigate the moderating effects of personal attributes. The third contribution is that the paper focuses on heavy-users of online communities who, despite their potential as a profitable segment for tourist companies, have been scarcely investigated.
The work is divided into two parts. The first, theoretical part is made up by the literature review, hypotheses and the methodology. The second, with an empirical study of a sample of 262 heavy-users of online communities, examines the impact of different types of social influence on the decision to make PeWOM communications on online communities.
Conceptual frameworkA central research question in social psychology concerns the degree to which the evaluations of others have consequences for information processing and behaviour (Ferguson et al., 2005). Group marketing is the use of the psychological mechanisms underlying group influence to drive behaviours that benefit companies (Harmeling et al., 2017). Groups have strong and pervasive effects on the behaviours of their members and alter how people decide which products to purchase and recommend. Social media groups are ubiquitous in the tourism industry. As an example, TripAdvisor has on average 390 million visitors to its online community each month and 465 million ratings across the globe (TripAdvisor, 2017), providing hotels and restaurants with more visibility and access to consumers.
Information exchanges through social media allow consumers to easily observe how many people have used tourist products and how many of them are satisfied with them. Therefore, these can originate associations in consumers’ minds, conditioning attitudes and behaviours (Bigné et al., 2015; Casaló et al., 2010). Informational influences play an important role in influencing consumer decision-making (Bigne et al., 2017; You et al., 2015). Informational influences refer to the tendency to accept information from mass media, from others with more knowledge and to be guided in searches for products, brands and stores. Consumers take into account the advice from the group to which they belong, give credibility to other consumers’ comments and make inferences about them (Chu and Kim, 2011). This paper analyses how the social climate on online communities and voluntary informational influences (Interpersonal Influences and External Influences) can affect eWOM. We focus on the segment of heavy-users of online travel communities conceptualised as Internet shoppers who usually visit online travel communities to book accommodation. These consumers have previous experience booking accommodation on online travel communities.
Drawing on Social Interaction Utility Framework, Group Marketing and Social Learning Theory, a conceptual model is developed below for the role of social climate and informational (voluntary) influences on PeWOM on online communities. Fig. 1 shows the conceptual model.
Social climate and eWOM
eWOM on hotel review websites differs from real life WOM situations in that traditional offline WOM regarding tourist services tends to occur in a spoken, interpersonal, strong tie communication setting, while eWOM messages on online communities can be viewed simultaneously by many other consumers via the Internet and are available to large global audiences. Moreover, consumers can decide when and how they receive content from other community members, having greater control over the information received than they would in offline settings, due to lack of anonymity in physical contexts.
According to the theoretical framework of Social Interaction Utility (Balasubramanian and Mahajan, 2001), different kinds of utility are derived by consumers from their communicative behaviour in social media. Social utility refers to the consumer value obtained from reading the contributions of the other travel community contributors, which can motivate the consumer to add comments. These utilities influence individual future behaviour and encourage further contributions. Social climate can be defined as the individual's desire to exchange information, belong a group, maintain on-going relationships, and establish relational bonds (Bock et al., 2005; Sun et al., 2016). Social climate in an online community depends on its overall atmosphere, of course created by its members, and reflects the nature of the entire group. This paper approaches the concept of social climate in the context of online communities using two variables: Social Presence and Social Identity.
Social Presence, Social Identity and PeWOMSocial Presence is defined as the degree to which users of an online community can feel the presence of others as the result of interpersonal interactions during a communication process (Walther, 1992; Yeh et al., 2011). It can be regarded as the degree to which a person feels the proximity of other people in social media (Gefen and Strauss, 2003).
Groups that have more social interactions and mutual activities are likely to exert greater social influence on members and shape their perceptions and behaviours (Tsai and Bagozzi, 2014). Online communities facilitate the interconnectivities of individuals and increase the availability of online social support through sharing information among members. In general, customers with less experience (light users) rely more heavily on peripheral cues, drawing meanings from them rather than engaging in the deeper levels of processing characteristics of the website than customers with more experience (heavy-users) (Bettman and Sujan, 1987; Suri et al., 2003; Chiou and Pan, 2009). Prior investigations show that individuals who perceive greater Social Presence feel stronger emotional connections and social support are more motivated to make purchases and share their experiences (Hajli and Sims, 2015; Kim et al., 2009; Qu and Lee, 2011). Following this reasoning, online interaction among heavy-users of an online community endorsing a hotel/restaurant, in a positive manner, can positively influence others to take the final step to make a booking and share their positive experiences. We posit that heavy-users engage in deep processing of the knowledge shared in the online travel community, feel strong emotional connections with other members and, therefore, are motivated to disseminate PeWOM. Therefore,H1 Social Presence positively influences PeWOM behaviour among heavy-users of online communities.
Social Identity Theory developed by Tajfel and Turner (1979) proposes that people tend to classify themselves into exclusive groups, constructing part of their identity on the basis of belonging to that group and creating barriers with groups other than their own. According to the Social Interaction Utility framework, community identification refers to a person's belief that he/she is an integral part of a community, collective, or group (Algesheimer et al., 2005; Koh and Kim, 2004). It reflects the perception of social integration, similarity and interdependency with others, which results in higher willingness to maintain long lasting relationships with the community. Therefore, Social Identity refers to individuals’ perceptions of belonging to certain social groups with which they share certain values that are important for the individual and the group (Tajfel and Turner, 1979; Lee et al., 2011).
In online communities, Social Identity can be channelled by reflecting one's self concept in terms of one's relationship with other consumers exchanging experiences on the website (Zhou, 2011). As a specific form of social attachment, community identification in an online group is close to the sense of identification observed in traditional communities (Qu and Lee, 2011). As Casaló et al. (2010) stated, if a consumer identifies with a group on an online travel community, participation in joint activities with the collective will be congruent with his or her personal values, which will motivate this member to participate actively by helping others in the online travel community. Recent research finds that community identification positively influences the individual's participation and loyalty to an online community (Bigné et al., 2015; Shen et al., 2010; Qu and Lee, 2011; Zhou, 2011; Oh et al., 2014). Social Identity may also increase the volume of knowledge sharing in virtual communities (Chiu et al., 2006; Shen et al., 2010). eWOM is a dimension of loyalty (Zeithaml et al., 1996) and a form of knowledge sharing, therefore, we expect consumers with higher Social Identity will be more willing to spread positive eWOM about the online community.
Identity appraisal mechanisms linked to groups can drive conforming behaviours in which the customer matches her or his attitudes and behaviours with the group's attitudes (Harmeling et al., 2017). Zhou (2011) found a strong relationship between Social Identity and social networking sites participation. Arenas-Gaitan et al. (2013) demonstrate the influence of Social Identity on online travel community use and positive eWOM among heavy-users of online travel communities.
We posit that heavy-users with higher Social Identity feel more affiliation to the group and greater social utility from their interactions. When heavy-users feel greater social utility, it indicates a long-term orientation in the relationship with the group derived from frequent and pleasing contacts with other members of the online travel community. Therefore, we posit heavy-users of online communities with higher Social Identity experience greater emotional connection and are more motivated to recommend the online community to other potential consumers.H2 Social Identity in an online community positively influences PeWOM behaviour among heavy-users of online communities.
Users with a greater degree of knowledge than other members, that is, those with greater Social Presence, are more likely to feel integrated in the online community. As heavy-users spend more time searching and disseminating information in the online travel community, they will develop stronger bonds with other consumers, which triggers higher Social Presence. For users with a stronger Social Presence in a group, perceived differences between members of the group (intragroup) are smaller than perceived differences with members of other groups (intergroup), which means there is a strong correlation between Social Presence and Social Identity (Shen et al., 2010; Lee et al., 2011). Harmeling et al. (2017) propose that the net effect of the group on consumer behaviour depends on the sum of its dynamically varying influences on information interactions (Social Presence) and identity appraisals (Social Identity), as the consumer's time on a website increases. Following this reasoning, we propose that as the consumer's experience on the online community deepens, the consumer will repeat behaviours learned there, which affects the value of group-provided interactions (Social Presence) and accepts the online travel community as part of his or her self identity, which in turns affects the relevance of group norms for identity management (Social Identity). Therefore,H3 Social Presence positively influences Social Identity among heavy-users of online communities.
Interpersonal Influence refers to the effect of the opinion of friends, colleagues, etc. (Ajzen, 1991; Taylor and Todd, 1995; Bhattacherjee, 2000; Hsu and Chiu, 2004; Roca et al., 2006; Woon and Kankanhalli, 2007). People accept the influence of others through processes of internalisation and identification if, by following their advice, they achieve their objectives which will maintain a satisfactory group relationships (Li, 2011). Peer groups are one of the most recognised socialisation agents and are found to be highly influential in shaping consumption-related decision-making (Hsu and Lin, 2008; Sledgianowski and Kulviwat, 2009; Shen et al., 2011; Zhou, 2011).
Heavy-users are subject to a higher informational influence than light users. Robinson et al. (2002) found that heavy Internet users were likely to spend more time communicating face-to-face and over the phone with family and friends than non-Internet users. Kraut et al. (1998) discovered that Internet heavy-users had larger increases in the size of their local and distant social circles and their face-to-face interaction with friends and family than light users. Zhao (2006) reported that heavy-users of the Internet for social purposes tend also to have more offline social ties than light users. Following Chu and Kim (2011), online community heavy-users are predicted to display a higher need to acquire information and guidance from knowledgeable contacts when searching for and contemplating purchase options which, in turn, will facilitate eWOM.
Casaló et al. (2011) posit that a traveller who is more easily influenced by information provided by others will give more weight to his or her perceptions regarding the advice obtained in the online travel community (social climate) in order to form their behavioural intentions than a traveller who is less susceptible to Interpersonal Influence. The conjoint effect of Interpersonal Influences and social climate on PeWOM is coherent with the Integrated Marketing Communication process (Schultz, 1992). This paper follows a multichannel approach to demonstrate that different social environments affect consumer intentions to communicate their opinions (PeWOM).
Therefore, we posit that the interpersonal interactions with peers that recommend using a specific online community will reinforce the impact of social climate on PeWOM for heavy-users of online communities.H4a The higher the Interpersonal Influence, the higher the impact of Social Presence on PeWOM behaviour among heavy-users of online communities. The higher the Interpersonal Influence, the higher the impact of Social Identity on PeWOM behaviour among heavy-users of online communities.
Social Learning Theory (Bandura, 1977) explains consumer socialisation, assuming that individuals develop their attitudinal and behavioural patterns as consumers in the marketplace partly as a result of their interactions and learning from external socialisation agents, such as parents, peers and the mass-media. Given the general assumption of the role of socialisation agents on both consumption attitudes and behaviours (Moschis and Churchill, 1978; Chu and Sung, 2015), we propose that opinion leaders, web-recognised experts and mass-media influence consumers’ tendency to conform to the expectations of others which, in turn, determines consumers’ eWOM behaviour.
External Influence is related to the mass media content and to the opinions voiced by recognised tourism experts both in the offline mass media and social media (Roca et al., 2006; Zhang et al., 2016). In a non-hierarchical environment, like the context of hotel bookings on websites such as TripAdvisor, Booking.com and Venere, users can seek the opinion of experts to help them make sense of conflicting information in online searches and to support their decision making (Hsu and Lin, 2008; Shen et al., 2011; Zhang et al., 2016). These experts could be advocates or opinion leaders (Flynn et al., 1996) who provide professional and instructive criticism of certain destinations, new trends and specific tourist accommodation based on their knowledge and experience.
According to Media Dependency Theory (Ball-Rokeach, 1985) mass media and social media provide a set of gratifications related to information, socialisation and entertainment for consumers, which enable them to achieve their personal objectives and, therefore, influence their behaviour. Prior research has also demonstrated that online customer behaviour is not only influenced by personal referents, such as family, friends and colleagues, but also by offline mass media (Bronner and De Hoog, 2011; Bigne et al., 2017), expert comments in restaurant guidebooks (Chossat and Gergaud, 2003) and reviews made by regular consumers, termed as “expert” by the third-party review platform because of the high-quality information they convey (Zhang et al., 2016). Ho and Dempsey (2010) suggest that an important antecedent of online information sharing is the consumption of electronic content from mainstream media like newspaper websites. Ruiz-Mafe et al. (2014) demonstrates social media dependency influence on consumers PeWOM towards Facebook fanpages.
External Influences influence individuals’ ways of life, modifying their values, attitude and perceptions (Bigne et al., 2017; Zhang et al., 2016). Consumers have strong motivations to comply with what their significant referents advise them and adapt their behaviour following their recommendations. Bigne et al. (2017) evidence that information received by heavy-users of online travel communities through mass media or tourism leaders (External Influences) is likely to be spread by means of eWOM.
Heavy-users of online travel communities process and disseminate information provided by the mass media and experts. This information may thereafter be considered by other consumers as more neutral than company (e.g., hotel) posted information. Moreover, this information can be presented by the heavy-user as a reference so he/she can position him/herself as a expert to his/hers thousands of followers.
Therefore, we expect that if an online travel community is recommended by the heavy-user's external social groups, he/she will engage in PeWOM about this online travel community.H5 External Influences have a positive effect on PeWOM behaviour among heavy-users of online communities.
Consistent with habitual research practices for collecting data about the Internet and social networking sites, we chose online surveys (e.g., Bagozzi and Dholakia, 2006; Steenkamp and Geyskens, 2006). Consumers responded to the survey through a web page that was designed specifically for this research project. Interviewed individuals belonged to an e-shopper database of an international market research company. They received an incentive for participating.
The population were Spanish Internet shoppers aged 18 or older who had used online communities in the last year (i.e., TripAdvisor, Booking.com, Atrapalo, Trivago or Venere) to book accommodation. A total of 902 individuals were initially contacted during the study and 415 finally agreed to participate. Hence, we have selected in our sample the 262 Internet shoppers who have been using online travel communities for more than 6 months and frequently visit online travel communities. Selection was made by choosing individuals scoring 5 or higher in a 1–7 Likert scale when they were asked how often they book accommodation through online travel communities (1=never; 7=always).
Selection bias was controlled by determining sampling quotas on the basis of gender and age. Our sample is composed by Internet shoppers who have used online travel communities to purchase accommodation services. According to IAB (2017), Internet shoppers are heavy-users of social media as 65% of the Internet shoppers in Spain have made a purchase decision influenced by social media. We compare the sample characteristics with available information about the population. However, without studies about online travel communities in Spain, we compare the sociodemographic characteristics of our sample with the most relevant study on the online Spanish-speaking population (ONTSI, 2015). The results are similar, as can be seen in Table 1. Thus, a good representativeness of the average online shopper was guaranteed. Table 1 summarises sample quotas.
Sample quotas.
| Quota (ONTSI, 2015) (%) | Quota (sample=262) (%) | |
|---|---|---|
| Gender | ||
| Male | 53.9 | 51.53 |
| Female | 46.1 | 48.47 |
| Age | ||
| 18–24 | 13.3 | 9.55 |
| 25–34 | 24.8 | 30.53 |
| 35–49 | 38.7 | 40.08 |
| 50–64 | 17.0 | 17.94 |
| >65 | 6.1 | 1.90 |
The questionnaire was pre-tested through 30 personal interviews with consumers with one year's experience as purchasers of online tourism services. As a result of the pre-test, some redundant questions were eliminated and the measurement scales adapted to facilitate understanding and avoid erroneous interpretations.
The study data were obtained through a single collection method, therefore, to prevent common method bias we followed the recommendations in Podsakoff et al. (2003) and MacKenzie and Podsakoff (2012) during data collection and analysis. Firstly, during data collection the anonymity of participants’ responses was guaranteed. Secondly, to avoid conditioning participants’ responses the exact aim of the study was not disclosed. Third, questionnaire items related to the dependent variables were placed after indicators that measured independent variables. Fourth, participants’ access to their responses to previous questions was limited so that their subsequent responses would not be determined by their previous answers. Finally, statistical corroboration of the absence of common method bias in the data was obtained using confirmatory factor analysis (CFA) (Conway and Lance, 2010). All variables were loaded on one factor to examine the fit of the confirmatory factor analysis model (Podsakoff et al., 2003). If common method variance is largely responsible for the relationship among the variables, the one-factor CFA model should fit the data well (Conway and Lance, 2010; Iverson and Maguire, 2000; Korsgaard and Roberson, 1995; Mossholder et al., 1998). The confirmatory factor analysis showed that the single-factor model did not fit the data well (χ2=435.1793, p=0.000, GFI=0.745; SRMR=0.0923), demonstrating the absence of common-method bias.
The purged sample of 262 individuals comprised 48.47% women and 51.53% men. The largest percentage of individuals is concentrated in the age groups 25–34 years (30.53%) and 35–49 years (40.08%). Many participants had studied at university level (32.8%). 77.3% of participants said that online communities were their main source of travel information. The most used online communities were Booking.com (30.1%), Trivago.com (22.2%) and TripAdvisor.com (11.5%), leisure being the main reason (70.5%) for booking accommodation through these channels.
The factors included in the study were measured using indicators adapted from previous studies, as Table 2 shows. All the variables were measured on 7-point Likert scales (1, totally disagree to 7, totally agree).
Measurement scales.
| Interpersonal Influence | |
| 1 My friends think I should use X for booking accommodation | Bhattacherjee (2000); Roca et al. (2006); Li (2011) |
| 2 My colleagues think I should use X for booking accommodation | |
| 3 My family think I should use X for booking accommodation | |
| External Influence | |
| 1 I read news and reports saying that using X was a good way of booking accommodation | Bhattacherjee (2000); Roca et al. (2006); Li (2011) |
| 2 Expert opinions depicted a positive sentiment for using X | |
| 3 Mass media reports convinced me to use X for booking accommodation | |
| Social Presence | |
| 1 I can inform others of my presence through X | Yeh et al. (2011); Lee et al. (2011) |
| 2 I can feel human sensitivity through X | |
| 3 I can notify others of my feelings through X | |
| 4 I can feel a sense of sociability in X | |
| Social Identity | |
| 1 I feel a sense of belonging towards the user group of X | Yeh et al. (2011); Lee et al. (2011) |
| 2 I have a feeling of togetherness or closeness in user group of X | |
| 3 I have a strong positive feeling towards the user group of X | |
| 4 I enjoy being together with the user group of X | |
| eWOM activity on social media | |
| 1 I say positive things on social media about X to other people | Zeithaml et al. (1996) |
| 2 I recommend X on social media to those who seek my advice | |
| 3 I encourage friends and relatives on social media to use X to book accommodation | |
X=My favourite online community.
The hypothesised relationships in the theoretical model were estimated applying partial least squares equation modelling (PLS). SmartPLS 3.2.4 software was employed to estimate the parameters and bootstrapping of 5000 samples was used to obtain their significance (Ringle et al., 2015). The choice of PLS is because this study focuses on the prediction of the dependent variable (Roldán and Sánchez-Franco, 2012). In addition, the academic literature recommends the use of PLS for the analysis of interaction effects, since the use of covariance methods, when the interaction involves Likert-scale variables, may be problematic because of the relevant degrees of shared variance (Hernández-Mogollon et al., 2010; Sanchez-Franco, 2010).
Measurement modelAs for the psychometric properties of the measurement model, the reliability of the individual items was assessed and the loadings were estimated. All the loadings show values greater than 0.7 (Carmines and Zeller, 1979), thus being significant (see Table 1). Construct reliability was evaluated using Cronbach's alpha coefficient (Cronbach, 1951) and composite reliability (ρc) (Werts et al., 1974). In all cases (Social Presence, Social Identity, PeWOM and External Influences), values above the recommended minimum of 0.7 were obtained (Nunnally and Bernstein, 1994) (see Table 3), thus confirming the reliability of the measuring instrument. The convergent validity was demonstrated when it was verified that the average variance extracted (AVE) of all the constructs of the model presented higher values than 0.5 (Fornell and Larcker, 1981) (Table 3).
Measurement model.
| Loading | t-value | Cronbach's alpha | Composite reliability | AVE | |
|---|---|---|---|---|---|
| Social Presence (SP) | 0.895 | 0.927 | 0.761 | ||
| SP1 | 0.846 | 35.112 | |||
| SP2 | 0.870 | 39.565 | |||
| SP3 | 0.887 | 52.013 | |||
| SP4 | 0.886 | 52.116 | |||
| Social Identity (SI) | 0.937 | 0.955 | 0.842 | ||
| SI1 | 0.897 | 53.243 | |||
| SI2 | 0.928 | 62.264 | |||
| SI3 | 0.944 | 108.740 | |||
| SI4 | 0.900 | 45.786 | |||
| External Influences (EEI) | 0.881 | 0.927 | 0.809 | ||
| EEI1 | 0.941 | 101.556 | |||
| EEI2 | 0.912 | 52.217 | |||
| EEI3 | 0.843 | 31.372 | |||
| PeWOM | 0.864 | 0.916 | 0.785 | ||
| PeWOM1 | 0.882 | 54.698 | |||
| PeWOM2 | 0.912 | 61.964 | |||
| PeWOM3 | 0.863 | 30.881 | |||
To test discriminant validity, two criteria were used: the Fornell–Lacker criterion and the HTMT85 (heterotrait–monotrait ratio of correlations) criterion (Hair et al., 2017). The discriminant validity was demonstrated by the fact that each construct load more strongly on its own measures than on the other constructs, with all correlation ratios below 0.90 (see Table 4).
Discriminant validity.
| Social Presence | Social Identity | External Influences | PeWOM | |
|---|---|---|---|---|
| Social Presence | 0.872 | 0.726 | 0.536 | 0.812 |
| Social Identity | 0.720 | 0.917 | 0.595 | 0.711 |
| External Influences | 0.477 | 0.543 | 0.899 | 0.576 |
| PeWOM | 0.720 | 0.644 | 0.504 | 0.886 |
Note: Diagonal values represent AVE square root; values below the diagonal reflect latent variable correlations; above the diagonal are HTMT ratios.
Bootstrapping (5000 re-samples) was used to obtain the results of the structural model (Hair et al., 2017) (see Table 5). Additionally, the R2 and Q2 values were obtained to verify the strength and predictive relevance of the model. All R2 values were higher than 0.10 (Falk and Miller, 1992), with Q2 values higher than 0, which leads to the conclusion that the model has strength and predictive relevance.
Structural model.
The first goal of this paper is to assess the impact of social climate and External Influences on eWOM. As can be seen from Table 5, the results show that the increase in Social Presence on an online community increases heavy-users’ PeWOM behaviour (H1 accepted; βSP-PeWOM=0.504; p<0.001), therefore interactions with other users on an online community increases the willingness of heavy-users to inform others about their own personal experiences. This result supports previous research stating the positive role of Social Presence in the consumer's attitude towards a website (Cui et al., 2010). Social Identity has a significant impact on PeWOM (H2 accepted; βSI-PeWOM=0.267; p<0.01) supporting findings reported by some authors in the tourism industry (Casaló et al., 2010, 2011; Qu and Lee, 2011). There is a strong relation between Social Presence and Social Identity (H3; βSP-SI=0.671; p<0.001), which means that on an online community Social Presence makes a significant contribution to social integration of its members (Shen et al., 2010; Lee et al., 2011). This result supports previous research stating that for social use of the Internet, heavy-users tend to have more social ties than light users do (Zhao, 2006).
In addition, it is verified that expert opinion diffused by the off-line mass media, blogs etc. (External Influence) influences PeWOM (H5 accepted; βEI-PEWOM=0.123; p<0.005). This finding supports previous research made in offline (Chossat and Gergaud, 2003) and online contexts (Zhang et al., 2016). Experts’ opinions play an important role in the tourist market for the following reasons: information is imperfect and very costly to acquire and its quality is, in large part, variable and consumers welcome expert validation. The conjoint effect of External Influences and social climate on PeWOM is coherent with the Integrated Marketing Communication process. This paper follows a multichannel approach to demonstrate that different social environments affect consumer intentions to communicate their opinions (PeWOM).
The contrast of the indirect effect was carried out through a mediation analysis (see Table 6).
As can be seen in Table 6, when the Social Identity mediator is introduced into the analysis, the direct relation between Social Presence and PeWOM weakens, from having a coefficient c of 0.683 to having a value c′ of 0.504, the indirect effect of the Social Presence in PeWOM being significant. This result allows us to conclude that Social Identity mediates the relationship between Social Presence and PeWOM (a1b1). However, we can say that this mediation is only partial, since the variant accounted for (VAF) index (Hair et al., 2017) has a value between 20% and 80%.
Moderation: interaction effectsThe second goal of this paper is to study the moderating effect (interaction) of the Interpersonal Influence variable (II) on SP-PeWOM and SI-PeWOM relations. For this, we compare the results of the model without and with the interaction construct, using PLS. We applied the procedure described by Chin (2003); Chin et al. (2003) (product indicators), where the predictor and moderator variables are multiplied to obtain the interaction terms.
The results of the model with no interaction effect indicate that all paths were significant at a level of 0.05. The standard SP (Social Presence) was 0.504 and the SI (Social Identity) at 0.267 with an R2 of 0.570 for PeWOM. The inclusion in the model of the moderator variable II allowed us to obtain the following results: 0.110 (SP*II→PeWOM), 0.042 (SI*II→PeWOM), increasing the R2 for PeWOM to 0.593 (see Table 7 and Fig. 2).
Interaction effect.
| Interpersonal Influences (II) | ||||
|---|---|---|---|---|
| Main effect model | Interaction effects model | t-value (bootstrap) | Supported | |
| Social Presence→PeWOM | 0.504*** | 0.462*** | Yes | |
| Social Identity→PeWOM | 0.267** | 0.230** | Yes | |
| SP*II→PeWOM | 0.110 | 3.142 | Yes | |
| SI*II→PeWOM | 0.042ns | 0.958 | No | |
| R-square PeWOM | 0.570 | 0.593 | ||
* p<0.05.
The path coefficient of the interaction term indicates to what extent the exogenous variable's influence on the endogenous variable changes depending on the moderating variable (Henseler and Fassott, 2007). According to our results, one standard deviation (SD) increase in the II will increase the impact of Social Presence from 0.462 to 0.572 (0.462+0.110). It is thus verified that II positively moderates the relationship between Social Presence and PeWOM, such that the greater the II, the stronger the impact of Social Presence on PeWOM. Additionally, we compared the R2 of the interaction model with the R2 of the main effects model, which excludes the interaction construct. The difference in R2 was used to assess the overall effect size ƒ2 for the interaction, where 0.02, 0.15, and 0.35 have been suggested as small, moderate, and large effects, respectively (Cohen, 1998). In the moderator variable II, the effect size ƒ2 for the interaction effect was 0.03, which in turn represents a small effect (Chin et al., 2003). It is, therefore, concluded that Interpersonal Influences positively moderate the impact of Social Presence on PeWOM (H4a accepted).
Results also show that Interpersonal Influences do not moderate the impact of Social Identity on PeWOM (H4b rejected). Information obtained through Interpersonal Influences (strong tie source) is usually more personal and private than information shared in online travel communities with higher levels of self-disclosure in a socially extensive environment (weak tie source), so it is transferred mainly in the same physical environment (WOM) and context in which it was received (Bigne et al., 2017). Moreover, Social Identity involves a feeling of belonging and identification with the other members of the community. Therefore, recommendations made by peers and relatives of the consumer do not reinforce the impact of Social Identity on PeWOM, as these peers and relatives do not belong to the online travel community.
ConclusionsTheoretical contributionsUnderstanding the role of social influences on PeWOM communications of heavy-users of online communities has strategic importance for tourism providers. An enormous volume of customers use review websites to get information and/or book accommodation and the heavy-users are key influencers on the other, lighter users.
Overall, the model enables us to explain 59.3% of the variance of positive recommendations on online communities. Unlike previous studies focused on eWOM intention, this study uses PeWOM behaviour as the outcome measure. This study offers three contributions to the literature. Firstly, this research adds strength to previous arguments, based on the Social Interaction Utility framework (Balasubramanian and Mahajan, 2001), by highlighting the key role of the social climate of the online community on the consumer's decision to spread PeWOM, with a focus on the segment of heavy-users of online communities. This study also extends Group Marketing Theory to the context of online communities, and proposes that users interacting with group members with similar interests respond positively to group-level social capital (social cues) embedded in within-group interactions. We demonstrate that interaction cues developed among heavy-users of an online community elicit consumer responses to other members of the online community (PeWOM). We would expect social factors to exert stronger effects on light-users, as they lack experience and are more susceptible to social influence; but this paper demonstrates that social climate is also important for heavy-users, although these effects operate differently. Findings show that interaction with other members of the online community (Social Presence) is the main predictor of eWOM, the effect of Social Identity and opinions of experts being less relevant.
The second contribution of this paper is the analysis of the moderating effects of consumer susceptibility to Interpersonal Influence on the impact of social climate on PeWOM. This finding contributes to this emerging literature body as until now most studies on online communities and user-generated content (Bigné et al., 2015; Hajli and Sims, 2015; Kim et al., 2009; Qu and Lee, 2011) have focused on the direct effects of social influence, neglecting other interactions that may induce the consumer to recommend an online community. We support the proposition that group marketing effectiveness (Harmeling et al., 2017) depends on the dynamic interplay of the group's influence on peer-informational influences and emotional bonds with other members of the online community. Finally, the study is based on statements made by actual heavy-users of online communities, rather than simply from an examination of consumer comments available online or from an experiment under controlled conditions.
Managerial implicationsFindings suggest that managers of online communities should be aware of the importance of social processes like interaction through Social Presence. The exchange of experiences (content creation by users) differentiate and characterise online communities. Therefore, managers of these websites might organise bulletins with news, opinions, offers and additions to tourism services and resources. In these bulletins, users could also provide topics for discussion, such as transport options, quality of establishment, prices and nearby tourism resources. They could also study other consumers’ evaluations, for example, through viewing consolidated data from responses to standard questions and provide areas for free comments.
Online communities can enhance Social Presence, facilitating communication among users. For example, when a consumer expresses interest in a specific hotel on the website, he or she should be able easily to track reviews from other consumers with similar interests, just by clicking a relevant button on the online community. Online communities might also develop Social Presence through non-verbal communication. Thus, in addition to text posts, travel websites might, e.g., develop emotions and allow consumers to post pictures and videos to express their opinions more easily. Finally, Social Presence can also be reinforced with active management by the accommodation providers controlling user activity: rapid, accurate responses to questions from users and their comments, direct communications through email and even personal telephone calls, if that possibility exists. In doing so, managers should avoid using the same language in their responses to different online reviews. Users’ email messages will have to be carefully monitored and there should be easy access to call lines.
Online communities should also try to promote the concept that customers act as social connectors or referents. Offline Interpersonal Influencers, such as friends, family members, colleagues and experts are critical for encouraging dissemination of PeWOM. Leading travel websites such as TripAdvisor and Expedia might feature links to tourism managers accounts. This would allow the managers and consumers to exchange information, e.g., positive pieces of news and good feedback. Related expert blogs might also be made available on the website.
Heavy Internet shoppers tend to join reward programmes more than light shoppers (Chiou and Pan, 2009). Thus, heavy-users might be compensated for sharing their positive purchasing experiences on the online community, by discounts or the offer of “freebies”. As heavy-users are the most attractive segment for online communities, tourism managers should provide them with reliable and useful contents in terms of product features, prices, awards, images, photos, and/or videos (Hur et al., 2017). This will encourage heavy-users of online communities to develop emotional bonds with the online travel community by reading others’ suggestions and by viewing photos and videos. This will also encourage heavy-users to communicate on content to their friends. As strength of Interpersonal Influence moderates the impact of Social Presence on PeWOM, managers might consider how to develop offline interpersonal interactions, e.g., by offering very attractive discounts for group bookings/company wide bookings.
Limitations and future research linesThis study has some limitations that should be dealt with in future research. First, we have not taken into account the type of bonds that exist between the consumers amongst whom information is exchanged. Some studies have considered these connections, showing that the effect of the information gathered differs, depending on the knowledge and opinion that the consumer has about its source (Chu and Kim, 2011). In future studies, the strength of bonds between consumers could be considered in order to differentiate the effect on eWOM. This research has focused on a specific tourist service (online communities) and on a specific target (heavy-users), which must be taken into account when generalising the conclusions. It is proposed that, to assess the applicability of the model to online tourism communities as a whole, the study be repeated using another types of service with different levels of perceived purchase risk, like flight tickets, holiday packages or restaurants and using a sample of light users or alternatively by potential influencers (Litterio et al., 2017). Similarly, analysis of negative comments is a field that would provide an integrative framework for the study of eWOM generation.
This work was supported by the Ministry of Economy (Spain) under Grant ECO2014-53837R.