Does eWOM influence destination trust and travel intention: a medical tourism perspective

Abstract Virtual community membership has a strong influence on a tourist’s behaviours and the way information is transmitted. Drawing on trust transfer theory as a theoretical framework, this study tests an empirical model that investigates the influence of electronic word-of-mouth (eWOM) on destination trust and travel intention. The study also diagnoses gender differences in the proposed model. Data were obtained from a sample of 216 tourists in Cyprus. The findings from regression analyses suggest that eWOM is positively related to travel intention; eWOM is positively related to destination trust; and destination trust is positively related to travel intention. Additionally, the impact of eWOM on destination trust was significant for both genders but stronger for men, while the impact of destination trust on travel intention was stronger for women. The paper extends the existing literature regarding new response variables associated with eWOM.


Introduction
Hypermedia has led to an increase in the number of people using the internet to access information about prospective travel destinations (Litvin, Goldsmith, & Pan, 2008). According to Travel Industry Association of America-TIA (2005), about 67% of American travellers accessed the internet to retrieve information concerning travel destinations. This suggests that online reviews have a strong influence in the tourism industry (Lin, Jones, & Westwood, 2009;Casaló, Flavián, Guinalíu, & Ekinci, 2015). Due to the intangible nature of tourism services, hindsight regarding service is only available upon consumption, hence there is a greater level of uncertainty and ambiguity. This intangibility encourages potential visitors to visit sites such as Trip-Advisor.com, LateRooms.com, and Hotels.com, which rank hotels in destinations according to how favourably they are reviewed.
The argument in this paper is tied to the medical tourism industry. Medical tourism (med-tour) is the process of combining medical and tourism products/services because visitors want to combine a vacation with health care (Lee, Heesup, & Tim, 2012). In particular, med-tour is divided into two units, beautification and health treatment (Lee, Soutar, & Daly, 2007). Among other things, the reasons people engage in med-tour tends to vary from the availability of well-qualified practitioners, the high cost of treatment, long queues to economic bargains (Keckley, 2008;Woodman, 2008). According to the Turkish Ministry of Health (2012), the country estimated a revenue of $31.4 million in 2011, and about $1 billion in 2015 (Anadolu Agency, 2015). These figures highlight the importance of this sector in the country, as well as the country's reputation in terms of med-tour.
Potential visitors consult online reviews because the tourism product and service is not accessible until the moment of consumption: thus, the risk and uncertainty associated with it is increased . For instance, a medical service such as plastic surgery is accessible to med-tourists only upon consumption. This paper is grounded in Trust transfer theory, which posits that trust transfer occurs when 'the unknown target [is] being perceived as related to the source of the transferred trust' (Stewart, 2003). Therein, it relies on the cognition of individuals based on certain factors such as relatedness, similarities, and closeness (Campbell, 1958). Researchers such as Ng (2013) have pointed out that 'trust in a social network community may also be transferred from trust among its members to the focal firm providing services' . In this sense, eWOM message trust can be transferred to the medical destination under consideration.
The motives behind this study are the lack of comparative quality and safety data, and that knowledge of overseas institutions, and the reporting of adverse events, is lacking. More practically, all medical treatments are entangled with an element of risk that poses danger to the patient's health. Therefore, med-tour signifies a new dynamic to this element of risk, due to the overseas travel involved. Given this, med-tour marketers and destination policy makers are interested in assuring patients' safety, which can be achieved by communicating destination trust to reduce fear and uncertainty among tourists, and to increase market share. Based on the aforementioned theoretical arguments, this paper attempts to clarify these behaviours in the med-tour industry by evaluating how eWOM will increase or decrease risk-taking activities, and how it could shape future travel intentions, the paper also considers how the impact may vary across gender. advertisements, affordable price, and personnel experience (Ye, Yuen, Qiu, & Zhang, 2008). Along with any of the other aforementioned motivating factors, logically, WOM is an inclusive category. Arndt (1967) defined WOM as a physical conversation with regard to a product or service that takes place between individuals who are not commercial actors. Accordingly, the power of WOM resides in the final purchase stage; because favourable WOM messages comfort customers, reduce fear and uncertainty, and enact assurance (Martilla, 1971). Technically, WOM messages are transmitted to reduce risk and uncertainty regarding products or services. Murray's (1991) influential study shows that consumers rely on WOM messages 'to reduce their perceived risk derived from the uncertainty inherent in service purchase decisions' . Settle and Alreck (1989) added that WOM messages are influential and popular as the primary uncertainty eliminator for risks and uncertainty accrued from the purchase of a product or service.
The advent of the internet has extended the concept to an online context known as the electronic word of mouth (eWOM), which has the ability to reach large audiences within a short period of time (Abubakar & Ilkan, 2013). eWOM communication refers to any positive or negative statements made by consumers in regard to products or services that are scripted and posted on the internet for individuals and institutions (Hennig-Thurau, Gwinner, Walsh, & Gremler, 2004, p. 39). The inherent impact of eWOM on consumers' behavioural intentions can be more powerful than the traditional WOM (Eunha & Soocheong, 2011). The eWOM is more reliable than WOM due to its anonymous nature, and the absence of incentives (Abubakar & Ilkan, 2013). Marketers can seize the opportunity in the platform to improve product/service quality, innovation and future product characteristics, install trust in the minds of consumers, which of course stimulate purchase intents.
Empirical findings have shown that WOM messages influence the receiver's awareness (Sheth, 1971), intentions and consideration (Grewal, Cline, & Davies, 2003). Prior literature concerning consumer's behaviours, such as purchase intention, argued that such behaviours take place after the consumers' general evaluation of a product, service or brand (Hsu, 1987). In other words, future consumption behaviours are shaped by their assessment of products or services entwined with external stimulating factors (i.e. eWOM). As such, researchers such as Engel, Blackwell, and Miniard (2001) suggested that purchase intention involves subjective judgement for future behaviour. Swan, Bowers, and Richardson (1999) added that customer trust may lead to positive attitude towards a brand, may also enhance loyalty, and further install purchase intents. Trust has a direct and indirect impact on consumers purchase intention (Grazioli & Jarvenpaa, 2000). According to , destination trust refers to a visitor's willingness to rely on the ability of a med-tour destination to perform its advertised functions. The authors added that destination trust has three dimensions, namely reputation (ability to provide consistent and improved medical service), credibility (the willingness to deliver promised medical service and attend to patients in case of adverse effect) and competence (ability to meet and satisfy tourists), signalling that destination trust can have a significant influence on tourists' travel intention. eWOM communication has received huge attention in recent years for a number of reasons such as its impact on marketing strategy (Smith, Coyle, Lightfoot, & Scott, 2007), tourists' choice of destination (Reza Jalilvand & Samiei, 2012), persuasive impact on decision-making (Sparks & Browning, 2011), persuasive impact on purchase intention (Gajendra, Ye, Sun, & Li, 2012;Michelle, 2006;Soares, Pinho, & Nobre, 2012), and its role in reducing travellers' perceptions of risk when booking accommodation (Gretzel, Yoo & Purifoy 2007). This paper argues that a positive eWOM could reduce med-tourists' risk and uncertainty perception when choosing a destination.
Intention to visit refers to the willingness of a potential visitor to visit the destination (Chen, Shang, & Li, 2014); it is the rational evaluation of the costs/benefits of a set of alternative destinations, which are derived from external information sources, including eWOM or travellers' blogs (Chen et al., 2014). Scholars have supported the notion that positive eWOM has the tendency to increase the likelihood of booking and room sales (Duverger, 2013;Mauri & Minazzi, 2013;Öğüt & Onur Taş, 2012). A vast amount of tourism research has advocated that eWOM has the ability to influence travel intentions (Arsal, Backman, & Baldwin, 2008;Filieri & McLeay, 2014;Vermeulen & Seegers, 2009;Ye, Law, & Gu, 2009). Based on the aforementioned empirical and theoretical arguments, the following hypotheses were proposed: H1: Medical tourists' eWOM is positively related to travel intention. H2: Medical tourists' eWOM is positively related to destination trust. H3: Medical tourists' destination trust is positively related to travel intention.
The impact of gender upon decision-making, and shopping intent has been a subject of special interest for a long time. Sun and Zhang (2006) noted that 'women are not pragmatic when compared to men; men experience less anxiety when compared with women who experience more, and women are more strongly influenced by their immediate environment' . A dominant finding suggests that 'women are clearly more risk averse than men in the arena of physical health and safety' (Harrant & Vaillant, 2008), thus women are more risk sensitive in the context of losses than gains (He, Inman, & Mittal, 2007).
Men and women differ on several dimensions in the online arena primarily due social, cultural, psychological, and environmental factors (Meyers-Levy & Loken, 2015). For instance, Dittmar, Long, and Meek (2004) posited that men differ from women with respect to eWOM messages and shopping behaviours. Rodgers and Harris (2003) also supported the notion; Byrnes, Miller, and Schafer (1999) added that this is because women are less likely to take risks than men. In the presence of social and technological hazards, women are more risk-averse (i.e. less risk-taking) than men (Olsen & Cox, 2001); even when the level of expertise and experience is the same (Dwyer, Gilkeson, & List, 2002;Harris, Jenkins, & Glaser, 2006). Based on the aforementioned theoretical and empirical arguments, it is proposed that the relationship between eWOM, destination trust and intention to travel will be moderated by gender in the med-tour context. Thus, the following hypotheses were proposed: H4a: Gender will moderate the relationship between eWOM and destination trust in the med-tour context.
H4b: Gender will moderate the relationship between eWOM and intention to travel in the med-tour context.
H4c: Gender will moderate the relationship between destination trust and intention to travel in the med-tour context.
All the hypothesized relationships in this study are presented in Figure 1.

Measures
The questionnaire in this study consisted of four units (i) eWOM; (ii) destination trust; (iii) travel intention; (iv) demographic data. eWOM was measured with six items adopted from Bambauer-Sachse and Mangold's (2011) study. A sample item is 'I often consult other medical tourists' online travel reviews to help me choose a good medical destination' . Destination trust was measured with eight items adopted from . A sample item is 'I could rely on Turkish hospitals to solve my medical problems' . Travel intention was measured with three items adopted from Kassem, Lee, Modeste, and Johnston's (2010) study. A sample item is 'I predict I will visit Turkish hospitals in the future' . Demographic variables used in the study include age, gender, marital status, education level and monthly income.

Sampling and procedure
The current study employs a random sampling approach, utilising cross-sectional survey data. The simple random sampling approach is a type of statistical sampling strategy in which a subset (study-sample) is chosen from the population of interest randomly, such that subjects within the target population have an equal chance of been chosen for participation. This type of sampling technique is not subjected to selective bias, and is free of classification error. The majority of the patients understand the Turkish language; as a consequence a Turkish versions of the questionnaire were developed by professional translators. Prior to the survey, seven participants were contacted to gauge potential ambiguity, and at the end necessary changes were made to the questionnaire. The questionnaires were distributed to local patients receiving medical care in public and private hospitals in Northern Cyprus, and were asked to fill the questionnaires in a self-administered manner. Data collection took four weeks in total (May-June 2014), and the respondents were potential med-tourists to Turkey, primarily due to the nature of their medical complications. There were two inclusion criteria for the participants (1) are you a member of any online community? (2) Do you read or post online reviews? Those who failed to meet the criteria were excluded from the survey.
Frequency analysis was used in generating respondent's profiles; correlation and regression analyses were used to examine the hypothesised relationships. Bootstrapping analysis was used to augment the indirect effect; finally, a multi-group moderation analysis was conducted to test whether gender moderated the paths in the proposed model. Out of the 400 questionnaires distributed, 243 valid samples were returned out of which 27 had missing data; thus, only 216 were used in analysis. This yielded a 61% response rate. According to descriptive statistics, 54% were females and the rest were males; similarly, 54% of the respondents were married and the rest single. An overwhelming number of the respondents' ages ranged between 31 and 40 (49%); and 21 and 30 (42%). The majority of the participants (46%) had a monthly income (in Turkish Lira) between 3000 and 3999; 33% had a monthly income over 4000; and 19% between 2000 and 2999.

Data analyses and results
The proposed model was estimated with SPSS program version 22. Factor analysis using the maximum likelihood method was conducted with a varimax rotation, and an eigenvalue greater than or equal to 1. Factor 1 (destination trust) explains about 26% of the total variance, Factor 2 (eWOM) explains an additional 25% of the variance, while Factor 3 covers nearly 19%. Two items from the destination trust scale, and one item from the eWOM scale were eliminated due to cross-loading and/or low loadings. The retained item loadings were all above the benchmark 0.50 as recommended by Hair, Anderson, Tatham, and Black (1998). Cronbach's alpha, composite reliability (CR) and average variance extract (AVE) were used to test the reliability of the scale items. Reliability is inversely related to the degree to which a measure is contaminated by random error, and this reports the degree of internal consistency. Cronbach's alphas were between 0.87 and 0.90 above the cut-off point of 0.60 as noted by Hair et al. (1998). CR ranged from 0.88 to 0.90; and average AVE ranged from 0.55 to 0.75 above the cut-off level of 0.50 (Hair et al., 1998), and they were assessed using the Stats Tools Package developed by Gaskin (2012). The current findings suggest evidence of convergent and discriminant validity as evident in Table 1. Furthermore, the inter-factor correlation among the study variables presented in Table 2 were all below 0.85, providing a confirmatory support for discriminant validity as suggested by (Kline, 2005).
Although this relationship was not hypothesised, the indirect effect of eWOM on travel intention through destination trust was assessed using bootstrapping analysis with a resample of (n = 2000) following recommendations from Preacher & Hayes (2004) and Shrout & Bolger (2002). The analysis shows that the standardised indirect effect of eWOM on intention to travel through destination trust was 0.270 (p < 0.001, 95% confidence interval: 0.18 -0.38). A practical interpretation of this finding, suggests that a 1 unit increase in eWOM will increase intention to travel by 0.27 units through destination trust (see Table 4).
Finally, a multi-group moderation analysis was conducted to see if there was a difference between genders. First, the dataset was divided into two (i.e. gender-wise), and then the impact of the study variables on one another were tested gender-wise. As a next step, the coefficients were observed, and the invariance was analysed based on these coefficients. Chi-square and significance tests were used in interpreting the moderating effect using Gaskin's (2012) Stats Tools Package. Hypothesis 4a posits that gender will moderate the relationship between eWOM and destination trust; the result shows that gender indeed moderates the relationship. More practically, the relationship between eWOM and destination trust is significant for both genders but stronger for males. As such, H4a received empirical support. Hypothesis 4b posits that gender will moderate the relationship between eWOM, and intention to travel; the result shows that gender does not moderate the relationship. The relationship between eWOM and intention to travel is not significant for both genders. As such, H4b did not receive empirical support. Hypothesis 4c posits that gender will moderate the relationship between destination trust, and intention to travel; the result shows that gender moderated the relationship. The relationship between destination trust and intention to travel was significant for both genders but stronger for females. Thus, H4c received empirical support (see Table 5).

Discussions and conclusion
This research extends prior work in the field of eWOM communication, both on methodological (i.e. the use of both procedural and sophisticated statistical approaches) and theoretical fronts (i.e. linking eWOM with trust transfer theory, and the application of gender variance in the med-tour industry). The present findings are theoretically and practically important because the study has provided a more nuanced understanding of   the relationships between eWOM, destination trust, and travel intention as moderated by gender. As noted earlier, previous studies noted that tourists are especially keen to post and read criticisms and/or complimentary experiences online; this may affect the behavioural outcomes of potential med-tourists. Nonetheless, the majority of tourists in this study were in leisure tourism and not in the med-tour industry, providing additional space for more exploration. This study asserted that eWOM has a great influence on travel intention; regression analysis suggested that eWOM has a positive and significant impact on med-tourist's travel intentions. This supports the notion that positive online reviews may increase visitors travel intention, which is in line with previous travel and leisure eWOM research (e.g. Duverger, 2013;Reza Jalilvand & Samiei, 2012b;Jani & Hwang, 2011). Next, the paper hypothesised that eWOM has a great influence on destination trust, regression analysis suggests that eWOM has a positive and significant impact on med-tourist's destination trust. This supports the notion that positive online reviews may increase visitors destination trust; similarly, prior research has documented a positive relationship between eWOM and destination image (Jalilvand, Samiei, Dini, & Manzari, 2012;Ko & Kim, 2011), attitudes (Vermeulen & Seegers, 2009), and destination choice (Zhu & Lai, 2009). Hypothesis 3 states that med-tourists' destination trust is positively related to travel intention, the findings show that med-tourists' destination trust significantly influenced their travel intentions. This is in line with prior findings that brand trust exerts a significant effect on purchase intention (e.g. Winch & Joyce, 2006;Wu & Wang, 2011).
This paper also hypothesised that the relationship between eWOM and destination trust will be moderated by gender. The finding suggests that gender moderates the relationship between eWOM and destination trust, and the linkage was stronger in men. The outcome of this study contradicts Feingold's (1994) findings, which claimed that women are more trusting than men. eWOM influenced the intention to travel to a medical destination, and the impact was about 1.2 times stronger for men. To this end, the finding highlights the independent, agentic and 'masterful' nature of men when it comes to risk-taking behaviour (Eagly & Wood, 1991;Taylor, Lewin, & Strutton, 2011).
Hypothesis 4b stated that the relationship between eWOM and travel intention will be moderated by gender. The finding suggests that gender failed to moderate the relationship between eWOM and travel intention. A first possible explanation for this null finding might relate to the fact that eWOM has different forms, e.g. eReferral which coexists among people who share strong ties, which tend to influence women more (Abubakar, Ilkan, & Sahin, 2016). It is asserted that eWOM will always impact travel intention irrespective of the gender. Finally, gender also moderated the relationship between destination trust and travel intention. The linkage was significant and stronger for females, and the impact was about 2.5 times stronger for females. This is because destination trust reduces risk perception, and the probability of adverse effects is lowered at this stage; this is in line with Byrnes et al. 's (1999) and Pascual-Miguel, Agudo-Peregrina, and Chaparro-Peláez's (2015) arguments, e.g. women are more risk-averse. Accordingly, from a destination marketer's point of view, the effects of eWOM on various response variables are deemed relevant in the med-tour industry. This paper enriches our understanding of eWOM in conjunction with destination trust and gender; and also draws implications on how medical destinations can better leverage this competitive advantage tool. Med-tour practitioners seeking to attract female med-tourists can use eWOM in the form of referrals for product and/or service recommendations. The greater the destination trust, the greater the tendency that a female med-tourist will visit a medical destination, whereas practitioners seeking to attract male med-tourists can also use eWOM for product and/or service recommendations, but they should bear in mind that men perceive the anonymity provided by the internet as an ideal venue in which they can openly share their concerns and thus break out of the traditional masculine role to ask for help (Mo, Malik, & Coulson, 2009;White & Dorman, 2000).
Relying on the outcomes from this study, the researcher recommends that destination marketers in the med-tour industry can utilise eWOM to maximise their market share by considering the type of medical care they provide in respect of the target gender. For instance, the outcome of this study delineates that positive eWOM enhanced destination trust in both genders but was stronger among male med-tourists; this is because men are agentic in nature. Women who trusted a medical destination have a higher intention to travel because their concern for both privacy and misuse of information is reduced. Another approach to enhance destination trust is by increasing the credibility of eWOM by including user identity, or by using referral systems, in line with trust transfer theory propositions. Previous studies have noted that women's concerns are abated when a website is recommended by a friend, but this is not the case for men (Garbarino & Strahilevitz, 2004;Meyers-Levy & Loken, 2015).
Among other things, the shortcomings in this paper include the data collection approach, sample size, and the cross-sectional design used in the study. In addition, the current findings are only applicable to North Cyprus. Thus, the outcome and interrelationship between the variables should be viewed with caution given the above limitations. Future studies should incorporate the effect of technologyadvancement in the model. A second plausible model advancement includes the incorporation of travel distance, destination image, and intention to revisit. Finally, the same model can be tested on other tourism industries, such as educational tourism.