117

The influence of social media
on tourists ’ behaviour

Evidence from China

Yingchan Luo1,a

Linjia Zhang2,a

Jingjing Du3,b

a International Business School Suzhou (IBSS) at Xi’an Jiaotong-Liverpool University

b Durham University

Tourism is playing an increasingly important role in Chinese economic development because of its rapid growth in recent years. The impact of internet technology on the traditional tourism industry has also become more apparent. The Chinese government has initiated two strategies to strengthen the combination of the traditional tourism industry and internet technology: “Mass entrepreneurship and Innovation” and “Internet plus”. Policy makers pay close attention to the transformation of the tourism industry, as well as the improvement of tourism quality. With those strategies put into effect, tourism has evolved to a new level. As a result, China has reported being in the fourth position ranked by international tourism arrivals and receipts in 2016 (World Tourism Organization, 2017). Data from the National Bureau of Statistics of China shows that the number of domestic tourists increased by 50% between 2010 and 2016, and foreign exchange income from 118international tourism increased by 50% as well. In 2017, the number of domestic tourists increased by 12.8% compared to the same period in 2016, which had already reached 5.01 billion RMB. Total tourism revenue reached 5.4 trillion RMB in 2017, achieving an increase of 15% in total. Generally, annual tourism income contributed to 11.04% of China’s GDP in 2017, and provided 10% of the employment. We also noticed that the Chinese online transactions of the tourism industry increased by 34% compared with 2015, which amounted to 602 billion RMB in 2016; e-tourism is also developing rapidly.

Previous studies have conducted surveys to research the correlation between websites’ information quality and tourists’ potential travel intentions. Most of the findings proved that the quality of the information provided by the destination websites has a positive influence on tourist behaviour. Additionally, various studies have proved that personal experience and evaluation positively affect online WOM (word of mouth) and intention to purchase, because people trust in personally shared information more than that from a formal source (Agag and El-Masry, 2015). However, fewer studies concentrate on the influence of social media on tourist behaviour. This study will investigate the factors that can affect the influence of social media on tourists’ travel arrangement.

We proposed several hypotheses based on previous studies and theories, and then built a structure to test those by partial least square structural equation modelling (PLS-SEM). A questionnaire was designed to collect data. The structure of this research is organized as follows. Section 1 introduces the theoretical model and the proposed hypotheses. Section 2 explains the methodology and describes the data collected by the questionnaire. Section 3 presents the results of the data analysis. Section 4 discusses the results and conclusion provides the limitation and contribution on practical application and future research.

119

1. Theoretical model and hypotheses

1.1 DeLone and McLean ’ s IS success model
and website quality model (WebQual)

With the development of information communication technology, the internet has become a major platform for users to share and search information. Numerous scholars have conducted research to identify the factors that actually measure the information systems’ success. Because of the different interpretation of success, researchers have a diverse emphasis on factor setting. To integrate those information system success theories, DeLone and McLean introduced an IS success model in 1992. The IS success model was proposed by reviewing 180 articles. In this model, DeLone and McLean (1992) provided six main dimensions, which are “system quality”, “information quality”, “use”, “user satisfaction”, “individual impact” and “organizational impact” (DeLone and McLean, 1992). The model structure is shown in Figure 1.

Fig. 1 – The DeLone and McLean’s IS success model (1992).

However, as IS success model has been widely used by researchers, some limitations and alterations also have been pointed out. Pitt et al. (1995) suggested that “service quality” should be considered more important than IS related products when seeking to determine the IS success’ efficiency (Chung, et al. 2015). In DeLone and McLean’s model (1992), ‘system use’ had been considered as a mediator between 120information design success and users’ satisfaction. However, the consequences of the use of information system should also be considered to measure the final success (Seddon, 1997). Meanwhile, research in behavioural economics suggested that the behavioural and perceptual variables of IS evaluation should be considered separately when determining the level of IS success. Thus, another modification of IS success model was pointed out by Seddon (1997), who adapted the original IS success model into two parts: the partial behavioural IS use model and IS (design) success model. The model structure is shown in Figure 2, where the perceived usefulness is used as a mediating variable which directly influences the user satisfaction (Chen, 2010).

Fig. 2 – Seddon reconstructed IS success model (1997).

DeLone and McLean (2003) updated their initial model (1992) by changing the major elements to make this model more suitable for e-commerce business structure. The main re-specifications of the updated structure shown in Figure 3 include the addition of ‘service quality’ and ‘intention to use’. Moreover, this reconstruction replaced 121individual impact and organizational impact by net benefit to measure satisfaction of using e-commerce systems. In this model, “system quality” is measured by several characteristics of the technology system, which are “reliability”, “data quality”, “flexibility”, “ease-of-use”, and “functionality”. “Information quality” is measured by “precision”, “timeliness”, and “reliability”. “Service quality” is measured by “reliability”, “intimate services”, “responsiveness”, and “meticulous care” (DeLone and McLean, 2003).

Fig. 3 – DeLone and McLean’s updated IS success model (2003).

The IS success model has been extensively used in social science research since 1993, in order to measure a website’s quality or an online system’s quality. Chen (2010) conducted research to examine the relationship between e-learning systems and job outcomes. He observed that both information quality and system quality positively impact usefulness and satisfaction of an e-learning system. In his research, perceived usefulness has been proved to positively affect users’ satisfaction. In addition, the IS success model has been used to evaluate knowledge sharing behaviour in the healthcare industry. Alali and Salim (2013) built a multidimensional model based on the IS success model and Technology Acceptance Model (TAM) to measure the VCoPs success. VCoPs (virtual communities of practice) is one of the learning channels used by health organizations to improve practitioners’ knowledge. In 122the model, they put VCoPs perceived usefulness as an independent variable. The results show that VCoPs members’ use satisfaction positively depends on the quality of shared knowledge, system and services. But the ease-of-use does not affect users’ satisfaction and system ease-of-use (Alali and Salim, 2013). In 2015, Hanae Roky and Meriouh conducted research to evaluate the information system of the automotive industry (XPPS) based on the IS success model. The results showed that information quality is a crucial factor that enhances users’ satisfaction and continues use intention. Both use and use intention affect individual impact on XPPS performance positively. Gao et al. (2015) adopted the IS success model to identify which elements in mobile purchases could attract potential consumption. They found that information quality and security are key factors to influence eTrust, service quality and consumption quality (Gao et al. 2015). Additionally, a similar article discussed online shopping platforms and consumers’ purchase intention, and demonstrated that an e-commerce platform’s quality influences consumers’ satisfaction more than continued using intention (Chen and Cheng, 2009). Some previous applications of the IS success model are shown in the Table 1.

Tab. 1 – Previous applications of the IS success model.

Authors	Method		Theory	Findings
Chen (2010)	partial least square (PLS)	Questionnaire (350 responses in Taiwan, China)	IS success model	The study examines the relation between e-learning systems use and overall job outcomes. And employees’ e-learning systems use is significantly associated with overall job outcomes.
Alali and Salim (2013)	PLS-SEM	Web-based survey (362 responses in Middle East)	IS success model	The quality of shared knowledge, the system, and service, and perceived usefulness of health VCoPs will affect user’ satisfaction, while perceived ease of use did not influence on a user’s satisfaction.
123 Hasanov and Khalid (2015)	ordinary least square (OLS)	Survey (304 reponses in Malaysia)	WebQual model	The results indicate that website quality has indirect impact on online purchase intentions of health foods in Malaysia through the full mediation of customer satisfaction.
Chung et al. (2015)	PLS-SEM	Questionnaire (169 responses in Korea)	IS & ECM	This paper examined co-relation between destination website quality and the intention to continuous use, and the subsequent effects on the intention to visit the destination. And Information quality highly affects website quality
HanaeRoky (2015)	Structural equation modeling (SEM)	Questionnaire (200 responses)	IS success model	Information quality plays a key role in user satisfaction and intention to use of the system; there exists a significant relationship between service quality and the use of the Information system; the effect of both intention to use and use on individual impact and finally the positive influence of the individual impact on the organizational performance of XPPS(a system).
Kim et al (2016)	PLS-SEM	survey & questionnaire (China)	information quality framework	The results suggest that various aspects of tourism IQ in social media are positively associated with different types of destination image.

An expanded application model of the IS success model is the WebQual model, which is also widely used to evaluate the quality of e-commence websites. In this model, websites’ usability, information quality and service interaction quality are three main elements that are measured to test a website quality. Hasanov and Khalid (2015) investigated the relationship between website quality and consumers’ repurchase intention of organic food in Malaysia. Their article found that there is no significant positive correlation between repurchase intention of organic food online in Malaysia and website quality (Hasanov and Khalid, 2015).

124

Though numbers of studies measure system efficiency, user satisfaction and intention to use according to the DeLone and McLean IS success model in the social science area, few of them use the IS success model to identify the quality of social media and its influence on tourists’ intention to travel (Hanae Roky and Meriouh, 2015). Chung et al. (2015) conducted a survey on the Korean creative travel economy. The results showed that website quality will finally positively affect tourists’ intention to travel and the quality of information provided on the Destination Management Organizations’ (DMOs) websites is the most important factor to influence consumers’ decisions. Perhaps this is the first article to use the IS success model to prove that the relationship between destination websites and people’s decision-making actually exists (Chung et al. 2015).

Though previous studies have found a positive correlation between destination websites and people’s decision making, few studies identify the influence of broad social media on tourists’ potential intention to travel and travel decisions making. Based on the above application of the IS success model, we proposed the following hypotheses. In this research, we mainly use information quality and online ease-of-use to measure social media platforms’ quality and hypothesize that these two factors affect the social media usefulness and satisfaction. Additionally, we concentrate on people’s attitudes towards online booking like e-ticketing and tourism products purchasing. Their evaluation is considered as one of the aspects of social media satisfaction, and we focus on finding whether the social media quality will affect the online booking satisfaction.

Hypothesis 1. The online information quality has a positive impact on the usefulness of the social media platform.

Hypothesis 2. The online information quality has a positive impact on the tourists’ online booking satisfaction.

Hypothesis 3. The system quality has a positive impact on the usefulness of the social media platform.

Hypothesis 4. The system quality has a positive impact on tourists’ online booking satisfaction.

Hypothesis 5.Usefulness of the social media platform has a positive impact on the tourists’ online booking satisfaction.

125

1.2 Expectation-Confirmation model (ECM)

The Expectation-Confirmation model was first proposed by Bhattacherjee in 2001, based on expectation-confirmation theory (ECT). Previous studies have examined the elements that actually influence one’s continued use intention of an information system. Based on those cognitive findings, ECT is widely used in the psychological area to study consumer behaviour, especially in consumer satisfaction and repurchase behaviour (Bhattacherjee, 2001). Oliver (1993) explained the logic mechanism of how satisfaction influences one’s continued purchase intention. The ECT framework can be summarized in five steps: Firstly, consumers have a pre-consumption expectation of products or services from others sharing experiences. Secondly, they buy the products or services and form their own judgment about the products’ performance. Thirdly, consumers will compare products’ performance with their initial expectation. Fourthly, based on their original expectation and actual performance, consumers will form their satisfaction of the whole transaction experience. Finally, people’s satisfaction will affect the repurchase behaviour. If their pre-consumption expectation is greater than actual performance, consumers will lose their trust and become unsatisfied with the transaction, and potentially will not use this online system anymore and vice versa. ECT insists that satisfaction from prior use of products or services has a strong impact on continuous use behaviour (Oliver, 1980).

In other words, the Expectation-Confirmation model is actually analysing the users’ continued use intention of information systems since IS users continuous use intention is similar to consumers’ repurchase behaviour (Bhattacherjee, 2001). Lee and Kwon (2010) conducted a survey to prove their suggestions about an extended ECM. They showed familiarity (how familiar users are with the website services based on personal experience and previous learning) and intimacy (a kind of feeling of liking and emotional bonding) positively affect users’ continued use behaviour (Lee and Kwon, 2010). ECM has been widely used by scholars in web-based areas, usually combined with TAM (Technology-Acceptance model) and the IS success model to analyse the interaction among website quality, users’ satisfaction and continuous use intention. For example, Chow and Shi (2014) proved that pre-adoption expectation and satisfaction will influence students’ intention to learn by the e-learning system (Chow 126and Shi, 2013). Also Chen et al. (2013a) built an integrated model with technology readiness and ECM to explain the factors that influence the mobile service users’ intention to use. The model structure is shown in Fig. 4 and all the hypotheses are supported (Chen et al. 2013b).

Fig. 4 – Chen et al. (2013) extended ECM.

ECM has also been used in tourism. Chung developed a study to prove that the quality of a destination’s website will actually affect tourists’ expectations before travel, by affecting their image according to the IS success model and ECM (Chung et al. 2015). Some reviewed articles based on ECM are shown in Table 2.

Tab. 2 – Previous studies of ECM.

Authors	Method		Theory	Findings
Lee and Kwon (2011)	Structural equation modeling (SEM)	Web-based survey	expectation–confirmation model (ECM)	This study proposed an extended ECM to investigate how cognitive factors like perceived usefulness, affective factors (familiarity and intimacy) actually do affect users’ continuance intention.
Chung et al. (2015)	PLS-SEM	Questionnaire (169 responses in Korea)	IS & ECM	This paper examined co-relation between destination website quality and the intention to continuous use, and the subsequent effects on the intention to visit the destination. And Information quality highly affects website quality
127 Chen et al. (2013)	confirmatory factorial analysis (CFA)	Questionnaire (368 responses in China)	expectation–confirmation model (ECM)	This study integrated technology readiness into the expectation–confirmation model (ECM) for explaining individuals’ continuance of mobile data service usage.

The ECT pointed out that pre-consumption expectation determined consumers’ satisfaction and then influences repurchase intention (Bhattacherjee, 2001). If a destination website exaggerates the description of their tourist attraction, tourists will have a higher expectation or image of that tourism destination. Their satisfaction with both the website and scenic spot will go down, and they will have a disappointed impression that affects their continuous use intention of the website and their revisit behaviour. Thus, it is reasonable for us to combine the IS success model with ECM to test social media platforms’ quality and the influence on tourists’ travel arrangements. Based on the previous studies, we proposed the following hypotheses.

Hypothesis 6. Usefulness of social media platform has a positive impact on tourists’ frequent use of online travel agencies (OTA) and destination websites.

Hypothesis 7. Usefulness of social media platform has a positive impact on an online platform’s continued use intention.

Hypothesis 8. Tourists’ online booking satisfaction has a positive impact on an online platform’s continued use intention.

Hypothesis 9.Tourists’ online booking satisfaction has a positive impact on tourists’ frequent use of OTA and destination websites.

Besides these two models, we also proposed hypotheses of electronic Word-Of-Mouth. Previous research has proven that eWOM will influence people on decision making and consumption or use intention. Also Abubakar and Ilkan (2015) confirmed that WOM does affect tourists’ intention to travel. Agag (2016) mentioned that to make a purchase decision, people prefer to rely on a personal shared 128experience rather than on a formal source. Based on this evidence, we propose the following hypothesis. We suppose that a person’s familiarity with social media will influence his/her tendency of searching information and using the Internet. Then it will affect the extent to which they are affected by online WOM. Additionally, “tourists’ degree of use” and “continued use intention” are the other two factors reflecting this extent. “The direct influence from social media on travel arrangements” represents the effectiveness of others’ experiences as an influence on one’s choice, which is the influence degree of WOM on travel arrangements.

Hypothesis 10. Tourists’ frequent use of online platforms has a positive impact on the direct influence from social media on travel arrangements.

Hypothesis 11. Tourists’ continued usage intention of online platforms has a positive impact on the direct influence from social media on travel arrangements.

Hypothesis 12.Tourists’ degree of familiarity of social media has a positive impact on the direct influence from social media on travel arrangements.

2. Research method and data description

2.1 Research structure

In this study, we aim to investigate the correlation between social media platforms and potential tourists’ travel intentions, as well as the influence path. We proposed a hypothesis that social media will influence tourists’ intention to travel through usefulness and users’ satisfaction. Furthermore, we mainly test information quality and ease-of-use to measure social media quality. We build a structure to test those hypotheses, which is shown in Figure 5. Each variable or construct in the structure is evaluated by several questions, which have been empirically designed according to previous theories and research. 129All questions in the structure test the OTAs efficiency and destination websites’ quality, and people’s satisfaction of both websites’ quality and tourists’ products. All measurement items are shown in Table 3.

Fig. 5 – Structure – really should be continued use intention.

Tab. 3 – Measurement items (questions).

CONSTRUCTS	MEASUREMENT ITEMS
information quality	21. I think the information update on the website is very timely and the information is very effective. 22. I believe that the video and pictures on social media are the same as the real scenes of the scenic spot, without exaggeration or over rendering. 23. I think the travel information provided by the third-party online platforms (ctrip, tuo, etc.) is timely. 24. I believe that the tourism products launched by third-party platforms (ctrip, tuo, etc.) are highly cost-effective. 25. I believe that the information provided by third-party platforms (ctrip, tuo, etc.) does not have any false elements. 26. I think the safety of tourism products launched by third-party platforms (ctrip, tuo, etc.) can be guaranteed.
ease-of-use	27. I can quickly find the information I need on the official website of scenic spot. 28. I can easily search for the sharing travel information I want on the Internet.
130 social media usefulness	13. I think the official website of the scenic spot can provide enough information for me. 14. I think the information provided by the official website of the scenic spot is of great reference value. 15. I think the travel tips shared by others on the social media is a valuable guide for me. 16. I believe that the information shared on social media is very helpful to my travel planning.
online booking satisfaction	11. I have bought travel products online and have no unpleasant experiences during my travels. 20. I believe that the travel products launched by third-party platforms (ctrip, tuo, etc.) have improved my travel experience. 17. I think online booking can save time. 18. I think the booking hotel online can guarantee the quality and safety of the hotel. 19. I think it’s very convenient to book air tickets and train tickets online.
continued usage intention	34. I will continue to pay attention to the sharing information of travel arrangement online. 35. I think online travel platforms will replace offline travel agencies. 36. I think I will continue to use online platforms to plan my own travel routes. 37. I think I will continue to use the online platform to book hotels. 38. I think I will continue to use the online platform to book tickets.
direct influence from social media	2. A friend’s recommendation will affect my choice of travel destination. 3. A friend’s recommendation will affect my travel route planning. 4. The travel tips shared by others on social media will affect the choice of my travel destination. 5. Some travel documentaries on the Internet can affect the choice of my destination. 6. I think some travel tips shared on the social media will become my motivation of traveling to one place. 32. I refer to other people’s travel arrangements to plan my own itinerary.
degree of familiarity	1. I think the travel route designed by myself is interesting. 10. I usually follow some microbloggers who are interested in sharing travel notes. 12. I’m familiar with MaFengWo (A Chinese website that shares travel tips) and Airbnb. 31. I usually pay attention to the topics of travel on platforms such as Weibo, Facebook and Instagram.
frequent use	7. I will book the hotel by reference to others’ share. 8. I often use the online platform (12306 website and other apps) to book air tickets and train tickets. 9. I often book hotels with online platforms. 29. I’m used to comparing the hotel prices provided on the various online platforms before booking. 30. I am used to browsing the official website of scenic spots before I travel.

131

Meanwhile, we divided the total 445 data into two control groups according to age and gender in this study. Through testing those hypotheses based on different groups of data, it is expected to see whether age and gender are two influencing factors that will affect our hypotheses. Previous research has pointed out that demographic characteristics may influence consumer behaviour towards social media and online consumption, which includes age, gender, income and education. Differences between female and male online purchasing behaviour are indicated in Mona and Seyed’s research (Mona and Seyed, 2013). Moreover, Hasanov and Khalid (2015) found that age difference will affect consumers’ online purchasing intentions with regard to e-ticketing in Malaysia (Hasanov and Khalid, 2015). They suggested that unlike the older generation, the younger generation prefers online shopping because they are familiar with the new technology and the internet (Hasanov and Khalid, 2015). In 2008, Cretzel found that gender difference causes different behaviour on searching and sharing travel information online. Females demonstrate a more active attitude for sharing information online and are influenced by WOM (Kavouraa and Stavrianea, 2015). Thus, we separated the data into two control groups to find out whether there exist different results for hypotheses and structure testing.

2.2 Data collection and description

Based on previous studies, a questionnaire is designed to investigate participators’ perception on social media. The questionnaire is divided into three sections. The first section focuses on collecting participators’ general information, such as age, income, education, and their travel preference and habits. The second section focuses on researching respondents’ attitude towards social media services, and their evaluation of social media information and usability. This includes people’s frequent use of social media and continued use behaviour. The last section is used to test their online consumption behaviour and their familiarity with social media platforms, as well as the degree of influence of social media on travel arrangements.

Most of the questions we designed as measurement items in this research were adapted from previous studies and the two models - IS success model and ECM. All of the answers to the questionnaire are designed to utilize a five-point Likert scale with strongly disagree (1) 132and strongly agree (5). We use the IS success model to design questions to measure information quality (6 items) and ease-of-use (2 items). Social media usefulness (4 items), online booking satisfaction (5 items), frequent use (5 items) and continue to use (5 items) are modified from Chung et al. (2015) based on ECM. Each question is shown in Table 3.

In total, 470 responses were collected; 445 of them were effective and used to conduct the analysis. The general information is shown in Table 4. The respondents were made up of approximately 40% male and 60% female. More than three-quarters of our respondents were people in the 18 to 25 age group. The younger generation seems more familiar with new technology than the older generation and they are more used to consume online frequently (Hasanov and Khalid, 2015). Most of our participators are undergraduate students or people with even higher degree and around 75% of them earn less than 5000 RMB per month. That is because the student respondents receive most of their “income” from their parents. Most of the participators travel around 3 times per year and they always travel with their parents and friends; only 25% of people take package tours. The majority of participators, which occupied 67% of the total number would like to make their own travel arrangements by searching for shared information online. Additionally, 20% of people choose to buy semi-self-help travel products, and the same number of people would like to join a tour group. The most common types of destination tourists prefer to visit are natural scenery and historical relics.

Tab. 4 – General information.

Profile	Category	Frequency	Percentage	Profile	Category	Frequency		Percentage
Gender	Male	180	40.45%	Travel	<1 time		69	15.51%
	Female	265	59.55%	Times/year	1-2 times		206	46.29%
					2-3 times		105	23.60%
Age	<18	2	0.45%		3-5 times		40	8.99%
	18-25	301	67.64%		>5 times		24	5.39%
	26-30	25	5.62%
	31-40	22	4.94%	Travel with	Parents		264	59.33%
	>40	95	21.35%		Friends		347	77.98%
133					Tour group		111	24.94%
Academic	Under Bachelor	69	15.51%		Alone		91	20.45%
Back-ground	Bachelor	356	80.00%
	Master	13	2.92%		Self-help travel		300	67.42%
	Doctor	7	1.57%	Travel ways	Tour group		87	19.55%
					Tourism product		12	2.70%
Most likely	Natural scenery	364	81.80%		Semi-self-help travel product		86	19.33%
to visit	Historical relics	213	47.87%
	Amusement park	133	29.89%	Income	<5000		338	75.96%
	Museum	152	34.16%	Level/month	5000-10000		81	18.20%
	Celebrity museum	52	11.69%		>10000		26	5.84%
	Other	43	9.66%

2.3 analytical method

In order to test the hypotheses and the model reliability, we use Partial Least Square (PLS) path modeling to analyze the collected data. SmartPLS 3.0 is used to do the PLS calculation and bootstrapping. Theoretically, PLS makes no assumption of data distribution, and the sample size could be small, which is appropriate for a comprehensive model with a large number of constructs. We evaluated the structure model by calculating R-square of each construct. The standard sample size of bootstrapping is 500. To evaluate the measurement model, which reflects the relationship between the constructs and indicators, we concentrate on validity: content validity, convergent validity and discriminant validity. We use composite reliability (CR), average variance extracted (AVE) and Cronbach’s alpha to measure the convergent validity. Generally, the constructs demonstrate convergent validity if CR is greater than 0.7, AVE is bigger than 0.5 and Cronbach’s alpha is greater than 0.5.

134

3. Analysis and results

3.1 Reliability and validity of measurement model

Table 5 shows the results of the confirmatory factor analysis. Most of our measurement items’ outer loading is higher than 0.6, CR of all constructs is greater than 0.7 and Cronbach’s alpha of all constructs is higher than 0.5. Thereby, the results show a satisfactory convergent validity. The Fronell-Larcker criterion shows the square root of AVE and the correlation of constructs, only when the square root of AVE is larger than the correlations with other constructs, the constructs can be discriminated as a good one. Table 6 shows an adequate discriminant validity that all the inter-construct correlation is smaller than the square root of AVE and lower than 0.8.

Tab. 5 – Confirmatory factor analysis.

	Measurement items	Outer Loadings	T Statistics (|O/STDEV|)	P Values	Composite Reliability	Average Variance Extracted (AVE)	Cronbach’s Alpha
CONTINUED USAGE INTENTION	cui1	0.571	9.177	0.000	0.838	0.528	0.756
	cui2	0.362	5.279	0.000
	cui3	0.834	39.942	0.000
	cui4	0.874	57.323	0.000
	cui5	0.851	46.182	0.000
DEGREE OF FAMILIARITY	dof1	0.615	9.785	0.000	0.758	0.442	0.582
	dof2	0.710	13.538	0.000
	dof3	0.563	7.663	0.000
	dof4	0.753	16.926	0.000
EASE OF USE	eou1	0.789	15.455	0.000	0.816	0.690	0.555
	eou2	0.870	37.585	0.000
INFPRMATION QUALITY	iq1	0.699	20.217	0.000	0.879	0.547	0.837
	iq2	0.769	30.998	0.000
	iq3	0.732	22.030	0.000
	iq4	0.781	28.069	0.000
	iq5	0.708	15.082	0.000
	iq6	0.745	21.882	0.000
135 ONLINE BOOKING SATISFACTION	sms1	0.305	3.798	0.000	0.776	0.432	0.654
	sms2	0.534	7.648	0.000
	sms3	0.821	30.822	0.000
	sms4	0.655	15.276	0.000
	sms5	0.823	37.922	0.000
SOCIAL MEDIA USEFULNESS	smu1	0.629	11.365	0.000	0.839	0.571	0.754
	smu2	0.669	12.311	0.000
	smu3	0.849	43.866	0.000
	smu4	0.849	42.315	0.000
DIRECT INFLUENCE FROM SOCIAL MEDIA	tm1	0.687	15.214	0.000	0.855	0.497	0.799
	tm2	0.681	15.792	0.000
	tm3	0.752	22.535	0.000
	tm4	0.731	21.809	0.000
	tm5	0.694	15.891	0.000
	tm6	0.679	19.353	0.000
FREQUENT USE	ub1	0.682	17.915	0.000	0.783	0.438	0.655
	ub2	0.780	27.947	0.000
	ub3	0.792	29.762	0.000
	ub4	0.632	12.720	0.000
	ub5	0.295	3.652	0.000

Tab. 6 – Adequate discriminant validity.

	Fornell-Larcker Criterion	1	2	3	4	5	6	7	8
1	online booking satisfaction	0.657
2	continued usage intention	0.646	0.727
3	degree of familiarity	0.407	0.507	0.665
4	ease of use	0.297	0.349	0.294	0.831
5	information quality	0.354	0.264	0.283	0.454	0.740
136 6	social media usefulness	0.497	0.412	0.421	0.474	0.439	0.756
7	direct influence from social media	0.436	0.513	0.447	0.258	0.172	0.467	0.705
8	frequent use	0.643	0.628	0.486	0.335	0.207	0.446	0.563	0.661
note: the ‘red’ number in the Fornell-Larcker Criterion represent the square root of AVE. For an adequate discriminant validity, the square root of AVE must be higher than correlation between the other constructs.

3.2 Model fitting and hypothesis testing

By using the bootstrap analysis technique, we estimate the path coefficient and the t-test to measure the proposed structural model and the results are shown in Tables 7, 8, and Figure 6. The results show a distinct correlation among information quality, social media usefulness, online booking satisfaction, frequent use, familiarity with social media and continued usage intention. Tables 7 and 8 present the results of model path testing and R-square of each constructs, which shows that all hypotheses 1-12 are supported except hypothesis 3 (ease-of-use will positively influence online booking satisfaction) if we suppose the confidence level is 95%. If we take a more accurate judgment of the data and take the confidence level to 99%, hypotheses 2 and 7 are also been rejected.

Hypotheses 1 and 4 are indicating that social media usefulness is significantly influenced by online information quality (β=0.282, p=0.000**) and ease of use (β=0.346, p=0.000**), with ease-of-use being one of the measurement factors of system quality in the IS success model. Thus, we have proved that information quality and system quality of social media platforms are crucial elements for ‘usefulness’. Hypothesis 2 shows that information quality (β=0.159, p=0.036*) has a positive impact on online booking satisfaction. Unfortunately, hypothesis 3 has been rejected, which means that the system quality seems to have no direct influence on people’s evaluation of online products and services. Hypothesis 5 shows a distinct positive relationship between social media usefulness (β=0.414, p=0.000**) and online booking satisfaction. Additionally, H1 and H5 have proved that information quality will both directly 137and indirectly affect the online booking satisfaction while social media usefulness acts as a mediator. Hypotheses 6 to 10 investigate whether social media usefulness and online booking satisfaction are two factors that influence people’s usage and continued use behaviour; the path coefficient has been shown in Table 7 and Figure 6. Hypotheses 11 and 12 are the keystone of our proposed structure. Hypothesis 10 indicates that the availability of social media will enhance the influence of social media on tourists’ travel arrangements. The results show continued use intention (β=0.205, p=0.000**) has a significant effect on the direct influence of social media on travel planning. Hypothesis 12 indicates that the familiarity of online platforms (β=0.173, p=0.000**), where people share individual experiences, has a positive impact on the degree of social media’s effect on travel arrangements.

R-square represents how the constructs can be explained by exogenous latent variables, and we calculate the value of R-square to evaluate the predictive power of the proposed structure.

Tab. 7 – Results of structural model (with total data).

Hypothesis	Path	T Statistics	P Values	Corr	Results	Result corr
H1	information quality will positively influence social media usefulness	6.35	0.00	+	supported	+
H2	information quality will positively influence online booking satisfaction	2.11	0.04	+	rejected	+
H3	ease of use will positively influence online booking satisfaction	0.51	0.61	+	rejected	+
H4	ease of use will positively influence social media usefulness	7.58	0.00	+	supported	+
H5	social media usefulness will positively influence online booking satisfaction	7.89	0.00	+	supported	+
H6	social media usefulness will positively influence frequent use	3.11	0.00	+	supported	+
H7	social media usefulness will positively influence continued usage intention	2.39	0.02	+	rejected	+
H8	online booking satisfaction will positively influence continued usage intention	12.09	0.00	+	supported	+
138 H9	online booking satisfaction will positively influence frequent use	10.58	0.00	+	supported	+
H10	frequent use will positively influence direct influence from social media	6.56	0.00	+	supported	+
H11	continued usage intention will positively influence direct influence from social media	3.75	0.00	+	supported	+
H12	degree of familiarity will positively influence direct influence from social media	3.86	0.00	+	supported	+

Tab. 8 – R-square (with total data).

	R Square	R Square Adjusted
Online booking satisfaction	0.27	0.27
Continued usage intention	0.43	0.43
Social media usefulness	0.29	0.28
Direct influence from social media	0.38	0.38
Frequent use	0.44	0.43

Fig. 6 – Overall model path estimation
(PLS analysis with total 445 data, note: β=path coefficients).

139

3.3 Comparison of control groups

In this research, we use two control groups to find whether the testing results of hypotheses will change with different age and gender. The data set has been divided firstly into two groups: those aged 30 and older and those younger than 30, to test age effect. Then the data is divided into male and female groups to investigate gender effect.

Table 9 presents the results of hypotheses path testing based on the proposed structure with different groups of data. It can be seen from the chart that only hypotheses 1, 4 and 9 are supported in all conditions, which indicates a distinct correlation between information quality and social media usefulness, ease-of-use and social media usefulness, online booking satisfaction and frequent use. However, most of the hypotheses are rejected in the male group.

140

Tab. 9 – Results for all groups.

Hypothesis	Path	Total data		AGE<30		AGE>30		FEMALE		MALE
		Results	Corr	Results	Corr	Results	Corr	Results	Corr	Results	Corr
H1	information quality -->social media usefulness	supported	+	supported	+	supported	+	supported	+	supported	+
H2	information quality -->online booking satisfaction	rejected	+	rejected	+	rejected	+	rejected	+	rejected	+
H3	ease of use --> online booking satisfaction	rejected	+	rejected	--	supported	+	rejected	--	rejected	+
H4	ease of use --> social media usefulness	supported	+	supported	+	supported	+	supported	+	supported	+
H5	social media usefulness --> online booking satisfaction	supported	+	supported	+	supported	+	supported	+	rejected	+
H6	social media usefulness --> frequent use	supported	+	supported	+	supported	+	supported	+	rejected	+
H7	social media usefulness-->continued usage intention	rejected	+	rejected	+	rejected	+	rejected	+	supported	+
H8	online booking satisfaction -->continued usage intention	supported	+	supported	+	supported	+	rejected	+	rejected	+
H9	online booking satisfaction --> frequent use	supported	+	supported	+	supported	+	supported	+	supported	+
H10	frequent use --> direct influence from social media	supported	+	supported	+	supported	+	supported	+	rejected	+
H11	continued usage intention--> direct influence from social media	supported	+	supported	+	rejected	+	rejected	+	rejected	+
H12	degree of familiarity --> direct influence from social media	supported	+	supported	+	rejected	+	supported	+	rejected	+

141

3.4 Discussion and evaluation

We proposed hypotheses 1-10 to test the relationship between social media quality, usefulness, people’s satisfaction, frequent use, and continued use intention. Additionally, we have proposed hypotheses 11 and 12 to test whether people’s attitude towards the availability of social media will enhance or reduce social media’s effect on travel arrangements. All the path coefficients and data analyses are included in tables 7 and 9. From the analysis, we demonstrated that our suggestions have different results for different groups of data. In other words, demographic characteristics like gender and age will influence the results of our hypotheses.

Hypothesis 2 is rejected in all the groups, which indicates that social media information quality has no effect on tourists’ satisfaction and social media usefulness. Meanwhile, hypothesis 3 shows that ‘ease-of-use’ also has no impact on tourists’ degree of satisfaction with social media. Li et al. (2017) conducted a research on the influence of economic hotel websites on consumers’ online booking behaviour. In their study, the results show that website usability and ease-of-use are insignificant in affecting consumers’ continuous booking intentions. The reason is that most of the economic hotels’ websites in China are similar in the information provided and website images. In our study, ease-of-use also insignificantly influenced people’s satisfaction of online booking. We suppose that the reason is similar to Li’s explanation. There is no difference in information provided and usability among travel blogs and tourism websites. Besides, information technology has improved quite fast, people already can search information consistently on the internet. If we set the confidence level to 99%, hypothesis 7 is rejected as well, which signifies that social media usefulness cannot affect people’s continuous use intention. However, this result is different from the previous studies. We speculate the reason for this conclusion is based on tourists’ behaviour knowledge. This phenomenon is usually common in tourism. In general, people usually visit the destination once only and they always choose to visit the local representative tourist attractions no matter whether it is worth a visit or not.

Table 9 shows the comparison of hypothesis path coefficients of different groups. It can be seen that hypothesis 3 has only been supported in the group ‘AGE>30’, which indicates that ease-of-use 142significantly affects people’s satisfaction when older than 30. Therefore, system quality is one of the important factors influencing people’s attitude towards social media and satisfaction of online shopping. We suppose that for those people who are older than 30, to adapt themselves to the new information technology like blogs and other social media platforms is harder than the younger generation (younger than 30). Besides, sometimes they are less active on social media and do not rely on the internet so much compared with young people. They need to learn how to use websites, search information online, and book tickets online. If it is hard for them to obtain useful information or the process of e-ticketing is complicated, they will feel unsatisfied with social media services. In short, older people’s cost of learning to use social media effectively is higher than the perceived benefits (Sulaiman et al. 2008). Moreover, elder people are more sensitive about the security of transactions online, so once they feel deceived, they may not use the website in the future. However, both hypotheses 2 and 3 are rejected in group ‘AGE<30’. This shows that there is no distinct correlation between information quality, ease-of-use and online booking satisfaction. We speculate that the younger generation has more passion for using new technologies such as the internet to purchase new products, browse websites and search for information. Wood (2002) conducted a study to discuss the retailing evolution from an angle of consumer behaviour. From the research, he suggested that those under 25 are more interested in using new technology. Additionally, a study conducted in 2008 that focused on testing consumers’ attitude towards e-ticketing in Malaysia pointed out that age is one of the demographic characteristics that influence people’s online consumption behaviour. The results of that study showed that people who are 26 to 35 years old purchased more tickets online (Sulaiman et al. 2008). Our findings have proved the results of those previous studies and suggested that age is a considerable factor that influences people’s perceptions of social media and their use of social media information for travel planning.

Moreover, there exists quite a large difference between the female and male groups. This can be seen in Table 9. Social media usefulness has no influence on male attitudes towards frequent use and satisfaction. But the social media usefulness has a positive influence on the continued 143use intentions of males. Mona and Seyed’s (2013) research indicates that online consumption behaviour is different for different genders; also that males spend more money on online shopping than females. According to this, we infer that no matter whether the information and services’ quality are good or not, males will always use social media to search travel information or complete online transactions, but once they feel they are unable to find effective information online, their continued use intention of social media will be influenced. For the female group, social media usefulness will have no influence on frequent use and continued use intention, while social media usefulness will influence males on continued use intention. However, social media usefulness will positively influence both online booking and frequent use for females, while for males these two hypotheses are rejected.

Hypothesis 11 is only supported in the group ‘AGE<30’, which means people’s continued use intention will enhance the degree of social media influence on travel arrangements. We suppose that the reason why this correlation is significant is that people younger than 30 have the motivation to use new technology, and more and more young people are used to using the internet and sharing information online. Also, we speculate the reason why this hypothesis is insignificant for older people is that middle-aged people seldom use social platforms in their life anyway, and their motivation for using social platforms is lower compared with young people. Hypothesis 12 is supported in the female group while it is rejected in the male group; this result indicates that the familiarity of various social platforms influences female travel planning more than male planning. Kavoura and Stavrianea (2015) pointed out that females are more likely to join in online travel communities and be affected by WOM. Based on this, we speculate that females are more sensitive about sharing travel information online, and consult others’ experiences more when making travel arrangements.

144

Conclusion

Our study investigated the correlation between social media usefulness and online booking satisfaction, as well as people’s behaviour in using social platforms and the influence of social media on travel arrangements. We proposed 12 hypotheses, and built a structural model to test these hypotheses and provided a robust check of our model. To analyze whether gender and age will affect the result of hypotheses, we test a structure with different groups of data. Path coefficient analysis reveals that hypotheses 1-12 are supported except hypotheses 2, 3 and 7. The results indicate a distinct correlation between social media usefulness and online booking satisfaction, and both usefulness and satisfaction will positively affect people’s frequent use and continued use of social media platforms to search travel information or purchase tourism products. Additionally, the results show that people’s continued use intention and familiarity of travel websites will enhance social media influence on travel arrangements. However, the comparison of each groups’ results demonstrate that age and gender are two influencing factors on tourists’ attitude towards social media. Previous studies have discussed the quality of official destination websites and their influence on people’s travel arrangements. However, with the development of digital social media and smart tourism, official websites are not the only channel for tourists to get information. This study may be the first one to investigate whether social media plays a key role in the new revolution of tourism, and how it will influence people’s decision making and travel planning. We found that word-of-mouth has a greater influence on female than on male decisions; a positive word-of-mouth will attract potential tourists. Additionally, for those of an older generation (more than 30 years old), ease-of-use (which is the websites’ system quality) is a crucial factor in affecting their satisfaction of social media platforms, while the younger generation (younger than 30) care less about the system quality. People who are younger than 30 are more sensitive to social media information. While social media usefulness influences females more than males, women are more interested in sharing information online. So we speculate that young women who are under 30 are the 145group the most sensitive to social media promotions and online sharing. For practical applications, these findings are useful for destination management organizations, online travel agencies and travel website operators when considering innovation for the tourism revolution and the changing of tourists’ consumption behaviour.

Our research only focuses on whether a better information and system quality of social media will enhance its influence on tourists’ travel arrangements. Future research can focus on other influential aspects of social media such as travel videos, celebrity charm and so on. Also, it is an interesting topic to search for the correlation between intimacy and the effectiveness of social media efforts on tourism, while intimacy is defined as a kind of emotional bonding or trust, which makes users dependent on social media.

There exists some systematic limitations when collecting data by questionnaire. First, the respondents’ quality cannot be controlled. Second, the data construction is also an inevitable problem. Of the 445 respondents, the younger generation (18-25 years old) represented around 50%, which impacted the results a great deal. On the basis of big data and programming, future research can obtain data from websites and achieve topical analyses.

146

References

Abubakar M., Ilkan M. (2015), “Impact of online WOM on destination trust and intention to travel: A medical tourism perspective”, Journal of Destination Marketing & Management, vol. 5, no3, p. 192-201.

Agag G., El-Masry A. (2016), “Understanding consumer intention to participate in online travel community and effects on consumer intention to purchase travel online and WOM: An integration of innovation diffusion theory and TAM with trust”, Computers in Human Behavior, vol. 60, p. 97-111.

Alali H., Salim J. (2013), “Virtual Communities of Practice Success Model to Support Knowledge Sharing Behaviour in Healthcare Sector”, Procedia Technology, vol. 11, p. 176-183.

Bhattacherjee A. (2001), “Understanding information systems continuance: An expectation–confirmation model”, MIS Quarterly, vol. 25, no3, p. 351-370.

Chen H. (2010), “Linking employees’ e-learning system use to their overall job outcomes: An empirical study based on the IS success model”, Computers & Education, vol. 55, no4, p. 1628-1639.

Chen C., Cheng C. (2009), “Understanding consumer intention in online shopping: a respecification and validation of the DeLone and McLean model”, Behaviour & Information Technology, vol. 28, no4, p. 335-345.

Chen S., Liu M., Lin C. (2013a), “Integrating Technology Readiness into the Expectation–Confirmation Model: An Empirical Study of Mobile Services”, Cyberpyshology, Behavior, and Social Networking, vol. 16, no8, p. 604-612.

Chen Y., Shang R., Li M. (2013b), “The effects of perceived relevance of travel blogs’ content on the behavioral intention to visit a tourist destination”, Computers in Human Behavior, vol. 30, p. 787-799.

Chung N., Lee H., Lee S. J., Koo C. (2015), “The influence of tourism website on tourists’ behavior to determine destination selection: A case study of creative economy in Korea”, Technological Forecasting & Social Change, vol. 96, no2, p. 130-143.

Chow W., Shi S. (2014), “Investigating Students’ Satisfaction And Continuance Intention Toward E-Learning: An Extension Of The Expectation– Confirmation Model”, Procedia - Social and Behavioral Sciences, vol. 141, p. 1145-1149.

DeLone W. H., McLean E. R. (1992), “Information systems success: the quest for the dependent variable”, Information Systems Research, vol. 3, no1, p. 60-95.

147

DeLone W. H., McLean E. R. (2003), “Model of information system success: a ten-year update”, Journal of Management Information Systems, vol. 19, no4, p. 9-30.

Gao L., Waechter K., Bai X. (2015), “Understanding consumers’ continuance intention towards mobile purchase: A theoretical framework and empirical study – A case of China”, Computers in Human Behavior, vol. 53, p. 249-262.

Hanae R., Meriouh Y. (2015), “Evaluation by users of an industrial information system (XPPS) based on the DeLone and McLean model for IS success”, Procedia Economics and Finance, vol. 26, p. 903-913.

Hasanov J., Khalid J. (2015), “The Impact of Website Quality on Online Purchase Intention of Organic Food in Malaysia: A WebQual Model Approach”, Procedia Computer Science, vol. 72, p. 382-389.

Kavouraa A., Stavrianeaa A. (2015), “Following and Belonging to an Online Travel Community in Social Media, its Shared Characteristics and Gender Differences”, Procedia - Social and Behavioral Sciences, vol. 175, p. 515-521.

Kim S. E., Lee K. Y., Shin S., Yang S. B. (2017), “Effects of tourism information quality in social media on destination image formation: The case of Sina Weibo”, Information & Management, vol. 54, no4, p. 687-702.

Lee Y., Kwon O. (2010), “Intimacy, familiarity and continuance intention: An extended expectation–confirmation model in web-based services”, Electronic Commerce Research and Applications, vol. 10, no3, p. 342-357.

Li L., Peng M. J., Jiang N., Law R. (2017), “An empirical study on the influence of economy hotel website quality on online booking intentions”, International Journal of Hospitality Management, vol. 63, p. 1-10.

Mona A., Seyed M. (2013), “Linking trust, perceived website quality, privacy protection, gender and online purchase intentions”, IOSR Journal of Business and Management (IOSR-JBM), vol. 13, no4, p. 63-72.

National data. Available at: http://www.cnta.gov.cn/zwgk/lysj/201711/t20171108_846343.shtml

Oliver R. L. (1980), “A cognitive model for the antecedents and consequences of satisfaction”, Journal of Marketing Research, vol. 17, no4, p. 460-469.

Oliver R. L. (1993), “A Conceptual Model of Service Quality and Service Satisfaction: Compatible Goals, Different Concepts”, Advances in Services Marketing and Management: Research and Practice, vol. 2, no10, p. 65-85.

Pitt L., Watson R. T., Kavan C. B. (1995), “Service Quality: A Measure of Information System Effectiveness”, MIS Quarterly, vol. 19, no2, p. 173-187.

Seddon P.B. (1997), “A Respecification and Extension of the DeLone and McLean Model of Is Success”, Information Systems Research, vol. 8, no3, p. 240-253.

148

Sulaiman A., Josephine N., Mohezar, S. (2008), “E-Ticketing as a New Way of Buying Tickets: Malaysian Perceptions”, Journal of Social Science, vol. 17, no2, p. 149-157.

Wang H. (2010), “Investigating the determinants of travel blogs influencing readers’ intention to travel”, The Service Industries Journal, vol. 32, no 2, p. 231-255.

Wood S. L. (2002), “Future fantasies: a social change perspective of retailing in the 21st century”, Journal of Retailing, vol. 78, no1, p. 77-83.