Exploring Educational Tourists' Destination Selection Behavior Using the Extended S-O-R Model

The education industry is under increased competition due to the rapid expansion of new educational tourist destinations and higher education. This study utilized the stimulus-organism response (S-O-R) theory to examine the relationship between extrinsic motivation, emotions, cognitive image, and the destination choice of educational tourists. Destination image is a reliable indicator of destination selection behavior in many research contexts. However, in educational tourism, destination image's role as a mediator within the stimulus-organism response (S-O-R) hypothesis is limited. Therefore, the researchers chose the S-O-R theory as the foundational theory and modified the Mehrabian-Russell model as the conceptual framework for this study. An empirical evaluation of the theoretical framework was conducted using a survey of 208 international educational tourists who are residing in Malaysia. The Partial Least Squares method (PLS-SEM) was employed to analyze the obtained data. The results corroborated the adapted Mehrabian-Russell model of environmental psychology theory within the realm of educational tourism. The study revealed seven significant correlations among the four constructs: extrinsic motivation, emotion, cognitive image, and destination selection behavior. A comprehensive understanding of the significance of external motivation and the perception of a location in the selection process of educational tourism destinations has substantial implications for the Malaysian government's endeavors to establish itself as a preferred choice for international educational tourists in the global market. The outcome of this study benefits future researchers by expanding the theoretical nature of the S-O-R model in other study setting.


Introduction
Rising wages and the establishment of new schools and universities in developing nations primarily drive the expansion of the global education market.This global demand has Vol 13, Issue 3, (2024) E- ISSN: 2226-6348 To Link this Article: http://dx.doi.org/10.6007/IJARPED/v13-i3/21676DOI:10.6007/IJARPED/v13-i3/21676 Published Online: 06 June 2024 further expanded international student enrollment in the postsecondary education industry (Ojo et al., 2016).In addition, the UNWTO Sustainable Development Goals (SDGs): Tourism in 2023 Agenda suggests that the tourism industry should support educational investment and facilitate labour mobility through cross-border agreements on credentials, standards and certification.Through educational tourism as a very significant strategy for national development, these international agreements will always contribute to the enhancement of human resources.In these instances, political or economic forces are at play, while international educational tourism can be read not just within the context of cultural interactions, but also within the context of political negotiation or the educational industry.
According to statistics provided by the Ministry of Higher Education Malaysia, the number of international students enrolled in public universities and private higher education institutions has been inconsistent between 2018 and 2021, resulting in the failure to meet the targeted number of international student enrollments.Comparing international students enrolling in Malaysian higher education institutions (HIs) from 2018 to 2021, Table 1 illustrates the population's erratic growth.Thus, a novel theoretical framework was established for investigating the relationships between four constructs: educational tourists' extrinsic motivations, emotions, cognitive image and destination selection behaviour.This research applied the Stimulus-Organism-Response (S-O-R) model by (Mehrabian and Russell, 1974).As a contribution to the body of knowledge, this study proposed an extended S-O-R model for educational tourism, which will greatly contribute to the field of educational tourism by examining the S-O-R model's influence on educational tourists.Jeong et al (2020) cited that the S-O-R model is among the most suitable frameworks for studying tourist behaviours by considering intangible products and services of tourism.The last stage of the model explains that organisms would influence their responses.Hence, educational tourists with a high level of motivation and a positive perception of the destination could have a higher willingness to choose and recommend Malaysia as a preferred educational tourism destination.Figure 1 illustrates the research framework of this study.

Research Framework and Hypotheses Development
Despite decades of research showing that intrinsic motivation is often better, extrinsic motivation remains important, especially in education and the workplace (Ryan & Deci, 2000).Hashiguchi et al (2021) discussed extrinsic motivation and exciting outcomes, including economic achievement, performance-based encouragement and societal recognition reflected in the individual's external stimuli.Additionally, extrinsically motivated activity obtains or receives something from others or some separable outcomes.Furthermore, Alipour et al (2020) stated that as tourism gained popularity, researchers began to incorporate social and environmental dimensions into the triple-bottom-line framework of sustainability (social, economic and environmental), which has been enhanced from a sustainable theory perspective (Social, Economic and Environmental Dimensions).As suggested by Rahman et al (2021), environmental awareness and the presence of other sources or additional information should also be considered a reason to learn in a different, exclusive and exceptional routine from one's home country to further study geographic closeness, which identifies with the geography in the Malaysian context.With that notion, four extrinsic motivations were tested in this study: education in the host country, sociocultural features, environmental features and related costs.Furthermore, destination image is an important indicator of tourist satisfaction and loyalty Mohamad et al (2019), intent to visit or revisit Kock et al (2016) and destination choice (Rasoolimanesh et al., 2021).Prior to an actual visit, the image of a destination is perceived differently by different tourists.A destination with a positive image is more likely to be selected than one with a negative image (Soonsan & Sukahbot, 2019).In the context of tourism destinations, Fyall (2019) argued that the emergence of many new destinations throughout the world heightens the competition in the marketplace and necessitates the development of tools and strategies for destinations to preserve and strengthen their competitive position.The author also said that more places are trying to market themselves as preferred places to live, work, invest and study, besides acting as tourist destinations.As a result, destination image dimensions in this study were evaluated through affective (emotions) and cognitive images.Based on the typology of destination images, the cognitive and affective images are not independent of one another.This study aims to investigate the relationship between constructs that ultimately resulted in behavioural intention; therefore, emotions and cognitive images were tested.
In the context of destination selection behaviour, Kamb et al (2021) claimed that tourists' destination selection is a complex process that can be studied in various stages, including deciding to take a vacation, going abroad and choosing a destination country.In this study setting, destination selection behaviour was measured based on the intention to recommend using word-of-mouth (WoM) and electronic word-of-mouth (eWoM) as a response to educational tourists' experiences after enrolling in Malaysia's higher educational institutions, which will reflect their intention to recommend it to others.The primary difference between WoM and eWoM is the media platform.In contrast to traditional WoM, eWoM is held online using popular platforms such as YouTube, WhatsApp and Instagram (Adil et al., 2020).Thus, this study filled the gaps in the research on destination selection behaviour and referral intention by incorporating traditional and electronic WoM as a platform for spreading behaviour intention among international educational tourists, particularly in Malaysia.Therefore, seven hypotheses developed for the study:

Research Methodology Instrument Development
The questionnaire in this study was divided into two sections: 1) the demographic profile of respondents; and 2) measurement items for five latent constructs mentioned in the research model.All construct items were adopted from other studies and modified to fit the current study.For the demographic data, the items were adopted from (Mun et al., 2018).The measurement items for extrinsic motivation were adopted from Ojo et al (2016) and Negm (2018).Moreover, items of emotion were adopted from Hong (2004), while items of cognitive image were adopted from (Kaur et al., 2016).Finally, items measuring destination selection behaviour were based on (Wang, 2015;Boo and Kim, 2013).All items from previous literature were tested with a reliability above 0.70, which is considered acceptable (Hair et al., 2020).
Additionally, the measurement method chosen for all the constructs (extrinsic motivation, emotion, cognitive image, and destination selection behaviour) was a 5-point Likert scale (1 = most negative to 5 = most positive score).

Sampling and Data Collection
This study used the purposive sampling method as it focused on international educational tourists living in Malaysia at the time of data collection.The survey was sent out online for two months using Google Forms, Electronic Direct Mail (EDM) and a WhatsApp group.Since this study used structural equation modelling with SmartPLS, the sample size should be based on the analysis's power and the bare minimum of samples required to test the study's developed hypothesis (Hair et al., 2017).The researcher must predetermine values for the significant level (α), statistical power 1 -(β) and population effect size (f2) (Cohen, 1988).Using the configurable statistical software's default settings or pre-existing values (0.05, 0.80, and 0.15) in G*Power version 3.1, a minimum sample size of 68 responses was deemed appropriate for this research.Ultimately, 208 completed and usable questionnaires were returned; hence, sample size was not an issue in this study.
The findings showed that 51% (n =106) of the respondents in the survey were female, while 49% (n =102) were male.The majority of respondents in this survey (73.4%) (n =152) were from Southeast Asian countries, while only 26.6% (n =55) were from non-Southeast Asian countries.Furthermore, in this study, the majority of respondents (50.5%) (n = 105) were pursuing their undergraduate programs, followed by respondents at the postgraduate levels (46.6%) (n = 97), and only about 2.9% (n = 6) were studying for a diploma program in Malaysia.In terms of institutional status, the findings revealed that 58.2% (n = 121) of respondents came from public universities, while 41.8% (n = 87) were from private higher education institutions.

Data Analysis
This study aimed to predict the relationship between variables in the research model; as a result, the data were analysed using Smart PLS Ringle et al (2015), a covariance-based structural equation model (SEM).This current study also referred to Hair et al (2020) and used a two-stage approach: 1) the measurement model, which assesses convergent and discriminant validity; and 2) testing the structural model for hypotheses using the bootstrapping method with 5,000 resampling techniques (Hair et al., 2019).

Evaluation of the Measurement Model
The measurement model must fulfil the convergent validity and discriminant validity requirements to be considered valid.Generally, convergent validity can be established if the loading reaches a value of 0.708 or higher Hair et al (2017), average variance extracted (AVE) reaches a cut-off value of 0.4 and composite reliability (CR) achieves a minimum value of 0.5 (Hair et al., 2017).Table 3 displays that the convergent validity was acceptable as the AVE and CR were higher than the threshold values, thus confirming that the convergent validity is not an issue in this study.After the convergent validity was fulfilled, the discriminant validity of the model was tested.Discriminant validity is confirmed if the heterotrait-monotrait (HTMT) values are lower than -.9 (Franke & Sarstedt, 2019).The results depicted in Table 4 satisfied the HTMT criterion, indicating that all the values were lower than the proposed 0.9.Hence, the results of this study proved that the model met the discriminant validity requirements of the tested constructs and items.

Evaluation of the Structural Model
Before evaluating the structural model, multicollinearity test was performed after the measurement model assessment to ensure no collinearity issues were present in the model.The variance-inflated factor (VIF) values, which must be less than the threshold value of 5, were used to assess collinearity (Hair et al., 2017).Table 5 shows that all VIF values were less than five, indicating no collinearity between the predictor variables.(Hair et al., 2020).Table 6 shows the results of the structural analysis carried out with Smart PLS, whereby all the direct effect hypotheses were supported.It shows the path coefficients () indicating the relationship between the structures, as well as the significance of these relationships.6, H1, which posited that extrinsic motivation (EM) has a positive influence on destination selection behaviour (DSB), showed a positive relationship ( = 0.306, t = 3.580, LL, UL = 0.180, 0.459, p = <0.001).Therefore, H1 was supported.For H2, it was suggested that EM positively influences emotion (EMO), and the results showed that EM had a positive relationship with EMO ( = 0.566, t = 6.950,LL, UL = 0.431, 0.701, p = <0.001).Therefore, H2 was supported.Regarding H3, it was proposed that EM positively influence cognitive image (COG), and the results showed that EM had a positive relationship with COG ( = 0.532, t = 5.828, LL, UL = 0.386, 0.687, p = <0.001).Therefore, H3 was supported.For H4 and H5 on the relationship between EMO, COG and DSB ( = 0.208, t = 2.375, LL, UL = 0.060, 0.349, p = <0.001) and ( = 0.334, t = 4.225, LL, UL = 0.203, 0.461, p = <0.001),respectively, the values confirmed positive relationship between EMO, COG and DSB, hence supporting H4 and H5.
For mediation analysis, Hair et al (2017) suggested bootstrapping the indirect effect to test the mediation effect.Preacher and Hayes (2008) stated that LL and UL do not straddle a 0 in between and indicated a mediation effect is non-existent between independent and dependent variables.The results in this study presented that ( = 0.118, t = 2.259, LL, UL = 0.034, 0.204, p = <0.001)for the relationship between EM → EMO → DSB, confirming that EMO has a sequential mediation effect on the relationship between EM and DSB, hence supporting H6.Meanwhile, for H7, the results indicated that COG sequentially mediated the relationship between EM and DSB ( = 0.178, t = 3.721, LL, UL = 0.104, 0.261, p = <0.001).Table 7 shows the results of the mediation effect in this study.As for the assessment of the structural model, the magnitude of the R 2 values indicates whether a significant amount of the variance in the dependent variables is explained.According to Hair et al (2017), the explained variance in the endogenous variable (R 2 ) should be greater than or equal to 0.1.The categories of large, moderate and small variance were based on the rule-of-thumb ranges of 0.75-1.0,0.50-0.74and 0.25-0.49,respectively (Hair et al., 2017;Ramayah et al., 2018).An increasingly common alternative to considering solely R 2 is to also use the predictive relevance criterion Q 2 as suggested by (Hair et al., 2017).According to the authors, Q 2 offers a measure of how well the studied values can be reconstructed by the model and its parameters.If Q 2 is greater than zero, the model has predictive relevance, and vice versa if it is less than or equal to zero.As shown in Table 8, the R 2 values were greater than 0.1 for all variables.Likewise, all the Q 2 values were greater than zero.Therefore, the predictive relevance of the model was confirmed.

Discussion
This present study has proven that extrinsic motivation positively affects educational tourists' destination selection behaviour.These findings suggest that extrinsic motivations have the greatest influence on destination selection behaviour when students have a fulfilling experience while attending higher education institutions in Malaysia.Furthermore, the findings from this study also revealed that extrinsic motivation had a positive relationship with emotion.The findings of this hypothesis are consistent with a study by Pestana et al (2020), which found a significant relationship between motivation and emotions in their investigation of tourist destination choice.The finding from the study on the relationship between extrinsic motivation and cognitive image has been observed in earlier studies, which concluded that the perceived image or cognitive image will be formed through the image projected by the destination and the individual's own needs, motivations, prior knowledge, preferences, and other personal characteristics (Xu et al., 2022).As for the effects of emotion and cognitive image on destination selection behaviour, the findings confirmed that the more positive their emotional and cognitive image of the chosen destination, the more emotionally influenced their intentions to return, recommend and spread the destination through WOM and e-WOM.These phenomena will allow for indirect referrals among peers to choose Malaysia as their preferred educational tourism destination.These findings are consistent with those reported by Prayogo and Kusumawardhani (2016), who found that the more favourable a destination's image, the greater the intention to return and share their experience.Additionally, emotions partially mediate the relationship between extrinsic motivation and destination selection behaviour since direct and indirect relationships are significant.The result of this study was also similar to a study carried out by Namkung and Jang (2009) in the extended S-O-R model by Mehrabian and Russell (1974), demonstrating positive emotions as a mediator in the relationship between atmospherics and future behavioural outcomes.Cognitive image also partially mediated the relationship between extrinsic motivation and destination selection behaviour.The effect of cognitive image as a mediator in this current study, supported by Llodra-Riere et al ( 2015), indicated that motivation influenced tourists' cognitive image formation of Bali as a destination.In line with the current study, Malaysia has a positive image, owing to its residents' warmth and friendliness.
The ideas of extrinsic motivations, emotion and cognitive image were introduced and developed as an extension of the S-O-R paradigm based on theories of environmental psychology.The Mehrabian-Russell modified version of the S-O-R environmental psychology model was used for the current investigation as it has been extensively used in earlier studies to investigate how the environment influences behaviour by affecting emotion and cognitive image (Jeong et al., 2020;Mun et al., 2018).International education tourists are critical to the global travel industry as they travel frequently to learn about different cultures and local languages.In addition, international education tourists spend a substantial amount of money on their trips since they spend more time, travel in groups and encourage friends and family to join them.It is also noteworthy that, from a tourism marketing perspective, education tourism is regarded as an important tourism product that contributes to numerous industry and community benefits.Additionally, this study can contribute to the Sustainable Development Goals (SDGs): Tourism in the 2023 Agenda by UNWTO (2015).According to UNWTO (2015), harnessing tourism's benefits will be critical to achieving sustainable development goals and implementing the development agenda.The contribution of this current study to the SDGs is specifically in Goal 8-decent work and economic growth.Goal 8 aims to give access to decent work opportunities in the tourism sector for all, particularly youth and women, who can benefit from increased skills and professional development.In line with the current study, education tourism is knowledge-intensive, has a high added value, and offers long-term socio-economic benefits.The emergence of the knowledge economy has significantly influenced the internationalisation of higher education globally.Future researchers are recommended to expand this study to a focused geographical area for better generalisation in the forthcoming research.Instead of focusing on overall international educational tourists, future research can be conducted by focusing on neighbouring countries, as different specialities can attract international educational tourists to choose the country as their education tourism destination.

Fig. 1
Fig. 1 Research Framework 1. Extrinsic motivation positively influences destination selection behaviour in the extended Stimulus-Organism-Response (S-O-R) model.2. Extrinsic motivation (EM) positively influences emotion in the extended Stimulus-Organism-Response (S-O-R) model.3. Extrinsic motivation positively influences cognitive image in the extended Stimulus-Organism-Response (S-O-R) model.4. Emotion positively influences destination selection behaviour in the extended Stimulus-Organism-Response (S-O-R) model.5. Cognitive image positively influences destination selection behaviour in the extended Stimulus-Organism-Response (S-O-R) model.6. Emotion mediates the relationship between extrinsic motivation and destination selection behaviour in the extended Stimulus-Organism-Response (S-O-R) model.7. Cognitive image mediates the relationship between extrinsic motivation and destination selection behaviour in the extended Stimulus-Organism-Response (S-O-R) model.

Table 1
International Students' Enrolment in the Higher Education Institutions 2018-2021

Table 8
Predictive Relevance of the Model